Feaecs sa : ay TRAE A IRS ne ALM St
era 3 i Sa telaten views,
i she Deere erates

estes
Tanta sczhesank

Pegs stes
Viana

ance nasa roe tcnevry
ARIA ONS MENTS PRA Sey
Sil en tena ned Aas

fi Ria te

Atta
n Grn Maieeinssin aa
Dade Metance ets
FO aM AMIR)
TA Base ea ate “
UA Be

De sn oko ie
Fae ane

a Avatmawsmeibnh chee tas
fnew ite | meses

Wayen aw a wos aia ahind stay eal
Hig eios ma ernie ectntn?
Ag ne Sk eara casa seataahans
Spy Aca NAR GAD Baten hs

SEs
Salas
aU Aan
Sramtihaeas antes ee

Sigh aaa
tink nam wane
Setanta

Sess
Bento tiat
Jaaan edd DAES me At Vey MICAS,
2 MAD wath
Pectin

RAs ree
RIT Mal el are

Cesta

eatery
cheated pat
pero ime te
iheesietet

Happier
LEM ierPaPre Sioa
Ltga replete:

peter Tks ot oaks
Hoenn

Co salecamesen
Sr AE eRe
Awienedssuccaten ye

ERE ee
Nat
CA epastets
ide missed ety ine tr tate gas
ret ee Feat
Speteasepgaqrcrue tetas Of ail Mekgeat ots ge

Je a reaty tae A A oUnNe portly aay

Powe thee nese lua ety ta s TASS (Lynde kg manne se chee
Panties > ? elag try oy ade

Vocus aterpeometarquirty ere
Katoree tr evarlay eyes
wie Pianl uy Raia ht ba Sue teed vies
iar eelurge ae sedaielenes thay Peeled fh
Wanaaren ne t 4 Rirarphrereitct
Spt eat
nt satin ya gew aus eas
inate : reeeenees y
Phir tee vir ; ragievindte
tau aie

Je tauraeemygetcr eye
Coe Ce eae
Hitieacae ead
De
very Vistar
haateaya eaeeueu lear.

oH
TE SAHG THT

Hepat ante

itty ies at
runs pi e ‘ i piu aay UR howe
YESSY revty ieausee tent 1 i Rear es

eariveny!

‘akeivinsacuny
Tees)

Care ca yea
sreaa cece nwitieys

en Y REF ter g a
veel ean re ra)

SH Ate a Ansara pain) Geth
‘ 2) igus Tieebecipee ov airnees

oR
Sia save

Hea
m pp retatea
ie eer esu Ne ean Uses
Dr ote : Pee AEC NER
By WP 8: Mee Weaae
aga gana frets

ort age eh
el eraeneoengeeees
Sea uaNe undead

a ee ce: = tra ) \ FS —_— {- y > - re ¥ = ww “7AN
Kaw) = eu) 2 SE et 2 Get GZ? es E ty 2°
@ J = 2 S E 2 eS Mey 2 Ene kK
Coke S b Xe) Sy . a NO, Y/Y - & ay a Ly —- Xe wb XO g
NDS “nse n <n4is8> m QS UG m CN.DE 2 ns
z z 7) = o =

}1YVYGIT_LIBRARIES SMITHSONIAN INSTITUTION NOILALILSNI NVINOSHLINS SSIYWYRIT_LIBRARIES SM

Ly,
NVINOSHLINS S3IYV
Y
Ly
Uj
NVINOSHLINS S3IYV
SAC

SMITHSONIAN

SMITHSONIAN

Y
NVINOSHLIWS
NG as
By
SMITHSONIAN

NVINOSHLIWS

SMITHSONIAN

%
»

LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI NV

*

STITUTION NOILMLILSNI NVINOSHLIWS saryvagi

2p) 4 w = ” > i!p) —

w & a 3 hae a o

oc = = “.¢Y ZB = oc =

< = < a Li YY, x < = 4

a c oc S a Yz- Ss a Cc!

a 5 2 5 a ““ 3 a 5

a z SOS) 2 3 4 z a 2
1kYVUGIT LIBRARIES SMITHSONIAN INSTITUTION NOILALILSNI NVINOSHLINS S3IYVYUdIT LIBRARIES SM

c ee Z Eee c z

os = Es = E a =

> = i> = = > =

2 =. ae = - 2 Ee,

m 2 m Zz Zz rm 2°

n = w = = w =
STITUTION NOILNLILSNI NVINOSHLINS S3Itu¥vudly LIBRARIES SMITHSONIAN INSTITUTION NOILOLILSNI NV

AN

NVINOSHLINS” S314YV4G17 LIBRARIES SM

NVINOSHLINS S31uVvudl

NVINOSHLINS
SMITHSONIAN
NVINOSHLINS

SMITHSONIAN
NVINOSHLIWS

SMITHSONIAN
TIT

atMvuaiy LIBRARIES SMITHSONIAN INSTITUTION

YY, f fa
Y A Wee 4

ISTITUTION NOILALILSNI NVINOSHLINS S3IYVYEIT LIBRARIES SMITHSONIAN

ES

NOILNLILSNI
NOILNLILSNI

LIBRARIES SMITHSONI

LIBRARIES
NOILNLILSNI

NOILNLILSNI

INSTITUTION NOIJLALILSNI NV

S3INVUSIT. LIBRARI

z = i = Zz
S) Ka ° = = , =
~ YS led 04
= = = ke =
= oe as “GHy,
JIYVYE!IT_LIBRARIES SMITHSONIAN NOILALILSNI NVINOSHLINS SSIYVYEIT LIBRARIES SN

Rey

NVINOSHLINS S31YVUdIT_ LIBRARIES

‘Lp
‘Lp
NVINOSHLINS S3IYVualT

SMITHSONIAN
NVINOSHLIWS
SMITHSONIAN
NVINOSHLIWS
SMITHSONIAN
iz
4 Le

SMITHSONIAN
oy ey 4
Vi

.. _ y ; »,
ISTITUTION NOILALILSN! NVINOSHLINS S3IuUvYUdIT LIBRARIES SMITHSONIAN INSTITUTION NOILONLILSNI Nv

o = rr) 7)
uu a uw f ‘ Ww
= | — Lap (RANG —
z = E PY : ‘
a =| «x Ye: a
a 5 ao i cs)
a 2 3 i =
J/¥vudi1 LIBRARIES INSTITUTION NOILALILSNI NVINOSHLINS SSlYVudit LIBRARIES Sh

x
SMITHSONIAN

4

8

INSTITUTION NOILNLILSNI

S3IINVUAIT LIBRARIES
INSTITUTION NOILNLILSNI

INSTITUTION NOILNLILSNI

Sdiuvudl
INSTITUTION
Sadiuvaugit

ISTITUTION NOILOLILSNI NVINOSHLINS S31YVYSIT_ LIBRARIES SMITHSONIAN INSTITUTION NOILALILSNI_ N\

NVINOSHLINS S31uVvudl

22) Zz 2) P = za wo =
= oy, = = = < = <
‘ss SS . W Yi 2) w 2) 7) RAO AT) AD F
oe DANS fe) Vif fis: = [e) ae a P S38 WS O i=: coe \i
“ QF = 2 ES Saas Ss z =
Ses 3 5 . pS 5
J!uYvudil LIBRARIES SMITHSONIAN INSTITUTION —NVINOSHLINS Saluvuds LIBRARIES SN
Zz e : 3 = z
am >) & (Ee LON Wl Vi & KS LUN wi LSD B w STO a & JONG ay Ww
LR 3 KEW = Coy = (RM) = (AM) 3 NQy = (A>) = AMD &

iS © J k = SHaityN y
Raby = ees D Wy Wf = Coe D > ree
i o = ra) =
STITUTION NOI TITSNUSNVINOSHIINS S43 luvaad rout BRARIES
sg = = = Ke
SS 5 = = ZNS
~a Sz 5 z ; 2 SAS
2 rs z =z NX,
> j = > =
Zz 7) 2 a >
LIBRARIES SMITHSONIAN _ INSTITUTION NOILNLILSNI
S o 2 to
(22) a” —
=! ec. =! c
=) <x } SI <
=I Ss s S
fo) = fo) =
z aj 3 a
STITUTION _ NOILNLILSNI_NVINOSHLINS Sa luvud 7 tl BRARIES
} aay fe S} =
5 2 5 2
E > E >
5 = 5 =
z o = o
YVUAIT_ LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI
; z ’ = z ® g Ay
5 = 5 Yj.
= Zz = Ze Vae
= SX > = >”
ow aati Dineee ds ee
NOILNLILSNI NVINOSHLINS S3IYVYSIT LIBRARIES
7) > 7) =
us w uJ w
= 4 = 4
« = ce =!
<x = _< =
[aa = a =|
m ro) es 3
=5 Zz a) 2
WIYvVudsiT LIBRARI ES SULIT USUI OA OMEN ES
{ peo
~ oO = (2)
= w =
2 = 2 =
> Fe: > e
z E a =
D z o z
NOILNLILSN! NVINOSHLINS SSIYVYEIT LIBRARIES

eee

HILIAS
Up,

NVINOSHLIINS
SMITHSONIAN
SMITHSONIAN

uvugia LIBRARIES SMITHSONIAN INSTITUTION NOILALILSNI

e

RARIES
INSTITUTION NOILALILSNI NVINOS

STITUTION NOILNLILSNI_NVINOSHLINS S3iuvudit LIBRARIES

\
\

SMITHSONIAN

SJIYVYUSIT LIBRARIES

HuVvUaIT_LIBRARIES INSTITUTION NOILALILSNI
“ b) @

NVINOSHLINS S3idvudII_LIB

STITUTION NOILAL

NVINOSHLIWS LIBRARIES

ARIES SMITHSONIAN INSTITUTION NOJLOALILSNI

ARIES SMITHSONIAN
ILILSNI NVINOSHLINS

ALILSNI

}
;

SMITHSONIAN

<h
ie}
a

t

0
<niyist

SS

SMITHSONIAN

INSTITU
Salyv?
INSTITL

NOILALILSNI NV

INSTITUTION

SONIAN

AASVAD

NVINOSHLINS S3IYV!

NVINOSHLIWS

LIBRARIES SMI

ES SMITH

rs
<
a
ro)
Ww
ac
=
=
w
at a — Pa, Y
=| < al ay
= ce Ss a %
re) 2 fe) aaa
2 Sage = a5
SMITHSONIAN INSTITUTION NOILNLILSNI NVI
z sy - S (=
[@) = — —
= wo = w
E = E ss
= = = =
Be = i -
z o z Oo
NVINOSHLINS SSIYVYEIT LIBRARIES SMI
z 7) ; Zam a)
< = Ly < =
z =] py, = =
5 * yi S ;
ae (2) SYA as (2)
ir i Ee z
5 z ze
SMITHSONIAN INSTITUTION NOILN _NVI
lJ wo uJ 7)
z = z =
< PS <x c
Land 4 oa 4
2 5 eo 3
3 ae al =z
NVINOSHLINS S3IYVYSIT LIBRARIES SM
Ue z - =
: : : :
rad A =} a =). t
> - > = ey
2 = 2 i
m me) = wn
a) See (a) 4
SMITHSONIAN INSTITUTION NOILALILSNI NVI
w Zz x: (2) z
= ae ee <=
= z <a AE
2 = g EX
> = > C =
2 7) pata g 77)
_NVINOSHLINS Saluvudit LIBRARIES SMI
2 St uJ = uJ
4 AN = z a
2 \N Ps S Ps G
pa i co = ao
S Oe z zi
_ SMITHSONIAN INSTITUTION NOILOLILSNI_NVI
re) es , = fe) = ‘
5 ily, 2 5 2
= Jala teas =
- &, 0% e 2
w* 7 m n m
= 7) z )
NVINOSHLINS Saluvyugl oLIBRARIES, SMI
= ats Si
+ z a fF
é 7s: BG
g iS 2X
>" = >
Z wo 2
SMITHSONIAN INSTITUTION _ NVI
nf ty 2 0 NO ws =
ss Yip 4 kK oD = a
< oY. =i (& a) < =

ia)

ies

ii

OCS STUDY
MMS 88-0021

Population Status of California
Sea Otters

Population Status of California Sea Otters
Pacific OCS Region
Minerais Management Service
U.S. Department of the Interior

Contract No. 14-12-001-30033

rey aD
“i Br MADR V aye

Ne
ete

Diy
ih TURE
ie i

1 on
nt it
aT
ean

ti nt
Hay

Mi
Orth
eit! WM

Population Status of California

Sea Otters

Edited by

Dg Io Salbere
Department of Ecology and Behavioral Biology
University of Minnesota
Minneapolis, MN 55455

and

K. Ralls
National Zoological Park
Smithsonian Institution
Washington, D.C. 20008

November 30, 1988

This study was funded by the Pacific Outer Continental Shelf
Region of Minerals Management Service, U.S. Department of the

Interior, Los Angeles, California under Contract No. 14-12-
001-30033

JUN No 1693 >

= niles

<r"

\ ghirnot htao to susase mokaetigen
a1e2t0° 2ar

wa beer cid

gm Mee ess

eee. theanbtiens sterue oes ties
sive om PoneIned sehen pipmoes

Disclaimer

This report has been reviewed by the Minerals Management
Service and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the Bureau, nor does mention of trade names of

commercial products constitute endorsement or recommendation
for use.

TABLE OF CONTENTS PAGE

ACKNOWLEDGEMENTS Vv
LIST OF TABLES AND FIGURES WALILSL
ABSTRACT xix
TECHNICAL SUMMARY OK
APPENDICES 275

CHAPTER 1. Overview of the study:
background and general methods
D. B. Siniff and K. Ralls a;

CHAPTER 2. Reproduction, survival and tag loss
in California sea otters
D. B. Siniff and K. Ralls 13

CHAPTER 3. Movement patterns and spatial use
of California sea otters
K. Ralls, T. Eagle, and D. B. Siniff 33

CHAPTER 4. Time budgets and activity patterns of
California sea otters
Ke Ralls and D2 BB. sinver 64

CHAPTER 5. Feeding patterns of California sea
otters
K. Ralls, B. Hatfield, and D. B. Siniff 38:4

CHAPTER 6. Age determination of California sea
otters from teeth
P. Pietz, K. Ralls, and L. Ferm 106

CHAPTER 7. Analysis of the precision and accuracy
of radiotelemetry equipment and
methods used in California
A. Mercure 116

CHAPTER 8. Movement patterns of adult female and
weanling sea otters in Prince William
Sound, Alaska
C. Monnett and L. Rotterman 133

CHAPTER 9. Sex-related patterns in the post-natal
development and survival of sea otters
in Prince William Sound, Alaska
C. Monnett and L. Rotterman 162

CHAPTER 10. A simulation model for assessing the
risks of oil spills to the California
sea otter population and an analysis of
the historical growth of the population
A. Brody 191

ACKNOWLEDGEMENTS

California Field Studies

We are extremely grateful to the many people who
contributed greatly to this research effort and made this
report possible. Jack Ames, Robert Hardy, and Fred Wendell
of the California Department of Fish and Game and James
Bodkin, Brian Hatfield and Ron Jameson of the USFWS provided
invaluable assistance by capturing otters and assisting during
catching operations. Dr. Thomas D. Williams implanted all the
transmitters, and was most patient and accommodating
throughout many rather frustrating moments during field
operations. Angela Doroff, Lisa Ferm, Brian Hatfield, Paul
Henson, Christopher Jordon, Alan Mercure, Steve Osmek, Pamela
Pietz, and Marian Skupski monitored otters for extended
periods. Allan -Brody, Colleen Baggot, Ken Halama, Leslie
Larson, Marianne Reidman and Christopher Logan also assisted
in the field for short periods. Nancy Black, James Bodkin,
Ron Jameson and Galen Rathbun of the USFWS and Marianne
Reidman provided occasional reports on the location of our
otters. Mary Faustini, Steve Mareck, Mike Henry and Stacy
Kawa observed otters in Morro Bay.

The following individuals gave us access to property
along the coastline so that we could monitor our otters: Cc.
Douglas Walling (Pfeiffer Point Mutual Water Company), Claire
and Sybil Chappellet and their foreman Dennis Krackenberg
(Rancho Rico), Evan Goldblatt (Big Creek Reserve), Doyle
Danley and Wayne Titus (Cambria Radar Station), and Robert
Smith (Diablo Canyon Nuclear Power Plant).

Dick Rodgers, Morro Bay Harbormaster, and Brooks Bowhay,
Monterey Harbormaster, gave us temporary docking facilities
for our boat and Clyde Clark allowed us to store it
temporarily in Morro Bay State Park. The McQueens of the Big
Sur Campground allowed our small trailer to be a most
important field camp in the Big Sur Area and also provided
much logistic and moral support for our people in the field.

Thomas Eagle wrote several helpful computer programs and
assisted with data management. Lisa Ferm, Alan Mercure, and
Marian Skupski completed most of the data entry. Dorothy
Bromenshenkel and Rachel Ayetey worked magic with computer
word processors, bringing this report through its many drafts
until its final emergence into the form presented here.
However, any errors that still exist, of course, rest with the
editors.

Robert L. Brownell, Jr. of the USFWS gave us access to
facilities at the Piedras Blancas Lighthouse for our field
base. Ron Jameson was most helpful throughout the project,
as he provided valuable insights about field operations and
sea otter biology from his extensive experience in California.
Cedar Creek bioelectronics lab (L. Kuechle, D. Reichle, R.
Schuster) provided telemetry equipment and advice about its
use. Lynn Rathbun provided us with excellent drawings of our
study area and a schematic drawing of the transmitter.

Alaska Field Studies

E. Birney, K. Ralls and D. Smith read the Alaska papers
in this report and made many useful comments. Many thanks to
A. DeGange, D. Garshelis, K. Schneider and T. Simon-Jackson,
who provided access to various unpublished manuscripts. D.
and J. Garshelis gave freely of their knowledge about the sea
otters of Prince William Sound. The following individuals
contributed during some phase of the field work: J. Bennett,
D. Carlquist, A. DeGange, A. Doroff, F. Foode (pilot), P.
Gullett (veterinarian), K. Hill (veterinarian), A. Johnson,
F. Koecher (veterinarian), J. Nelson, P. Rosenberg (pilot),
J. Ross, L. Rotterman, J. Sarvis, T. Simon-Jackson, M.
Sorenson, D. Traun and T. Williams (veterinarian).

For the Alaska studies, logistical support and/or
equipment for field work were provided by the United States
Fish and Wildlife Service (USFWS: M. Blenden, C. Dau, A.
DeGange, A. Johnson, S. Lawrence, L. Pank, J. Sarvis, T.
Simon-Jackson, D. Taylor), the Minerals Management Service
(MMS: B. Hughes, G. Reetz, S. Treacy), the National Oceanic
and Atmospheric Administration (NOAA: L. Jarvela, G. Lapine,
M. Meyer), and the U.S. Forest Service (C. Nelson, K.
Giezentanner, L. Keeler), the National Marine Fisheries
Service, the Peter Pan Cannery (W. Bright), the Alaska
Department of Fish and Game (H. Griese, J. Reynolds), and the
University of Minnesota. The University of Alaska Marine
Advisory Program (R. Steiner and G. Ference) provided office
space and a contact point.

Two Cordova businesses subsidized this project and
contributed significantly to its successful completion. The
Reluctant Fisherman Hotel (M. Johnson and R. Borer) generously
provided lodging in Cordova at times when it was sorely
needed. The Eyak Corporation (L. Borer) permitted us to
establish a field camp on Native Alaskan lands.

Model

Dr. L. L. Eberhardt, Battelle Institute, Richland,
Washington, developed the stochastic model for the females,
derived the parameter estimates required for its operation and

vi

was invaluable in helping with the data analysis. Dr. Tony
Starfield was also very helpful and stimulating in helping
with, and formulating ideas about, the population model
development. Jack Ames, Robert Hardy, and Fred Wendell of
CDF&G gave us access to the raw data from past surveys for sea
otters in California. CDF&G, USFWS, and the museums listed
in Chapter 6 generously provided us with teeth for age
determination.

Finally we acknowledge the patience, confidence and
general support shown to us by Gordon Reetz, our Contractor
Officer's Technical Representative. Throughout the project
we were faced with delays and difficult decisions as to how
to proceed. His support at these times was greatly
appreciated.

LIST OF TABLES AND FIGURES PAGE

TABLE 2.1 -- Summary of sea otters captured 15
in California during 1984 and 1985.

TABLE 2.2 -- A list of sea otters that were
instrumented with implanted radio transmitters. 17

TABLE 2.3 -- Reproductive and age data, and
length of monitoring period for adult
female sea otters. 21

TABLE 2.4 -- Reproductive information on
eight adult female sea otters and data
on 6 otters that were used to calculate
the inter-birth interval. 24

TABLE 2.5 -- Annual survival rate estimates
for the four sex/age categories of adult
females, adult males, juvenile females and
juvenile males. 25

TABLE 2.6 -- Estimates of annual tag survival
rates based upon the method of Heisey and
Fuller (1985). 26

TABLE 2.7 -- A comparison of the age of pups at
separation from the female in California and
Alaska. 27

TABLE 2.8 -- Numbers of male and female carcasses
in good condition compared to those expected
if the sex ratio was equal. 28

TABLE 3.1 == Average distance (km) between
successive locations, recorded between 18
and 36 hours apart for each instrumented otter. 37

TABLE 3.2 -- Average distance (km) between successive
locations, recorded more than 36 hours apart,
for each instrumented otter. 38

TABLE 3.3 - Area (ha) of daily home ranges based on
data obtained during 24-hour watches. 47

TABLE 3.4 - Average monthly distance deviations (km)
from the harmonic mean center for instrumented
California sea otters. 53

TABLE 3.5 - The average location along the five
fathom line for each instrumented sea otter. 55

viii

TABLE 3.6 - The average distance (km) between
extreme locations. for four sex and age
categories of California sea otters.

TABLE 3.7 - Comparison of home range areas (ha)
for sea otters in California and Alaska.

TABLE 4.1 - A comparison between activity data
obtained visually with that obtained using the
quality of the telemetric signal.

TABLE 4.2 - A comparison between time budgets
calculated from observing activity visually
and judging activity using the quality of the
telemetric signal.

TABLE 4.3 - A comparison among the methods of
calculating time budgets.

TABLE 4.4 - Analysis of variance testing for
differences in percent of time spent feeding
between the various sex/age classes.

TABLE 4.5 - A comparison of time budgets between
females with small pups and those with large

pups.

TABLE 4.6 - A comparison of the percent of time
spent in each of the three activity categories,
resting, feeding and other, for activity data
over the entire 24-hour period.

TABLE 4.7 - A comparison of the activity budgets
for sea otters calculated in this study and
those in the literature.

TABLE 5.1 - A comparison among the average dive
times (sec) for feeding sea otters for various
type of prey.

TABLE 5.2 - The average length of the feeding dives
(sec) for sea otters for prey of different
sizes.

TABLE 5.3 - A comparison among the surface times
(sec) required to consume the various prey
items.

TABLE 5.4 - A comparison among the average surface
times (sec) required to consume various size
prey.

TABLE 5.5 - A comparison of the percent of successful

ix

56

59

73

74

76

76

UN

77

79

88

88

89

89

feeding dives for various sex/age categories. 90

TABLE 5.6 - A comparison of the average dive times
(sec) for sea otters during feeding bouts. 91

TABLE 5.7 - Analysis of variance testing for
differences in the length of the dives made
by otters belonging to the various age/sex
classes. 94

TABLE 5.8 - The average surface times (sec) for
individual sea otters during feeding bouts. 95

TABLE 5.9 - A comparison among the average surface
times (sec) for five sex/age categories that
were recorded during feeding bouts. 96

TABLE 5.10 - The average lengths of the feeding bouts
(min) for individual sea otters. 97

TABLE 5.11 - The lengths of feeding bouts (min),
grouped by time intervals, for five sex/age
categories. 98

TABLE 5.12 - The average lengths of intervals between
feeding bouts (min) for individual sea otters. 99

TABLE 5.13 - The frequency of the lengths of the
intervals between feeding bouts (min), grouped by
time intervals for five sex/age categories. 100

TABLE 5.14 - A comparison of day and night dive
lengths (sec) for the individual sea otters,
for five age/sex classes. 101

TABLE 5.15 - A comparison of the mean lengths of
surface intervals made during the day and
night by the individual instrumented otters
in the various age/sex classes. 102

TABLE 6.1 - Tooth age estimates for sea otters of
minimum known age. 108

TABLE 6.2 - A comparison of animal ages determined
from tooth cementum by Gary Matson: (1) with
animals of known age; (2) with duplicate
determinations based on a different tooth from
the same animal; and (3) with ages estimated
from the degree of tooth wear (reprinted from
Matson's Tooth Cementum Age Analysis, Progress
Report No. 9, Spring 1987, Table 1). LL?

TABLE 6.3 - Comparisons of sea otter age assignments
based on counts of incremental lines in tooth
cementum. 113

TABLE 7.1 - Summary statistics for the 897 hand-held
compass bearings to radio transmitters on buoys
off the California coast. 127

TABLE 7.2 - Summary of the calculated precision and
accuracy of locations of radio transmitters. 128

TABLE 7.3 - Comparisons between the field method of
plotting data and the Andrews estimator for the
same set of data. 129

TABLE 9.1 - Comparison of growth rates for large vs.
small pups. 170

TABLE 9.2 - Error is estimation of birth dates from
growth rate assumptions. 172

TABLE 9.3 - Sex of dependent Prince William Sound
sea otter pups. 172

TABLE 9.4 - Survival rates of sea otter pups in
Prince William Sound. 182

TABLE 10.1 - Estimates of annual survival rates of
telemetered sea otters in California, as
determined by the method of Heisey and Fuller
(1985), 1983-1986. 205

TABLE 10.2 - Default parameters used in OTPOP and
- LESLIE. 212

TABLE 10.3 - CDFG and USFWS censuses used in analysis
of California sea otter distribution. 225

TABLE 10.4 - Analysis of variance in CDFG and USFWS
California sea otter census data, 1968-1985.
Dependent variable is the proportion of census
total along a 10 km section of coast. 226

TABLE 10.5 - California sea otter sex ratios of
recovered carcasses (1968-1985) and as
subjectively estimated by field biologists, by
season and CDFG carcass recovery area. 232

TABLE 10.6 - Parameters used in short-term otter
movement model. AR and CE are regression
parameters discussed in text. The symbol sd
is standard deviation of regression errors,

xi

R* is given for the regressions. Vmax is mean
maximum daily movement, derivation discussed
Ty Cext 246

TABLE 10.7 - Parameters giving the best fit of
OTRANGE to historical data. See text for

explanation of parameters. 267
FIGURE 1.1 - A map of the study area in California. 4
FIGURE 1.2 - A map of the study area in Alaska. 5

FIGURE 1.3 - Schematic drawing of the upper jaw
of a sea otter showing vestigial first pre-
molar that was sometime removed from captured
animals and sectioned for determination of
age. 7

FIGURE 1.4 - A schematic drawing of the implanted
radio transmitter used during this study. 9

FIGURE 2.1 - The percent of the adult females with
pups during each month of the year. 22

FIGURE 2.2 - A comparison of tag-loss rates between
this study and the study of Ames, et al., 1983. 30

FIGURE 3.1 - A comparison of the average distance
between successive locations for four age/sex
categories, for locations made 18-36 hours apart
and those made more than 36 hours apart. 39

FIGURE 3.2 - A plot of the 20 longest trips made
between successive locations that were 18-36
hours apart for four age/sex categories. 41

FIGURE 3.3 - The average location of each instrumented
sea otter along the California Coast. 42

FIGURE 3.4 - The general north-south movement pattern
of individual adult males. 43

FIGURE 3.5 - The general north-south movement pattern
of individual adult females. 44

FIGURE 3.6 - The general north-south movement pattern
of individual juvenile females. 45

FIGURE 3.7 - The general north-south movement pattern
of individual juvenile males. 46

FIGURE 3.8 - The distribution of distances offshore
for four age/sex categories of California sea
otters.

49
FIGURE 3.9 - The average distance offshore while

resting and feeding for individual otters for five
age/sex categories. 50

FIGURE 3.10 - The average distance offshore by hour
of the day for juvenile males and females. 51

FIGURE 3.11 - The average distance deviation from the

harmonic mean center of monthly home ranges for
four age/sex categories.

52
FIGURE 3.12 - The distance between extreme locations

for instrumented sea otters in California. 54
FIGURE 4.1 - The locations along the California coast
of watches for collecting time budget data on sea
otters instrumented with radio transmitters. 68
FIGURE 4.2 - The percent of time that adult male sea
otters spent in various activities at the various
hours of the day. 70
FIGURE 4.3 - The percent of time that juvenile sea
otters (male and female spent resting,

feeding and
in other activity for various hours of the day.

71
FIGURE 4.4 - The percent of time that adult female sea
otters (with and without pups) spent resting,

feeding and in other activity for various times
of the day.

UZ
FIGURE 5.1 - The distribution of dive times during

feeding bouts for five groups of sea otters in
California.

92
FIGURE 5.2 - The distribution of the length of time

of the surface intervals during feeding bouts
for five groups of sea otters in California.

93
FIGURE 6.1 - Comparison of age estimates based on

teeth to age estimates using skull features.

110
FIGURE 6.2 - Distribution of age estimates based on

incremental lines in tooth cementum.

114
FIGURE 7.1 - Illustration of the method used to

determine the location of a buoy by taking
compass bearings.

118

xiii

FIGURE 7.2 - Distribution of compass bearings from
sightings through a telescope (Questar). 123

FIGURE 7.3 - Distribution of hand-held compass
bearings to a prominent landmark. 124

FIGURE 7.4 - Compass bearings to a radio transmitter
on a buoy off the California coast. SLABS

FIGURE 7.5 - Distribution of hand-held compass
bearings to the signals from radio transmitters
off the California coast. 126

FIGURE 8.1 - Study area in Prince William Sound,
Alaska, 1984-1987. 136

FIGURE 8.2 - Distances between extreme locations of
eight adult female sea otters in eastern Prince
William Sound, Alaska. The number of fixes and
total monitoring intervals are given:

# fixes / # days. 139

FIGURE 8.3 - Movements of an adult female sea otter in
Prince William Sound, Alaska, during a 20 month
interval, June 1984 - February 1986. Summers
were spent in the western portion of the study
area and winters in the eastern portion, near
the Cordova male area. 140

FIGURE 8.4 - Division of study area in Prince
William Sound, Alaska, into numerically
designated habitat zones and superzones. Zones
correspond to major bays or passages. 141

FIGURE 8.5 - Use of habitat zones in Prince William
Sound, Alaska, by eight radio-instrumented adult
female sea otters. 142

FIGURE 8.6 - Seasonal changes in the use of portions
of eastern Prince William Sound by eight radio-
instrumented adult female sea otters. Superzones
are delineated on Figure 4. 143

FIGURE 8.7 - Distances between extreme locations of
26 female sea otters in Prince William Sound,
Alaska, that were accompanied by dependent pups.
Most observations are based on females
accompanying radio-instrumented pups. 144

FIGURE 8.8 - Changes in the home ranges of sea otter
female-pup pairs in Prince William Sound, Alaska,

Xiv

that occur as the pups approach weaning age. The
distances between extreme locations of pairs are
compared for the last 30 days before weaning and
for the earlier period when the pup was younger. 147

FIGURE 8.9 - Distances traveled from the site of

weaning in Prince William Sound, Alaska, by male
and female weanling sea otters. Monitoring

interval varied from a few days to approximately
18 months. Short monitoring intervals resulted

when pups died during their travels.

FIGURE 8.10 - Distance between weaning location of
sea otters location in Prince William Sound,

Alaska, and their first post-weaning home range.

The distance was traveled in a single
relatively rapid trip.

FIGURE 8.11 - Relative size of weanling male and
female sea otter home ranges in Prince William

Sound, Alaska, during the first winter following

weaning. Only weanlings with well defined home
ranges are included.

FIGURE 8.12 - Tendency for weanling sea otters in

Prince William Sound, Alaska, to leave the natal

female area after being weaned. Female
weanlings usually do not leave the natal female
area, whereas males usually do. Female area
consists OF, ZONES alte e2),. 1S rate: Ole 7). Sn and: tt
on Figure 4.

FIGURE 9.1 - Location of study area in Alaska and
in Prince William Sound.

FIGURE 9.2 - Growth rates of dependent male and
female sea otter pups.

FIGURE 9.3 - Estimated birth dates and capture

dates of sea otter pups in Prince William Sound,

Alaska.

FIGURE 9.4 - Weaning dates of instrumented sea
otter pups in Prince William Sound, Alaska,
1985-1986.

FIGURE 9.5 - Chronology of dependency periods of 27
sea otter pups in Prince William Sound, Alaska,
1984-1986.

FIGURE 9.6 - Sea otter pups instrumented in Prince
William Sound, Alaska, that died or with which

xv

149

151

152

153

165

173

174

175

176

radio contact was lost. 179

FIGURE 10.1 - Schematic representation of the
interrelation of the submodels used to predict
the potential effects of oil spills on California
sea otter population dynamics. 195

FIGURE 10.2 - Hypothetical survivorship curve
depicting the relationships of the 3 competing
risks of Siler (1979) and Eberhardt (1985). 200

FIGURE 10.3 - Hypothetical reproductive curve
depicting the relationship between prime
reproductive rate and senescence, after
Eberhardt (1985). 202

FIGURE 10.4 - The effect of the value of b on the
non-linearity of the density dependence function
used in OTPOP and LESLIE. K is the carrying
capacity. 203

FIGURE 10.5 - Distribution of ages of California sea
otters estimated by tooth cementum. 207

FIGURE 10.6 - Relative average number of small pups
and large pups, by month, in the CDFG index
areas, 1977-1984. 209

FIGURE 10.7 - Age-specific female survivorship curves
and annual survival rates at different per
capita growth rates. 214

FIGURE 10.8 - Age-specific reproductive rates under
the default population parameters used in
the population model. 215

FIGURE 10.9 - Age-specific female California sea
otter annual survival rates calculated from
certain model parameters. 216

FIGURE 10.10 - Age-specific male California sea otter
annual survival rates calculated from certain
model parameters. 217
FIGURE 10.11 - Male and female California sea
otter survivorship curves from certain
population parameters. 218

FIGURE 10.12 - Pattern of monthly pup abundance
obtained by simulation. 221

FIGURE 10.13 - Contour diagram indicating annual

xvi

changes in sea otter density in California
from 1968-1985. 227

FIGURE 10.14 - Contour diagram indicating monthly
changes in sea otter density in California from
1968-1985. 228

FIGURE 10.15 - Density functions used to obtain
locations of independent sea otters in
California in June (dashed line) and December
(solid line). 230

FIGURE 10.16 - Local proportion of California sea
otters that are female in June (dashed line) and
December (solid line), used in the population
models. BSL

FIGURE 10.17a - Density functions used to obtain the
location of male (dashed line) and female
(solid line) sea otters in California in June. 234

FIGURE 10.17b - Density functions used to obtain the
location of male (dashed line) and female
(solid line) sea otters in California in
December. 235

FIGURE 10.18 - Schematic representation of the
methods to predict sea otter densities in
expanded range. 237

FIGURE 10.19a - Daily locations of a juvenile
female California sea otter as determined by
telemetry, 1985-1986. 242

FIGURE 10.19b - Daily locations of a juvenile male
California sea otter as determined by telemetry,
1985-1986. 243

FIGURE 10.20a - Daily locations of a juvenile female
California sea otter, as in Fig. 10.19a. 244

FIGURE 10.20b - Daily locations of a juvenile male
California sea otter, as in Fig. 10.19b. 245

FIGURE 10.21a - Simulated movements of sea otters
around an oil spill in Monterey Bay beginning
1 December and lasting 15 days. 249

FIGURE 10.21b - Simulated movements of sea otters
around an oil spill in Monterey Bay beginning
1 December and lasting 15 days, as in
Fig. 10.21la, except that in this simulation

xvii

movement parameters were different than in
aleyy UO GAINS i 250

FIGURE 10.21c - Simulated movements of sea otters
around an oil spill in Monterey Bay beginning
1 December and lasting 15 days, as in Fig.
10.21la, with different parameter values. Pysyal

FIGURE 10.22 - Age-specific reproductive values of
female California sea otters for default
parameter settings. 252

FIGURE 10.23a and 10.23b - Printed output from a run
of the model introducing a large oil spill for
10 days along a 50km section of coast between
Marina and Yankee Point. 254-7

FIGURE 10.23c - Trace of the total simulated
population size for runs in Fig. 10.23a. 258

FIGURE 10.23d - Trace of the total simulated
population size for the control (no oil spill)
runs in Fig. 10.23a. 259

FIGURE 10.23e - Trace comparing the mean values from
Figs. 10.23c and 10.23d. 260

FIGURE 10-23£ - Mean cumulative number of otter
deaths due to oiling for the run in Fig. 10.23a.261

FIGURE 10.24 - Schematic representation of the computer
program OTRANGE. 264

FIGURE 10.25 - Density dependence functions used in
the computer program OTRANGE. 265

FIGURE 10.26 - Fit of computer model to historical
data using the "best estimate" parameters
without density independent mortality. 268

FIGURE 10.27 - Fit of computer model output to
historical data using the "best estimate"
parameters incorporating density independent
mortality. Solid line traces population size,
dashed line traces carrying capacity. 269

Xviii

ABSTRACT

The main objective of the contract was to develop a
simulation model to facilitate analysis of the risk of oil
spills to the threatened California sea otter population.
Existing data on the dynamics and demography of the population
were reviewed and synthesized. The additional data needed for
model development were collected through radiotelemetry
studies of sea otters in Alaska and California.

Our field work indicated that the California population
had a high reproductive rate but many pups did not survive to
weaning. /Adult females had the highest survival rates and
adult males the lowest. Juvenile females had lower survival
rates than adult females and spent more time foraging than
other otters. Otters tended to stay within a small area for
an extended period and then suddenly move for a considerable
distance. They made more long-distance movements than
expected. Juvenile males tended to travel more extensively
and range farther offshore than other otters.

The simulation model contains four interrelated
stochastic submodels: a short-term population model, a long-
term population model, a sea otter distribution model, and a
sea otter movement model. This report includes a detailed
description of the model, the data on which it is based, and
an operating manual. The computer program for the model has
also been provided to MMS.

TECHNICAL SUMMARY
Chapter 1.

The objectives of the contract were to review and
synthesize existing information on the dynamics and demography
of the threatened California sea otter population, to design
and conduct field studies to collect the data needed to fill
data gaps identified through this process, and to develop a
Simulation model to facilitate analysis of the risk of oil
spills to this population.

We computerized past survey data collected by the
California Department of Fish and Game and the U.S. Fish and
Wildlife Service and aged teeth from the salvaged otters found
dead along the California beaches over the last 20 years.

The additional data needed for model development were
collected through radiotelemetry studies of sea otters in
Alaska and California. Because of the sensitive nature of
hands-on field work on the threatened California population,
we tested procedures and equipment in Alaska before applying
for permits to use them in California. We developed and used
a radio transmitter that could be implanted within the
abdominal cavity of sea otters. The use of these transmitters
enabled us to make a number of new discoveries about sea
otters in both Alaska and California.

Chapter 2.

We observed 40 California sea otters, representing all
four major age/sex groups, that were flipper-tagged and
instrumented with implanted radio transmitters.

The proportion of females accompanied by a pup peaked in
the spring, with a secondary peak in the fall. Two methods
of estimating the annual reproductive rate gave comparable
values of 0.88 and 0.90 pups per adult female. The average
inter-birth interval was 416 days. Eight of the 19 pups born
did not survive to weaning.

Among the four major age/sex classes, adult females had
the highest estimated survival rates and the adult males the
lowest. Juvenile females had lower survival rates than adult
females but juvenile males had higher survival rates than
adult males.

The estimated annual loss rate for the flipper-tags was
0.26. More individuals lost two tags than would be expected
by chance. It is unlikely that accurate estimates of sea
otter survival rates can be derived from observations of
tagged individuals.

XX

Chapter 3.

We obtained a detailed picture of sea otter movement
patterns in California by attempting to locate each
instrumented otter, by radiotelemetry, on a daily basis. [In
general, otters tended to stay within a small area (1-2 km of
shoreline) for an extended period and then suddenly move for
a much longer distance. Our daily monitoring revealed that /
individual otters of all age/sex classes make a surprising |
number of long-distance movements at all times of year. There
was substantial variation in movement patterns among
individuals within all age/sex classes but there were also
Significant differences between classes. Juvenile males were
the most extensive travelers and also ranged farther offshore
‘than otters of the other age/sex classes.

Chapter 4.

Radiotelemetry is particularly useful for collecting time
~budget and activity data on sea otters because radio signals
are not transmitted through sea water. Three general
categories of activity can be distinguished by listening to
the radio signal from an otter: resting, feeding, and
"other". Otters of all age/sex classes tended to be active
and feed for a large proportion of the time during the late
afternoon and early evening but there were differences in the
activity patterns of the various groups. Juvenile females
and adult females with pups spent more time foraging than
other otters. Differences in the ability of members of
different age/sex classes to compete for food resources are
common in vertebrates. [In the California sea otter
population, the juvenile females spent almost half of their
time foraging, suggesting that they are poor competitors for
food.

Chapter 5.

Although we collected some data on uninstrumented otters,
we focused on the foraging patterns of individual instrumented
sea otters as indicated by radio-telemetry. Our telemetry
data indicated that visual observations of otter foraging

~patterns tend to underestimate mean dive lengths. There was
a striking degree of individual variation in foraging
patterns, Many individuals displayed differences in diurnal
and nocturnal dive-length patterns that may reflect a tendency
to specialize on different prey species by day and night.
However, there was no general tendency for longer dive lengths
or surface intervals during the day or night. Juvenile males
often fed far from shore where they could not be seen.
Juvenile females had longer feeding bouts than otters of the
other age/sex classes.

XxXi

Chapter 6.

In an effort to gain insight into the age structure of
the California population, we studied a sample of premolars
from more than 580 dead sea otters salvaged from beaches. We
counted bands in the cementum of the sectioned teeth to
estimate age. We were able to examine teeth from ten otters
of known minimum age and the age estimates based on these
teeth compared quite favorably with those made by field
biologists. Age estimates based on teeth also compared well
with those based on skull features. Teeth that had been
boiled were more difficult to interpret than those that had
not been boiled. There was excellent agreement between
successive age estimates by the same reader and good agreement
across readers.

Chapter 7.

We evaluated the accuracy and precision of the
radiotelemetry methods we used to locate otters in California
with a radio transmitter on a buoy anchored off the coast.
We established the location of the buoy with visual methods
and took a series of compass bearings on the buoy's radio
signal. Signal bounce was not a significant problem. The
accuracy of our bearings compared quite favorably with that
of those taken in other radio-telemetry studies. For otters
located within about 800 meters from shore, precision was
estimated at 0.03 to 0.06 hectares and accuracy at 51 to 110
meters. The results obtained by hand-plotting points, which
was our usual field procedure, compared well with those
obtained with the Andrews estimator calculated by the computer
program TRIANG.

Chapter 8.

This chapter focuses on relatively long-term movement
patterns of adult female and juvenile sea otters in Alaska.
Adult females were much more mobile than had previously been
suspected but their movements were greatly reduced in the
month before weaning. Male weanlings left the area in which
they were born shortly after weaning, so that spatial
segregation of the sexes occurred at a very young age. Sea
otters used different portions of the available habitat for
different purposes, such as for weaning pups and over-
wintering. Hence, movements and habitat use varied
seasonally.

Chapter 9.

Birth dates, growth rates, dependency periods, weanling
behavior and survival of male and female otter pups in Alaska

Xxii

were compared. Many of these factors did not vary between
males and females. However, dependent male pups grew more
rapidly than dependent females and weaned females had lower
survival rates than weaned males.

Chapter 10.

A stochastic simulation model of California sea otter
population dynamics was constructed to be used in the analysis
of the risk of oil spills to the legally threatened
population. The model consists of four submodels: iL)
population model that iterates on a monthly basis; 2) a
population model that iterates on a yearly basis; 3) a
spatially explicit population distribution model; and 4) a
sea otter movement model. Simulated population dynamics are
density-dependent but the model has the flexibility to allow
investigation of density-independent reproduction and
mortality. The monthly population submodel operates for four
Simulated years before the simulated oil spill. At the time
of the spill, individual animals are distributed along the
coast by the distribution submodel. In the movement submodel,
individual animals then either avoid or are killed by the
spill. Population recovery can be simulated for up to 50
years after the spill using the monthly and annual population
models.

Age and sex specific survival and reproductive rates are

the core of the population submodel. These rates are
estimated using telemetry and other data in a "competing
risks" theoretical framework. Data from the semi-annual

censuses of the population conducted by the California
Department of Fish and Game and the U.S. Fish and Wildlife
Service are incorporated in the distribution model. [In the
movement submodel, daily movements are modeled with regression
equations, using parameters estimated from the radiotelemetry
data on the California animals.

Sensitivity analysis of the population model indicated
that the recovery time after a spill depends on the percentage
of the female population killed, the status of the population
in relation to its carrying capacity at the time of the spill,
and the amount of environmental stochasticity in annual
survival rates.

Xxiii

CHAPTER 1

OVERVIEW OF THE STUDY: BACKGROUND AND GENERAL METHODS

D. B. SINIFF AND K. RALLS

November 30, 1988

BACKGROUND

The outer continental shelf of the Pacific coast is
believed to contain extensive oil and gas reserves. The Santa
Maria and Santa Cruz Basins, off the coast of central
California, are potentially some of the most active areas of
oil exploration and development. Areas to the south of Point
Conception have already been developed into productive fields,
and more will be developed in the future. The Minerals
Management Service (MMS), U.S. Department of Interior, is the
federal agency responsible for administering leases of
submerged federal lands. Amendments to the Outer Continental
Shelf Lands Act of 1953 set MMS objectives for managing
development of outer continental shelf lands, including
protection of human, marine, and coastal environments.

A first step in making decisions about leasing,
exploration, and development that protects the marine and
coastal environment is risk analysis. MMS has directed and
funded a number of studies of the risk of off-shore oil
development, particularly of 011 spills resulting from leasing
activities, to wildlife populations (e.g. Ford 1985, Reed, et
al., 1986).

One of the most sensitive wildlife species, from both
political and biological perspectives, that could be impacted
by an accidental spill development in the Santa Maria and
Santa Cruz Basins is the California sea otter (Enhydra
lutris). Commercial exploitation during the 18th and 19th
centuries reduced the aboriginal population of perhaps 20,000
otters along the California coast to probably less than 100
in 1911 (USFWS 1986). Protection provided by international
treaty and federal and state legislation allowed the
population to recover, at a rate of about 5% per year, to its
present size of approximately 1500 animals (Ralls, et al.,
1983). In 1976 the southern sea otter was officially listed
as "threatened" under the Endangered Species Act; a major
reason for the designation was the potential risk of oil
spills to the small and geographically isolated population.
To obtain information on the southern sea otter, MMS issued
a request for proposal in late April 1983. It outlined a
series of objectives for studies on "The Population Status of
California Sea Otters". Three objectives were central to this
request for proposal: 1) to consider the existing information
on the dynamics and demography of the California sea otter
population and determine what additional information would be
necessary to predict the effects of oil spills, of various
sizes in different parts of the sea otter range in California
(Fig. 1.1), on the population; 2) to design and conduct
studies needed to fill the identified data gaps; and 3) to
develop a population model that would help to determine the
way in which the size and productivity of the population would

2

be likely to be affected by oil spills in different parts of
the range.

In October, 1983, we received the contract to carry out
this work and immediately started to obtain the necessary
federal and state permits to conduct field studies of sea
otters in California. We proposed the use of implanted radio
transmitters to monitor otters along the California coast.
In March, 1984, we received the necessary permits to implant
five otters and began field work in California. We
subsequently obtained additional permits and have implanted
a total of 40 sea otters in California and monitored them
through late December, 1987. In this report, we analyze the
data collected on these individuals and the results are
presented in Chapters 2 through 5.

We collected teeth from the salvaged otters found dead
along the California coast over the last 20 years. The teeth
were sectioned to allow estimation of the ages of individuals.
This information was used in the development of the population
model. The methods used and the results of the tooth analysis
is presented in Chapter 6. In the development of the model,
the data from the field studies were used primarily for
establishing a basis for movement patterns along the coast and
estimates of reproduction and survival rates. Since the
precision and accuracy of the telemetry positions along the
California coast were important in our analyses, an evaluation
was carried out using transmitters at fixed locations that
simulated floating otters. This evaluation is presented in
Chapter 7.

Because the California sea otter population is classified
as threatened under the Endangered Species Act, we tested
procedures and equipment on sea otters in Prince William
Sound, Alaska (Fig. 1.2), before using them in California.
Data on sea otters in Alaska were collected while equipment
and procedures were being developed. These data resulted in
significant new findings, particularly with respect to events
during the period of pup dependency. They are presented in
two chapters of this report, "Movement patterns of adult
female and weanling sea otters in Prince William Sound,
Alaska" (Chapter 8) and "Sex-related patterns in the post-
natal development and survival of sea otters in Prince William
Sound, Alaska" (Chapter 9).

The final chapter of this report describes the population
model that we developed. A significant body of data from
diverse sources has been integrated into this effort. These
data have come from our current field studies as well as the

FIGURE 1.1 - A map of the study area in California showing the
approximate range of sea otters during the period of the study
in 1984, 1985, and 1986. The portion of the range in which
reproduction occurs is indicated by cross-hatching; the
northern and southern areas occupied mostly by males are
indicated by diagonal lines.

San Nicolas @

FIGURE 1.2 - A map of the study area in Alaska showing the
general range of the sea otters that were studied during 1984
and 1985.

“eee a” PRINCE WILLIAM
SOUND

[i] stuoy AREA

Ati YW ALDEZ
iQ mu

—zZ—>

AR Se “gl
BG

‘ i
i jut

: iat

ng! ior

“2 7

i

oi aT iy | ig a ae

past efforts of both the California Department of Fish and
Game and the U.S. Fish and Wildlife Service. We organized all
these past survey data into a data base and used it
extensively in the development of the model.

GENERAL METHODS

Capture and release

We captured otters in three ways: with floating gill-
nets, the Wilson trap developed by the California Department
of Fish and Game (Ames, et al., 1983), and dip-nets, which
have been used extensively by the U.S. Fish and Wildlife
Service in California. Floating gill-nets were used mostly
in Alaska. We used them initially in California but wind and
fog made it difficult to check the nets frequently. The
majority of the adult otters in California were captured with
the Wilson trap and most of the juveniles with dip-nets. We
also used dip-nets to capture pups and newly independent young
in Alaska. All animals were released near the point of
capture.

Teeth

Sea otters have a small, vestigial premolar directly
behind each canine (Fig. 1.3). Schneider (1973) developed an
aging technique based upon the number of cementum layers found
in stained sections and Garshelis (1984) applied this
technique to teeth extracted from anesthetized sea otters in
Alaska. We collected this premolar from many of the
California animals. However, our sample was incomplete due
to lack of permission to extract a tooth during the early
portion of the study and some breakage during extraction of
the teeth.

Transmitters

The transmitters were developed by the University of
Minnesota's Cedar Creek Bioelectronics Laboratory. The first
models, used in Alaska in 1982 before this project began,
measured about 6.8 x 4.8 x 1.8 cm and weighed about 70 g.
Although it had been shown that implanting transmitters in
the abdominal cavity had no deleterious effects in other
species (Smith, 1980; Eagle, et al., 1984), no information on
this point was available for sea otters. Therefore we set out
to compare the results of implanting transmitters beneath the
skin (subcutaneous) and within the abdominal cavity

FIGURE 1.3 -- Schematic drawing of the upper jaw of a sea
otter showing the vestigial first premolar that was sometimes
removed from captured animals and sectioned for determination
of age.

VESTIGIAL FIRST PREMOLAR

(intraperitoneal). Because some of the 1982 implants were to
be subcutaneous, we used small, flat batteries on which we had
no previous performance data. Neither method of implantation
appeared to have significant deleterious effects on the
otters; however, the subcutaneous implantation procedure left
a noticeable lump so we decided to use the intraperitoneal
method in the future. Unfortunately, the small batteries
proved to be unreliable and most of them failed within four
months after the transmitters were implanted (Garshelis and
Sime, IGS)

We then designed a new transmitter using lithium
batteries developed by the Medtronic Corporation. These
batteries were developed for use in medical devices implanted
in humans and were known to be extremely reliable in these
applications. However, at the request of the U.S. Fish and
Wildlife Permit Office, these transmitters were subjected to
four months of extensive testing prior to use in California.
They were operated under simulated physiological conditions
for this entire period, except when undergoing tests at
extreme temperature and pressure conditions, and successfully
passed all tests.

These transmitters (Figure 1.4) measured about 7.6 x 5
x 2.5 cm and weighed about 120 g in air. This weight ranges
from about 1.8 percent (in an 18-lb juvenile) to about 0.4
percent of sea otter body weight (in a 70-lb adult male), thus
these transmitters were smaller, relative to body weight, than
those used successfully in other species of mammals. For
example, Eagle, et al., (1984) used transmitters that ranged
up to 3.7 percent of body weight in mink and about 8 percent
of body weight in ground squirrels.

They were coated with medical grade Energy Technology
Urethane to ensure that they would not produce adverse
reactions in biological tissues. They were gas-sterilized and
sealed in plastic surgical bags for storage until implanted.

These new transmitters were first used in five animals
in California, beginning in March 1984. Their reliability
proved to be excellent and their lifespan approached the
maximum expected battery life of 700 days. They had a rather
limited range of approximately one mile from surface
monitoring stations. Engineers at the Cedar Creek
Bioelectronics Laboratory then reconfigured the placement of
the internal components of the transmitter and redesigned the
antenna. These’ improvements increased the range
significantly, up to five miles from surface stations and 10
miles from aircraft. This new design was used in subsequent
transmitters, beginning in Alaska in the summer of 1984 and
in California in the spring of 1985.

FIGURE 1.4 - A schematic drawing of the implanted radio
transmitter used during this study showing its component parts.

Encapsulating material

Antenna

Lithium battery
Urethane coating
Lithium Expansion spacer
battery

Lithium battery

Expansion spacer

Drugs

The otters were immobilized using the methods given in
detail in Williams, et al., (1981). Fentanyl was given
intramuscularly at dosages of 0.5-0.1 mg/kg of body weight in
combination with azaperone at dosages of 0.010 to 0.053 mg/kg.
This combination produced a safe, short-acting, and easily
reversible immobilizing agent suitable for use under field
conditions. The combination of anesthetic and tranquilizer
was given to the otters while they were entangled in the gill-
net or held in the Wilson trap or dip-net.

Surgical procedures

All surgery in California, and the initial surgery in
Alaska, was carried out by Dr. Thomas D. Williams, who
developed both the anesthetic procedures (Williams and Kocher,
1978) and the surgical techniques (Williams and Siniff, 1983).
In 1984, Dr. Williams trained two other veterinarians to do
the operation and they performed some of the operations in
Alaska.

Surgery was carried out on a specially constructed
operating table, either on board the capture boat or on the
beach near the capture site. After an initial health
screening procedure, the anesthetized otter was secured to the
table with the ventral surface up. The status of the animal
was monitored by capillary perfusions, color of the mucous
membranes, respiratory rate and depth, temperature, and heart
beat. A 50-50 mixture of KY jelly and betadine solution was
applied to the ventral midline below the umbilicus and rubbed
down to the skin. A comb was used to part the pelage and
betadine solution was sprayed over the part. A sterile drape
was placed over the ventral abdomen and thorax. Sterile
gloves were used for each operation and all instruments were
sterilized in benzol.

Taggin

In California, all instrumented animals were tagged on
the hind flippers with colored Temple tags. In Alaska, both
Temple tags and small button tags were used. The tagging
methods and color/location coding system used in California
were those used by both the California Department of Fish and
Game and the U.S. Fish and Wildlife Service for many years
(Ames, et al., 1983). The particular color combination used
on each animal was selected in consultation with CDF&G
personnel.

Monitoring procedures

We used different monitoring techniques and schedules in
California and Alaska because of the climatic and geographical
differences between the two areas. It was possible to monitor
the animals all year in California but not in Alaska. The
usual field season in Alaska extended from late April to mid-
September. However, we visited Alaska occasionally during the
winter and conducted aerial searches for instrumented animals.

Routine daily monitoring in California was done from the
ground. When animals could not be located from the ground,
aerial searches were conducted using a small, ized wing plane
with antennas attached to each wing.

There were no roads in the Alaska study area. Some
monitoring was done from a small boat but it was impossible
to locate each animal every day due to the large size and
complicated geography of the study area. Much of the
monitoring was done from the air.

In California, we evaluated the accuracy and precision of
the sea otter locations that were obtained by telemetry
triangulation. Transmitters were placed in floating buoys and
positioned along the coast so that readings could be taken
according to the established procedures we used in our
monitoring of instrumented otters. The results of this
evaluation are presented in Chapter 7.

LITERATURE CITED

Ames, J.A., R.A. Hardy, and F.E. Wendell. 1983. Tagging
materials and methods for sea otters, Enhydra lutris.
Calif. Fish and Game 69:243-252.

=== ——————— - 1986. A simulated translocation of sea otters,
Enhydra lutris, with a review of capture, transport, and
holding techniques. Marine Resources Technical Report
NOs BB, NY jyas

Eagle, T.C., J. Choromanski-Norris, and V.B. Kuechle. 1984.
Implanting radio transmitters in mink and Franklin's
ground squirrels. Wildl. Soc. Bull. 12:180-184.

Garshelis, D.L. 1984. Age estimation of living sea otters.
J. Wildl. Manage. 48:456-463.

Garshelis, D.L. and D.B. Siniff. 1983. Evaluation of radio-
transmitter attachments for sea otters. Wildl. Soc.
Bull. 11(4) :378-383.

Schneider, K.B. 1973. Age determination of the sea otter.
Alaska Dept. Fish and Game, Fed. Aid in Wildlife
Restoration, Final Report, Proj. W-17-4 and W-17-5, Job
8.10R. 23 pp.

Smith, H.R. 1980. Intraperitoneal transmitters in suckling

white-footed mice, Peromyscus leucopus. Biotelemetry
Patient Monitoring 7:221-230.

Williams, T.D., A.L. Williams, and D.B. Siniff. 1981.
Fentanyl and azaperone produced neuroleptanalgesia in the
sea otter, (Enhydra lutris). J. Wildl. Dis. 17:337-342.

Williams, T.D. and F.H. Kocher. 1978. Comparison of
anesthetic agents in the sea otter. J. Am. Vet. Med.
Assoc. 173:1127-1130.

Williams, T.D. and D.B. Siniff. 1983. Surgical implantation
of radiotelemetry devices in the sea otter. J. Am. Vet.
Med. Assoc. 183:1290-1291.

CHAPTER 2

REPRODUCTION, SURVIVAL AND TAG LOSS IN CALIFORNIA SEA OTTERS

D. B. SINIFF AND K. RALLS

November 30, 1988

INTRODUCTION

The California sea otter population is listed as
threatened on the U.S. Endangered Species list and its status
and management are of concern to several state and federal
agencies. A population model is a basic tool for the
understanding and management of any wildlife population. The
development of a population model requires reliable estimates
of reproductive and survival rates; no estimates of these
rates are available for the California sea otter population.

Early knowledge of the general biology of the sea otter
reproductive cycle was gained mostly by the examination of
reproductive tracts from animals collected in the U.S.S.R. or
Alaska (Barabash-Nikiforov, 1947; Sinha, et al., 1966; Kenyon,
1969; Schneider, 1973). These studies showed that the litter
size is typically one, with maternal care extending at least
four months after parturition, and that a birthing peak occurs
in the spring, although birth can occur at any time of year.
These early studies generally placed the inter-birth interval
at about two years. However, subsequent observations of
tagged sea otters in both California and Alaska have indicated
that the inter-birth interval is closer to one year (Jameson
and Johnson, 1979; Loughlin, et al., 1981; Wendell, et al.,
1984). Considerable data on tagged individuals are available
for the California sea otter population (Estes and Jameson,
1983; Wendell, et al., 1984) but it has not been possible to
obtain good estimates of survival rates from these data for

several reasons. For example, information on tag loss rate
and difficulty in resighting tagged individuals greatly
complicates this estimate. The only available data on tag

loss rate is based on the resighting of tagged individuals
(Ames, et al., 1983). It is thus an estimate of the rate at
which tagged individuals disappear from the pool of regularly
re-sighted animals, rather than a direct estimate of the rate
at which individual tags are lost.

In this chapter, we present estimates of the proportion
of adult females accompanied by pups throughout the year; the
inter-birth interval; the period of pup dependency; and
reproductive, survival and tag loss rates for the California
sea otter population. All estimates are based on observations
of telemetry instrumented, flipper-tagged sea otters.

METHODS

We captured 49 otters (Table 2.1). Females known to be
pregnant and small pups were not implanted with transmitters.
We implanted radio transmitters in 40 otters, which were
assigned to age/sex classes on the basis of their weight,
estimated age (sometimes from teeth annuli), and, in the

TABLE 2.1 -- Summary of sea otters captured in California
during 1984 and 1985.

OTTER CAPTURE CAPTURE SEX WT. TRANS.

NO. DATE AREA LBS) FREQ. LEFT;RIGHT TAG

1 7Mar84 Morro Bay 74 723 4/5 red;1/2 silver

2 16Mar84 Morro Bay 65 955 4/5 white;1/2 silver

3 21Mar84 Morro Bay 44 545 4/5 chartreuse;1/2 silver
4 21Mar84 Morro Bay 53 784 4/5 1t. blue;1/2 silver
6+ 3Ju184 San Simeon 36 807 1/2 orange;1/2 roy.blue

7 15Feb85 Big Sur R. 54 333 4/5 pink;4/5 purple

8 1Mar85 Rancho Rico # 49 none 1/2 1t. green;1/2 purple
9 1Mar8s85 Wreck Beach 30 233 1/2 silver;1/2 purple

10 1Mar85 Wreck Beach 60 041 1/2 1t. blue;4/5 purple
11 15Mar85 Molera Point 35 417 4/5 red;1/2 purple

12 16Mar85 Grimes Point # 45 none 4/5 chartreuse;1/2 purple
13 16Mar85 Torre Canyon 25 461 4/5 white;4/5 purple

14 16Mar85 Torre Canyon 41 217 4/5 1t. blue;1/2 purple
15 20Mar85 Torre Canyon 43 842 1/2 purple;4/5 orange

16 3Aprs5 Wreck Point 39 884 4/5 pink; 1/2 purple

17 3Aprs5 Grimes Point 53 230 1/2 1t. green;4/5 purple
18 3Aprs5 Torre Canyon # 57 none 1/2 silver;1/2 purple

19 3Apr85 Rancho Rico 36 373 4/5 white; 1/2 purple
20 3Apr8s5 Rancho Rico none 4/5 purple;1/2 purple

21 10Apr8s5 False Sur 45 133 1/2 dk. green; 1/2 purple
22 10Apr8s5 Big Sur R. 42 562 1/2 roy. blue; 1/2 purple
23 10Apr8s5 False Sur 63 062 1/2 roy. blue; 4/5 purple
24 10Aprs5 Point Sur * -- none ear tag 217
-- 10Apr85 Point Sur -- none none
25 13Apr8s5 Anderson Crk 56 933 1/2 orange; 1/2 purple
-- 13Apr8s5 N. of slide -- none none
-- 13Aprs5 N. of slide * -- none none
26 8May85 Little Sur R. 46 680 4/5 1t. green; 1/2 purple
27 40ct85 Dolan Rock 46 121 4/5 1t. green; 1/2 purple
28 40ct85 Esalen 36 256 4/5 orange; 1/2 purple

29 110ct85 Dolan Rock SiG 6S 5ieel /2iesa vex 2p anik<

30 110ct85 Dolan Rock 34 475 1/2 white; 4/5 purple

31 110ct85 Buck Cr. 40 904 4/5 silver; 1/2 purple

32 180ct85 Buck Cr. # 50 none 4/5 gold; 1/2 purple

33 180ct85 Big Slide 46 970 1/2 gold; 1/2 purple

34 190ct85 J.P. Burns 80 625 4/5 1t. blue; 4/5 purple
35 8Nov85 Ragged Pt. 35 960 1/2 1t. blue; 4/5 chartreuse
36 8Nov8s5 County Line 37 433 4/5 red; 1/2 chartreuse
37 22Nov85 Beckets Rf. 25 380 4/5 orange; 1/2 chartreuse
38 22Nov85 Cypress Ovrl. F 25 603 4/5 pink; 1/2 chartreuse
39 22Nov85 San Carpoforo F 30 587 4/5 yellow; 1/2 chartreuse

ed) testes od eed Pred 1) Ta) Feed ea) Ua) 00a) aed Pee) 5p tesa) ae) Ua od og) 19 td Pood) eal ead cena) ed tee
|
fey)

15

TABLE 2.1 (continued)

OTTER CAPTURE CAPTURE SEX WT. TRANS.

NO. DATE AREA LBS) FREQ. LEFT;RIGHT TAG

40 17Dec85 Piedras Blan. 31 405 4/5 1t. blue; 1/2 chartreuse
41 17Dec85 Piedras Blan. 28 152 1/2 orange; 4/5 chartreuse

42 17Dec85 San Carpoforo 35 273 4/5 1t. green; 1/2 chartreuse
43 18Dec85 Piedras Blanc 353 1/2 pink; 4/5 chartreuse

44 18Dec85 Piedras Blanc 33 534 4/5 white; 1/2 chartreuse

45 18Dec85 Piedras Blanc 27 493 4/5 purple; 1/2 chartreuse

46 18Dec85 San Simeon Pt 25 301 4/5 silver; 1/2 chartreuse

47 30Dec85 Lover's Pt. 30 028 1/2 pink; 1/2 orange

hy hy hy ey Ss oS
uJ
| aed

+ Otter 5 was captured and tagged by the U.S. Fish and Wildlife
Service

# Pregnant female

* Pup

case of females, reproductive history during the monitoring
period. Our sample consisted of nine adult males, five
juvenile males, 16 adult females, and 10 juvenile females
(Table 2.2).

Adult females were located on an almost daily basis by
radiotelemetry. We then attempted to observe them visually
through binoculars and a Questar spotting scope (up to 80
power) and record the presence or absence of pups and flipper
tags. This was often impossible due to weather conditions,
such as fog or rough seas, or difficult lighting conditions.
Some individuals were more difficult to observe then others,

depending upon their location along the coast. Thus the
length of time between visual observations varied across
individuals.

To determine the proportion of females accompanied by
pups each month of the year, we tallied whether or not each
female was accompanied by a pup for every month she was
monitored (Appendix 2.1). We considered that a particular
female had been accompanied by a pup for a given month if we
knew she had been accompanied by a pup for more than a 15 day
period that month. If she had been accompanied by a pup for
less than a 15 day period, we considered that she had not been
accompanied by a pup that month. Months where the status of
a particular female -was unknown were not counted. The
variation in the number of days between sighting of the
individual female otters created a problem in calculating the
inter-birth interval and the number of days a pup remained
with a female. In the case of an inter-birth interval (the
time from birth of one pup to birth of the next pup), we
usually did not know the exact birth dates of the two pups,

16

TABLE 2.2 -- A list of sea otters that were instrumented with
implanted radio transmitters, their estimated age when
available and other vital statistics.

OTTER SEX WEIGHT ESTIMATED PUPPED? AGE/SEX

NUMBER LBS AGE CLASS

1 M 74 25 SF ADULT MALE

2 M 65 oS QS ADULT MALE

3 M 44 2S SS ADULT MALE

4 M 53 2S 2S ADULT MALE

6 F 36 SO NO ADULT FEMALE

7 M 54 7* SO ADULT MALE

9 F 30 5* YES ADULT FEMALE
10 M 60 6* ae ADULT MALE
11 F 35 == YES ADULT FEMALE
13 M 25 SO am JUVENILE MALE
14 F 41 6* YES ADULT FEMALE
15 F 43 5% YES ADULT FEMALE
16 F 39 S52 YES ADULT FEMALE
17 M 53 6* oS ADULT MALE

19 F 36 SS YES ADULT FEMALE
21 F 45 15* NO ADULT FEMALE
22 F 42 8 NO ADULT FEMALE
23 M 63 5 aS ADULT MALE
25 F 56 7 YES ADULT FEMALE
26 F 46 2S YES ADULT FEMALE
27 F 46 10 YES ADULT FEMALE
28 F 36 13* NO ADULT FEMALE
29 F 36 3 NO JUVENILE FEMALE
30 M 34 2* a JUVENILE MALE
31 F 40 11 YES ADULT FEMALE

33 F 46 10 YES ADULT FEMALE

34 M 80 8 == ADULT MALE
35 M 35 2 oo JUVENILE MALE
36 F 37 9 YES ADULT FEMALE

37 F 25 2 NO JUVENILE FEMALE
38 F 25 1* NO JUVENILE FEMALE
39 F 30 2 NO JUVENILE FEMALE
40 F 31 3 NO JUVENILE FEMALE
41 M 28 2 =D JUVENILE MALE
42 1p 35 2 NO JUVENILE FEMALE
43 M 31 <1 o> JUVENILE MALE
44 F 33 2 NO JUVENILE FEMALE
45 F 27 1 NO JUVENILE FEMALE
46 F 25 2 NO JUVENILE FEMALE
47 F 30 <1 NO JUVENILE FEMALE

*The teeth from these individuals were damaged so that only
a minimum age could be estimated.

although we knew that each birth occurred within some time
period. For example, if we saw an instrumented female without
a pup on 1 April and saw her again with a pup on 10 April, the
pup had obviously been born between these two dates. Suppose
we continued to monitor this female throughout the period of
pup dependency and her next pregnancy. We last saw her
without a pup on 15 March of the following year but she had
a pup when next seen on 10 April. There were thus two
intervals when her reproductive status was unknown. The first
interval was 10 days and the second was 26 days. To calculate
the inter-birth interval, we added one-half of each of these
intervals to the number of days between the first sighting of
the female with a pup one year and the first sighting of this
female with her next pup. For our example, then, the
inter-birth interval was 383 days -- 365 days from 10 April
of the first year to 10 April of the second year plus five
days from the interval between 1 April and 10 April spanning
the birth of the first pup and 13 days from the interval
between 15 March and 10 April spanning the birth of the
subsequent pup. To make this calculation, we arbitrarily
included for the interbirth interval estimates, only data for
which neither of the unknown status intervals was more than
51 days.

The same problem arose in calculating the number of days
the pup remained with the female. Again, there were two
intervals when the status of the pup was unknown, one spanning
its birth and a second spanning its disappearance. As before,
we used only data in which both intervals for which the status
of the pup was unknown were less than 52 days. We divided
each interval in half and added it to the period between the
first and last dates the pup was seen with the female.

We estimated the annual reproductive rate in two ways.
The first method was based on the total number of days all the
adult females were monitored and the known number of pups born
to them during this period. This calculation again included
intervals when the adult females were not seen and their
status was unknown. The longest such intervals were 10 months
for one female (otter 11) and four months for another (otter
16) (Appendix 2.1). We summed the total number of monitoring
days across 13 adult females, and divided this figure by the
total number of pups produced to obtain the average number of
days required to produce a pup. This average number of days
was then divided into 365 to obtain an estimate of the annual
pupping rate.

In the second method, we considered only the five females
used to calculate the inter-birth interval and the period the
' pup remained with the female. For this data set, we divided
the average inter-birth interval into 365 to obtain an
estimate of the annual pupping rate.

18

Our first method of estimating the annual reproductive
rate is probably most similar to that which one would use for
tag-resight data (Wendell, et al., 1984), where females are
observed periodically and their reproductive status noted.
This method also produces gaps where the status of the female
is unknown; the length of these gaps varies with the frequency
of the attempts to re-sight the females and the difficulty of
observing each female.

The estimation of survival rates from radiotelemetry data
may be approached somewhat differently than that from data
derived: from resighting or recapture of tagged animals.
Instrumented animals are known to be alive or dead on a daily
basis. Thus the procedure that has been developed for
radiotelemetry data is to estimate a daily survival rate and
expand it to the time period desired, usually one year,
assuming that the daily rate remains constant over this period
(Heisey and Fuller, 1985). The formulation recommended by
Heisey and Fuller is:

Annual Survival = (Transmitter days - deaths) *°

Transmitter days

When days are used as the basic time interval, it is
necessary to assume that the status of each individual (dead
or alive) is known for each day. This assumption was not
fulfilled for our animals when individuals disappeared and we
were unable to determine their fate. These individuals were
classified as missing (Appendix 2.2), and these individuals
might have died or their transmitters might have expired.
Animals classified as transmitter failed, transmitter expired,
or transmitting were known to be alive at the number of days
indicated in Appendix 2.1. The way in which these missing
animals are treated can affect the survival rate estimate.
One way to handle this problem is to assume that transmitters
are extremely reliable for some number of days, and animals
lost prior to this time have died. We used 450 days for this
decision point for missing animals. Thus, we assumed that
otters missing in less than 450 days from the capture date had
died and that those missing after 450 days were alive on the
date they became missing. This criterion was based on the
average life span, 485 days, of the five transmitters in which
expiration was verified. We also calculated survival rates
for each age/sex class by using the number of individuals in

each sex and age class as the basis of calculations. This
means following the binomial model:
Seal - yn where:
S = Estimate of survival
D = number of animals that died
n = sample size for the particular sex/age

category.

For this model an estimate of variance is easily calculated
by the standard:

s* = "47n where:
p = survival rate estimate
q = D/n, or the proportion dying

It only remains to specify the time interval, which is
normally taken to be one year. However, in this case we
followed the same 450 day criterion. Thus the binomial model
for an estimate of the annual survival rate becomes:

Ss = (1 - "yn 365/450

We estimated annual survival rate each of these ways for the
four sex and age classes of adult females, adult males,
juvenile females and juvenile males.

The pup survival rate was based on the ratio of the
number of pups that died during the period of dependency to
the number of pups born, over the entire monitoring period.
We assumed that pups that remained with the female less than
150 days died, based on an estimated pup dependency period of
six months (Payne and Jameson, 1984; Wendell, et al., 1984).

One of the advantages of being able to locate animals by
radiotelemetry is that they are still identifiable after tag
loss has occurred. All of our instrumented animals were
tagged with one Temple tag in each hind flipper, following a
procedure developed by the California Department of Fish and
Game (Ames, et al., 1983). In addition, the tags were drilled
so that a small nylon or stainless steel screw could be used
to hold the two sides of the tag together. A small amount of
glue was dripped into the hole before the screw was inserted.
Our data on the presence or absence of tags were similar to
our data on reproduction. However, tags were more difficult
to see than pups, so many of the intervals between the last
date a tag was seen and the first date it was seen to be
missing were longer than the corresponding intervals for pups.
We analyzed these data using the same method we used to
analyze the data on the survival of individual otters.
However, we assumed that a tag had survived until the date it
was seen to be missing and thus calculated only the maximum
possible tag survival rate.

20

RESULTS

Although some proportion of adult females were accompanied
by pups throughout the year, this proportion peaked in the
spring, with a secondary peak in the fall (Fig.2.1). Our
total reproductive data set includes information on the 13
adult females that were monitored at least 355 days (Table
2.3). However, our best data for the determination of the
inter-birth interval and the period the pup remained with its
mother came from five females for which the length of all
three intervals in which their status was unknown (those
spanning the birth of the first pup, the disappearance of the

Table 2.3 -- Reproductive and age data and length of
monitoring period for adult female sea otters.
OTTER AGE DAYS PUPS PUPS
NUMBER YRS MONITORED BORN DIED

6 = 355 0 (0)

9 5* 544 2 1
11 = 621 1 (0)
14 6* 744 2 (0)
15 5% 725 1 0
16 = 608 2 (0)
19 = 587 3 (0)
22 8 585 (0) (0)
25 7 555 1 (0)
27 10 540 1 1
31 11 637 2 2
33 10 630 2 2
36 9 609 2 2
TOTALS 7740 19 8

*Only a minimum age could be estimated from these individuals
due to tooth damage.

first pup, and the birth of second pup) was less than 52 days

(Table 2.4). For this data set, the average inter-birth
interval was 416 days. Our first estimate of the annual
reproductive rate is based on the information summarized in
Table 2.3. Nineteen pups were produced during 7,740

monitoring days; thus, an average of 407 days was required to
produce one pup, and the estimated annual reproductive rate
is 0.90 pups per adult female. Our second estimate is based
on the data in Table 2.4. The average inter-birth interval
is 416 days, which corresponds to an annual reproductive rate
of 0.88 pups per adult female per year.

Eight of the 19 pups almost certainly died before

21

FIGURE 2.1 - The percent of the adult female study animals in
California that were with pups during each month of the year.

Percent

O Females with Pups

39 4 Females with Pups,
Running Average

28 | Oo

Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

‘Month

Die

weaning, giving an estimated pup survival rate of 0.57 (Table
PE) B We estimated that these eight pups (Table 2.4B)
remained with their mother for less than 65 days. Our
estimate assumes that pups born shortly before the females'
transmitters expired lived, when in fact we were unable to
determine their fate, and it may exclude some pups that lived
only a short time and were thus undetected (Appendix 2.1).

Considering all 80 flipper tags in our sample (Appendix
2.3), the estimated annual tag survival rate was 0.74 (Table
2.6A). Survival rates for right and left tags were not
Significantly different. However, more otters lost two tags
than would be expected by chance, based on the assumption that
tag loss follows a binomial distribution. Using the number
of tags lost over the period of monitoring to derive the
expected probabilities of losing one, two, or no tags and
comparing these expected values to the data, we see that
significantly more otters lost both tags than expected (Table
2.6C). This result is perhaps not surprising, as some
individuals have been seen to bite and manipulate their
flipper tags, ultimately removing them, while other
individuals appear to ignore the presence of tags. If the
otters that lost two tags are excluded, the estimated annual
tag survival rate increases to 0.91 (Table 2.6B).

Both methods for estimating survival rates for the four
main age/sex classes, indicated the highest survival for adult
females and lowest for adult males (Table 2.5). Estimates for
juvenile females were lower than those for adult females,
while estimates for juvenile males were higher than those for
adult males. The survival rate of juvenile females was lower
than that of juvenile males for both methods of estimation.
The data on which we based these estimates are presented in
Appendix 2.2.

DISCUSSION

Our data on the proportion of females accompanied by pups
over the yearly cycle, which show a peak in the spring and a
smaller peak in the fall, are similar to data on the
proportion of pups to independent otters recorded by the
California Department of Fish and Game in their monthly counts

of index areas (Brody, Chapter 10). The independent otters
in these index areas are probably largely adult females along
with a few territorial males and juvenile females. This

bimodal pattern could be the result of females that
successfully raise pups having a longer inter-birth interval
than those that do not, thus placing the birth of their next
pup at one year plus three to four months. Our data are in
agreement with the emerging consensus that many female sea
otters pup on an approximately annual basis (Loughlin, et al.,

23

Table 2.4 -- Reproductive information on adult female sea
otters that were known to have given birth to at least one
pup. The table gives the number of days during which the
reproductive status of these females was known and unknown
(A). Calculated periods with and without pups and inter-birth
intervals for those animals where the unknown intervals were
not more than 51 days are shown below (B).

A. DATA

Otter Un- Known Un- Known Un- Known Un-
known Days known Days known Days known
Status With Status Without Status With Status
(Days) Pup (Days) Pup (Days) Pup (Days)

One Two
9 13 36 11 330 51

ALI a 113 1 318

14 11 149 20 390

15 2 81l**

16 41 117 104 101 120 28%*

19 68 85 80 91 57 39 5

25 2 176 8

27 11 24 26

31 8 16 8 313 28 41 20

33 10 31 2 376 5 8 19

36 37 26 2 258 17 1 5

B. CALCULATED VALUES

OTTER PERIOD WITH PERIOD UNTIL INTER-BIRTH
NUMBER PUP (DAYS NEXT BIRTH (DAYS) INTERVAL (DAYS
g* 48 361 409

14* 165 421 586

25 181

27 43

31% 24, 65 331 355

33% 37, 20 380 417

36* 46, 12 266 312

MEAN = 64 MEAN = 416

* Five females used to calculate the inter-birth interval.
** Still with pup when last seen.

24

Table 2.5 -- Annual survival rate estimates. Two estimates
were made for the age/sex categories of adult females, adult
males, juvenile females and juvenile males. The first was
obtained by converting the daily survival rate of radio-
instrumented otters to an annual rate (Heisey and Fuller,
1985). The second used the standard binomial model (see text
for explanation) and the status of individual sea otters as
a basis for the estimate. The estimated pup survival rate was
based on the pups born to the instrumented females.

TRANSMITTER DAYS INDIVIDUALS
AS THE BASIS* AS THE BASIS*

: STANDARD

SURVIVAL SURVIVAL DEVIATION
ADULT FEMALES 0.91 0.89 0.088
ADULT MALES 0.61 0.52 0.167
JUVENILE FEMALES 0.80 0.75 0.145
JUVENILE MALES 0.88 0.85 0.179

*Animals that were classified as missing before 450 days were
assumed to have died at the time they became missing, while
those classified as missing after this time were assumed to
have been alive when they became missing.

PUPS TO WEANING 0.57
1981; Estes and Jameson, 1983; Wendell, et al., 1984;
Garshelis, Johnson, and Garshelis, 1984). Our average

inter-birth interval of 416 days is clearly within the range
of the 17 intervals recorded by Wendell, et al., (1984) based
on observations of tagged otters. Although our sample size
is smaller than theirs, the intervals when the reproductive
status of the female was unknown tended to be smaller in our
data. One interesting aspect of our data is the rather large
degree of variation, with a range of 312 to 611 days. The
reason for this variation may be correlated with the length
of the dependency period. It is generally agreed that female
otters rarely, if ever, mate when accompanied by a pup
(Kenyon, 1969; Calkins and Lent, 1975; Garshelis, et al.,
1984). Clearly more data are needed on the relationship
between the inter-birth interval and the period of pup
dependency, which probably varies with the age and condition
of the female.

Minimum reproductive rates for sea otters in Alaska have
been suggested by the reproductive condition of females in
samples killed in late winter and spring (Kenyon, 1969;
Schneider, 1973). Kenyon (1969) found that 71% of animals
collected in this period were pregnant and 17% had recently

25

Table 2.6 -- Estimates of annual tag survival rates based upon
the survival rate estimation procedures for instrumented
animals as outlined by Heisey and Fuller (1985). The only
difference here is that the "death" of tags rather than
individuals is the unit of measure (A and B). A comparison
of tag loss to expected tag loss based on the binomial
distribution is shown in C.

A. ALL OTTERS

TAG ANNUAL NUMBER OF
SURVIVAL TAGS LOST
RATE (OF 80)

RIGHT 0.70 13

LEFT 0.78 10

ALL 0.74 23

B. EXCLUDING OTTERS THAT LOST TWO TAGS

TAG ANNUAL NUMBER OF
SURVIVAL TAGS LOST
RATE (OF 64)
ALL 0.91 ~ 6

C. COMPARISON OF TAG LOSS TO THE BINOMIAL DISTRIBUTION
ESTIMATED PROBABILITY OF LOSING ONE TAG = 23/80 = 0.2875

NUMBER OF NUMBER OF OTTERS EXPECTED FROM
TAGS LOST BINOMIAL MODEL
NONE 25 21
ONE 8 16
TWO 7 3
P<0.05, CHI-SQUARE = 61, 1 D.F.
given birth. Schneider (1973) found that for samples

collected in May, 59% were pregnant and 14% had recently given
birth. Combining these values gives approximate annual rates
of .88 and .73 for these two studies, respectively. No
estimates are available for the California population. Our
estimates, based on the number of pups born during the
monitoring period and the average inter-birth interval are,
as expected, slightly higher than the minimum rates available

26

for Alaska since ours include the complete annual cycle.
Because of the sea otter's ability to produce pups throughout
the year, and the probable relationship between the
inter-birth interval and the period of pup dependency, it
seems likely that estimates of annual reproductive rates in
sea otters will be quite dependent upon conditions during the
particular interval during which the data are collected.

Our estimated survival rate for pups from birth to
weaning, 0.57, while possessing the potential for bias, as
mentioned under results, is close to the 0.50 estimate needed

TABLE 2.7 --A comparison of the age of pups at separation from
the female in California and Alaska, based on our data and
other published data.

DAYS PUP ALASKA* CALIFORNIA**
WITH FEMALE

0-50 0) 8
51-100 3 3
101-150 1 2
151-200 4 5

* Data from Garshelis (1983).
** Data from Table 2 plus eight pups from Loughlin, et al.,
(1981).

SSaaa———aBa=a=ESeeeeeee————— eee

to produce a zero population growth rate in our population
model when combined with our other survival rate estimates
(Chapter 10). Surveys suggest that the population has been
stable over the past decade (Estes and Jameson, 1983). Table
2.7 contrasts survival patterns of dependent pups in
California and Alaska. Early mortality appears to be more
frequent in California. This pattern, if confirmed by
additional data, may be a result of storm patterns in
California, because the Alaskan data were collected in Prince
William Sound, which provides more shelter during periods of
inclement weather.

The annual survival rate estimates we obtained by
expanding the daily survival rate based on transmitter-days
also appear reasonable. The relatively close correspondence
between the estimates calculated by two different methods is
encouraging, as is their general agreement with other aspects
of the data collected on our instrumented animals, such as the

movement patterns, time budgets, and feeding patterns of the
different age/sex groups.

Our adult females, which had the highest survival rates,
were in areas of the range where human activities were minimal
and they traveled the least of the four sex/age groups. Adult
males had lower survival rates than adult females and
juveniles of both sexes. A low survival rate for adult males
is also suggested by the sex ratios in the California
Department of Fish and Game database on dead sea otters.
Considering only the carcasses in good condition, where the
sex of the carcass was known in almost all cases,
significantly more dead adult males than adult females washed
ashore (Table 2.8). Our adult males tended to be particularly
vulnerable during periods of travel. One was shot after he
had moved a considerable distance from his capture location
and two others disappeared when they were moving through areas
in which gill-net boats were operating. Juvenile males also
traveled extensively but tended to remain farther offshore,
which may provide some degree of protection from human
activities, such as shooting (Wild and Ames, 1974; Morejohn,
et al., 1975) and incidental capture in gill-nets (Wendell,
et al., 1986), and contribute to their higher survival rate.
The only juvenile male that died during our study was almost
certainly attacked by a shark.

Table 2.8. - The California Department of Fish and Game
maintains a data-base on the dead sea otters that wash ashore.
Each carcass is given a condition rating, aged and sexed if
possible. Numbers of male and female carcasses in good
condition are compared to those expected if the sex

ratio was equal.

SEX
AGE CLASS
(after Morejohn MALES FEMALES
et al. 1975)
Observed Expected Obs. Expected

Pups/immatures 52 53 54 53 ns
Subadults 23 21.5 20 21.5 ns
Adults 84 69.5 55 69.5 p<0.05*
Old adults 29 28 27 28 ns

*Chi-square = 6.1, 1 d.f. :

28

Juvenile females had low survival rates compared to adult
females. They also had slightly lower survival rates than
juvenile males, as did the juvenile females in Alaska (Chapter
9). However, in the CDF&G database on dead sea otters, again
considering only the carcasses in good condition, the number
of juvenile male and female carcasses was about equal,
suggesting similar mortality rates for the two sexes of
juveniles in California. Juvenile females in California
clearly foraged longer than other otters to obtain sufficient
food (Chapters 4 and 5) and one juvenile female in Alaska
starved to death (Chapter 9). Taken as a whole, our data
strongly suggest that the age/sex groups are differentially
affected by the various sources of sea otter mortality.

Our observations of tags on instrumented individuals
provide estimates of annual tag loss rates that exclude the
possibilities that mortality and movements of tagged
individuals out of the study area added to the perceived tag
loss. The annual tag survival rate of 0.74 represents a good
estimate for the Temple tag. The only other estimates of tag
loss rates are those of Ames, et al., 1983, who used three
methods of tag application and used resightings of individuals
over time to estimate loss rate. The use of resighting
information to estimate tag loss has the potential to include
loss due to death and movement of individuals out of the study
area in addition to tag loss, and thus one would expect it to
overestimate actual tag loss. Curiously, however, using our
annual tag survival of 0.74 (which excludes the other sources
of loss of mortality and movement of the area), we found the
loss rate to correspond almost exactly to the loss rate for
double-anchored tags in Ames, et al., 1983 (Fig. 2.2). When
we used our data and plotted the number of tags remaining on
the otters that were still being located on a regular basis
against time (subtracting from the original 80 tags all those
that were on otters with failed or expired transmitters and
on missing and dead otters plus those that were lost by
animals with functioning transmitters), the estimated tag loss
was of course higher and corresponded very closely to that for
the other two methods of tag application (single anchor,
unglued; and single anchor, glued) in Ames, et al., 1983 (Fig.
Davie

Ames, et al., found that the apparent loss rate of
double-anchored tags was less than the apparent loss rate of
single-anchored tags (whether glued an unglued). Our data
suggest two possible reasons for this difference. The first
possibility is that the retention of double-anchored tags was
very high and the observed loss rate was almost entirely due
to mortality and/or movements out of the study area. The
second is that the small number of animals on which this
method was used rarely, if ever, moved out of the study area
and did not suffer significant mortality during the study

29

FIGURE 2.2 - A comparison of tag-loss rates between this study
and the study of Ames, et al, 1983. (See text for explanation.)

oD) pete tte...

Gaal eta Sees ee ee

s Q oS ccoe,
i)

Eo.

@

ai:

H

a:

Fo

rs) od —— Single Anchor, No glue

Eo --— Single Anchor, Glued

A collage --- Double Anchor, Screwed

a oe

<= ®O A This Study, Log of Tags

eu Eo Remaining Out of 80 Tags.

Ore Seis ces ~ This Study, Using Estimated

— Annual Survival Rate (0.74).

0 100 200 . 300 400 500 600

Time (Days)

30

period. These conditions could occur, for example, if the
tagged individuals were mostly adult females.

Considerable data can be obtained by following tagged
animals and tag-resight data can potentially be used to
estimate annual survival rates for the various age/sex classes
of a species. However, mark-recapture techniques, using the
appropriate models, must be used for these survival rate
estimates (Seber, 1973). For sea otter tag-resight data, the
combined effects of unequal probability of sighting among
individuals, movement of animals out of the intensive study
area, and differential mortality patterns among age/sex groups
make the application of such methods extremely difficult.
When tag loss is added to these complications it becomes
rather unlikely that accurate estimates of annual survival
rates for sea otters can be derived from such data.
Furthermore, the comparison of our data with those of Ames,
et al., (1983) suggests that for sea otters, even under the
best conditions, it would be difficult to separate tag loss
from actual mortality and movement away from the study area.
In our study, known tag loss, added to verified mortality,
would have produced unrealistically low survival rate
estimates.

LITERATURE CITED

Ames, J. A., R. A. Hardy, and F. E. Wendell. 1983. Tagging
materials and methods for sea otters, Enhydra _ lutris.
Calif. Fish and Game 66: 196-209.

Barabash-Nikiforov, I. I. 1947. Kalan. (The Sea Otter, pp.
1-174) Soviet Ministron RSFSR. Glavhoe upravlenie po
zapovednikam. (In Russian.) (Translated from Russian
by Dr. A. Birron and Z. S. Cole. Published for the
National Science Foundation by the Israel Program for
Scientific Translations, Jerusalem, 1962).

Calkins, D. G. and P. C. Lent. 1975. Territoriality and
mating behavior in Prince William Sound sea otters. J.
Mammal. 56: 528-529.

Estes, J. A. and R. J. Jameson. 1983. Summary of available
population information on California sea otters.
Minerals Management Service, POCS Technical Paper No.
83-11, 29 pp.

Garshelis, D. L., A. M. Johnson and J. A. Garshelis. 1984.

Social organization of sea otters in Prince William
Sound, Alaska. Canadian J. of Zoology 62: 2648-2658.

31

Heisey, D. M. and T. K. Fuller. 1985. Evaluation of survival
and cause-specific mortality rates using telemetry data.
J. Wildl. Mgt. 49: 668-674.

Jameson, J. R. J. and A. M. Johnson. 1979. Evidence of
annual reproduction among sea otters. Abstract, Third
Biennial Conference on the Biology of Marine Mammals,
7-11 October 1979, Seattle, WA.

Kenyon, K. 1969. The sea otter in the eastern Pacific Ocean.
N. Amer. Fauna 68: 1-352.

Loughlin, T. R., J. A. Ames, and J. E. Vandevere. 1981.
Annual reproduction, dependency period and apparent
gestation period in two California U.S.A. sea otters
Enhydra lutris. Fish. Bull. 79: 347-349.

Payne, S. F. and R. J. Jameson. 1984. Early behavioral
development of the sea otter, Enhydra lutria. J. Mammal.
65: 527-531.

Seber, G. A. F. 1973. The Estimation of Animal Abundance and
related parameters. Hafner Press, N.Y.

Sinha, A. A., C. H. Conway, and K. W. Kenyon. 1966.
Reproduction in the female sea otter. J. Wildl.Manage.
30: 121-130.

Schneider, K. B. 1973. Reproduction in the female sea otter.
Federal Aid in Wildlife Restoration Project W-17-4.
Project Progress Report. Alaska Department of Fish and
Game. 36 pp.

Wendell, F. E., J. A. Ames, and R. A. Hardy. 1984. Pup
dependency period and length of reproductive cycle:
estimates from observations of tagged sea otters, Enhydra
lutris in California. Calif. Fish and Game 70: 89-100.

32

CHAPTER 3

MOVEMENT PATTERNS AND SPATIAL USE OF CALIFORNIA SEA OTTERS

K. RALLS, T. EAGLE, AND D. B. SINIFF

November 30, 1988

33

INTRODUCTION

Sea otters tend to be sexually segregated in "male
areas", occupied largely by males, and "female areas",
inhabited by adult females and their young (Kenyon, 1969;
Schneider, 1978). Breeding males maintain territories, either
seasonally (Garshelis and Garshelis, 1984) or all year
(Loughlin, 1980), within the "female areas". Male territories
are often smaller than the home ranges of adult females
(Loughlin, 1980; Garshelis and Garshelis, 1984), although
life-time home ranges of males may exceed those of females
(Kenyon, 1969; Garshelis and Garshelis, 1984).

In Alaska, home ranges consist of extensively used areas
connected by travel corridors (Garshelis and Garshelis, 1984;
Chapter 8). Long-distance seasonal movements between "male
areas" and "female areas" have been documented in both Alaska
and California: four adult males moved over 100 km in Alaska
(Garshelis and Garshelis, 1984) and three males moved over 80
km in California (Ribic, 1982). Movement patterns of juvenile
sea otters in Alaska have been studied by Monnett and
Rotterman (Chapter 8); few data are available on the movement
patterns of juveniles in California.

Sea otters are a coastal species, although they may be
found quite far from shore in shallow-water areas (Kenyon,
1969). Loughlin (1980) and Ribic (1982) reported that they
rarely venture beyond the outer limits of the kelp canopy in
California.

We report here on the movement patterns and spatial use
of 38 California sea otters representing all four major age
and sex classes: adult females, adult males, juvenile
females, and juvenile males. The otters were instrumented
with implanted radio transmitters. Because these transmitters
had a much longer life span than those used in previous
studies and we attempted to locate each individual every day,
our data provide a more detailed picture of sea otter movement
patterns and spatial use than previously available.

METHODS

Otters were assigned to sex and age classes based on
their weight at capture, estimated age based on the
examination of cementum layers in the vestigial premolar
extracted for this purpose, and, in the case of females,
reproductive performance. All juveniles were judged to be no
more than two years of age (Chapter 2).

We usually attempted to locate each instrumented otter
on a daily basis by listening for their radio signals from
points along the shore. Individual otters were located at

34

various times of day, depending upon their movements from the
previous day and our searching pattern. Sometimes, usually
when an animal had moved a considerable distance from its
previous location, an individual could not be located for
several days. We searched for missing individuals from the
air; this was generally successful, as the radio signal could
be heard from a greater distance from the air than from the
shore. Location data were also recorded once an hour during
24-hour watches.

Three methods were used to estimate the position of a
otter once its radio signal was detected: visual observation
of the otter, triangulation on the radio signal, and, when
neither of these was possible, the best judgement of the
researcher based on the direction and strength of the radio
Signal. The accuracy of the locations determined by
triangulation was estimated by triangulating on radio signals
from transmitters attached to buoys at known locations
(Chapter 7). The method by which each location was estimated
was coded in the data, enabling us to analyze only those
locations determined by a particular method when appropriate.
Unless otherwise noted, all location data were included in an
analysis. Triangulations were plotted on topographic maps of
the study area; locations were recorded in the form of x-y
coordinates based on the UTM grid.

We used several measures in our analyses. The average
distance between successive locations of each individual
otter, separated into those recorded between 18 and 36 hours
apart and those recorded more than 36 hours apart, was used
to compare the usual distance traveled on a short-term basis
among the age/sex classes. This distance was measured on the
UTM grid and is the straight line distance between the two
locations. The path taken by the otter was probably longer.

The area used by individual otters on a daily basis was
calculated from the hourly triangulated locations recorded
during 24-hour watches, using the minimum convex polygon
method commonly used for terrestrial mammals (Hayne, 1949).

To portray the movement patterns of individuals over the
entire monitoring period, we moved each daily location for an
individual to the nearest point along the 5-fathom contour.
We then calculated the deviation of each daily location from
the mean location for that otter along this contour and
plotted these daily deviations against time.

Only daily locations determined by triangulation were
used to estimate distance offshore. The coastline was
digitized in UTM coordinates and a BASIC program was written
to calculate the distance from each otter-location to the
nearest point on the shore.

35

Monthly movement patterns were examined using the average
general harmonic mean distance (Hp) (Neft, 1966). This
measure, which is calculated from the distance between all of
an otter's locations for a given month, is insensitive to a
few long distance movements, and thus, may better reflect
Burt's (1943) conception of the home range. At least seven
locations in a month were required before Hp was calculated.

The "distance between extreme locations" (DBEL) was used
to compare the length of coastline frequented by individuals
over the entire monitoring period. This measure was first
used by Garshelis and Garshelis (1984) in their studies of sea
otters in Prince William Sound, Alaska, and is the distance
between the two farthest-apart locations of an individual
otter over the period of time it was monitored. It can he
considered an approximation of an otter's range during the
monitoring period. It is particularly useful for comparisons
between California and Alaska data, as field conditions in
Alaska preclude collection of the daily locations of each
otter that enabled us to calculate other measures for the
California otters. This distance was measured on the UTM
grid.

Statistical comparisons among age and sex classes were
performed using analysis of variance, controlling for
variation among individuals within classes. We performed a
log (base 2) transform on the data to reduce heterogeneity of
variances. All statements that differences are statistically
significant are based on the 0.05 probability level.

RESULTS
Distance between successive locations

Otters of all age and sex classes were usually found
within a comparatively short distance of their location on the
previous day. Data on the distance between successive
locations were divided into two categories: those recorded
within 18-36 hours of the previous location of that individual
and those recorded after an interval of more than 36 hours
(Tables 3.1 and 3.2). Analysis of variance (Appendix 3.1)
indicated that significant variation occurred among
individuals within all the age and sex classes in both
categories except for the juvenile males in the greater than -
36 hour category. This suggested that individual movement
patterns from one day to the next were different for young
males but that their long term patterns were similar.
However, juvenile males were the most similar in both the 18-
36 and more than 36 hour categories. The variation among the
sex/age classes was much greater than that within classes.
For all these data, even though a log transformation was used

36

to improve the homogeneity of the within individuals
"variance", Bartletts test for homogeneity showed Signitvean’
differences in all cases.

TABLE 3.1 - Average distance (km) between successive
locations, recorded between 18 and 36 hours apart, for each
instrumented otter along the California coast. AF = Adult
Female, AM = Adult Male, JF = Juvenile Female, JM = Juvenile
Male.

OTTER AGE/ STANDARD
NUMBER SEX MEAN N DEVIATION
15 AF 0.355 342 0.428
28 AF 0.434 10 0.124
4 AM 0.489 115 1.004
7 AM 0.517 326 0.485
10 AM 0.711 282 1.661
1 AM 0.717 229 4.119
2 AM 0.724 89 1.939
33 AF 0.757 272 1.056
17 AM 0.773 138 1.147
46 JF 0.823 375 0.731
42 JF 0.825 345 1.806
11 AF 0.897 311 0.806
19 AF 0.900 194 1.493
16 AF 0.909 248 1.396
36 AF 0.918 390 0.855
9 AF 1.018 265 1.113
3 AM 1.044 183 3.152
6 AF 1.091 101 1.522
25 AF 1.105 284 1.629
38 JF 1.134 18 0.928
45 JM 1.244 164 1.836
37 JF 1.260 117 1.677
40 JF 1.284 285 3.206
31 AF 1.369 308 - 1.463
27 AF 1.453 180 1.692
47 JF 1.536 66 1.378
39 JF 1.680 244 3.187
44 JF 1.716 77 3.314
34 AM 2.051 47 5.946
29 JF 2.165 237 2.547
43 JM 2.170 186 2.159
41 JM 2.299 229 2.335
22 AF 2.354 187 2.944
26 AF 2.401 12 1.796
14 AF 2.409 381 3.282
13 JM 2.569 83 5.488
35 JM 2.744 241 4.821
30 JM 2.961 230 3.339

TABLE 3.2 —- Average distance (km) between successive
locations, recorded more than 36 hours apart, for each
instrumented otter along the California coast. AF = Adult
Female, AM = Adult Male, JF = Juvenile Female, JM = Juvenile
Male.

OTTER AGE/ STANDARD
NUMBER SEX MEAN N DEVIATION
15 AF 0.417 108 0.462
28 AF 0.912 6 1.155
33 : AF 0.998 149 1.364
46 JF 1.077 119 1.115
44 JF 1.120 36 0.971
36 AF 1.177 105 0.977
42 JF 1.302 107 2.301
11 AF 1.388 104 1.301
31 AF 15393), Tau 1.411
7 AM 1.405 101 5.453
38 JF 1.407 10 1.323
9 AF 1.496 97 2.080
25 AF 1.579 96 2.465
1 AM 1.587 70 7.452
37 JF 1.608 70 2.389
19 AF 1.646 136 2.398
2 AM 1.652 31 3.181
10 AM tba ve) alas) 8.601
47 JF 1.936 178 1.928
16 AF 1.973 123 3.325
6 AF 2.017 66 1.903
26 AF 2.126 7 1.191
27 AF 2.336 92 2.428
4 AM 2.343 115 9.117
3 AM 2-442 94 7.626
45 JM 2.514 102 4.740
29 JF 3.346 110 3.980
41 JM 3.499 82 3.726
14 AF 3.546 133 4.533
17 AM 4.524 52 10.050
40 JF 4.562 115 14.330
22 AF 5.137 124 6.954
35 JM 5.295 84 9.836
43 JM 5.424 129 9.360
13 JM 6.220 136 12.030
39 JF 6.552 76 11.470
30 JM 8.075 118 15.740
34 AM 28.070 55 46.420

To illustrate the relatively short-term movement patterns
of the four age/sex classes, we plotted the average movement
between successive locations for all eight data sets, e.g.

38

FIGURE 3.1 - A comparison of the average distance between
successive locations for the age/sex categories of adult females,
adult males, juvenile females, and juvenile males, calculated for
locations made 18-36 hours apart and those made after more than

36 hours.

6.0

5.0

4.0

3.0

Distance (km)

2.0

1.0

(643)

(1409)

Time between locations

18-36 hours apart

[_] >36 hours apart

(620)
(1036)

(1413)
(3143) .

| |
ADULT JUVENILE
FEMALES FEMALES

39

(399)

(801)

|
JUVENILE
MALES

adult males, adult females, juvenile females and juvenile
males for both the 18-36 hour data set and the >36 hour data
set, plus or minus two standard errors of the mean (Fig. 3.1).
Within the 18-36 hour data set, the distance moved between
successive locations was relatively consistent, averaging
around 1 km, the exception being the juvenile males, who
averaged about 2.3 km, significantly greater than the other
age/sex classes. Adult males showed much more variation in
the >36 hour data set than in the 18-36 hour data set. This
variation reflects the occasional long-distance movements of
adult males, which often resulted in the animals not being
located for several days. Juvenile males and juvenile females
also made these long-distance movements but there was more
consistency within these classes with respect to the distance
traveled on such trips. —

In general, otters tended to stay within a small area for
an extended period and then suddenly move for a considerable
distance. Tables 3.1 and 3.2 do not illustrate the distance
which an individual otter can travel from one day to the next
during these periods of rapid movement. To give a sense of
this distance, we plotted the 20 longest movements (again
using the UTM grid) made within 18 to 36 hours for otters
belonging to each age/sex class (Fig. 3.2). The longest daily
movements were 47.5 km for an adult male, 40.1 km for a
juvenile female, 38.8 km for a juvenile male, and 17.5 km for
an adult female. However, movements of more than 10 km a day
were infrequent.

Daily convex polygon areas

Areas used by individual otters over a 24-hour period
ranged from 10 to over 1000 ha (Table 3.3). Adult males were
not included in the analysis of variance (Appendix 3.2)
because of insufficient sample size. For the other three
age/sex classes, there was no significant variation within
age/sex classes but variation among classes was significant.
Juvenile males tended to travel over a larger area than
individuals of the other age/sex classes.

Long-term movement patterns along the coast

The average location along the California coast for each
instrumented otter is shown in Fig. 3.3. Figures 3.4, 3.5,
3.6, and 3.7 illustrate both the substantial degree of
variation between individuals within age/sex classes in the
extent to which they travel away from this average location
and the way in which many otters tend to remain within small
areas for extended periods of time and then suddenly move a
much greater distance.

40

FIGURE 3.

2 - A plot of the 20 longest trips made between

successive locations that were 18-36 hours apart for the four
age/sex categories of adult females, adult males, juvenile

females,

DISTANCE (KM)

and juvenile males.

50

>
ro)

® ADULT FEMALES

+ JUVENILE FEMALES
° ADULT MALES

4 JUVENILE MALES.

ie
(=)

20

1 3 5 U 9 11 13 15 17 19

TRIPS IN ORDER FROM LONGEST TO SHORTEST

41

FIGURE 3.3 - The average locations of the instrumented sea otters
along the California coast.

Santa Cruz

Monterey

42

FIGURE 3.4 - The general north-south movement pattern of
individual adult males, in relation to their average north-south
location, over the monitoring period.

80

DISTANCE (KM)
:

80
OTTER 10
160
SOs
OTTER 17
OTTER 34
I a TET a a ae Card )
1 JAN 85 "1 JAN 86 1 JAN 87 —- 1: JAN 88

43

FIGURE 3.5 - The general north-south movement pattern of
individual adult females, in relation to their average north-
south location, over the monitoring period.

OTTER 6 , : .

or;

1 JAN 84 1 JAN 85 1 JAN 86
OMMERES
OMERR I

160 Oineieat

OIA 1S)

OMEN We ay Ane

OnMER ets

Co
O

(@)

OME 22

DISTANCE (KM)

Co
O

OTTER 25 H -
sage COMMER 27
SOUTH i"
OTTER 31
$$ ry aarti igen pe tte $$

OMA IS)
ANA rp

OTTER 36
rrr prterni atrcgereeninarncmcrnerngan Q e————

Sr) (reel) beara rere) eo ray Poa eae air an «UL a a
1 JAN 85 1 JAN 86 1 JAN 87 1 JAN 88

44

FIGURE 3.6 - The general north-south movement pattern of

individual juvenile females, in relation to their average north-

south location, over the monitoring period.

NORTH
160

0.)
(e)

DISTANCE (KM)
0a

160
SOUTH

OMER 23

OMER: 57
ee Cee 10 ee Ss

OMNES OS)

OTTER 40

OTTER 42

spa a eee

OTTER 44

ae OE a Te lame

OTTER 45

OTTER 46
pe ee eee eee ee eee

OTTER 47
I tempt OO err rm

Sa Maa aE Ta AGE TCE in TG ante ieaMlLa Gla nenae hae

1 JAN 85 1 JAN 86 1 JAN 87 1 JAN 88

45

FIGURE 3.7 - The general north-south movement pattern of
individual juvenile males, in relation to their average north-
south location, over the monitoring period.

OTTER 13
NORTH
160
OTTER 30
= ao
x
4
Lu
OQ
Z
a OTTER 35
O 80
OTTER 41
160
SOURG Wi tf
OTTER 43

SSS ST ee ee
1 JAN 85 1 JAN 86 1 JAN 87 1 JAN 88

46

Four general movement patterns were apparent from these
figures: 1) remaining within a small area throughout the
study (for example, otters 11 and 46); 2) generally remaining
within a small area but making occasional long-distance trips
(otters 1 and 10); 3) shifting of centers of activity for
extended periods of time (otters 17 and 30); and 4) frequent
travel over long distances (otters 22 and 34).

Adult males captured in both "male areas" (otters 1-4) and
"female areas" (otters 7, 10, 17, 23, and 34) sometimes made
long-distance movements. These were often relatively brief
"trips" to a new location, followed by a return to the

TABLE 3.3 - Area (ha) of daily home ranges based on data
obtained during 24-hour watches in which locations were
recorded once per hour. The areas were determined using the
minimum area home range method (Hayne, 1949). AF = Adult
Female, AM = Adult Male, JF = Juvenile Female, JM = Juvenile
Male.

OTTER AGE/SEX AREA
NUMBER (HA)
16 AF 10.33
36 AF 13.08
16 AF 22.52
36 AF 34.10
27 AF 58.23
19 AF 68.97
27 AF 227.59
29 AF 1166.35
34 AM 6.88
34 AM” 12.75
7 AM 223.37
42 JF 31.69
42 JF 32.88
40 JF 107.03
40 JF 118.81
45 JF 119.29
47 JF 165.36
39 JF 212.53
47 JF 213.75
41 JM 221.24
35 JM 258.52
43 JM 302.14
35 JM 359.62
41 JM 379.49
30 JM 511.40
35 JM 570.23
30 JM 625.82
35 JM 666.38

41 JM 759.32

47

original location. No seasonal pattern was apparent in these
"trips". Adult females tended to be more sedentary. However,
two of them (otters 14 and 22) often moved distances on the
order of 10 km within a short time.

Juvenile females tended to move more extensively than
adult females and two of them (otters 39 and 40) made long-
distance trips. Juvenile males tended to travel more than the
other age/sex classes.

Distance offshore

Adult males and females were usually found relatively
close to shore (Fig. 3.8). Adult females with pups were
particularly close to. shore. There waS no apparent
correlation between age of pup and distance offshore. The
three females with pups in Figure 3.9 are (from left to right)
otters 16, 27 and 36. The potential ages of their pups were
in the ranges of 81 to 101, 6 to 17 and 0 to 37 days,
respectively. Juveniles ranged farther from shore, and the
tendency for the juvenile males to be much farther off-shore
than individuals of the other age/sex classes was particularly
striking. About four percent of our locations of juvenile
males were over three km from shore.

In general, otters tended to feed slightly closer to shore
than they rested but this was not true of all individuals
(Fig. 3.9). Distance offshore was not closely related to time
of day, although juvenile males were often relatively close
to shore about 6 to 7 a.m. (Fig. 3.10).

Monthly harmonic mean ranges

The average monthly deviations from the harmonic mean
center of the locations of the instrumented otters presented
in Table 3.4 are plotted in Figure 3.11. These data reflect
the general monthly movement patterns of the four age/sex
classes. Monthly harmonic mean home range sizes of juveniles
(Fig. 3.11) appeared to increase during the peak pupping
months of February, March, and April, and during the secondary
peak in August, September, and October (see Fig. 2.1).
However, when monthly Hp were grouped into these six months
of peak pupping vs. non-peak pupping seasons, no significant
seasonal differences were detected by analysis of variance
(Appendix 3.3). This lack of statistical significance,
particularly among juvenile males, is probably because we
monitored only a few juvenile males, there was great variation
within individuals, and they had very small home range sizes
aligl JAKES (Haale Bo alal)) o

48

FIGURE 3.8 - The distribution of distances offshore for the four
age/sex categories of adult females, adult males, juvenile
females, and juvenile males.

Adult Females Adult Males

Juvenile Females Juvenile Males

Percent of Locations

300 1300 2300 300 1300 2300

Distance (Meters)

49

FIGURE 3.9 - The average distance off-shore while resting and
feeding for individual otters partitioned by five age/sex
categories of adult females, adult females with pups, adult
males, juvenile females, and juvenile males.

Resting Juvenile Males
MM Feeding

ine)

q

Distance Offshore (km)

lh Juvenile Females aan
_| | = Females
dt a oly Adult

| j Females

=| ae eee | with Pups

50

JUVENILE MALES

JUVENILE FEMALES

(AM) GHOHSYAO AONVLSIG

FIGURE 3.10 - The average distance off-shore for the various
hours of the day for juvenile males and juvenile females.

16 20 24

12

TIME OF DAY

51

FIGURE 3.11 - The average distance deviation from the harmonic
mean center of monthly home ranges for the four age/sex

categories of adult females, adult males, juvenile females, and
juvenile males.

ADULT FEMALES
JUVENILE FEMALES

ADULT MALES
JUVENILE MALES,

~)
O

—s
Oo

ty}

DEVIATION (KM)

Bye:

TABLE 3.4 - Average distance deviation (km) from the harmonic
mean center of all locations for the four sex and age
categories of adult males, adult females, juvenile females and
juvenile males for each month of the year. AF = Adult Female,
AM = Adult Male, JF = Juvenile Female, JM = Juvenile Male.

MONTH AM AF JF JM
MEAN N MEAN N MEAN N MEAN N

JAN 0.207 7 0.623 13 0.810 10 1.418 5
FEB 0.079 7 0.472 13 1.170 10 2.288 6
MAR 0.059 10 0.478 17 0.984 10 1.433 5
APR 0.080 12 0.371 20 0.791 9 1.527 6
MAY 0.081 12 0.454 20 0.653 10 1.677 5
JUN 0.145 12 0.354 20 0.396 10 0.993 5
JUL 0.138 12 0.459 21 0.735 9 1.143 5
AUG 0.131 10 0.404 21 0.532 9 0.702 5
SEP 0.099 9 0.328 20 0.368 9 1.638 6
OocT 0.112 9 0.315 23 0.436 9 1.559 6
NOV 0.336 8 0.27 21 0.389 9 1.038 7
DEC 0.031 6 0.188 18 0.367 9 0.816 6

There was significant variation in monthly -Hp among
individuals of both adult sex classes but not for juveniles,
whether seasonal effects were considered (Appendix 3.3) or not
(Appendix 3.4). When seasonal effects were disregarded,
variation of Hp among age/sex classes was much greater than
variation among individuals within age/sex classes and was
Significant (Appendix 3.4). Adult males tended to move within
a small area; adult females utilized areas slightly larger
than those of adult males. Juveniles of both sexes traveled
over larger areas than adults throughout the year, and
juvenile males used strikingly larger areas than individuals
of the other age/sex classes.

Distance between extreme locations

The average and two extreme locations, and the distance
between these two extreme locations, for each otter are shown
in Table 3.5 and plotted in Figure 3.12. Unlike Hp, the
average distance between extreme locations is extremely
sensitive to a few long-distance movements. Analysis of
variance indicated that differences among age/sex classes were
Significant (Appendix 3.5), with juvenile males having the
largest distances and adult females the smallest (Table 3.6).
Distances for adult males, which occasionally take long-
distance trips, were greater than those for adult and juvenile
females.

These data are consistent with our other analyses of the
movement data, indicating that juvenile males are the most
extensive travelers.

53

FIGURE 3.12 - The distance between extreme locations for all of
the instrumented sea otters in California, partitioned by the
four age/sex categories of adult females, adult males, juvenile
females, and juvenile males.

280

aa) Juvenile Males

200 Adult Males

160

Juvenile Females
120

Distance (km)

Adult Females

80 F

40

15 36 1133 31 19 27 6 2516 9 1422 23 17 4171034 46 374742 29444539 40 41 35 43 13 30

Otter Number

54

TABLE 3.5 - The average location along the five fathom line, and the
northern-most and southern-most location along this line, that were
recorded during the period of monitoring for each instrumented sea

otter. AF = Adult Female, AM = Adult Male, JF = Juvenile Female, JM =
Juvenile Male.
OTTER AGE/ MEAN NORTHERNMOST SOUTHERNMOST Distance
NO. SEX LOCATION LOCATION (A) LOCATION (B) Between
A&B(km
15 AF COAST GALLERY GRIMES POINT PARTINGTON POINT 5.0
36 AF SALMON CREEK REDWOOD GULCH RAGGED POINT 10.0
11 AF FALSE SUR VENTURA ROCKS BIG SUR RIVER al GS)
33 AF PARTINGTON PT GRIMES POINT ESALEN 15.5
31 AF BUCK CREEK TORRE CANYON DOLAN ROCK 15.5
19 AF WRECK BEACH FALSE SUR GRIMES POINT 18.5
27 AF DOLAN ROCK ANDERSON CANYON LUCIA 21.5
6 AF ARROYO LAGUNA CHINA GULCH LITTLE PICO CRK 21.5
25 AF ANDERSON CR GRIMES POINT SQUARE BLACK ROCK 22.5
16 AF PFEIFFER BEACH LITTLE SUR RVR GRIMES POINT 23.5
9 AF BUCK CREEK PFEIFFER POINT DOLAN ROCK 26.5
14 AF BUCK CREEK GRIMES POINT LUCIA 30.0
22 AF ROCKY POINT POINT PINOS ANDERSON CANYON 86.0
2 AM MORRO ROCK CAYUCOS POINT MORRO BAY 12.5
3 AM MORRO ROCK ARROYO LAGUNA MORRO BAY 54.5
1 AM MORRO ROCK ARROYO LAGUNA MORRO BAY 56.0
7 AM POINT SUR POINT PINOS COOPER POINT 60.5
4 AM MORRO ROCK RAGGED PT INN MONTANA DE ORO 79.5
17 AM PFEIFFER BEACH MOSS LANDING COAST GALLERY 112.0
10 AM PFEIFFER BEACH SOQUEL POINT TORRE CANYON 140.5
34 AM SOBERANES PT GREYHOUND ROCK BUCK CREEK 181.0
46 JF ARROYO LAGUNA BECKETS REEF SAN SIMEON POINT 13.0
37 JF ARROYO LAGUNA BECKETS REEF SAN SIMEON POINT 15.0
47 JF POINT JOE MUSSEL POINT SUNSET POINT 18.0
42 JF RAGGED PT INN COUNTY LINE ARROYO LAGUNA 22.0
29 JF DOLAN ROCK COAST GALLERY LUCIA 30.5
44 JF BECKETS REEF GORDA ARROYO LAGUNA 34.5
45 JF PLASKETT ROCK DOLAN CREEK BECKETS REEF 62.5
39 JF SODA SPRING CR POINT SUR PIEDRAS BLANCAS 103.5
40 JF CRUZ ROCK VENTURA ROCKS SAN SIMEON PT 120.0
41 JM CRUZ ROCK REDWOOD GULCH SAN SIMEON CREEK 44.0
35 JM COUNTY LINE PACIFIC VALLEY BECKETS REEF 64.0
43 JM BECKETS REEF SALMON CREEK SAN SIMEON PT 82.0
13 JM CARMEL BEACH SUNSET BEACH RAGGED PT INN 191.5
30 JM JADE COVE SOBERANES POINT PIEDRAS BLANCAS 258.0

55

TABLE 3.6 - The average distance (km) between extreme
locations, measured along the five fathom line, for the four
sex and age categories of adult males, adult females, juvenile
females and juvenile males. AF = Adult Female, AM = Adult
Male, JF = Juvenile Female, JM = Juvenile Male.

AGE/SEX MEAN N VARIANCE RANGE
(KM) (KM)
AF 23.65 13 368.28 5- 86
AM 97.71 7 2042.204 54-181
JF 46.56 9 1426.024 13-120
JM 127.90 5 6840.04 44-228

Because these data are based on the most extreme distances
moved, they also show the tendency for adult males to make
fairly long-distance trips. The one exception in Figure 3.12
is male number 2. However, he was monitored for only 100 days
before he disappeared and it seems probable that he simply
happened not to make a long trip within this relatively short
monitoring period.

DISCUSSION

Radiotelemetry studies have shown that sea otter home
ranges in Alaska consist of several heavily used areas
connected by travel corridors (Garshelis and Garshelis, 1984;
Chapter 8). The results of our studies in California agree
with this general picture: otters tended to stay within a
small area for an extended period and then suddenly move for
a much greater distance.

Distance between successive daily locations

Otters of all age and sex classes were usually found
within one or two km of their location on the previous day.
There are no comparable data on the locations of individuals
from one day to the next in the literature but Ribic (1982)
measured the distance between successive locations of the same
individual at 3.4-5 hour intervals. Some of Ribic's sample
sizes were extremely small. However, if we compare our data
to those data where she had samples of at least 10 locations,
the distances reported in the two studies are similar.

This similarity suggests that sea otters have a tendency
to move fairly quickly and directly between locations where
a considerable amount of time is devoted to more sedentary
activities such as resting or feeding. Such direct movements
have been reported (Loughlin, 1980) and we often observed them
during course of our study.

56

The longest distance between successive daily locations
recorded in our study was about 47 km. This is similar toa
movement of 48 km in 22 hours mentioned by Ribic (1982).
Kenyon (1969) estimated that sea otters can swim at sustained
speeds of about 2.5 km per hr., thus the long daily movements
we recorded probably involved nearly constant swimming.

Long-distance movements

Other investigators have documented that sea otters make
occasional long-distance movements (Kenyon, 1969; Ribic, 1982;
Garshelis and Garshelis, 1984), but these were thought to
largely reflect seasonal movements of males between summer and
winter ranges. Our daily monitoring revealed that individual
otters of all age/sex classes make a surprising number of
long-distance movements at all times of year.

Although some individuals remained within a small area for
extended periods, e.g. otter 7, an adult male that remained
within a very small area for 18 months (Fig. 3.4), and otter
15, an adult female that visited only 5 km of coastline during
the entire monitoring period (Fig. 3.5; Otter 15), it became
evident to us that the longer an individual was monitored, the
more likely it was to travel a significant distance. Only a
few "trips" have been previously documented in the literature,
probably because such long distance movements are less likely
to be detected in tag-resight studies and radio telemetry
studies of short duration. Although the reasons for these
“trips" are unknown, it seems likely that they vary. Adult
males may be seeking mating opportunities or areas with high-
quality food resources. Juvenile males may be displaced by
older males and may seek the company of other young males as
well as high-quality food resources. Females, both juvenile
and adult, that take extended "trips" may be looking for areas
where they can become resident.

Seasonal patterns

We were unable to detect a significant seasonal pattern
in the frequency of long-distance movements for any age/sex
class, including adult males. We also looked for seasonal
trends in the size of monthly harmonic mean home ranges. This
method eliminates the possibility of greatly overestimating
the area utilized due to a few long-distance movements. Once
again, we failed to find any seasonal pattern in the size of
the area used by adult males. However, we monitored only a
small number of adult males. It also seems possible that
predictable seasonal movements of adult males occur in some
but not all areas. Juveniles appeared to show a seasonal
pattern in harmonic mean home range size, with the peaks
during the peaks of parturition period and pup dependency.

57

Although this effect was not statistically significant in our
data set, it may nevertheless be a real phenomenon.

Variation within and among age/sex classes

We found substantial variation in movement patterns among
individuals within all age/sex classes. For example, most
adult females tended to be relatively sedentary but two of the
15 we monitored often traveled for considerable distances.
In spite of this extensive individual variation, some
generalizations on the movement patterns characteristic of the
different age/sex classes are possible.

Measures that minimize the effects of long-distance
movements, such as the distance between successive daily
locations and monthly harmonic mean home ranges, indicate
that, over the short-term, adult males tend to utilize smaller
areas than adult females. This agrees with the findings of
Loughlin (1980) and Ribic (1982) that adult females have
larger home ranges than territorial or "resident" males.
However, measures that are sensitive to long-distance
movements, such as the distance between locations recorded at
intervals of more than one day and the DBEL, show that adult
males are more likely than adult females to make long-distance
movements and thus tend to visit greater lengths of coastline
over the long-term.

On a daily basis, movements of juvenile females were
similar to those of adult females. However, juvenile females
were more likely than adult females to make long-distance
movements. Juvenile males tended to move for greater distances
than the otters belonging to other age/sex classes on both a
short= and long-term basis.

Distance offshore

It has been reported that otters in California rarely
travel far offshore (Loughlin, 1980; Ribic, 1982). Estes and
Jameson (1988) found that 90% of the otters observed from
shore were within 600 m of the coast and that the probability
of a shore-based observer sighting an otter was constant over
observation distances of 50-850 m from shore, although it
declined to zero by 1300 m. The majority of our locations of
adult males and females and juvenile females were within 800
m of shore; however, the majority of our locations of juvenile
males were over 800 meters from shore and over half of them
were more than 1300 m (Figure 3.8). Although Estes and
Jameson (1988) concluded that “few otters occurred at
distances from shore beyond the observers' viewing ranges",
our juvenile males were frequently located at distances beyond
the viewing range of a shore-based observer.

58

Furthermore, our data underestimated the extent to which
these juvenile males traveled offshore. The signal from an
otter must be detectable from at least two shore locations to
enable estimation of its distance offshore through
triangulation. The signals of the juvenile males were often
only faintly detectable from a single shore location; their
distance offshore could not be estimated on these occasions.
In addition, we were unable to locate the juvenile males a
much larger proportion of the time than the other otters,
probably because they often moved so far offshore that we
could not receive their radio signals.

Home range size

Estimates of the area utilized by an individual otter are
greatly influenced by whether or not long-distance movements
by that individual were detected during the study period,
whether or not they were included in the analysis if detected,
and the specific method used to estimate the area utilized.
The minimum convex polygon method is appropriate for measuring
the areas used by otters in the intervals between long--
distance movements; we used it to estimate the area used
within a 24-hour period.

TABLE 3.7 - Comparison of home range areas (ha) for sea otters
in California and Alaska calculated in this and previous
studies.

AGE/SEX CLASS _ LOCATION AREA (HA) SOURCE

Nonterritorial males AK 400-1440 Garshelis&Garshelis 1984
Territorial males CA 18-58 Loughlin 1980

Resident males CA 80-460 Ribic 1982
Nonterritorial males CA 29-138 Loughlin 1980

All males CA 80-2980 Ribic 1982

Adult males CA 7-223 This study

Juvenile males CA 221-759 This study

All females AK 20-960 Garshelis&Garshelis 1984
All females CA 28-198 Loughlin 1980

Resident females CA 470-680 Ribic 1982

All females CA 470-2110 Ribic 1982

Adult females CA 10-1166 This study

Juvenile females CA 32-214 This study

ape

Our estimates of the area used on a daily basis overlap
estimates of home range size over longer periods made by other
investigators (Table 3.7). This suggests that, in the
intervals between long-distance trips, otters tend to travel
on a daily basis over much of the area they are currently
using. Because of the variety of methods used to calculate

59

the area utilized and the different periods of time included
in the analyses, these estimates show great variance of home
range size.

Early investigators suggested that there should be a
relationship between home range size and metabolic rate
(McNab, 1963). Because sea otters are known to have a high
metabolic rate and are unable to fast for long, there may be
a relationship between area used and prey availability in sea
otters. However, such a relationship would be impossible to
detect unless there is some standardization in methods of data
collection and analysis. Greater consistency could be
achieved by considering a rather short period, such as 24
hours, and excluding those periods in which individuals spent
much of their time traveling between locations. Such
comparisons might not be appropriate between areas in which
the near-shore communities differed greatly.

The distance between extreme locations has been used as
an index of home range size, particularly in Alaska (Garshelis
and Garshelis, 1984; Chapter 8). In general, our values for
adult males in California are greater than those reported for
Alaska. There have been no reports of DBEL's over 100 km for
instrumented otters in Alaska; three of our adult males in
California had DBEL's of 112, 141, and 181 km. (Table 3.5).
However, most of the Alaska data are from Garshelis and
Garshelis (1984), who monitored instrumented otters for
relatively short periods during the non-winter months, and it
seems clear that this index, like measures of home range area,
tends to increase with the length of time individuals are
monitored. Although Garshelis and Garshelis (1984) did not
document long-distance movements by any of their instrumented
adult males, they observed that four tagged males moved about
100 km between their summer territories and a "male area"
occupied during other times of the year. This suggests that
the greatest distances visited by males in California and
Alaska may be more similar than current telemetry data
indicate.

DBEL's for adult females seem to be comparable in the two
areas. Ours ranged from 5 to 86 km and those reported by
Monnett and Rotterman for Alaskan females in Chapter 8 range
from 28 to 80 km. Again, those reported by Garshelis and
Garshelis (1984) were smaller, the largest being about 20 km.

Our data for juveniles are also similar to those of
Monnett and Rotterman from Alaska (Chapter 8), in that males
tended to move greater distances than females. Our DBEL's for
juveniles of both sexes are much greater than for Alaska but
the data are not entirely comparable. The Alaska data
represent the movements of individuals instrumented as
dependent pups and monitored through the early period of

60

independence, for a maximum of 21 months after weaning. Our
data represent the movements of individuals captured after
weaning. Some of these individuals appeared to be quite young
but others were estimated, based on the cementum lines in
their premolars, to be up to two years of age. In Alaska, the
longest distance moved by a juvenile female was less than 50
km and that by a juvenile male was approximately 120 km
(Chapter 8). Our longest distances were 120 km for a juvenile
female and 258 km for a juvenile male. Although our juveniles
moved farther, they were older.

In Alaska, juvenile males tended to move relatively long
distances, which took them out of the areas occupied by
reproductive adults, within a few weeks after weaning. It is
not known if young males make similar movements in California
soon after weaning, but the juvenile males we monitored
remained in the area occupied by breeding adults for most of
the study period. Many of them did associate with a male
group that formed near Ragged Point, well within the "female
area" for several months. Two of them finally moved into the
"male area" in the southern part of the California range
towards the end of the monitoring period.

Sex differences in dispersal patterns

Although the timing may differ, sex differences in
dispersal patterns appear similar in California and Alaska.
Sea otters exhibit the dispersal pattern typically found in
polygynous mammals (Greenwood, 1980): juvenile males tend to
move farther than females. The juvenile males ultimately join
male groups, usually within "male areas", while juvenile
females remain within their natal "female area".

Aggression from territorial males may play a role in the
initial departure of the juvenile males from the kelp beds
frequented by breeding adults. However, it is likely that,
on the average, juvenile males ultimately benefit from this
long distance dispersal in terms of increased reproductive
success.

The extensive travels of juvenile males probably enable
them to become familiar with a large area; this may be an
advantage later in life, when, as young adults, they return
to a "female" area and search for an available territory
(Loughlin, 1980, Ribic, 1982). By participating in the
frequent social interactions, such as various forms of play-
fighting, that occur in male groups, juvenile males may gauge
their strength relative to other males and develop the
fighting skills needed to acquire a territory. Because male
groups are often located in areas that have been occupied by
sea otters for a shorter period than the "female" areas,
juvenile males that join these groups may tend to derive

61

nutritional benefits. Increased prey availability during the
male's growth period could have important benefits later in
life, as the sexual dimorphism characteristic of sea otters,
with males considerably larger than females, suggests that
large body size is likely to be an advantage to males when
fighting with other males.

In general, "female areas" have been occupied by otters
for many years, and prey availability there is reduced
compared to the "male areas". By remaining within these
areas, juvenile females are forced to compete with larger,
older otters for good foraging locations in an area that has
already been exploited for some time. Our California data on
time budgets and activity patterns (Chapter 4) and feeding
patterns (Chapter 5) indicate that juvenile females tend to
be poor competitors and their survival rates are rather low
compared to other age/sex classes in both California (Chapter
2) and Alaska (Chapter 8 and 9).

Why then, do juvenile females remain in these "female
areas"? Theory suggests that males will tend to behave so as
to maximize mating opportunities and females so as to maximize
the food resources available to them and their offspring.
Juvenile females may ultimately benefit from acquiring a
detailed familiarity with the distribution and availability
of prey within a small area. In many mammals in which young
females tend to remain within their natal area, female young
may benefit by acquiring all or part of their mother's home
range-~ The extent to which this occurs in sea otters is
unknown. If adult females are intolerant of strange females,
juvenile females that disperse from the "female" areas and
subsequently return may have reduced chances of acquiring a
good quality home range within these areas.

Regardless of the reasons for its occurrence, the sex
difference in movement patterns of juvenile sea otters may
have important consequences for sea otter population dynamics
by decreasing survival of juvenile females (Chapter 2).

LITERATURE CITED

Burt, W. E. 1943. Territoriality and home range concepts as
applied to mammals. J. Mammal. 24:346-352.

Estes, J.A. and R. J. Jameson. 1988. A double-survey
estimate for sighting probability of sea otters in
California. J. Wildl. Manage. 52: 70-76.

Garshelis, D: L. and J. A. Garshelis. 1984. Movements and
management of sea otters in Alaska. J. Wildl. Manage. 48:
665-678.

62

Greenwood, P. J. 1980. Mating systems, philopatry, and
dispersal in birds and mammals. Anim. Behav. 28: 1140--
1162.

Hayne. 1949. Calculation of size of home range. J. Mammalogy
30: 1-18.

Kenyon, K. W. 1969. The sea otter in the eastern Pacific
Ocean. North American Fauna. No. 68. Bureau of Sport
Fisheries and Wildlife. U.S. Government Printing Office,
Washington, D. C. 352 pp.

Loughlin, T. R. 1980. Home range and territoriality of sea
otters near Monterey, California. J. Wildl. Manage. 44:
576-582.

McNab, B. 1963. Bioenergetics and the determination of home
range size. Am. Nat. 97: 133-140.

Nees, Do | Siro 1966. Statistical analysis for area
distributions. Regional Science Research Institute
Monograph Series 2. Philadelphia, PA. 172 pp.

Ribic, Coy Acar 1982). Autumn movement and home range of sea
otters in California. J. Wildl. Manage. 46: 795- 801.

Schneider, K. B. 1978. Sex and age segregation of sea otters.
= Fed. Aid in Wildlife Restoration Project W-17-4 and W-17-
5. Final Report. Alaska Department of Fish and Game.

45 pp.

63

CHAPTER 4

TIME BUDGETS AND ACTIVITY PATTERNS OF CALIFORNIA SEA OTTERS

K. RALLS AND D. B. SINIFF

NOVEMBER 30, 1988

64

INTRODUCTION

The California sea otter population was reduced to a small
number of animals by fur hunters in the 18th and 19th
centuries. The remnant population grew at approximately five
percent per year until sometime in the mid 1970's, when growth
apparently ceased (Ralls, et al., 1983; Estes and Jameson,
1983; Estes, et al., 1986). A central question relating to
the dynamics of the California sea otter population, and hence
the development of a model for this population, is whether or
not the recent lack of growth is due primarily to density
independent factors, such as entanglement in gill nets
(Wendell, et al., 1986) or attacks by sharks (Ames and
Morejohn, 1980), primarily to density dependent factors, such
as competition for food or other resources (Miller, 1980), or
to both. However, the most recent surveys suggest that the
population may have resumed growth (Jameson and Estes, 1988).

Several authors have proposed that time budgets might be
useful as indicators of population status, assuming that food
is an important limiting resource (Eberhardt, 1977; Estes, et
al., 1982; Estes, et al., 1986). The prey available to sea
otters varies with location and the length of time the area
has been occupied by sea otters. Typically, most of the
population initially consumes large prey items with high
caloric content; as the availability of such prey decreases,
the diet of the population diversifies to include smaller
items and less preferred species (Estes, et al., 1981;
Garshelis, et al., 1986). An otter should have to spend more
time foraging to obtain a constant amount of energy in
habitats with reduced abundance, size, or quality of prey than
in those where high-quality food is abundant. Estes, et al.,
(1982) contrasted foraging patterns between two islands of the
Aleutians; one where otters have existed for many years
(Amchitka) and another that was recently colonized (Attu).
The difference in time spent foraging (as determined by visual
observations during the day) was dramatic as otters at
Amchitka foraged during 55% of the daylight hours while those
at Attu foraged only 17% of the daylight hours. Garshelis,
et al., (1986) working in Prince William Sound, Alaska,
assumed food was more abundant in areas that were recently
invaded by sea otters and, under this assumption, confirmed
- the predicted relationship between food availability and the
proportion of time devoted to foraging: otters spent about 10
percent more time foraging at Green Island, where sea otters
had been present for many years, than in Nelson Bay, an area
only recently reoccupied by sea otters where high quality prey
items were abundant.

Time budgets in California have been determined primarily
by visual observations of unidentified otters. The most
recent study (Estes, et al., 1986) concluded that food was

65

probably not limiting further growth of the California
population because sea otters there apparently foraged less
than at Amchitka, there is unoccupied habitat at both ends of
the range, and other mortality sources (e.g. gill nets) have
been impacting the population. The use of radio-telemetry to
collect time budget data has a number of advantages
(Garshelis, et al., 1986) and is helpful in evaluating the
many sources of mortality. Data can be collected over the
entire 24-hour period and the time spent foraging can be
estimated for individuals and age/sex classes, and direct data
on sources of mortality are sometimes available.

One potential limitation of telemetry data is that the
number of individuals monitored is often small. Thus, when
the assumption is made that the instrumented individuals are
representative of the population, the potential for bias,
brought about by small sample size, exists. However, the
ability to evaluate individual variation and compare age/sex
groups allowed by this method greatly enhances’ the
understanding of sources of variation. Further, visual scan
samples also have biases because of differences in the spatial
distribution among age/sex groups. For example, we found that
juvenile males are often too far offshore to be seen, even
with spotting scopes (Chap.3).

There have been two studies of sea otter time budgets in
California based on telemetry. However, Ribic (1982) did not
separate feeding from other kinds of activity and Loughlin's
(1980) sample was small. Our data provide more extensive
information on time budgets and activity patterns of
California sea otters based on telemetry data.

METHODS

Radiotelemetry is particularly useful for collecting time
budget and activity data on sea otters because radio signals
are not transmitted through sea water. Because of this
characteristic, the radio signal pattern varies according to
an otter's activity. Three general categories of activity
have been distinguished by listening to the radio signal from
an otter: resting, feeding, and "other" (Loughlin, 1980;
Garshelis, et al., 1986). The radio signal from an otter
resting on the surface of the water is constant. Most
foraging otters alternately dive to obtain prey and return to
the surface to consume their catch, or if they were
unsuccessful, breathe before diving again. Thus the radio
signal of a feeding otter is usually a characteristic pattern
of alternating periods of signal and silence. When otters are
engaged in activities other than resting or feeding, such as
swimming, vigorous grooming, or social interactions, the
signal is variable, with the strength of each signal depending
upon the orientation of the otter's body with respect to the

66

surface of the water. We called this category "other". When
the otter could be seen, specific behaviors such as swimming,
active grooming, or social interactions were recorded. These
behaviors were later combined into the "other" category so
that the data collected visually would be comparable to data
based solely on the radio signal. A potential bias in
California was the very shallow feeding within the kelp canopy
that might be confused with the "other" category. This was
evaluated by comparisons with visual observations.

In preliminary analyses, we distinguished a fourth
activity category, "unknown", not used in other telemetry
studies of 24-hour activity (Loughlin, 1980; Garshelis, et
al., 1986), for those data points where the observer could
tell from the radio signal that the otter was active but was
not certain whether or not it was feeding, and those in which
the radio signal was so weak that the activity could not be
identified.

We directly compared activity data collected visually with
those obtained only by listening to the radio signal without

seeing the otter. Two observers, one using each method,
simultaneously recorded data on the same animal at five-minute
intervals. These data were collected by several different

pairs of observers on a variety of individual otters.

A team of 3 to 4 people, each taking a 6 to 10 hour
shift, monitored the activity of individual otters for one or
more 24-hour periods. Most data-collection periods were
either 24 or 48 hours in length (Appendix 4.1). Data were
recorded at 10-min intervals. We developed these methods by
trying various alternatives on otters 1-4 in Morro Bay. Our
activity data on these otters are thus not directly comparable
to the majority of our data and are not included in our
analyses of time budgets. Thus, the data that were included
in our analyses of time budgets mostly came from the center
of the California sea otter range. Fig. 4.1 shows the
relative locations of the watches. Considering the locations
of these watches and the number of individuals that we were
able to monitor for each age/sex class, we feel that we had
a representative sample of sea otter activity in areas that
had been occupied for a long period of time.

RESULTS

Activity in relation to time of day

Otters of all age/sex classes tended to be active and feed
for a large proportion of time during the late afternoon and
early evening but there were differences in the activity
patterns of various groups. For adult males in Morro Bay, we
plotted only periods of activity and inactivity: these animals

67

FIGURE 4.1 -- The locations along the California coast of
watches for collecting time budget data on sea otters
instrumented with radio transmitters.

HALF MOON BAY

0310 20 40 60
|

SCALE IN KILOMETERS
o.6S (10 20 30 40
eee!

SCALE IN MILES
> ANOQ NUEVO ISLAND

SANTA CRUZ
OQUEL POINT

MONTEREY MOSS LANDING
BAY
/SEASIOE
\AMONTEREY—PACIFIC GROVE
POINT PINOS

POINT LOBOS
-YANKEE POINT

BIXBY CREEK
POINT SUR

—SIG CREEK
LOPEZ POINT

CAPE SAN MARTIN
SALMON CREEK

POINT PIEDRAS BLANCAS
=<—SAN SIMEON

Location of watch Adults Juveniles
Females Males Females Males

Santa Cruz - Moss Landing
Moss Landing - Point Sur
Point Sur - Cape San Martin

Cape San Martin - San Simeon

68

had a clearly bimodal activity pattern with a second major
activity period in the morning, peaking about 8 a.m. (Fig.
4.2a). A similar early morning peak, in the percentage of
time feeding, was noticeable in the juvenile males (Fig.
4.3a), the adult females (Fig. 4.4a), and the adult females
with pups (Fig. 4.4b). The evening feeding peak for the adult
males in the Big Sur area was so large that their pattern was
almost unimodal (Fig. 4.2b). The juvenile females had a much
broader feeding peak than any of the other groups, feeding
about 50% of the time even in the middle of the day when the
other age/sex groups rested (Fig. 4.3b). In contrast to the
other groups, the juvenile females often rested during the
night, from midnight to seven a.n.

There was a good deal of variation between individuals,
but also variation from one day to the next, at the same time
of day, for each individual. Our sample sizes, at a specific
time of day, were not large enough to statistically test the
significance of these sources of variance, since most
individuals had only three to four data points for a given
hour of the day.

Comparison of visually and telemetrically collected data

Our data provide the first formal comparison of visual and
telemetric estimates of otter activity in which feeding was
distinguished as a separate category, as neither Loughlin
(1982) nor Garshelis, et al., (1986) undertook such a
comparison. Ribic (1982) found good agreement between visual
and telemetric data when only active and inactive were scored.

We found the highest agreement between the methods when
an otter was resting: when the visual observer indicated that
an otter was resting, the telemetric data agreed 93 percent
of the time (Table 4.1). The telemetric observer never scored
resting as feeding or "unknown" but occasionally scored it as
"other". This error tended to occur when sea conditions such
as high swell caused the otter to move about even though it
was resting.

When the visual observer indicated that an otter was
feeding, the telemetric data agreed 88 percent of the time
(Table 4.1). The telemetric observer rarely scored feeding
as resting or "other" but did sometimes indicate that the
otter's activity was "unknown". This tended to occur when an
otter was feeding without making regular dives. Otters in
California can forage in kelp for prey items such as kelp
crabs or obtain small items such as mussels from rocks without
remaining submerged for more than a few seconds.

When the visual observer indicated that an otter was
engaged in "other" activities, the telemetric data agreed only

69

FIGURE 4.2 -- The percent of time that adult male sea otters
spent in various activities at the various hours of the day.
Males in the Morro Bay area of California are shown in the
upper graph; the males located in the Big Sur area are shown
in the lower figure.

100

O Inactive
4 Inactive

100

GC Resting

4 Feeding
80 © Other

Percent of time

40

0 4 8 12 jen) 20 24
Time of Day

70

FIGURE 4.3 -- The percent of time that instrumented juvenile

sea

otters in California spent resting, feeding and in
"other" activity at the various times of the day.

Data for

juvenile males are shown in the upper graph and those for
juvenile females in the lower graph.

Percent of Time

100:

80

60

40

20

100 j
Resting

Feeding

80 Other

60

40

20

0 4 8 12 16 20 24
Time of day

71

FIGURE 4.4 -- The percent of time that adult female sea
otters in California spent resting, feeding and in "other"
activity at the various hours of the day. Data for adult
females without pups are shown in the upper graph and data
for adult females with pups in the lower graph.

100
80 |

60 |

100

Percent of Time

80 oO Resting
4 Feeding
O Other

Time of day

72

TABLE 4.1 - A comparison between activity data obtained
visually with those obtained using the quality of the
telemetric signal. The numbers in this table represent the
number of 5-minute sampling periods where the activity of an

otter was determined. These observations were taken
simultaneously by two independent observers. AF = adult female
without pup; AFP = adult female with pup; JF = juvenile

female; JM = juvenile male.

NUMBER OF FIVE-MINUTE SAMPLE PERIODS

Visual Percent
Observer Telemetry observer - Agree.

DATE REST REST FEED OTHER UNKNOWN

31-May-85 AF 38 36 2

04-Feb-87 AF 32 31 1

05-Feb-87 AFP 15 15

25-Feb-87 AF 7 6 1

27-Feb-87 JM 1 1

27-Feb-87 JF 24 20 4

03-Mar-87 AF 1 1

30-Mar-87 AFP 12 12

TOTAL 130 121 93

FEED REST FEED OTHER UNKNOWN

31-May-85 AF 33 33

14-Feb-87 AFP 8 1 6 1

25-Feb-87 AF 19 11 8

27-Feb-87 JM 3 3

27-Feb-87 JF 3 3

02-Mar-87 JF 43 41 2

03-Mar-87 AF 5 5

07-Apr-87 AF 8 8

TOTAL 122 107 88

OTHER REST FEED OTHER UNKNOWN

31-May-85 AF 2 1 i!

14-Feb-87 AFP 2 2

25-Feb-87 AF 2 1 1

27-Feb-87 JM 10 9 1

27-Feb-87 JF 7 2 5

30-Mar-87 AFP 1 1

TOTAL 24 ALS) 63

63 percent of the time. Activities in the "other" category
were rarely recorded as feeding by the telemetric observer but
were sometimes scored as resting or "unknown". "Other" was
most commonly recorded as "resting" when the otter was
grooming fairly vigorously without submerging the main part
of its body.

73

TABLE 4.2 - A comparison between time budgets calculated from
observing activity visually and judging activity using the
quality of telemetric signal.

Number of 5-min periods Percentage of time

Visual Telemetry Visual Telemetry
REST 130 121 47 44
FEED 122 108 44 39
OTHER 24 25 9 9
UNKNOWN (0) 16 (0) 6

A comparison of the overall time budgets based on the
visual and telemetric data collected simultaneously on the
same otters (presented in Table 4.1) indicated that the
telemetric data underestimated resting by 3 percent and
feeding by 5 percent for this particular data set (Table 4.2).
These data provide only an indication of the differences
between estimates based on the two methods. The magnitude of
the differences will vary with such factors as the sea state
during the period of data collection, the extent to which the
otter being studied feeds without making regular dives, and
the individuals recording the data. Although these data
suggest that we underestimated the proportion of time spent
feeding, it would be inappropriate to simply increase our
estimates. of the overall proportion of time spent feeding by
five percent because of this variation and because our
estimates are based on a combination of visual and
telemetrically collected data.

Time budgets

Using the entire data set collected during the 24 to 72-
hour watches (Appendix 4.1), we assessed the extent to which
different ways of handling the "unknown" data might affect our
results by comparing four ways of treating them. These were:
1) including them in the "other" category, as in our previous
reports (Ralls, et al., 1985; Siniff and Ralls, 1986); 2)
keeping them as a separate category; 3) excluding them from
the analysis; and 4) including them in the "feeding" category
to get an estimate of the maximum possible feeding time, as
our data suggest that much of the "unknown" category may be
feeding. Because the "unknown" category was small, ranging
from 1 to 7 percent of the total time, the various ways of
treating this category had relatively little effect on our
estimations of the percentages of time the different age/sex
classes of otters spent resting and feeding (Table 4.3). We
therefore included the "unknown" data in the "other" category
in subsequent analyses.

74

We tested differences in average percent of time spent
feeding among sex and age classes using analysis of variance
for the percent of time feeding for each individual monitored

for 24-hour activity data. The results of this analysis,
using Scheffe's multiple comparison test for difference among
sex/age classes, showed that juvenile females fed

significantly more than adult females, adult males and
juvenile males, but not more than adult females with pups (p
< .05) (Table 4.4). Females with small pups fed more than
females with large pups but this difference was not
Significant (Table 4.5).

To facilitate a comparison of our data to data collected
visually (Estes, et al., 1986), we tabulated the number of
ten-minute periods devoted to "resting", "feeding" and "other"
for each observation period in three ways: 1) for all data
recorded during that observation period (Appendix 4.1); 2) for
those data recorded during daylight hours (defined as 1/2 hour
before sunrise to 1/2 hour after sunset) (Appendix 4.2); and
3) for those data recorded while the otter could be seen
(Appendix 4.3).

The various age/sex classes of otters spent about the
same percentages of time resting and feeding during daylight
hours (Table 4.6) as they did over the entire 24-hour period.
When only the data collected visually were considered,
juvenile females still fed more than adult males and adult
females (Table 4.6); juvenile females and adult males fed a
greater, and adult females a smaller percentage of the time
than indicated in the other two data sets. Insufficient
visual data were collected on juvenile males and adult females
with pups to allow a comparison of these groups. Comparisons
of coefficients of variation (for the means of individuals
for each sex/age category for the given method of data
collection) indicated that the data collected visually were
the most variable and those collected over the complete 24--
hour period the least variable (Table 4.6).

DISCUSSION

Activity in relation to time of day

Sea otter activity patterns can be affected by a variety
of factors including geographical location, weather, season,
available prey, and age/sex class (Garshelis, 1983, Estes et
al., 1986). Otters in California tend to be crepuscular,
resting mainly in the middle of the day (Ribic, 1982;
Loughlin, 1980; our data); although Estes, et al., found that

75

TABLE 4.3 - A comparison among the methods of calculating time
budgets. The unknown data are classified by four different
methods: including them in the "other" category, separating
them as unknown, excluding them, and including them in
"feeding". OP = observation periods, AM = adult male. Other
abbreviations as in Table 4.1.

TREATMENT OF AGE/SEX PERCENTAGE OF TIME SAMPLE SIZES
UNKNOWN DATA CLASS Rest Feed Other Unknown Otters Hours OP
AS OTHER AF 48 37 15 i 8 830 28
AS UNKNOWN 48 36 8 7

EXCLUDED 52 37 9 ——

AS FEEDING 48 43 8 --

AS OTHER AFP 45 39 16 -- 6 264 8
AS UNKNOWN 44 39 11 4

EXCLUDED 46 41 12 —=—

AS FEEDING 44 43 11 i

AS OTHER AM 50 36 14 i 4 216 7
AS UNKNOWN 50 37 9 4

EXCLUDED 52 38 9 i

AS FEEDING 50 41 9 —=

AS OTHER JF 40 48 12 -- 9 417 12
AS UNKNOWN 40 49 10 1

EXCLUDED 40 49 10 --

AS FEEDING 40 50 10 --

AS OTHER JM 34 37 29 == 5 218 8
AS UNKNOWN 34 37 24 4

EXCLUDED 35 39 26 ==

AS FEEDING 34 41 24 Oe

TABLE 4.4 - Analysis of variance testing for differences in
percent of time spent feeding between the various sex/age
classes.

Source d.f. Sos. M.S. F
Among sex/age
classes 4 956 426.8 10.7

Error 24 1707 39.8
Test of means (% of time feeding)

JF AFP — AM JM AF

48 39

36 37 37

Means underlined by the same line are not significantly
different (p<.05; Scheffe's test).

TABLE 4.5 - A comparison of time budgets between females that
were accompanied by small pups and those accompanied by large

pups.

ACTIVITY MEAN VARIANCE NUMBER OF
PERCENTAGE OBSERVATION
OF TIME PERIODS
FEMALES WITH SMALL PUPS
Rest 40.60 257.30 4
Feed 42.71 194.80 4
Other 16.53 12.29 4
FEMALES WITH LARGE PUPS
Rest 48.45 17.97 3
Feed 36.02 104.70 3
Other 15.53 39.38 3

Table 4.6 - A comparison of the percent of time spent in each
of the three activity categories, resting, feeding and other,
for activity data collected over the entire 24-hour period,
for daylight data only, and for visual data, for the age/sex
categories of adult females, adult females with pups, adult
males, juvenile females, and juvenile males.

ENTIRE 24-HR PERIOD DAYLIGHT ONLY VISUAL DATA
MEAN NO.OBS.* MEAN _NO.OBS.* MEAN _NO.OBS. *
ADULT FEMALES
Rest 48.2 28 47.72 28 54.4 24
Feed 36.8 28 38.10 28 28.0 24
Other 15.0 28 14.18 28 17.7 24
ADULT FEMALES WITH PUPS
Rest 44.52 8 46.99 8
Feed 39.39 8 39.29 8
Other 16.09 8 13.71 8
ADULT MALES
Rest 50.43 7 47.54 7 43.6 5
Feed 35.8 7 38.06 7 41.2 5
Other 13.77 7 14.40 7 14.6 5
JUVENILE FEMALES
Rest 39.76 12 34.83 12 38.5 11
Feed 47.81 12 51.61 12 57.8 11
Other 12.43 12 13.56 12 3.7 11
JUVENILE MALES
Rest 33.96 8 28.44 8
Feed 36.84 8 37.15 8
Other 29.20 8 34.42 8

*NUMBER OF OBSERVATION PERIODS

77

the otters in one of the areas they sampled had no apparent
24-hour pattern. Garshelis (1983) found deviations from the
crepuscular pattern when locally preferred prey tended to be
active at night. Also, he suggested that, in Alaska, short
day lengths and poor weather conditions, combined with poor
resources, may have made it impossible for otters to maintain
body temperature during long rest periods. It might seem less
likely that this would occur in the milder California climate;
however, one of our juvenile females was observed shivering
on several occasions before she disappeared and presumably
died. Estes, et al., (1986) suggested that environmental
factors such as wind and waves may disrupt the 24-hour pattern
in California. Although we found that otters of all age/sex
classes in California did some feeding at night, none fed
primarily at night as did male otters at one location in
Alaska where the preferred prey was dungeness crabs, which are
thought to be more active, and hence more vulnerable, at night
(Garshelis, et al., 1986). Because we did not collect
activity data in extremely bad weather, we were unable to
determine whether or not otters in California rest for shorter
periods at midday under such conditions.

In contrast to the other age/sex classes, juvenile
females tended to rest for the greatest proportion of the time
from midnight to eight in the morning rather than feed. Thus,
for our data the juvenile females departed most from the more
usual crepuscular pattern. Sea otters are known to steal food
from other individuals (Fisher, 1939; pers. obs), and we
observed that juvenile females were the group from which food
was often taken. Perhaps they tend to feed at different times
than the majority of their conspecifics to reduce the risk of
losing prey in this manner.

Time budgets in relation to other studies

As Ribic (1982) did not separate feeding from other kinds
of activity, Loughlin's (1980) telemetry data on six otters
in Monterey Bay are the only California data comparable to
ours. Our data on adult males and females are very similar
to Loughlin's data in spite of his small sample size, the
different study areas and the number of years between the two
studies (Table 4.7). Garshelis, et al., (1986) present time
budget data based on telemetry for sea otters from two
localities in Prince William Sound, Alaska: Green Island and
Nelson Bay. Sea otters had been present in the Green Island
area since the 1950's or earlier. Females remained in this
area throughout the year; males visited for various periods
of time during the breeding season. In contrast, otters had
moved into Nelson Bay fairly recently. Most of the otters
there were males, some of which moved seasonally to breeding
areas such as Green Island. Both adult males and females fed
Significantly less time at Nelson Bay, where large, high

78

quality prey such as clams and crabs were easily available,
than at Green Island, where such items were rare.

TABLE 4.7 - A comparison of the activity budgets for sea
otters calculated in this study and those in the literature.

AGE/SEX LOCATION METHOD PERCENTAGE OF TIME REFERENCE

CLASS Rest Feed Other

AM CALIF. TELEMETRY 57 33 10 Loughlin 1979
AF CALIF. TELEMETRY 50 36 14 Loughlin 1979
AM CALIF. TELEMETRY 50 36 14 This study
AF CALIF. TELEMETRY 48 37 15 This study
AFP CALIF. TELEMETRY 45 39 16 This study
JM CALIF. TELEMETRY 34 37 29 This study
JF CALIF. TELEMETRY 40 48 13 This study

UNKNOWN CALIF. VISUAL 53-63 21-28 9-22 Estes,
et al., 1986

M AK** TELEMETRY 50 47 3 Garshelis***
AF AK** TELEMETRY 50 47 3 Garshelis***
AFP AK** TELEMETRY 43 53 3 Garshelis***
PUPS* AK** TELEMETRY 45 51 4 Garshelis***
M AK*¥*** TELEMETRY 49 37 aS Garshelis***
AF AK**** TELEMETRY 51 37 - 12 Garshelis***

*Independent pups

**GREEN ISLAND (Alaska)

***Garshelis, et al., 1986 (reference)
***k*NELSON BAY (Alaska)

Our data were collected in the central portion of the sea
otter range in California, in areas where sea otters have been
present for many years. The general patterns in our data were
the same as those at Green Island, Alaska, where otters have
also been established for a long time. At Green Island, adult
males and females fed for about the same percentage of time
and females with pups fed slightly more than females without
pups. At Green Island, recently weaned, independent pups fed
more than adults. Juvenile males in California fed about the
same amount of time as adults, but juvenile females fed more
than adults, except for females with pups.

Estes, et al., (1986) estimated the proportion of time
spent foraging by scan sampling (Altmann, 1974) the otters
visible along the California coastline from dawn to dusk at
1/2 hour intervals and recording the activity of each otter
observed. Estimates of the proportion of time spent feeding
based on this technique are lower than those based on

79

radiotelemetry (Table 4.7). If otters in California spend a
higher proportion of their feeding during the night than
during the day, then scan samples during the day would tend
to underestimate the proportion of time spent feeding.
Garshelis, et al., (1986) found that, in some areas of Alaska,
otters do feed mainly during the night. However, we found no
difference in the proportion of time spent feeding during
daylight hours and that over the entire 24-hour day. The most
probable explanation for the difference between the scan
sample and telemetry data is that feeding otters have a lower
probability of being seen than resting otters (Estes and
Jameson, 1988), and are thus more often missed during scan
samples.

Time budgets as indicators of population status

Estes, et al., (1986) concluded that further growth, in
recent years, of the California population was not because of
food limitation, primarily because their estimates of the
proportion of time the population spent foraging were similar
to those of populations in Alaska known to be below
equilibrium density. However, when all the available data are
considered (Table 4.7) it seems likely that their scan
sampling data underestimated the time spent feeding.
Certainly, other factors than food availability, such as
accidental capture in gill-nets, have contributed to the
reduced growth of the California .sea otter population
(Wendell, et al., 1986). Further, time budgets may be
affected by factors other than the prey availability, such as
weather conditions, prey type, and study methods that obtain
data in different ways. It seems likely that the best
comparisons of activity data are those between studies based
on telemetry methods. We did not collect data during bad
weather. Garshelis, et al., (1986) were able to use automatic
recording to obtain data during bad weather in Alaska, and
found that otters fed more during such periods. Thus, we
probably underestimated the proportion of time California
otters spend feeding. Nevertheless, we found juvenile females
fed at least 48 percent of the time; which represents a
substantial effort when compared to any existing activity
data.

They also tended to have longer feeding bouts than otters
in the other age/sex classes, although the intervals between
these bouts were about the same as in the other classes
(Chapter 5). Two of our juvenile female otters (numbers 44
and 46) had their prey stolen repeatedly. The prey stealing
was selective: only large desirable items were stolen. Otters
whose prey is stolen may temporarily stop foraging, move to
another location, or capture apparently less desirable species
of prey. Otter 44 subsequently died: her stomach was empty
although there were shells in her intestine. The only otter

80

in our study that regularly hauled out was another juvenile
female (otter 42). Hauling-out is a behavior that may help
to conserve energy (Garshelis, 1983).

Differences in the ability of members of different
age/sex classes to compete for food resources are common in
vertebrates (Sutherland and Parker, 1985; Clutton-Brock and
Albon, 1985; Dunbar, 1985). Because of these differences in
competitive ability, the effects of food shortage are usually
concentrated on particular individuals. If these individuals
cannot obtain sufficient food, they must emigrate or starve.

After constructing models of vertebrate populations
composed of individuals with varying degrees of competitive
ability, Sutherland and Parker (1985) argued that "the average
individual in the population can be doing very well despite
the population being at carrying capacity". They concluded
that the proportion of time spent feeding at the population
level is not a good measure of whether the population is
limited by food supply and that it will probably be necessary
to concentrate on the factors affecting the poorest feeders
in the population to understand the carrying capacity and
population dynamics of many vertebrates. In sea otters in
California, the poorest feeders appear to be the juvenile
females; also, many pups die before weaning (Chapter 2). In
Alaska, some females apparently abandoned pups prior to the
age of weaning (Garshelis and Garshelis, 1987). These authors
hypothesized that the abandonment was related to the poor
health of the female due to nutritional stress.

The exact mechanisms that are operating to slow the
growth of the California sea otter population remain unclear.
However, we believe that additional research on mortality
factors in pups and independent juveniles, particularly
juvenile females, would be likely to provide a better
understanding of the dynamics of the California sea otter
population.

LITERATURE CITED

Altmann, J. 1974. Observational study of behavior: sampling
methods. Behaviour 49: 227-267

Ames, J. A. and G. V. Morejohn. 1980. Evidence of white
shark, Carcharodon carcharias, attacks on sea otters,
Enhydra lutris. Calif. Fish and Game 66: 96-209.

Clutton-Brock, T. H. and S. D. Albon. 1985. Competition and
population regulation in social mammals. Pages 557-575

in R. M. Sibly and R. H. Smith, eds. Behavioural
Ecology. Ecological Consequences of Adaptive Behaviour.

Blackwell Scientific Publications, Oxford, U.K.

81

Dunbar, R. I. M. 1985. Population consequences of social
structure. Pages 507-519 in R. M. Sibly and R. H. Smith,
eds. Behavioural Ecology. Ecological Consequences of
Adaptive Behaviour. Blackwell Scientific Publications,
Oxford, U.K.

Eberhardt, L. L. 1977. "Optimal" management policies for
marine mammals. Wildl. Soc. Bull. 5: 162-169.

Estes, J. A., K. E. Underwood and M. J. Karman. 1986.
Activity-time budgets of sea otters in California. J.
Wildl. Manage. 50:626-636.

Estes, J. A. and R. J. Jameson. 1983. Summary of available
population information on California sea otters. POCS
Tech. Pap. 83-11 for Interagency Agreement 114-12-001.
U. S. Fish and Wildlife Service and U. S. Minerals
Management Service. 29 pp.

Estes, J. A. and R. J. Jameson. 1988. A double-survey
estimate for sighting probability of sea otters in
California. J. Wildl. Manage. 52:70-76.

Estes, J. A., R. J. Jameson, and A. M. Johnson. 1981. Food
selection and some foraging tactics of sea otters. Pages
606-641 in J. A. Chapman and D. Pursley, eds., Worldwide

Furbearer Conference Proceedings. August 3-11, 1980.
Frostburg, Maryland.

Estes, J. A., R. J. Jameson, and E. B. Rhode. 1982. Activity
and prey selection in the sea otter: influence of
population status on community structure. Am. Nat. 242-
258.

Estes, J. A., K. E. Underwood, and M. J. Karmann. 1986.
Activity-time budgets of sea otters in California. J.
Wildl. Manage. 50:626-636.

Fisher, E. M. 1939. Habits of the southern sea otter. J.
Mammal. 20:21-36.

Garshelis, D. L. 1983. Ecology of sea otters in Prince William
Sound, Alaska. Ph. D. Thesis, University of Minnesota,
Minneapolis, Minnesota. 321 pp.

Garshelis, D. L. and J. A. Garshelis. 1987. Atypical pup

rearing strategies by sea otters. Marine Mammal Science
33:263-270.

82

Garshelis, D. L., J. A. Garshelis, and A. T. Kimker. 1986.
Sea otter time budgets and prey relationships in Alaska.
J. Wildl. Manage. 50: 637-647.

Jameson, R. J. and J. A. Estes. 1988. Status of the California
sea otter population. Abstracts, American Society of
Mammalogists Annual Meeting, Clemson, South Carolina,
June, 1988.

Loughlin, T. R. 1980. Radio telemetric determination of the
24-hour feeding activities of sea otters, Enhydra lutris.
Pages 717-724 inc. J. Amlaner, Jr., and D. W. MacDonald,

eds. A Handbook on Biotelemetry and Radio Tracking.
Pergamon Press, Oxford, U. K.

Miller, D. J. 1980. The sea otter in California. Calif.
Coop. Oceanic Fish. Invest. Rep. 21: 79-81.

Ralls, K., D. B. Siniff, C. Monnett, T. Eagle, and L. Ferm.
1985. Summary of information pertaining to California
permit to capture sea otters for scientific research.
Report to California Fish and Game Commission, 104 pp.

Ralls, K, J. Ballou, and R. L. Brownell, Jr. 1983. Genetic

diversity in California sea otters: theoretical
considerations and management implications. Biol. Consv.
25:3:209-232.

Ribic Ce JAG 1982. Autumn activity of sea otters in

California. J. Mammal. 56:701-703.

Siniff, D. B. and K. Ralls. 1986. Summary of information
obtained on sea otter for MMS study on population status
of California sea otters. Report to California Fish and
Game Commission and Office of Sea Otter Coordination
(USFWS). 86 pp.

Sutherland, W. J. and G. A. Parker. 1985. Distribution of
unequal competitors. Pages 255-273 in R. M. Sibly and
R. H. Smith, eds. Behavioural Ecology. Ecological

Consequences of Adaptive Behaviour. Blackwell Scientific
Publications, Oxford, U.K.

Wendell, F. E., R. A. Hardy and J. Ames. 1986. An assessment
of the accidental take of sea otters, Enhydra lutris, in
gill and trammel nets. Calif. Dept. Fish Game, Mar. Res.
Tech. Rep. No. 54, 31 pp.

83

CHAPTER 5

FEEDING PATTERNS OF CALIFORNIA SEA OTTERS

K. RALLS, B. HATFIELD AND D. B. SINIFF

NOVEMBER 30, 1988

84

INTRODUCTION

Because sea otters can often be easily observed from
shore, their diet and foraging patterns have been studied in
various parts of their range (Estes, Jameson, and Johnson,
1981; Ostfeld, 1982; Garshelis, 1983; Lyons, 1987). Most
existing data have been obtained by visual observations of
foraging otters. Relatively little is known about night-time
foraging patterns, although studies using radio-telemetry have
shown that sea otters in California do forage by night as well
as by day (Loughlin, 1977; Ribic, 1982).

Early studies primarily yielded information about diet
and foraging patterns at the population level. They showed
that the California population feeds almost entirely on
macroinvertebrates, although some populations in Alaska and
the U.S.S.R. also feed on epibenthic fish, and that there
appeared to be a great deal of variation in the diet and
foraging patterns of individual otters (Estes, et al., 1981).
Recently, studies on otters that could be individually
identified by flipper-tags have confirmed that individuals
tend to specialize on one to three or more of the many
available types of prey and shown that these patterns of
specialization may be maintained for three or more years
(Lyons 1987).

Although we collected some data on uninstrumented otters,
we focused on the foraging patterns of individual instrumented
otters as indicated by radio-telemetry. Foraging sea otters
alternatively dive to search for prey and return to the
surface to breathe and consume their catch. Because radio
Signals are not transmitted through sea water, we could
measure the length of dives and surface intervals whether or
not we could see the instrumented otters. We present data on
dive and surface intervals, the length of feeding bouts and
the intervals between them, diurnal and nocturnal foraging
patterns, and variation in foraging patterns within and
between age/sex classes. We also compare data collected by
listening to the radio signal with those collected visually
and discuss the way in which variation in foraging patterns
appears to be related to competition among individuals.

METHODS

The majority of data were collected during watches
intended primarily to obtain information on time budgets and
activity patterns (Chapter 4). Most of these watches were 24
to 48 hours in length. Shorter watches, designed specifically
for the collection of feeding data, were conducted during
morning and evening feeding periods on the otters in Morro Bay
(otters 1-4). Data on the feeding activities of unidentified,

85

uninstrumented otters were also collected during one two-week
period to obtain additional information on dive and surface
interval lengths in relation to the size and type of captured
prey.

The length of dives and surface intervals was measured
from the radio signal from the instrumented animals and by
visual observation of the uninstrumented otters. As visual
observations indicated that the occasional dives shorter than
10 seconds were rarely feeding dives and that surface
intervals shorter than five seconds were almost always the
result of interruption by another otter, these were excluded
from analyses. When possible, foraging individuals were
observed from the shore through a high resolution telescope
(Questar, 50x or 80x magnification), and the number, size, and
species of captured prey were recorded.

Lengths of feeding bouts and the intervals between bouts
were measured to the nearest minute. If more than 30 minutes
elapsed between two feeding dives, these two dives were taken
as the end of one feeding bout and the beginning of a second
feeding bout, respectively. Records with a large amount of
activity recorded as "unknown" and those where there was any
ambiguity as to the end of a feeding bout were not used for
the determination of bout and interval length. Only complete
feeding bouts were used for the calculation of feeding bout
length but intervals between bouts were measured as long as
the ending of the first bout and the beginning of the next
were known. We defined the day as the period from 1/2 hour
before sunrise to 1/2 hour after sunset.

Otters were assigned to sex and age classes based on
their weight at capture; estimated age, often based on the
examination of cementum layers in a vestigial premolar
extracted for this purpose; and, in the case of females,
reproductive performance. All juveniles were judged to be no
more than two years of age (Chapter 2).

Statistical comparisons among age and sex classes were
performed using analysis of variance, controlling for
variation among individuals within classes. We performed a log
(base 2) transform on the data to reduce heterogeneity of
variances. All statements that differences are statistically
significant are based on the 0.05 probability level.

RESULTS
Observations from the shore

Because our instrumented otters often foraged at times
when or in areas where they could not be easily observed, many

of our visual observations were made on uninstrumented otters.

86

The data presented here consist of all visual observations on
both instrumented and uninstrumented otters. The mean length
of observed dives was 52.14 seconds (n = 712). Dive length
varied with prey type to some extent, being least for mussels
and greatest for octopus (Table 5.1) but was not related to
prey size (Table 5.2). Surface times were clearly related to
both prey type (Table 5.3) and prey size (Table 5.4). They
were longest for large prey such as crabs, abalone, and
octopus that often took an otter several minutes to eat.

Success rate varied with prey type. Otters foraging on
mussels and small, hard-bodied prey that they pounded with a
rock had the highest success rates while those foraging on
large, calorically rich prey such as clams, abalone, and crabs
of the genus Cancer had the lowest success rates. These
relationships were evident even within age/sex classes for a
sample of the instrumented animals (Table 5.5).

Data collected from the radio signal

The data presented here consist of dive lengths, surface
intervals, feeding bout lengths, and the lengths of intervals
between feeding bouts, all of which were collected from the
instrumented otters regardless of whether or not the animals
were seen.

Dive length -- The unweighted mean dive length for all
instrumented otters was 73.56 seconds (n = 8254). Mean dive
lengths for the individual instrumented otters ranged from 41
to 149 seconds (Table 5.6).

Analysis of variance on the log-transformed data
indicated that there were significant differences among the
lengths of the dives made by individuals within age/sex
classes (Appendix 5.1). There were significant differences in
dive length between age/sex classes (Table 5.7). Scheffe's
test showed that adult males made shorter dives than adult
females, who in turn made shorter dives than juvenile females
and adult females with pups. Juvenile males made the longest
dives, with a mean length of 116 seconds. The short dive
times for adult males reflect the large proportion of dives
from adult males in the relatively shallow waters of Morro Bay
in our sample for this age/sex class.

The distributions of dive lengths for the different
age/sex classes showed that adult males and adult females with
pups had more individual variation in dive lengths than the
members of the other age/sex groups (Fig. 5.1). The
distribution of dive lengths for juvenile males indicates
considerable internal consistency, even though the average
dive length for this group was by far the longest (Fig. 5.1).

87

TABLE 5.1 - A comparison among the average dive lengths (sec)
for sea otters prior to the capture of various types of prey
in California.

PREY DIVE LENGTH

TYPE MEAN (SEC) N VARIANCE
MUSSEL 33.81 296 DATO
CRABS (All) 56.07 261 831.76
CRABS (Pugettia) 56.27 60 1186.86
CLAMS 58.09 220 517.94
CRABS (Cancer) 59.10 79 790.88
SEA STAR 64.65 23 1361.44
ABALONE 71.69 39 557.34
POUNDED WITH ROCK 78.66 196 1576.49
TUNICATE 79.18 28 230.29
OCTOPUS 101.71 28 2493.63

TABLE 5.2 - The average length of the feeding dives (sec) made
by sea otters in California prior to the capture of prey of
different sizes.

PREY DIVE LENGTH

SIZE MEAN N VARIANCE
(SEC)

NONE 63.34 1149 741.06

SMALL 57.91 636 1245.61

MEDIUM 60.39 214 716.18

LARGE 64.90 229 882.29

EXTRA LARGE 61.37 147 674.06

Surface interval -- The unweighted mean surface interval for
all otters was 64.50 seconds (n = 7944). Mean surface
intervals for individual otters ranged from 25.5 to 155.3
seconds (Table 5.8). Analysis of variance on the log-

transformed data indicated that there were significant
differences among the lengths of surface intervals for the
individuals in all age/sex classes except the juvenile males
(Appendix 5.2). There were also significant differences among

88

TABLE 5.3 - A comparison among the lengths of the surface
interval (sec) that were required to consume the various prey
items taken by sea otters in California.

LENGTH OF SURFACE INTERVAL

PREY MEAN

TYPE (SEC) N VARIANCE
TUNICATE 33.25 28 98.76
SEA STAR 54.00 23 675.65
MUSSEL 58.18 291 1290.32
CRABS (Pugettia) 94.22 55 6498.75
CLAMS 95.14 219 4398.47
POUNDED WITH ROCK 97.60 195 3860.78
CRABS (ALL) 120.67 267 19198.31
OCTOPUS 132.89 28 21610.38
ABALONE 150.92 39 25252.22
CRABS (Cancer) 213.39 85 34860.80

TABLE 5.4 - A comparison among the average lengths of the
surface intervals (sec) that were required to consume various
sizes of prey taken by sea otters in California.

PREY LENGTH OF SURFACE INTERVAL
SIZE MEAN N VARIANCE
(SEC)
NONE 30.36 1095 1050.30
SMALL 55.56 635 2839.58
MEDIUM 62.34 216 2436.76
LARGE 122.59 226 14734.20
EXTRA LARGE 177.66 148 22886.74

the age/sex classes (Table 5.9). Adult females with pups
tended to have the longest surface times, followed by juvenile
males, and then a group consisting of the juvenile females,
adult males, and adult females (Table 5.9). These trends were
also apparent in the distributions of surface intervals for
the age/sex classes (Fig. 5.2).

89

TABLE 5.5 - A comparison among the instrumented individuals
in the various sex/age categories for the percentage of
successful dives during feeding bouts. AF = adult female
without pup; AFP = adult female with pup; AM = adult male; JF
= juvenile female. No data were obtained for the juvenile
males.

OTTER AGE/SEX NUMBER NUMBER SUCCESS PRINCIPAL
NUMBER CLASS SUCCESSFUL UNSUCCESSFUL RATE PREY
DIVES DIVES PERCENT TYPE
15 AF 102 115 47 ABALONE/OTHER®
6 AF 173 59 75 CRABS
22 AF 41 2 95 SHB**
16 AFP 34 32 52 CRABS
25 AFP 109 5 96 MUSSELS
1 AM 244 616 28 CLAMS
3 AM 33 63 34 CLAMS
7 AM 47 12 80 CRABS
Vi AM 82 15 85 CRABS |
/SHB**
4 AM 162 18 90 MUSSELS
44 JF 31 40 47 CRABS#*
37 JF 25 24 51 CRABS
40 JF 44 27 62 CRABS
45 JF 31 15 67 CRABS
46 JF 142 8 95° SHB** @
* -- Otters with two prey types listed caught approximately
equal numbers of each type.
SHB** -- unidentified small, hard-bodied prey that were
pounded with a rock.
# -- Four of the prey items captured were stolen by other
otters.
@ -- Three of the prey items captured were stolen by other
otters.
-- Pugettia
@ -= Cancer

90

TABLE 5.6 - A comparison of the average dive lengths (sec)
recorded for individual instrumented sea otters during feeding
bouts. AF = adult female without pup; AFP = adult female with

pup; JF = juvenile female; AM = adult male; JM = juvenile
male.
OTTER AGE/ DIVE LENGTH
NUMBER SEX MEAN N VARIANCE
(SEC)
46 JF 41.48 427 485.47
4 AM 41.50 161 255.80
9 AF 45.66 772 323.87
15 AF 52.28 774 653.49
1 AM 57.05 733 647.24
37 JF 57.77 70 258.72
3 AM 58.03 88 650.03
34 AM 61.46 69 3694.60
16 AFP 61.73 194 487.53
16 AF 62.69 179 1224.25
25 AFP 66.91 487 1297.79
45 JF 71.52 406 689.53
39 JF 71.85 239 369.36
19 AF 73.94 48 330.10
2 AM 74.91 22 626.45
6 AF 76.89 1006 818.20
7 AM 77.48 101 1298.72
17 AM 80.13 210 3411.17
36 AFP 83.20 119 976.18
44 JF 83.52 326 540.65
42 JF 92.33 271 1062.93
40 JF 92.86 308 1129.69
13 JM 95.91 107 1861.67
43 JM 100.50 22 2245.70
27 AFP 100.56 179 1077.39
47 JF 104.93 160 2580.98
30 JM 115.64 119 1183.44
41 JM 132.43 127 231.01
29 JF 132.46 118 1409.91
35 JM 135.82 123 1717.04
22 AF 140.18 153 1019.31
14 AFP 149.40 136 1106.62
Feeding bouts -- Mean feeding bout lengths for individual

otters ranged from 77 to 373 minutes (Table 5.10). The
shortest mean bout length was for an adult male feeding
primarily on clams in Morro Bay and the longest was for a
juvenile female feeding on small, hard-bodied prey items in
the Piedras Blancas area. Tallying the number of individuals
within the age/sex classes according to the mean length of
feeding bout suggested that juvenile females, and to a lesser

91

FIGURE 5.1 -- The distribution of dive times during feeding
bouts for adult males, adult females, adult females with
pups, juvenile females and juvenile males in California.

40

Adult Males

30

20

Adult Females

Adult Females With Pups

Juvenile Females

Percent of Dive Times

Juvenile Males

60 150 240 >315

Seconds
92

FIGURE 5.2 -- The distribution of the length of time of the
surface intervals during feeding bouts for adult males, adult
females, adult females with pups, juvenile females and
juvenile males in California.

40

Adult Males

30

20

Adult Females

Adult Females With Pups

Juvenile Females

Percentage of Surface Times

20

10

Juvenile Males

20

10

1
60 150 240 >315

Seconds
93

TABLE 5.7 -- Analysis of variance testing for differences in
the length of the dives made by otters belonging to the
various age/sex classes. Abbreviations as in Table 5.6.

Source df Mean square F p
Total 7937
Among age/sex classes 4 128.37 242.2 <0.001
Error 7933 6.53
Test of means:

AM AF Bye Nn JM
Mean dive length (sec) 64 75 83 92 116

The means for the age/sex classes underlined by the same line
are not significantly different from each other (P<0.05,
Scheffe's test).

extent, adult females, tended to have long feeding bouts
(Table 5.11A). The two individuals with mean feeding bout
lengths over 250 minutes were both juvenile females (otters
45 and 46). When we tallied the number of bouts of various
lengths for each age/sex class (Table 5.11B), we noted that
juvenile females tended to be the most different from the
others. Chi-square analysis testing for shifts in the
distribution of lengths of feeding bouts among sex/age classes
(Table 5.11B) showed a highly significant difference (p<0.01).
When we eliminated the juvenile females from the analysis, the
Chi-square was no longer significant, even at the 0.05
level,indicating no difference in bout length among the
remaining age/sex classes.

Interval between feeding bouts -= The mean interval between
feeding bouts was 187.7 minutes (n = 228). This is close to
the approximately 180 minutes required for food to pass
through the digestive system (Stulken and Kirkpatrick, 1955;
Costa, 1982). Values for individuals ranged from 80.9 to 300.9
minutes (Table 5.12). Juvenile females and females with pups
appeared to have shorter intervals between feeding bouts than
the other age/sex classes. However, Chi-square analysis of the
data in Table 5.13B indicated that there was not a significant
difference among age/sex classes in the pattern of the
distribution of lengths of the time between feeding bouts.

Comparison of day and night foraging patterns

We were able to compare day and night foraging patterns
in four age/sex groups: adult females, adult females with
pups, juveniles males and juvenile females. Unfortunately, we
had insufficient data on the nocturnal foraging patterns of
adult males to include them in this analysis.

94

TABLE 5.8 - The average surface times (sec) for individual
instrumented sea otters that were recorded during feeding bouts.
Abbreviations as in Table 5.6.

ES

OTTER AGE/ SURFACE TIME
NUMBER SEX MEAN N VARIANCE
(sec)

37 JF 25.52 69 1137.81
46 JF 38.72 424 1835.33
19 AF 39.86 44 696.03
9 AF 43.43 744 1972.60
39 JF 44.19 242 1016.50
3 AM 49.23 86 4547.46
15 AF 49.43 732 4359.00
1 AM 50.17 701 3140.20
45 JF 50.56 396 7165.18
6 AF 52.14 985 3300.86
7 AM 52.52 99 3119.66
2 AM 53.40 20 1745.54
4 AM 59.99 158 2359.77
25 AFP 65.98 453 2467.77
44 JF 67.27 324 11250.17
42 JF 69.38 276 6846.35
40 JF 77.71 301 8128.68
16 AFP 78.22 188 6867.25
41 JM 84.26 125 293.02
35 JM 84.47 113 5525.51
17 AM 88.36 190 9636.31
34 AM 91.17 63 3750.24
47 JF 92.47 158 9386.40
27 AFP 93.54 163 10118.73
30 JM 94.43 115 6304.85
16 AF 96.74 170 16644.03
43 JM 102.29 21 7415.98
13 JM 106.17 103 9990.12
29 JF 106.93 109 6643.02
14 AFP 131.90 119 1697.86
36 AFP 139.50 105 10092.75
22 AF 155.33 148 7428.96

TABLE 5.9 - Analysis of variance testing for differences
between the mean length of the surface intervals made by
otters of the various age/sex classes. Abbreviations as in
Table 5.6.

Source af Mean square F Pp
Total 7645 168.98 132.01 <0.001
Among age/sex classes 4
Error 7641
Test of means:

JF AM AF JM AFP
Mean surface
interval length (sec): 63 64 WS 94 102

Dive length -- Analysis of variance of the log-transformed
data indicated that there were significant differences between
the length of day and night dives for the adult females with
pups and the juvenile females but not for the adult females
without pups and the juvenile males (Appendix 5.3). However,
there were significant interactions between individuals and
the length of day and night dives for all four age/sex groups.

The mean lengths of the day and night dives of the
individual otters are compared in Table 5.14. Some individuals
made longer dives at night, some during the day, and others
made dives of about the same length during both periods. Only
two of the four adult females with pups tended to make longer
dives at night, even though the analysis of variance indicated
a significant difference between day and night dive length for
this group. The juvenile females tended to make longer dives
during the day but again there was no consistency within the
group (as indicated by the significant interaction terms in
the ANOVA), with only four of the seven individuals making
significantly longer dives during the day.

Surface intervals -- Analysis of variance of the log-
transformed data indicated that there were significant
differences between the length of the day and night surface
intervals for the juvenile males and females but not for the
adult females or the adult females with pups (Appendix 5.4).

There were significant interactions between individuals and
the length of day and night surface intervals for all age/sex
classes except the adult females with pups. The mean lengths
of the day and night surface intervals of the individual
otters are compared in Table 5.15. The general pattern was
similar to that for the dive length data in that some otters

96

TABLE 5.10 - The average lengths of the feeding bouts (min)
that were recorded for individual instrumented sea otters.
Abbreviations as in Table 5.6.

OTTER AGE/SEX BOUT LENGTH

NUMBER CLASS MEAN N
(MIN)

1 AM 77.33 15
13 JM 88.00 5
36 AFP 90.00 9
“41 JM 97.50 24
30 JM 99.50 8
27 AFP 103.90 10

9 AFP 106.60 5
11 AF 109.50 6
34 AM 111.57 7
29 JF 118.67 6
22 AF 119.23 13
16 AF 126.27 15
43 JF 130.40 10
14 AFP 133.60 5
35 JM 133.64 17
17 AM 134.63 11
16 AFP 138.29 7

7 AM 139.71 7

6 AF 142.66 18
43 JM 145.67 3
25 AFP 145.77 9
39 JF 146.50 9
19 AF 166.56 16
15 AF 171.06 16
42 JF 176.43 7
37 JF 186.60 5
40 JF 195.70 10

9 AF 201.25 4
36 AF 237.33 3
47 JF 244.42 7
45 JF 368.50 4
46 JF 373.00 3

97

TABLE 5.11 - The lengths of feeding bouts (min), grouped by
time intervals, classified by the five age/sex categories of
adult males, adult females, adult females with pups, juvenile
females, and juvenile males. Part A is a tabulation of the
average feeding bout lengths of each instrumented otter. Part
B is a tabulation of the lengths of the individual feeding
bouts.

A. MEAN FEEDING BOUT LENGTHS OF INDIVIDUAL OTTERS WITHIN
CLASSES

BOUT ADULT JUVENILE FEMALES ADULT JUVENILE |
LENGTH MALES MALES WITH PUPS FEMALES FEMALES

(MIN) ———————— EEE

<150 4 5 6 3 3
150-250 0) (0) 0 4 4
>250 0 0) 0 0 2

B. LENGTH OF INDIVIDUAL FEEDING BOUTS WITHIN CLASSES

BOUT ADULT JUVENILE FEMALES ADULT JUVENILE
LENGTH MALES MALES WITH PUPS FEMALES FEMALES
(MIN)

<150 19 41 30 40 29
150-250 6 14 13 23 14
>250 2 2 2 10 17

had longer surface intervals at night, some during the day,
and others had surface intervals of about the same length
during both periods. However, fewer otters (nine) had
significant differences between the length of day and night
surface intervals than between day and night dive lengths (15)
and values of the test statistic, K, tended to be smaller for
surface intervals for dive lengths.

Feeding bouts -- The mean length of entirely nocturnal feeding
bouts (119.16 minutes, n = 43) was similar to that of those
that occurred entirely during daylight hours (120.67 minutes,
n = 62) but bouts that spanned the transition period from day
to night or night to day tended to be considerably longer
(209.80 minutes, n = 25).

DISCUSSION

Comparison of data with existing data sets

The results of our visual observations agree with those
of the principal previous study in California (Estes, et al.,
1981). Our overall mean dive time was 52 seconds; Estes, et

98

TABLE 5.12 - The average lengths of the intervals between
feeding bouts (min) that were recorded for the individual
instrumented sea otters. Abbreviations as in Table 5.6.

OTTER AGE/SEX LENGTH OF INTERVAL BETWEEN BOUTS
NUMBER CLASS MEAN N
(MIN)
27 AFP 80.89 9
7 AM 96.67 3
14 AFP 103.60 5
37 JF 113.50 4
29 JF 121.60 25
41 JM 121.91 22
44 JF 122.89 9
25 AFP 139.77 9
39 JF 143.57 7
47 JF 146.57 7
9 AFP 160.20 5
36 AF 167.33 3
42 JF 170.00 5
35 JM 170.29 14
40 JF 175.38 8
45 JF 176.00 4
17 AM 177.78 9
16 AFP 179.17 6
36 AFP 184.00 6
6. AF 222.35 14
16 AF 233.54 11
13 JM 236.25 4
11 AF 242.80 5
9 AF 244.33 3
22 AF 254.50 12
30 JM 254.86 7
34 AM 272.60 5
19 AF 287.46

PR
Nu

15 AF 300.91

99

TABLE 5.13 - The lengths of the intervals between feeding
bouts (min), grouped by time intervals, classified by the five
sex/age categories of adult males, adult females, adult
females with pups, juvenile females, and juvenile males. Part
A is a tabulation of the average interval between feeding
bouts for each instrumented otter. Part B is a tabulation of
the individual intervals between feeding bouts.

A. AVERAGE INTERVAL BETWEEN BOUTS FOR INDIVIDUAL OTTERS

INTERVAL ADULT ADULT ADULT JUVENILE JUVENILE

LENGTH FEMALES FEMALES MALES MALES . FEMALES
(MIN) WITH PUP

<150 0) 3 1 1 6
150-250 4 3 2 2 3
250-350 3 io) 1 2 0
350-450 0) 0) 0 ce) (0)
450-550 (0) 0) 0 0) fe)
>550 0) te) 0) 0) 0)

B. INDIVIDUAL BOUTS

INTERVAL ADULT ADULT ADULT JUVENILE JUVENILE

LENGTH FEMALES FEMALES MALES MALES FEMALES
(MIN) WITH PUP

<150 19 25 7 30 35
150-250 16 10 6 7 7
250-350 8 3 9 4 1
350-450 7 0) (0) 5 4
450-550 8 0) 1 2 3
>550 3 3 (0) 1 0)

al. (1981) reported mean dive times of 50 to 60 seconds. Like
Estes, et al., (1981), we also found no relationship between
dive length and prey size and that dive length was not greatly
affected by prey type, with the exception of dives for a few
prey types such as mussels, which tended to be short, or
octopus, which tended to be long. We found that surface
times were longer for large, calorically rich prey items and
that success rates were higher for small prey items such as
mussels; again this agrees with previous studies in both
California (Estes, et al., 1981) and Alaska (Garshelis, 1983).

Although Loughlin's (1977) mean dive length of 57 seconds
based on telemetry data was similar to that derived from
visual observations (Estes, et al., 1981), the relatively few
individuals he studied appear to have foraged close to shore

100

TABLE 5.14 - A comparison of the mean lengths of dives made
during the day and night by the individual instrumented otters
in the various age/sex classes. The Kruskal-Wallis test was
used to test for differences between day and night means, at
the 0.05 probability level.

OTTER DAY NIGHT LONGER
NUMBER MEAN N MEAN N DIVES
Adult females
15 62.8 433 47.8 285 day
9 45.1 736 56.4 36 night
6 71.3 538 83.3 468 night
22 141.2 62 139.5 91 ns
36 66.9 52 95.9 67 night
Adult females with pups
25 53.9 334 95.3 153 night
16 63.1 286 59.1 87 ns
27 94.9 75 104.6 104 night
14 147.7 80 151.8 56 ns
Juvenile males
13 105.7 72 75.8 35 day
30 114.7 68 116.9 51 ns
43 40.6 7 128.7 15 night
41 131.7 105 137.9 22 ns
35 129.3 91 154.3 32 night
Juvenile females
45 72.9 272 68.7 134 day
39 71.7 97 72.8 141 ns
40 102.3 206 73.8 102 day
46 49.9 217 38.9 210 day
44 84.6 266 78.8 60 day
42 90.2 124 94.2 147 ns
47 94.0 116 133.9 44 night

Adult males
17 62.5 161 138.2 49 night

101

TABLE 5.15 - A comparison of the mean lengths of surface
intervals made during the day and night by the individual
instrumented otters in the various age/sex classes. The
Kruskal-Wallis test was used to test for differences between
day and night means, at the 0.05 probability level.

OTTER DAY NIGHT LONGER
NUMBER MEAN N MEAN N SURFACE
INTERVAL
Adult females
15 48.6 406 Bra) B/S) night
9 43.3 709 46.6 35 ns
6 54.0 522 50.1 463 ns
22 160.9 62 151.3 86 ns
36 119.4 46 155.2 59 ns
Adult females with pups
25 66.6 312 64.7 141 ns
16 86.2 273 89.6 85 ns
27 78.6 67 104.0 96 night
14 130.6 73 134.0 46 ns
Juvenile males
13 132.0 70 51.4 33 day
30 81.4 66 111.9 49 night
43 112.8 6 98.1 15 ns
41 87.2 102 Yakoil 23 ns
35 93.9 83 53.8 30 day
Juvenile females
45 49.2 266 Bs} oS} als}(o) ns
39 49.4 96 40.7 146 day
40 78.5 206 75.9 95 ns
46 47.4 213 30.0 211 day
44 69.6 271 55.4 53 ns
42 77.8 127 62.2 149 day
47 80.6 114 123.3 44 ns
Adult males
17 Gio aay 145.7 43 night
in the Monterey area. Our more extensive telemetry data
indicated that visual observations of otter foraging in
California tend to underestimate mean dive lengths. As

Garshelis (1983) found that dive length was correlated with
water depth in Alaska, this is probably because feeding otters
can only be observed easily when they are foraging close to
shore in relatively shallow water. The mean dive length for
our instrumented animals was about 13 seconds longer than our
estimate based on visual observations of both instrumented and

102 °

uninstrumented otters. Although Estes, et al., (1981)
reported, based on visual observations, that "dives longer
than 125 seconds almost never occurred" in California, we
found that five of 31 otters had mean dive lengths that
exceeded 125 seconds and that twelve otters had maximum dive
lengths of over 200 seconds. The longest dive we timed was
246 seconds; previous reports of maximum dive lengths are 200
and 275 seconds for California (Estes, et al., 1981 and
Loughlin, 1979, respectively), and 205 seconds in Alaska
(Garshelis, 1983).

Visual observations have given the impression that
adults forage in deeper water than juveniles (Estes, et al.,
1981). Although our otter with the shortest mean dive length
was a juvenile female, another juvenile female had a mean dive
length of 132 seconds. Furthermore, juvenile males spent much
of their time far offshore beyond the kelp beds and tended to
forage farther offshore, and hence probably in deeper water,
than otters of other age/sex classes (see Chapter 3). Because
our juvenile males tended to forage so far from shore, we were
unable to observe them feeding, but the radio signal indicated
that they tended to make longer dives than otters in the other
age/sex classes, with a mean length of 116 seconds. The
behavior of such juveniles is clearly not reflected in
previous data sets on sea otter diet and feeding patterns.

Diurnal and nocturnal foraging patterns

Many individual otters displayed differences in diurnal
and nocturnal dive length patterns that may reflect a tendency
to specialize on different prey species, that may tend to
occur at different mean depths, by day and night. However,
there was no general tendency for longer dive lengths during
either time period. Some of the many prey items available to
the California population may be more vulnerable at night.
For example, crabs belonging to the genus Cancer and octopuses
are generally thought to be nocturnal (Estes, et al., 1985;
Ricketts, et al., 1986; Barr and Barr, 1983). Individual
otters also vary in the extent to which they tend to feed at
night. Since the mean length of diurnal and nocturnal feeding
bouts was similar, differences in the distributions of diurnal
and nocturnal surface intervals were less frequent than those
for dive lengths, and the length of surface intervals is
related to the size of prey consumed, most individuals may
have similar diurnal and nocturnal rates of caloric intake.

Our data on the length of surface intervals were
generally similar to those in other studies, in that the
length of surface intervals increased with the size of the
captured prey. However, the time required to consume captured
prey is not the only factor affecting the length of surface
intervals. Adult females with pups and juvenile males had the

103

longest surface intervals. Visual observations showed that
these were, in part, the result of social interactions of the
adult females with pups and of the juvenile males with other
juvenile males.

Individual variation in foraging patterns

The degree of individual difference in foraging patterns
among California sea otters is striking. Data presented in
this chapter indicate that individuals vary with respect to
the size and species of prey consumed, dive length, surface
interval length, feeding bout length, and the degree of
difference between diurnal and nocturnal feeding patterns.
Data presented in Chapter 4 indicated that individuals also
vary in the total amount of time spent feeding per 24-hr day
and the proportion of time they forage at night.

Taken as a whole, these individual differences suggest
that prey items are not equally available to all individuals
in the California population. Juvenile females appear to be
at a disadvantage compared to adults. They tended to feed for
long periods and for a higher proportion of time than the
other age/sex classes (Chapter 4). Much of their prey
consisted of items that were too small to be identified and
when they were successful in capturing a large, desirable prey
item such as a crab belonging to the genus Cancer, this was
often stolen by another otter. They often fed during the day
when most otters were resting (Chapter 4), which probably
helped them avoid competition with older animals.

Juvenile males often fed farther off-shore than the
other age/sex groups (Chapter 3), on unknown prey species.
As a group, juvenile males had longer surface intervals than
juvenile females and their feeding bout lengths were similar
to those of adults. In general, our results on foraging, time
budgets and activity, movements, and survival strongly suggest
that juvenile females tend to be at a disadvantage in the
portion of the range where we studied.

LITERATURE CITED

Barr, L. and N. Barr. 1983. Under Alaskan Seas: the
Shallow-water Marine Invertebrates of Alaska. Northwest
Publishing Company, Anchorage, AK. 208 pp.

Costa, D. P. 1982. Energy, nitrogen, and electrolyte flux
and sea-water drinking in the sea otter Enhydra lutris.
Physiol. Zool. 55: 35-44.

Estes, J. A., R. J. Jameson, and A. M. Johnson. 1981. Food
selection and some foraging tactics of sea otters. Pp.
606-641 in J. A. Chapman and D. Pursley (eds.). Worldwide

104

Furbearer Conference Proceedings. August 3-11, 1980.
Frostberg, Maryland.

Estes, J. A. and G. VanBlaricom. 1985. Sea-otters and
shellfisheries. Pp. 187-235 in J. R. Beddington, R. J.
Beverton and D. M. Lavine (eds.). Marine Mammals and

Fisheries. George Allen and Unwin, London.

Garshelis, D. L. 1983. Ecology of sea otters in Prince
William Sound, Alaska. Ph. D. Thesis, University of
Minneapolis, Minnesota. 321 pp.

Loughlin, T. R. 1977. Activity patterns, habitat
partitioning, and grooming behavior of the sea otter,
Enhydra lutris, in California. Ph. D. Thesis, University
of California, Los Angeles, 110 pp.

Lyons, K. 1987. Abstract. Individual variation in diet and
foraging strategy in the female California sea otter,
Enhydra lutris. Animal Behavior Society, 21-26 June 1987,
Williamstown, Mass.

Ostfeld, R. S. 1982. Foraging strategies and prey switching
in the California sea otter. Oecologia 53: 170-178.

Ribic,» C..5 A. 1982. Autumn activity of sea otters in
California. J. Mamm. 63:702-706.

Ricketts, E. F. and J. Calvin. 1986. Between Pacific Tides.
4th edition, revised by J. W. Hedgpeth. Stanford
University Press, Stanford, CA. 614 pp.

Stulken, D. E. and C. M. Kirkpatrick. 1955. Physiological
investigation of captive mortality in the sea otter
(Enhydra lutris). Trans. 20th N. Amer. Wildl. Conf.:
476-494.

105

CHAPTER 6

AGE DETERMINATION OF CALIFORNIA SEA OTTERS FROM TEETH

P. PIETZ, K. RALLS, AND L. FERM

NOVEMBER 30, 1988

106

INTRODUCTION

Determining the age of individuals by counting incremental
lines in tooth cementum has proved to be a useful tool for a
wide variety of mammal species (for a review see Grue and
Jensen 1979). It has been used to estimate ages of both
salvaged and living sea otters in Alaska (Schneider 1973 and
Garshelis 1984, respectively).

Accurate age estimates of living animals offer potential
insights into many aspects of sea otter biology, such as the
age of first reproduction in females, the ages of territorial
males, and age-related differences in movement patterns.
Accurate age estimates of salvaged animals may also be useful
as a means of estimating population age structure and, thus,
for constructing models that can detect and predict changes
in population parameters.

METHODS

In an effort to gain insight into the age structure of the
California population, we have studied a sample of premolars
from more than 580 sea otters. We extracted 30 premolars
(PM,, as recommended by Schneider 1973) from animals that were
captured and radio-tagged; we collected the rest from skulls
of dead otters salvaged by the California Department of Fish
and Game and the U.S. Fish and Wildlife Service. Skulls of
many salvaged animals had been deposited at numerous
institutions and agencies. The majority of our teeth were
taken from skulls in collections at the following
institutions: Santa Barbara Museum of Natural History,
California Polytechnic State University (San Luis Obispo),
California Department of Fish and Game (Monterey, Morro Bay),
U.S. Fish and Wildlife Service (Piedras Blancas), University
of Puget Sound, and San Jose State University.

The teeth were decalcified, sectioned, and stained by Gary
Matson, P.O. Box 308, Milltown, Montana. From teeth mounted
in paraffin, he cut longitudinal sections 14 m thick and
stained them with Wolbach's Giemsa. Basic procedures (Luna
1968) were modified in conjunction with advice from Aleta Hahn
of the National Marine Fisheries Service (S.W. Fisheries
Center, P.O. Box 271, La Jolla, California). To determine
ages, we counted bands in the cementum of the sectioned teeth
using criteria outlined by Schneider (1973) for Alaskan
otters.

RESULTS AND DISCUSSION
Despite the extensive analyses of sea otter teeth
conducted by Schneider (1973) and Garshelis (1984), there are

presently only a few teeth available from known-age otters.

107

This makes it difficult to evaluate the relationship between
cementum lines and annual time intervals with certainty.
However, we have been able to examine teeth from ten
California otters of known minimum age, and the age estimates
from these teeth compare quite favorably with the age
estimates of the otters made by field biologists (Table 6.1).
Comparisons of ages determined from tooth cementum with known
ages for animals of a variety of other species are shown in
Table 6.2. Although there is exact agreement in most cases,
there is a difference of more than one year in a few

S———SESE>E>E>S>S>SEeS=S=E=ES|"=a=Lh)™)™“|["Sl=S]=e@a|*[*i[is| _is{s(7(“os)sS>S> Si i i Sass ———————,

WAIBINESs) Gra Tooth age estimates for sea otters of minimum known age.
Univ. FUS* CDF&G Sex Date Est. age Date Est. age Tooth
Minn. no. no. tagged when found at death age
no. tagged” dead
42 116 1142 F 10-82 {2 2 1-83 1° 2 1+
65 074 1012 F 10-80 0.5 - 1 9-81 Weds) 2 2
68 115 1182 F 10-82 0.5 - 2 5-83 1-3 265 2
617 097 1106 M 10-81 0.5 - 1 8-82 ino © 2 2
41 = 1577 Mientec38)5 OS onl 4-87 A 2 B55 3
47 090 1170 M 10°81 O.5 - 2 4-83 B95) Sob &
616 : 1573 M 6-82 < 0.5 3-87 5 5
55 051 1269 F 7-79 4+ 1-84 8.5+ 6
440 : 715 F 12-72 i+ 8-79 7.5% 8
615 = 14964 F 10-79 3+ 5-86 9.5+ 11

A

e All animals with FWS numbers were initially tagged by FWS. The

others were tagged by CDF&G.

2 Ranges include estimates by R. Jameson and J. Bodkin of FWS, to
the nearest half-year.

LLLL—h—w—haha Eh —L— LL

cases. Judging by the data in Tables 6.1 and 6.2, the
accuracy of the technique for sea otters may be similar to
that for other species.

Our optimism regarding the value of the cementum technique
for age determination for sea otters arises from three
sources: (1) the correlation between cementum lines and
yearly intervals in known-age animals of numerous other
species (Grue and Jensen 1973, Grue and King 1984); (2) the
correlation between cementum lines and age estimates made by
field biologists in our ten sea otter teeth of "known"-age
(Table 6.1); and (3) the correlation between age class
assignments determined from sea otter skulls and from teeth
Cmnigis Bese

108

Age estimates based on skull features (e.g. suture
closure, ridge development, tooth wear) were available for
over 200 of the animals for which we have sectioned teeth.
These estimates were provided by Jack Ames of the California
Department of Fish and Game. He assigned skulls to five age
categories: pups (< 6 months old); immatures (probably about
1/2 to 11/2 years); subadults (probably 1-4 years); adults
(probably at least 4 years), and old adults (probably at least

10 years). Comparing tooth age estimates from our most
experienced reader to these categories, we found that 96-100%
of our readings for animals in the "pup", "immature", and

"Subadult" categories fell within the appropriate age ranges
(Fig. 6.1). As expected for the broader and less well defined
categories of "adult" and "old adult", the range of tooth ages
within each category was greater and the overlap between
categories was greater. Nevertheless, 72% of teeth from
animals in the "adult" category were assigned ages from 4-10
years and 71% of those in the "old adult" category, ages from
7-16 years.

Ages of teeth from older animals are the most difficult
to assess. Garshelis (1984) noted that it was difficult to
determine exact ages when more than about 10 cementum annuli
were present because annuli were spaced so closely together
in older animals. Teeth obtained from skulls stored in
museums may offer an additional problem; according to
Schneider (1973), teeth allowed to air-dry developed dark
edges which made it difficult to differentiate outer annuli.

Despite these potential difficulties, our age assignments
for teeth from "old adult" skulls were not unreasonably low.
Male and female sea otters in Alaska have been estimated to
live 10-15 and 15-20 years, respectively (Calkins and
Schneider 1984). In California, the oldest tagged otters
under observation are a 13-14 year-old female and two > 12
year-old males (M. Riedman and J. Estes, pers. comm.). In our
sample of 580 animals, the oldest age estimates based on tooth
cementum were 16 years for two females and 15 years for two
males.

The technique of determining age by counting cementum
lines can best be evaluated with an extensive reference
collection of teeth from known-age sea otters. We are now in
the process of developing a reference collection, but this
requires long-term effort and inter-agency cooperation. In
the meantime, we have attempted to evaluate other aspects of
the technique: (1) variation within and among readers, (2)
variation among premolars of the same individual, and (3)
variation in readability due to different methods of preparing
skulls and teeth.

109

Percent in Each Age Class

FIGURE 6.1 -- Comparison of age estimates based on teeth to
age-class assignments for the same sample of otters made by
the California Department of Fish and Game using skull
features.

Subadults

Adults

234567891011 12
Old Adults

4° 8-456 7 BUononane ise liserc

Estimated Ages (Years)

110

A total of 614 teeth were sectioned, stained, and "read"
by at least one reader. A sub-sample of nearly 200 teeth was
examined once by two readers and twice by one reader. About
100 of these teeth were examined once by four readers.
Statistical analysis of the sub-sample of 100 showed no
significant differences among the age distributions determined
by three of the four readers or between readings made by the
same reader. In two trials by the same reader, exact
agreement between age assignments occurred in over 77% of 177
teeth. This reader showed a 0-1 year difference between age
assignments for 98% of the teeth. None of this reader's age
assignments were more than two years apart for a given tooth.
Age assignments made by two different readers showed exact
agreement for 52% of 179 teeth, a 0-1 year difference for 87%,
and < 2 year differences for 96% of the teeth. Age
assignments among three readers showed a 0-1 year difference
for 73% of 85 teeth, and < 2 year differences for over 93% of
these teeth. Table 6.3 gives more detailed results of
comparisons between and within readers.

Multiple teeth from about 40 otters were sectioned,
stained, and read by four readers in order to help us evaluate
variability among teeth of individual otters. All four first
premolars were available for 14 otters (boiled and broken
teeth were excluded). Comparing ages assigned by individual
readers with the modal age (across readers and teeth) for an
individual otter, we found that over 67% of the assigned ages
for 406 teeth were exactly the same as the mode, over 90% were
0-1 year away from the mode, and 96% were < 2 years away from
the mode (Table 6.3). These results compare well with age
estimates of duplicate teeth from other species (Table 6.2).

Boiling skulls to facilitate cleaning has been a standard
practice in many museums and laboratories. Schneider (1973),
however, noted that extensive boiling of teeth made cementum
lines less distinct. We compared the definition or
distinctness of lines noted for boiled and unboiled teeth in
our sample. Among the 516 teeth for which treatment during
preparation was known, there were 60 teeth for which the
reader noted "indistinct lines" or "poor definition." Half
of these 60 teeth had been boiled, whereas only 34% of the
other 456 teeth had been boiled. This difference was
statistically significant (Chi? Saya, Che = il, 1) <— Ms Ons) -
thus agreeing with Schneider's findings. We also examined
pairs of teeth from individual otters that were purposely
processed in different ways; i.e., of two teeth taken from the

same animal, one was boiled and the other was not. Tooth
sections were rated within pairs on quality of cementum line
definition. For the six pairs available, the boiled tooth

was always rated as poorer quality than the unboiled tooth
(sign test, P = 0.03). We concur with Schneider in recom-

111

TABLE 6.2 -- A comparison of animal ages determined from
tooth cementum by Gary Matson: (1) with animals of known
age; (2) with duplicate determinations based on a different
tooth from the same animal; and (3) with ages estimated from
the degree of tooth wear (reprinted from Matson's Tooth
Cementum Age Analysis, Progress Report No. 9, Spring 1987,
Table 1).

Period Kind of Species Numberlin Exact 1 Year More than § . Average Age

Comparison Sample Agreement a/ Difference Year Difference
1978-82 Known Age Various 28 22 4 2 5.0
1983-85 Known Age Various 14 8 5° q 5.4CA b/
54 KA
1986 Known Age 8T Deer i] 1 0 0 4.5CA
4.5 KA
1986 Known Age WT Deer 25 22 1 2 33CA
3.5 KA
1986 Known Age Caribou 14 12 1 0 29CA
29KA
1986 Known Age Kit Fox 9 7 2 (0) 25CA
; 21KA
1986 Known Age 8.H. Sheep 2 0 2 0 85CA
95 KA
1986 Known Age R. Oster 7 7 (3) 0 A5CA
Q5KA
1986 Known Age Raccoon 19 14 5 (0) ZICA
1.9KA
1984-85 Duplicate c/ Elk, Coyote, 172 8&9 271 2 29 od/
Bobcat 27
1986 Ouplicsate Bobcat 100 93 7 0 25
: 26
1986 Ouplicate Black Bear 138 9 34 13 52
5.2
71986 Ouplicate Pronghorn 10 & 2 (¢) &2
5.0
1983-86 Duplicate®/ — Elk 1,915 1,730 152 33 5.8
5.8
1986 Duplicate Ek 88 45 36 7 4.9
; 5.0
1986 Comparison M. Deer 192 123 58 10 23CA
with wear 3.5 WAI

& Matson’s age compared with age from 8 second source.

6. CA = Cementum sage; KA = Known age.

c A “duplicate” age determination was the analysis of 2 teeth from the same individual mammal without Matson’s knowlecge
of the identity of the duplicates.

Oo. The first number given is the average age according to the first analysis, and the second number is according to the
second analysis.

€. Two primary incisors were aged together to obtain greatest accuracy for each elk, Matson’s knew of the duplication
while aging, and the table shows the number and size of changes meade when the ages of paired incisors differed,

f. Upper canine teeth. Sample collected and cementum aged over 8 period of several years and aged again in 1986.

9 WA = Aged by tooth wear and replacement, :

mending that boiling be avoided in future specimen
preparation.

TABLE 6.3 -- Comparisons of sea otter age assignments based
on counts of incremental lines in tooth cementum.

Age assignments No. of Age assignments
compared teeth in :
between/among: sample * no 1-yr 2-yr >3-yr

Cbised, we ehisaig wen Colbie ohbsay

Trials A and B
by reader 2 177 137 36 4 0

Readers 1
and 2A 179 93 62 16 8

Readers 1,
2A, and 3 85 37 25 17 6

Readers 1, 2A, I
3, and 4 85 16 29 14 26

Teeth from the
same otter 406 274 93 23 6 oaks)

* No boiled teeth were included in these comparisons.

Finally, we evaluated the effects of air drying teeth on
our ability to read outer annuli. For a group of otters from
the same skull age category ("old adult"), we compared the
average ages assigned to teeth collected "fresh" and stored
frozen until slide preparation and teeth collected from museum
specimens that were stored dry at room temperature. We found
no statistically significant difference between the age
distributions of these two groups (Chi* = 3.04, df = 3, P >
0.25). Our sample size was small (47 teeth), however, and
thus should not be considered an endorsement for dry shelf
storage of teeth.

We believe that counting cementum lines in teeth is a
useful technique for determining ages of sea otters. We used
tooth ages assigned by our most experienced reader for the age
estimates of radio-tagged otters in California given in
Chapter 2 and to obtain the age distributions of males and
females, based on 425 teeth from dead animals of known sex in
Fig 6.2.

113

Percent in Each Age Class

FIGURE 6.2 -- Distribution of age estimates for 425 dead sea
otters, based on incremental lines in tooth cementum.

[] Males

Mi Females

Estimated Age (Years)

114

Literature Cited

Calkins, Donald G., and Karl B. Schneider. 1984. Species
account: the sea otter (Enhydra lutris). Alaska Dept.
Fish and Game. 14 pp.

Garshelis, David L. 1984. Age estimation of living sea
otters. J. Wildl. Manage. 48(2) :456-463.

Grue, Helen, and Birger Jensen. 1973. Annular structures in
canine tooth cementum in red foxes (Vulpes vulpes L.) of
known age. Danish Rev. Game Biol. 8(7):1-12.

Grue, Helen, and Birger Jensen. 1979. Review of the
formation of incremental lines in tooth cementum of
terrestrial mammals. Danish Rev. Game Biol. 11(3):1-48.

Grue, Helen E., and Carolyn M. King. 1984. Evaluation of age
criteria in New Zealand stoats (Mustela erminea) of known
age. New Zealand J. Zool. 11:437-443.

Luna, Lee G., ed. 1968. Manual of histologic staining
methods of the Armed Forces Institute of Pathology.
McGraw-Hill, New York. 258 pp.

Morejohn, G. Victor, Jack A. Ames, and David B. Lewis. 1975.
Post mortem studies of sea otters, Enhydra lutris L., in
California. Calif. Dept. Fish and Game, Marine Resources
Technical Report No. 30. 82 pp.

Schneider, Karl B. 1973. Age determination of sea otter.
Alaska Dept. Fish and Game, Fed. Aid in Wildlife
Restoration, Final Report, Proj. W-17-4 and W-17-5, Job
8.10R. 23 pp.

115

CHAPTER 7

ANALYSIS OF THE PRECISION AND ACCURACY OF RADIOTELEMETRY

EQUIPMENT AND METHODS USED IN CALIFORNIA

A. MERCURE

NOVEMBER 30, 1988

116

INTRODUCTION

The positions of the sea otters, instrumented with radio
transmitters, off the California coast in this study were
estimated, when possible, by triangulating on the radio
signals. The accuracy and precision of this method were
studied by placing a radio transmitter on a buoy, establishing
the location of the buoy by visual methods, and taking a
series of compass bearings on the buoy's radio signal. In
this chapter, I consider the field techniques associated with
the estimation of the precision and accuracy of the
triangulation measurements on these buoys, and provide
quantitative estimates of error for otter-location data
collected by field personnel.

METHODS

General procedures

Radio transmitters were placed inside a styrofoam buoy
to simulate a resting sea otter and the buoy was anchored off
the California coast in two locations (south and north
locations) near Piedras Blancas.

One of the first necessities of this work was to
determine the map location of the two buoy locations. Both
the south and the north buoy locations were determined by
sighting through a Questar telescope from three positions
along the shore. The coordinates of the three shore positions
were obtained from our UTM grid on topographic maps of the
study area. Compass bearings to the buoy were obtained by
holding a Silva ranger compass against the Questar and reading
the magnetic bearing from the compass. These bearings, along
with the map coordinates of the three positions from which
they were taken, were then used to plot the map location of
the buoy. This method is illustrated in Fig. 7.1.

The accuracy of this method was initially tested by
sighting on a prominent landmark (Piedras Blancas Lighthouse)
and recording 12 compass bearings from the Questar to the
landmark. Subsequently, the UTM topographic map coordinates
of the Questar position and the landmark location (in UTM
coordinates) were obtained and the true bearing between the
two locations was determined from these coordinates. This
bearing was then compared to the mean of the 12 Questar
bearings.

Our normal field procedure when locating otters was to
hold the compass by hand when taking-bearings, rather than
holding it against the Questar scope. To determine any
additional variance that may have been contributed by holding

117

FIGURE 7.1 -- Illustration of the method used to determine
the location of the buoy by taking compass bearings, from
three locations along the California coast, to the signal
from the radio transmitter on the buoy.

————

1 kilometer

118

the compass by hand, an additional 25 bearings to the Piedras
Blancas Lighthouse were taken using this method.

To check for possible effects of tides on the location
of the buoys, we collected data on the location of the south
buoy over several tidal cycles. One hundred twenty-one compass
bearings using the Questar scope were taken from one position
over a period of 300 hours.

Once the locations of the buoys had been determined in
UTM coordinates, our next task was to use these known
locations of the buoys to measure our ability to record
accurate locations of sea otters. To do this, we used the
same techniques and equipment we normally used to record
bearings on the instrumented otters to record bearings on
these buoys. The telemetry equipment consisted of two Ford
vans with directional four-element yagi antennas and Cedar
Creek Bioelectronics Laboratory radio receivers. The antenna
was pointed towards the direction of the strongest radio
signal, usually taken as mid-way between the two points at
which the signal disappeared ("nulls"), and a compass bearing
of this direction was determined by sighting along the antenna
with the hand-held compass. Four different observers helped
collect bearings toward the stationary buoys. Observers took
bearings in blocks of 50. In order to disrupt the tendency
to mechanically repeat prior readings, the mobile receiving
unit was moved about one meter after every two readings and
was completely turned around after every 10 readings.
Bearings were taken from three different positions along the
shore to the south buoy location and from two positions to the
north buoy location.

The shore positions used for the south buoy location were
Similar to those used for the collection of actual sea otter
location data, in that the angles of the intersections of the
bearings from these positions were generally 90 degrees or
less. However, the two shore positions used for the north
buoy location were chosen to test for effects of signal bounce
and estimate the extent to which error was related to the
distance of the shore position from the transmitter. Here,
bearing measurements were taken from positions four and six
kilometers from the buoy and the angle between the bearings
taken from these two positions was almost 180 degrees.

Establishing measures of precision and accuracy

Several methods have been developed to estimate the
accuracy and precision of radio signal locations determined
from bearing data. Lenth (1981) described three methods for
estimating the location of a radio signal from _ the
intersection of three bearings: the maximum likelihood
estimator, the Huber estimator and the Andrews estimator. All

119

three are based upon probability distributions. The Huber and
Andrews estimators are designed to be relatively insensitive
to outlying points. Precision is defined by Lenth (1981) on
the basis of an iterative algorithm and a variance co-variance
matrix. Computer program TRIANG (Garrott, et al., 1986)
calculates these estimators and defines the area of a 95%
confidence ellipse around them, using the methods described
by Lenth. Detailed mathematical descriptions are presented
in Lenth (1981). Garrott, et al., (1986) also did an
empirical test of the three estimators and found the Andrews
estimator to be superior to the other two estimators.
Confidence ellipses generated from the Andrews estimator were
more likely to include the actual transmitter location and
were found to be more accurate, i.e., the plotted point was
usually closer to the actual transmitter location.

Our work with the buoys gave me a sample of bearings on
a known location from several positions along the coast.
These bearings did not correspond exactly to those used to
locate otters, since they were repeated bearing measures on
a known location. To more closely simulate field data, I took
random samples of these bearings and used them to represent
bearings that would have been recorded during field
operations, if the buoy had been a resting sea otter.

One hundred bearings from each of the three shore
positions for the south buoy were randomly chosen and one
bearing from each position was randomly chosen and combined
into a set of three. These 100 sets of three bearings were
then used to triangulate the location of the transmitter and
estimate precision and accuracy for the south buoy location.

The angle generated from bearings taken from the two
shore positions for the north buoy approached 180 degrees,
thus these data could not be used, as recorded, to estimate
precision and accuracy. However, I was able to use the
distribution of errors (degrees difference from the actual
bearing) of the bearings taken on the north buoy. These error
measures were placed in a random order and 50 sets of three
were randomly drawn. Three positions along the California
coast near the north buoy location were then selected so that
the distance between them represented the maximum distance
between the positions normally used when locating otters. The
actual bearings from these positions to the north buoy were
calculated from our topographic maps and the sets of errors
were then added to the actual bearings to simulate field data.

Precision was defined as the 95% confidence ellipse, and
accuracy was estimated by calculating the distance from the
plotted point to the actual transmitter location.
Calculations were made using the Andrews estimator and
computer program TRIANG (Garrott, et al., 1986).

120

Comparisons of methods for estimating location from bearings

After the bearing data to the two buoy locations were
generated, it was then necessary to compare the location
estimates produced by our field method of plotting data and
the Andrews estimator. Our field method consisted of plotting
at least two, but preferably three, bearings to an otter on
the topographic maps and then assigning a position in UTM
coordinates to the point at the intersection of two bearings
or in the center of the triangle formed by three bearings. To
compare this method with the Andrews estimator, the 100 sets
of bearings toward the south buoy and the 50 sets of simulated
bearings toward the north buoy were plotted by the field
method. The resulting location estimates and the Andrews
estimators for the same sets of bearings were then compared
to the actual buoy locations.

A direct comparison of the individual location estimates
produced by the field method and the Andrews estimator was
made using 50 otter locations, randomly chosen from our otter
position data plotted in the field. These locations were
recalculated by entering the bearings obtained in the field
into a computer and again using program TRIANG to obtain the
Andrews estimator.

RESULTS

Accuracy of determining buoy locations visually

The difference between the mean of the bearings to the
lighthouse taken with the compass held against the Questar and
the calculated true bearing was 0.6 degrees, with a standard
deviation of 0.4 degrees. The difference between the mean of
the hand-held compass bearings toward the lighthouse and the
calculated true bearings was 1.6 degrees, with a standard
deviation of 1.2 degrees. These data are summarized in
Figures 7.2 and 7.3. It was not possible to read the compass
to an accuracy greater than one degree.

The bearings taken to evaluate the possible effects of
buoy movement due to changing tides are plotted against time
aligy ARSC PG eI TAG CNG No movements of the buoy over time were
apparent.

Accuracy of bearings towards the radio signal from the buoys

The difference between the mean of the 897 hand-held
compass bearings to the radio signals on the buoys and the
calculated true bearing was 0.5 degrees (Table 7.1). The
bearing error did not differ significantly with buoy location,
shore position, van, or observer (Table 7.1). The

121

distribution of the differences between these bearings taken
by field personnel and the calculated true bearings is shown
nlp SEKe fa PY/6 Sn

Precision and accuracy of triangulations on buoys

Table 7.2 summarizes calculations of the precision and
accuracy of our simulated locations of the north and south
buoys through radiotelemetry. Precision, defined as the 95%
confidence ellipse, was estimated as between 0.03 and 0.08
hectares. The mean accuracies of our estimations of the
locations of the buoy were 51 meters for the south location
and 110 meters for the north location.

Comparison of field-method estimates with Andrews estimator

The results obtained by hand-plotting the 100 sets of
bearings to the south buoy and the 50 sets of bearings to the
north buoy and the Andrews estimator for these same data sets
are compared in Table 7.3. Although the two methods gave
similar results, in that, on the average, the estimated
locations were about equally close to the true location of
the radio signal, the actual data points calculated by the two
methods were different. The mean difference between the 50
field data points and the Andrews estimator for the same
bearings was 162 meters with a standard deviation of 104
meters. The reason for this difference is that the two
procedures use different methods of weighting the bearings.
The Andrews estimator is designed to be robust to outliers and
thus assigns different weightings to individual bearings based
upon a probability distribution. However, field personnel
usually plotted the location of the otter in the middle of the
triangle formed by the three bearings and thus gave equal
weight to each of three bearings.

DISCUSSION

Errors in determining the true locations of the buoy by
visual sightings with the Questar were minimal and appeared
to be the result of our inability to read the compass to an
accuracy of less than one degree and/or the limitations of our
ability to locate points precisely on the topographic maps.
The positions of the buoys did not change with the changing
tides.

The accuracy of our bearings compared quite favorably

with that of those taken in other radiotelemetry studies.
Several studies have reported mean errors and standard

122

Number of Bearings

PLCUREOW (20 ou Disticlome tony ote | 2) compass) sbeatingse toma
prominent landmark sighted through a telescope (Questar).
The actual bearing of the landmark (zero degrees in the
figure) was calculated from the positions of the telescope
and the landmark on a topographic map.

-20 -16 =| =3) -4 0 4 3 12 16 20

Visual Questar Bearings Deviation From Actual

In Degrees

12%

FIGURE 7.3 -- Distribution of 25 hand-held compass bearings
to a prominent landmark. The actual bearing to the landmark
was calculated as in Figure 7.1.

Number of Bearings

-20 -16 S122 -8 4 0 4 8 12 16 20

Hand Held Compass Bearings Deviation From Actual

In Degrees

124

Degrees

226

225

224

_ 225

FIGURE 7.4 -- Compass bearings to a radio transmitter on a
buoy off the California coast, taken over a three-hour period
when the tide was changing.

100 200 300 400

Time in Hours

Number of Bearings

FIGURE 7.5 -- Distribution of 1125 hand-held compass bearings

to the signals from radio transmitters on buoys off the
California coast.

120

100

50

60

40

207

=20%m =16 12) @Oe-2) 1 ea 0 4 er Hebi2 Ke O
All Telemetry Bearings Deviation From Actual

In Degrees

TABLE 7.1 - Summary statistics for the 897 hand-held compass
bearings to the signals from the radio transmitters on buoys

off the California coast.

N Mean difference Standard
from true bearing deviation
(degrees) (degrees)

All bearings 897 0.5 4.2
By buoy location:

South 447 -0.7 4.4

North 450 1.6 4.0
By van:

1 597 0.5 4.3

2 300 0.5 4.1
By observer:

1 50 -3.6 1.7

2 250 (0) 6 al 3.8

3 297 1.6 4.4

4 300 0.6 4.2
By shore position:

1 (south buoy) 247 -1.4 4.1

2 (south buoy) 100 0.3 4.4

3 (south buoy) 100 0.3 Siew

4 (north buoy) 300 2.0 4.1

5 (north buoy) 150 0.8 365

127

TABLE 7.2 - Summary of the calculated precision and accuracy
of the methods used to judge the locations of radio
transmitters on buoys, based on hand-held compass bearings to
the direction of the radio signal and details of the locations
of the transmitters on the buoys and the mobile receivers near
which the bearings were taken.

A. Precision

South buoy location

N 100
mean size of the 95% confidence ellipse -03 hectares
standard deviation -O1 hectares

Simulated North buoy location

N 50

mean size of the 95% confidence ellipse -08 hectares

standard deviation -02 hectares
Accuracy

South buoy location

N 100
mean deviation from actual transmitter

location 51 meters
standard deviation 32 meters

Simulated North buoy location

N 50
mean deviation from actual transmitter

location 110 meters
standard deviation 66 meters

B. Distance to transmitter

South buoy location

receiver location 1
receiver location 2
receiver location 3

Simulated North buoy location
receiver location 1
receiver location 2
receiver location 3

Degrees between locations

South buoy location

receiver location 1-2
receiver location 2-3

Simulated North buoy location

receiver location 1-2
receiver location 2-3

700 meters
600 meters
613 meters

1642 meters
780 meters
1262 meters

34 degrees
78 degrees

74 degrees
52 degrees

—aaaaaaaBnBnBnBnBanBnBnDBanBn9n9nBnBRaRe

TABLE 7.3 - Comparisons between the field method of plotting
data and the Andrews estimator for the same set of data.

Field
method
South buoy location
N 100
mean deviation from actual
transmitter location 51 meters
standard deviation 37 meters
North buoy location
N 50
mean deviation from actual
transmitter location 119 meters
standard deviation 57 meters

129

100

51

32

50

110
66

Andrews
estimator

meters
meters

meters
meters

deviations greater than we observed. For example, in flat
terrain, Hupp and Rati (1983) recorded mean errors between 0.4
and 3.2 degrees, with standard deviations of 1.3-5.0 degrees.
In areas with mountains and trees, their estimates of mean
error were greater: 4.5 to 28.2 degrees, with standard
deviations of 52.7-83.8 degrees. Lee, et al., (1985), taking
bearings from fixed towers, reported mean errors between 1.76
and 5.27 degrees (after removing all bearing errors greater
than 10 degrees from the sample); Brewer (1983) found that
25%-40% of his bearings were unusable due to the inability to
distinguish between direct and reflected signals, and Garrott,
et al., (1986) noted that 52% of transmitter locations that
were not along the line-of-sight to the position from which
bearings were taken, resulted in bearings with large mean
errors and/or large standard deviations because of signal
reflection.

As shown by the results of these other studies, signal
bounce and interference from rugged terrain and obstructions
are often major problems during field studies using telemetry.
The preferred position to take bearings toward a signal is
from a topographically elevated site with a direct and
unobstructed line-of-sight to the transmitter. In our study,
we often had nearly ideal conditions for telemetry, as many
of the positions along the coast from which we took bearings
were located at the top of shoreline cliffs, above the
transmitters in the otters on the surface of the ocean, and
there were no obstructions between the receiver and the
transmitter. These conditions minimized the possibility for
signal bounce. The bearings which were taken toward the north
buoy were taken from positions chosen to be most likely to
produce signal bounce. Specifically, the positions of the
tracking vans with the receivers were 4 and 6 kilometers from
the buoy, so that the bearing direction to the buoy was almost
parallel to the general direction of the coast and provided
the maximum possibility for interference from intervening land
forms. As the error of bearings taken under these conditions
was not significantly different from those taken towards the
south buoy location, where the bearing direction was towards
the open ocean and there were no obstructions, we believe that
signal bounce was not a significant source of error in the
collection of telemetry location data on the California sea
otter.

The accuracy and precision that we calculated are
applicable to otters located within approximately 800 meters
from shore. At least 75 percent of the plotted locations of
adult male and female and juvenile female otters fell into
this category (Chapter 3, Fig. 3.8). However, since juvenile
males were often located more than 800 meters from shore, the
accuracy and precision of our locations for this class of

130

otters are probably considerably worse than our calculated
values.

It should also be realized that our calculations of
accuracy and precision represent the error present under ideal
conditions, when the radio signals are clear, continuous, and
strong and observers are attempting to take optimum readings.
Under field conditions, signals are frequently interrupted or
weak -- this is particularly likely to be true for those from
juvenile males far offshore -- and observers are sometimes
tired or hurried.

The locations estimated by the field method of hand-
plotting, using bearings from the buoy data sets, were as
close to the actual buoy location as the Andrews estimator
calculated by program TRIANG (Table 7.3). However, the
estimate of the average distance between the hand-plotted
locations and the Andrews estimator, using bearings on actual
otters, was 162 meters. However, this mean difference cannot
be used to quantify the error of the otter locations we
estimated in the field, as there are a variety of possible
spatial relationships between the two estimated locations. For
example, if the actual location of the sea otter is in between
the location estimated by the field method and the one
indicated by the Andrews estimator, then the difference
between the location plotted in the field and the one
resulting from Andrews estimator will exceed the distance from
either of these estimated points to the actual otter location.
Conversely, if both estimated points lie on the same side of
the otter's actual location, then the distance between the
Andrews estimator and the field estimate could be less than
the distance of either to the otter. The distributions of
both the points produced by the Andrews estimator and those
produced by the method we used in the field in relation to
actual otter locations are unknown. The distribution of the
Andrew estimator would be dependant upon the differential
weighting to bearings given by program TRIANG. We believe,
therefore, that it would be inappropriate to add the 162-meter
mean difference between these two estimates to our estimate
of accuracy.

LITERATURE CITED
Andrews, D.F., P.J. Bickel, F.R. Hampel, P.J. Huber, W.H.

Rogers and J.W. Tukey. 1972. Robust Estimates of
Location: Survey and Advances, Princeton University Press.

Brewer, L.W. 1983. Radio tracking the spotted owl in
Washington state. A discussion of equipment and
technique. In Proceedings 4th International Wildlife

Biotelemetry Conference, Ed. D.G. Piniock, Applied
Microelectronics Institute, Halifax, Nova Scotia.

131

Garrott, R.A., G.C. White, R.M. Bartmann and D.L. Weybright.
1986. Reflected signal bias in biotelemetry triangulation
systems. J. Wildl. Manage. 50:747-752.

Lee, J.E., G.C. White, R.A. Garrott, R.M. Bartmann and A.W.
Alldredge. 1985. Accessing the accuracy of a
radiotelemetry system for estimating animal locations. J.
Wildl. Manage. 49:658-674.

Lenth, R.V. 1981. On finding the source of a signal.
Technometrics 23:149-154.

Springer, J.T. 1979. Some Sources of Bias and Sampling Error
in Radio Triangulation, J. Wildlife Mgmt. 43:4 pp 926-
935.

132

CHAPTER 8

MOVEMENT PATTERNS OF ADULT FEMALE AND WEANLING

SEA OTTERS IN PRINCE WILLIAM SOUND, ALASKA

C. MONNETT AND L. ROTTERMAN

NOVEMBER 30, 1988

133

INTRODUCTION

The tendency for sea otters (Enhydra lutris) to exhibit
spatial segregation of the sexes is a well established feature
of their social system (Lensink 1962; Kenyon 1969; Peterson
& Odemar 1969; Schneider 1978; Garshelis & Garshelis 1984;
Garshelis, Johnson & Garshelis 1984). Males and females
segregate into geographically discrete portions of habitat
that are generally referred to as "male areas" and "female
areas" (Kenyon 1969, p. 208) or "breeding areas" (Garshelis,
Johnson & Garshelis 1984, p. 2648). Male areas are occupied
almost exclusively by males of all ages (e.g. Kenyon 1969;
Garshelis, Johnson & Garshelis 1984) whereas, female areas
tend to contain a mixture of mature males and females of all
ages.

The movement patterns of mature males and/or males in the
male areas are relatively well understood as a result of
tagging and short-term radio-telemetry studies. Young males
are born in female areas. In Alaska they apparently leave
their natal female areas shortly after weaning. They move to
and reside within a single male area or travel among several
male areas until maturity (e.g. Kenyon 1969; Garshelis,
Johnson & Garshelis 1984), which occurs at about 5-6 years
of age (Green 1978; Schneider 1978; Garshelis 1983). As
adults, males may re-enter the female areas, wherein they may
employ one of two non-mutually exclusive reproductive
strategies. The most conspicuous of these two strategies is
that of males that defend territories. Territorial males may
copulate in serial fashion with females that enter their
territories (Vandevere 1970; Calkins & Lent 1975; Loughlin
1977; Garshelis & Garshelis 1984; Garshelis, Johnson &
Garshelis 1984; but see Kenyon 1969). Other males, or
possibly the same males on _ other occasions, may
opportunistically search for and attempt to pair and/or
copulate with females (Kenyon 1969). Reproductive activities
normally are concentrated during the fall but some males may
remain on their territories year-around. Others return to the
male areas (Garshelis, Johnson & Garshelis 1984) where they
rejoin the male aggregations and remain until the following
breeding season.

Less is known about the movement patterns or distribution
of females that reside within the female areas. Short-term
studies of individuals, using radio-telemetry, suggest that
females are somewhat less mobile and less gregarious than
their male counterparts (e.g. Garshelis, Johnson & Garshelis
1984). However, available data have been inadequate to
indicate how the female areas are used by individuals or to
permit evaluation of variation in females' movement patterns
associated with seasonal or functional (e.g. breeding,
pupping, wintering, etc.) needs. Also, little information has

134

been available on the movements of weanlings and on the manner
in which they become established within their respective male
and female areas.

This chapter describes the movement and habitat use
patterns of mature female and immature male and female sea
otters in Prince William Sound, Alaska. The movements of
adult females are examined at different stages of the
reproductive cycle. The movement patterns of weanling males
and weanling females are contrasted and discussed in the
context of the evolution of dispersal patterns. The
relationships between the observed movement patterns and the
sea otter's social system are considered.

STUDY AREA AND METHODS

The study was located in the eastern portion of Prince
William Sound, in south-central Alaska (Figure 8.1). During
the past decade, sea otters have recolonized the deep bays,
mud flats and channels that are located to the west of the
fishing community of Cordova. The local population of sea
otters, its history and its habitat have been described by
various authors (Gabkinsycs bent, 19757 sSinith, (Williams!
Johnson & Garshelis 1982; Garshelis & Siniff 1983; Garshelis
1983; Garshelis & Garshelis 1984; Garshelis, Johnson &
Garshelis 1984; Garshelis, Garshelis & Kimker 1986).

Data were collected during 18 months between June, 1984
and October, 1986. The subjects included 8 adult females and
35 pups (and indirectly their mothers) from 2 cohorts; 14
during 1984 and 21 during 1985. All otters were captured in
Sheep Bay or Simpson Bay. Adults and a few dependent pups
were captured in floating tangle nets (91 m long by 5 m deep
with a 22 cm stretch mesh) during June, August or September
1984. Most dependent pups were captured in dip nets during
August or September, 1984 or 1985. Pups ranged in size from
8 - 20 kg and all still accompanied their mothers. Upon
capture, otters were brought aboard a 5.5 m skiff and
immobilized with a combination of fentanyl (0.05 mg/kg) and
azaperone (0.20 mg/kg) (Williams, Williams & Siniff 1981).
Each otter was weighed and its sex was recorded. One or more
nylon tags were inserted through the interdigital webbing of
one, or both, hind flippers for identification. A
radio-transmitter was surgically implanted in each animal's
peritoneal cavity by a veterinarian, as described in Chapter
abe

Radio-instrumented sea otters were monitored during
daylight in August-October and December 1984; April-December
1985; February, May-June and October 1986. Visual
observations were made from a skiff or from the shore with
binoculars or 50-80X telescopes (Questar Corp., New Hope, PA,
USA 18938). Instrumented otters were normally monitored from

135

1984-1987.

ALASKA
S

PRINCE WILLIAM
SOUND
? fi stupy area
I }
| 7 Dafa, VALDEZ

‘fe &

+ ae
’ rs fii . is WE

——— 4
| = ©
—— 5 yi

ETrra, SP
%s,
Ee)

a skiff that was traveling at 20-30 knots. fThe skiff was
equipped with 2 yagi antennas mounted on 4-m aluminum masts.
Antennas were attached at 60 and 300 degree angles from the
plane of the boat. Periodically, (approximately 250 total
flight hours), instrumented sea otters were monitored from
small aircraft that were equipped with 4-element yagi antennas
mounted under each wing (Gilmer et al., 1981). Aircraft
speed was set at about 100 knots and preferred altitude was
600-750 mn.

Radio-transmitter frequencies were scanned on a 2000
channel programmable scanning receiver (Cedar Creek Lab).
Radio fixes were determined by triangulation or by moving the
boat in the direction of the radio signal until the individual
was observed. Otter locations (fixes) were recorded either
as coordinates of the Universal Transverse Mercator Grid
System, or marked directly on large scale maps or tracings of
the various bays and channels. The latter were used
predominantly during aerial surveys. Distances were measured
on U.S.G.S. scale 1:250,000 or 1:63,360 contour maps.

As Garshelis & Garshelis (1984) pointed out, the annual
home range of Prince William Sound sea otters is composed of
numerous centers of activity connected by long travel
corridors. The area of any portion of the annual home range,
or rather, any cluster of fixes, can be estimated by measuring
the area of the minimum convex polygon enclosing the fixes
(Odum & Kuenzler 1955; Garshelis & Garshelis 1984). In Prince
William Sound, sea otter travel corridors often cross, and
enclose, deep, broad, and presumably, inhospitable expanses
of water. As a consequence, the same procedure, when applied
to estimation of annual home ranges, drastically over
estimates areas. The large number of fixes required for
characterization of such habitat utilization patterns, at
least 40 per activity center (Garshelis & Garshelis 1984),
makes an accurate measurement of annual, or longer-term, home
range impractical. Garshelis & Garshelis (1984) suggested an

index of home range: "distance between extreme locations"
(DBEL). Herein, it is used to describe the magnitude of the
movements of individuals. The distance between extreme

locations is the minimum distance an otter would have to swim
to go between its two most widely spaced fixes during some
time interval. It is approximately equivalent to the maximum
dimension of the home range (Garshelis & Garshelis 1984).

RESULTS

Adult female home ranges

Eight adult females were implanted and monitored for
periods ranging from 15-20 months. All eight females survived
the duration of the study. Four gave birth to pups. All

137

study females traveled extensively throughout the eastern
Prince William Sound. The median distance between extreme
locations of the eight females was 41 km (range 27 - 85 km)
(Figure 8.2).

Some females made long, circuitous trips which crossed
major bodies of water. During the summer of 1985, one female
(84001) traveled beyond the limits of the area that was
routinely monitored (Figure 8.3). Contact was lost in
mid-May and reestablished on October 31. On that date she was
near Green Island. She had returned to Sheep Bay by November
7. The short time interval between sightings, at two distant
locations, suggests that Hinchenbrook Entrance was traversed.
Hinchenbrook Entrance is a channel that spans 11.5 km at its
narrowest and is over 300 m deep at its shallowest crossing.
It has rapid tidal currents and intemperate conditions. The
only alternative to crossing that channel, or other comparably
deep, broad channels would have been for female 84001 to have
circumnavigated Prince William Sound, a minimum trip in excess
of 200 kn. All study females traveled between major bays.
Most traveled across large expanses of deep water. However,
if other study females journeyed beyond the limits of the
regularly monitored portions of the study area, they must have
done so for only brief periods, since all were located
regularly with no comparable periods of lost contact.

The study area was divided into 12 habitat zones in order
to illustrate the movements of the eight study females
(Figure 8.4). All of the females traveled in four, or more,
zones (range 4 - 9) during the time they were monitored
(Figure 8.5).

Females tended to occupy the western portions of the
study area during the late spring and summer but to travel to
the easternmost area where they spent the late fall and winter
(Figure 8.6). The eight radio-implanted females used
superzone A heavily during May, June and July. At that time
most females were pupping or tending small pups (Chapter 9).
Females were aggregated into rafts (often containing over 100
individuals) in shallow, protected coves and over shoals.
During the fall, many females moved into the bays on the north
side of the study area (superzone B). They formed less dense
aggregations, weaned their pups and presumably, in some cases,
mated with resident males. As winter approached, females
became rare in superzone A. This may have been in response
to winter storms which often batter those coastlines from the
east or northeast. During the late fall and winter, females
became abundant in zones 5, 8, 10 and the western portion of
zone 9.

138

ADULT FEMALES

84001 V7 31/604
soos 93/546

84005 50/486
soos, 301848

4007 | re

saoiol WJ“, 80542
sis} 35481

asoi7 J 4788

i 20. Ao so eo TOI co
DISTANCE BETWEEN
EXTREME LOCATIONS (km)

FIGURE 8.3 -- Movements of an adult female sea otter in
Prince William Sound, Alaska, during a 20 month interval,
June 1984 - February 1986. Summers were spent in the western
portion of the study area and winters in the eastern portion,
near the Cordova male area.

31 LOCATIONS ya
v7
7
7
7
?
?
Y
? ae
ye ee)
7 ge
a a
7 ys
Vie
7 oO
12

GREEN We
ISLAND
S cb

3

7) se oo 10 20
_———————S ee |

Kilometers

140

FIGURE 8.4 -- Division of study area in Prince William Sound,
Alaska, into numerically designated habitat zones and
superzones. Zones correspond to major bays or passages.

~~ SUPER ZONES
eS

141

FIGURE 8.5 -- Use of habitat zones in Prince William Sound,
Alaska, by eight radio-instrumented adult female sea otters.

1 Be oh SH ts Selva

cen | 7/7
4004 JV yy

84005 |e FF

84006 end

e007 | L777, LZ
84010) V77/77/7|—

aos | (7, V7

e417 | U7 | |
12345678 9101112
HABITAT ZONES

FIGURE 8.6 -- Seasonal changes in the use of portions of
eastern Prince William Sound by eight radio-instrumented
adult female sea otters. Superzones are delineated on Figure
8.4.

100%
DEC 84
a fe)
100%
APRIL 85
— O
100%
MAY 85 pee
0
100% ,,
z
100% i
JULY 85 as ee 5
0 2
100% &
AUG 85 i Zz
=
0 se
‘ 100% ©
0 O
100% 32
OCT 85 Pe wy.
9)
100% &
100%
DEC 85
| 0
100%
FEB 86
io... te)

AneaB ihe Cinti2)
SUPER ZONES

143

FIGURE 8.7 -- Distance between extreme locations of 26 female
sea otters in Prince William Sound, Alaska, that were
accompanied by dependent pups. Most observations are based
on females accompanying radio-instrumented pups.

FEMALE PUP PAIRS
p>

"o 10 20 30 40 50 60
DISTANCE BETWEEN
EXTREME LOCATIONS (km)

144

Females with pups

Females traveled extensively while they were accompanied

by a pup (Figure 8.7). The median distance between extreme
locations for females with pups born during 1984 (m = 17.25
km, range = 6.5 - 38.5, n = 10) was shorter than that of

females with pups born during 1985 (m = 33 km, range = 15 -
6275),-ene =" ay) This apparent difference probably resulted
from the fact that there are only limited data available on
the movements of several of the individuals from 1984.
Monitoring was discontinued between late September - mid
December, 1984. Consequently, fewer telemetry fixes were
available for assessing individual movements during that
period than in 1985.

Females apparently only occupied a portion of their
annual home range while they were accompanied by pups. Based
on data collected on radio- implanted females during 1985, the
DBEL of females during the time interval when they were
accompanied by pups were smaller than the annual DBEL of the
radio-implanted adult females (m = 33 km vs. m = 41 kn,
Mann-Whitney U test, p < .02). This was probably because the
formers' home ranges did not include the wintering areas in
the eastern portion of the Sound. Accompanied females
confined their movements to trips between the western nursery
areas and the north-central bays where most weaning took
place. Weaning occurred before movement into the wintering
areas.

Movements during the last month before weaning.--During
the last month before a pup was weaned, the female and pup
usually occupied a relatively small, shallow cove or channel.
It can be inferred that the pair's home range was smaller,
since the distance between extreme locations was shorter than
it was during the earlier portion of the dependency period
(Figure 8.8). Data for both sexes of pups are pooled for
analysis, since data for male and females are similar (last
30 days: t = -0.17, N.S.; earlier interval t = -0.91, N.S.).

-Potentially, there are two ways that the observed
differences in home range size could be an artifact of
sampling design. The first relates to the relative sample
sizes, the second to differences in timing between the
respective samples.

First, it has previously been shown that home range area
is correlated with the number of fixes analyzed when sample
sizes are small (under 40 fixes) (Garshelis & Garshelis 1984).
Those authors also found estimated home range area to be
correlated with monitoring interval. Thus, if the number of
fixes and monitoring intervals were not about the same in the
"last 30 days" and "earlier" samples, any differences could

145

be an artifact. Two arguments can be made against such a
problem distorting patterns in the data from this study. One
is that the sampling interval and the number of fixes were
similar in the two treatments described; sampling interval:
last 30 days, mean = 25.6 days (SD = 4.5), earlier, mean =
28.6 days (SD = 11.6); number of fixes: last 30 days, mean
= 8.9 (SD = 3.0), earlier, mean = 10.2 (SD = 4.0). The other
is that no correlation existed between extreme locations
within the pooled samples (r = -0.03, N.S.; r = .21, N.S.,
respectively).

Second, since most weaning took place in the late fall
(Chapter 9), one reasonable argument could be that shrinking
home ranges result from the tendency for adult sea otter
females to move into protected areas and to restrict their
movements at the onset of winter weather patterns. MThis does
not appear to be the case. Independent females continued to
travel, and thus, had large home ranges during the late fall
and winter. The distance between extreme locations was
longer for independent females, at that time, than it was
during the last 30 days before weaning for female-pup pairs
(m = 19.5 km, n = 8 vs. 7.5 km, n = 19, respectively;
Mann-Whitney U-test p < .02).

Sexual differences in adult home range size

A direct comparison can be made between the extent of
movements of adult males and adult females, within
northeastern Prince William Sound, by combining the results
of this study with those of the earlier studies of Garshelis
and Garshelis (1984, p. 674, Fig. 7). Those authors argued
that male home ranges were larger than female home ranges in
Prince William Sound. However, they pointed out that at
least part of the difference could have been due to females'
movements being constrained by geographic boundaries. That
is, female areas at Green Island, in central Prince William
Sound, were smaller than male areas in Nelson Bay, in
northeastern Prince William Sound. Thus, a comparison of
males and females in the same general area (i.e. within
northeastern Prince William Sound) should be a better test for
sex differences in movements.

As mentioned above, the relative number of fixes during
each study and the durations of the studies could affect the
results. Thus, an attempt was made to ensure that the data
from the two studies were directly comparable. Only data from
the July - Sept. interval, 1-3 months of monitoring, are
considered. However, data from the latter study includes that
on both independent females and females with pups.

To test for differences in movements of adults, the
proportions of males and females in 2 distance categories

146

FIGURE 8.8 -- Changes in the home ranges of sea otter female-
pup pairs in Prince William Sound, Alaska, that occur as the
pups approach weaning age. The distances between extreme
locations of pairs are compared for the last 30 days before
weaning and for the earlier period when the pup was younger.

ener" yyy
“... Wy
en ...... AA
Se LLL
eS Yj:
a GUM
MH LAST 30. DAYS
aE al LE
ZZ, t-6.05 , p<.001
A WY
S BMEIZIZZZZZ
ee = ede

OM oes Ores Seere0 seo 30). G0. 40 40

DISTANCE BETWEEN
EXTREME LOCATIONS (km)

147

were compared: individuals with DBEL < 15 km and those with
DBEL > 15 km. The DBEL of adult females were larger than
those of adult males (respective ranges 15 -60 km vs. 4.8 -
37 km; Chi square = 14.31, 1 DF, p < .001).

The sea otter population of Prince William Sound is still
increasing after near extirpation by fur traders during the
end of the 18th century (Lensink 1962). The history of the
local remnant population is fairly well documented. Simpson
Bay, in zone 4, and superzone C went through a transition from
being a male area to being a female area in the early 1980's
(Garshelis & Garshelis 1984; Garshelis, Garshelis & Kimker
1986; this study). By 1986 the male area was entirely
contained within superzone D. The eastern Prince William
Sound contained but a single, well consolidated male area
which was surrounded by female areas. At its eastern edge,
in zone 10, the population was dominated by females. That
area was heavily used by females, some of which were
accompanied by dependent pups, during the winter and spring
1984-1987.

Movements of weanlings

Most pups were born during May and weaned during the
subsequent fall (Chapter 9). Two characteristic movement
patterns were exhibited by sea otters during their first year
following weaning. Some weanlings stayed within, or very
close to, the home range they had occupied during the month
preceding weaning. However, most weanlings immediately made
a relatively large movement, then occupied a small home range
until spring. During spring, they expanded their home range.
Their movements within the extended home range took them still
further from their site of weaning.

By the end of monitoring (maximum 21 mo.) many of the
weanlings had traveled far from the site at which they
separated from their mothers (Figure 8.9). Males had moved
longer distances than their female counterparts. Mortality
was high (Chapter 9). Many of the weanlings' trips had
culminated with their death. The next few sections deal with
sea otter behavior during their first year of independence.

If weanlings moved significantly from their site of
weaning, they almost always began their travels abruptly,
within 2 weeks after weaning. For example, 14 weanlings in
the 1985 cohort traveled 20 km or more. Within their first
two weeks of independence, 13 of the 14 had moved at least 20
km from their weaning location. The 14th weanling also moved
abruptly but did not do so until about two months had passed.
Most weanlings departed almost immediately following
separation. Since departures were abrupt and movements tended
to be fairly long, contact was usually lost until a search

148

FIGURE 8.9 -- Distances traveled from the site of weaning in
Prince William Sound, Alaska, -by male and female weanling sea
otters. Monitoring interval varied from a few days to
approximately 18 months. Short monitoring intervals resulted
when pups died during their travels.

BB Mates

AW FEMALES
MANN-WHITNEY

P<.002

WEANLINGS

0 10 20 30 40 50 60 70 ,80 90 100 110 120
DISTANCE (km) FROM WEANING LOCATION

could be made from an aircraft. By the time that the first
post-weaning radio fix was taken, the weanling's travels were
usually completed.

Most traveled at least 20 km from where they were weaned
to their post-weaning home range. During their first trip
males tended to travel further than females (Figure 8.10).
Weanlings usually completed their first trip quickly. Data
on three weanlings that were weaned during the first week of
November, 1985, illustrate this point. Two of these weanlings
were males that were weaned on, or at most, a few days before
11/4 and 11/6, respectively. The third was a female that was
weaned during the night of 11/6. All departed from their
pre-weaning home ranges in Sheep Bay or Simpson Bay on the
night rote 1 7/6). Those bays were searched thoroughly from a
skiff on 11/7. None of the three were found within the 10 km
search radius. An aerial survey was flown on 11/9. The
female was found 38 km to the southwest on the far side of
Orca Bay. The two males were not located on 11/9. At that
time they were not within 50 km of their weaning location.
The search area was expanded during a second aerial survey on
11/16. One male was found near Valdez, at a distance of 123
km. The other male had traveled about 80 km along the same
coastline. On the next aerial survey of that area, a few
weeks later, he was also near Valdez, 109 km from the place
where he was weaned.

The extent of weanlings' movements that defined the
post-weaning home ranges appeared to vary substantially
between individuals. The distance between extreme locations
was greater for males than females, however, not significantly
(Figure 8.11).

The process of segregation of sea otters into male and
female areas figures prominently in early sea otter behavior.
When weanling males travel to their first post-weaning home
ranges they usually leave the female areas in which they were
reared (Figure 8.12). Conversely, young females usually do
not leave. In this study, three young females had home ranges
outside the female area during their first winter. Two of
these females survived until spring, at which time both
returned to the female area. Only two of twelve males known
to have survived their first winter were not within the male
area.

DISCUSSION

The data given in this paper indicate that the extent of
movement by sea otters in eastern Prince William Sound varies
with age, sex and reproductive status. Relatedly, otters use
specific portions of their habitat for different purposes.
Thus, densities within a given area can change dramatically

150

FIGURE 8.10 -- Distance between weaning of sea otters
location in Prince William Sound, Alaska, and their first
post-weaning home range. The distance was traveled in a
single relatively rapid trip.

BB mates
FEMALES

MANN-WHITNEY

P<.009

WEANLINGS
(oe)

7 N

0 10 20 30 40 50 60 70 80 90 100 110 120
DISTANCE (km) FROM WEANING LOCATION

151

FIGURE 8.11 -- Relative size of weanling male and female sea
otter home ranges in Prince William Sound, Alaska, during the
first winter following weaning. Only weanlings with well
defined home ranges are included.

MALES

MANN-WHITNEY
P > 0.05
N.S.

FEMALES

A 10 20 30 40 50 60
DISTANCE BETWEEN
EXTREME LOCATIONS (km)

152

FIGURE 8.12 -- Tendency for weanling sea otters in Prince
William Sound, Alaska, to leave natal female area after being
weaned. Female weanlings usually do not leave natal female
area, whereas males usually do. Female area consists of
zones 1, 2, 3, 4, 6, 7, 8, and 11 on Figure 8.4.

xX? = 12.7, P<0.01

sy
q 12 INSIDE
=) 10) FEMALE
OQ g AREA
= 6
(a)
=>

0
le OUTSIDE
<q 12 FEMALE
=2)) 3100 AREA
QO ¢s
= 6
O 4
=i

0

MALES FEMALES

IL5)3)

over the course of a year. These data also show that sexual
segregation occurs very early in life as a result of
differences in the behavior of male and female weanlings.
Weanlings of both sexes were competent and capable of making
considerable movements as soon as they became independent.
These observations are discussed below.

Adult movements

The observed movement patterns of the adult female sea
otters in Prince William Sound were remarkable for several
reasons. Overall, adult females were more mobile than had
been anticipated. Individual females used a considerable
portion of the total female area. However, both the extent
of their movements and their destinations changed seasonally
and with the age of their dependent offspring. Adult females
became quite sedentary when accompanied by large dependent
pups that were close to the age of weaning. They were also
more gregarious and exhibited more pronounced seasonal habitat
use patterns than suggested by previous investigations.

The median distances between extreme locations (DBEL) of
independent females (41 km) and females with dependent pups
(33 km) were longer than those measured in previous studies.
Kenyon (1969) suggested that the home range of females usually
included less than 17 km of coastline. Garshelis and
Garshelis (1984) studied independent females at Green Island,
which is located about 80 km southwest of the current study
area. They measured DBEL that ranged from 2.6 - 15 km. The
early studies in California reported female home ranges that
were comparatively smaller than those observed by Garshelis
and Garshelis (Loughlin 1977; Ribic 1982). None of those
studies reported female movements in excess of about 17 kn.
However, the results of ongoing studies at other locations in
Alaska and California (Monnett, 1987) (Chapter 3) suggest that
DBEL as long as those observed in this study may be common.

Garshelis and Garshelis (1984) recognized that females
were leaving their study area and that their data did not
necessarily reflect annual or lifetime home ranges. Ie ahs}
possible that some of the studies that have reported
relatively sedentary behavior by females have obtained that
result as a function of the study design. If individuals are
monitored for periods of a few months or less, large distance
movements between functional sub-habitats might not be
identified (e.g. breeding vs. wintering area). Seasonal
movements have also been suggested for the population that
inhabits the Bering Sea along the Alaska Peninsula (Lensink
1962; Cimberg & Costa 1975; but see Monnett 1987b).

Based on a limited number of observations at Green
Island, Garshelis & Garshelis (1984) concluded that females

154

occupied larger home ranges when accompanied by young pups
than they did when their pups were nearly ready for weaning.
They posited that the change in behavior, as the pups became
older, was necessitated by the need for pups to gain
experience in self feeding. Such experience could only be
gained if movements were restricted to a limited portion of
the habitat where water depths were shallow and suitable prey
were available. Their hypothesis is consistent with the
findings of this study. However, there are no data from this
study on pup feeding behavior in weaning, or other, areas.

It is likely that real variation in the movement patterns
of adult females exists between the populations of sea otters
at separate study areas due to differences in population
status, geography or genotypes.

In the eastern portion of Prince William Sound, range
reoccupation is still occurring, with subsequent changes in
distribution. The Green Island area, where Garshelis and
Garshelis (1984) conducted most of their work on female
movements, has been occupied much longer and, hence,
distributions per se are probably not changing because of
recolonization.

In areas where coastlines are complex, the extent of
movements could also be constrained by local geography
(Garshelis & Garshelis 1984). Relatedly, individuals that
inhabit calm waters such as Prince William Sound might find
travel less physically demanding and/or less risky than
individuals that inhabit more exposed waters such as the Gulf
of Alaska or the Pacific Ocean off California. For example,
females with pups might find travel difficult if they inhabit
waters along unprotected coastlines and, thus, restrict their
movements to only the most protected areas (D.B. Siniff
personal communication).

Authors have not agreed on the relative extent of
movements of males and females. Kenyon (1969) and Garshelis
and Garshelis (1984) have argued that males have larger home
ranges, whereas Loughlin (1977) and Ribic (1982) reported
larger home ranges for females. The significance of these
reported differences should be viewed conservatively. It is
likely that some comparisons reflect differences in movements
that are contained within single male areas or female areas
and exclude movements between major habitat units that are
of functional significance (e.g. the criticism by Garshelis
and Garshelis (1984) of Ribic's (1982) result). It is well
documented that individuals of both sexes can make very long
movements (> 80 km) between breeding areas and wintering areas
(reviewed in Garshelis & Garshelis 1984; this study).
Conversely, individuals of both sexes can be relatively
sedentary (e.g. territorial males (Garshelis & Garshelis 1984)

155

vs. females with dependent pups just before weaning (this
study)). Mobile and sedentary periods may not coincide in
males and females. As a consequence, measures of home ranges
are likely to be biased if studies are short term and do not
take into account sex-specific seasonal differences in habitat
use patterns. General conclusions about sex differences are
premature until longer-term data on both sexes are available.

From observations at Green Island, Garshelis and
Garshelis (1984) concluded that females formed considerably
smaller groups than did males. Indeed, only relatively small
groups of females (< 50) were encountered in the vicinity of
Green Island during field work conducted during 1985 and 1986
(personal observation). However, recent observations in other
parts of Prince William Sound and elsewhere suggest that this
conclusion may not be general. Females frequently form large
aggregations, well in excess of 100 individuals. Such
aggregations have been seen in Prince William Sound (this
study), in the Bering Sea along the Alaska Peninsula (Monnett
1987) and in the vicinity of Kodiak Island, Alaska (A. DeGange
personal communication). As Garshelis and Garshelis (1984)
point out, group size is likely to be related to the type of
sea otter activity and local density.

Weanling Movements

Observations during this study indicated that weanlings
established home ranges in their respective male and female
areas shortly after separation from their mothers. Most
weanlings moved a long distance from their site of weaning
before doing so. Since pups were weaned in female areas,
female weanlings were not required to travel as far as males.
Garshelis, et al. (1984) observed a similar rapid departure
of male weanlings from natal female areas. However, they also
noted that during May-August the age structure of males in
male areas had an excess of 2 and 3 year old males (by two to
three fold) compared to yearling males. Because of this age
structure they suggested that some weanling males may delay
entering male areas until after their first year.
Observations on radio-instrumented pups in this study do not
support that contention. Another way to interpret the age
structure observation is that they are observing variation
in cohort recruitment. Two observations support this latter
alternative. First, there was significant variation in the
sex ratio of dependent pups caught in the same area between
1984-1986 (Chapter 9). Yearly differences in male birth
rates, or survival rates, could result in corresponding cohort
differences in the male rafts. Moreover, the delayed entry
hypothesis does not explain the relative lack of 4-5 year old
males in the male area observed by Garshelis, et al. (1984).
Four and five year old males are generally assumed to be
immature and should have been present in the male area in

156

their actual proportions. The relative scarcity of 4 and 5
year old males is consistent with the notion of varying cohort
sizes.

Evidence from other investigations also suggests that
weanlings in Alaska segregate into respective male and female
areas relatively soon after weaning. Kenyon (1969) reported
more juvenile females than juvenile males (53:31) in female
areas near Amchitka, in the western Aleutian Islands.
Lensink (1962) examined dead juveniles at male hauling grounds
and found the sex ratios to favor males (30:6). This ratio
seems too extreme to be due to the easier identification of
male carcasses in poor condition (see Chapter 6).

The recently independent otters in this’ study,
particularly males, apparently were capable of traveling long
distances in a fairly short period of time. Young male sea
otters have been observed to make similar long distance
movements to male areas in previous studies. Garshelis, et
al., (1984) observed a male weanling that left a female area
and traveled over 100 km to a male area in Prince William
Sound.

Garshelis, et al. (1984) suggested that young male sea
otters may move into the aggregations found in male areas
because of certain benefits that may be derived from
gregariousness. Those authors suggested that benefits may be
derived from social facilitation, opportunities for assessment
of conspecific competitors, safety from predation and
metabolic advantages gained from hauling out on sandbars that
exist in those areas. Data on survival rates of weanlings
from this study supports the hypothesis that movement from
the natal area to a male area may be generally beneficial.
Male weanlings, that spent their first winter following
weaning in the male area, were more likely to survive than
those that remained in the female areas (Monnett 1987).

The relationship between movement data and dispersal

For this discussion we follow the terminology of
Greenwood (1980:1141) who distinguished between several common
usages of the term dispersal. He defined natal dispersal as

that dispersal "...from birth site to first breeding or
potential breeding site...". He contrasted natal dispersal
with breeding dispersal: "...movement of individuals, which
have reproduced, between successive breeding sites..." and

effective dispersal: natal or breeding dispersal followed by
successful breeding. These definitions are not universally
accepted. Rather, the terminology of dispersal has been used
inconsistently between authors (e.g. Greenwood 1980 cf.
Lidicker 1975; Gaines and McClenaghan 1980; Dobson 1982).

157

The movement data given herein should not be taken as
accurate measurements of natal, effective or breeding
dispersal as defined by Greenwood (1980). That is, measures
are incomplete because data are not given on the distribution
of mature individuals or their breeding sites. These data are
of post-weaning movement patterns.

From observations made during this study, it does appear
that young females are more conservative in their dispersal

movements than males. Several observations support this
contention. First, males traveled further than females during
the weeks immediately following weaning. The very long

movements were all made by males. Moreover, the sex ratio of
the individuals with which contact was lost strongly favored
males. That suggests that some males may have left the
monitored region. Second, males usually left the female area
of birth when weaned, whereas females rarely did. Third,
juvenile males remained outside the female area as long as
they were monitored. Conversely, those females that left the
female area following weaning, and that survived the winter,
returned to the female area in the spring. Fourth, as
juveniles, most males continued to travel further from their
weaning location. Females restricted their travels to sites
within the female areas.

The observations made during this study do suggest that
future interpretations of the relationship between breeding
sites and natal sites in sea otter populations are likely to
be problematic. In order to evaluate an individual's
dispersal, a starting point has to be determined, as well as
an ending point. Howard (1960) referred to this starting point
as "...its point of origin..." which Greenwood (1980) took to
mean its place, or group, of birth. There is little
confusion when the young remain at a single location
throughout dependency (e.g. nests, dens or burrows) or when
they remain within a cohesive social group (e.g. prides, packs

or troops). However, as shown in this study, sea otter
females travel extensively while accompanied by dependent
pups. Consequently, the sites of conception, birth and
weaning may not coincide. Any, or all, of those locations

could be regarded as the point of origin, depending upon the
question under consideration (e.g. inbreeding avoidance vs.
site experience).

Available evidence seems to indicate that females stay
within their natal female areas, near the areas they inhabited
when they accompanied their mothers. However, many leave the
immediate location where they were weaned. Conversely,
juvenile male departure from the same areas appears to be
almost obligate. Only two males out of the 12 that survived
their first winter had not left the female area by January.
In order to exhibit natal philopatry equal to that exhibited

158

by female weanlings, males would have to re-enter natal female
areas after becoming sexually mature and breed in areas they
inhabited with their mothers. However, from the perspective
of the juvenile male, any attraction to the natal female area
that might result as a consequence of benefits derived from
site familiarity or individual recognition (Greenwood 1980)
are likely to have been diminished by the length of tenure
in the male area. Male sea otters mature and apparently begin
to seek breeding opportunities after five or more years of
residence in the male area. It would seem questionable
whether what was learned about the natal female area during
dependency could be retained until the age of potential
re-entry. Moreover, even if retained, some of that knowledge
would be likely to be obsolete because of changes in habitat
conditions during that period (e.g. food abundance,
conspecific distributions).

LITERATURE CITED

Cimberg, R.L. and Costa, D.P. (1985). North Aleutian Shelf
sea otters and their vulnerability to oil. Proceedings:
Oil spill conference, Los Angeles, CA.

Calkins, D.G., and Lent., P.G. (1975). Territoriality and
mating behavior in Prince William Sound sea otters.

Journal of Mammalogy, 26, 528-529.

Dobson, F.S. (1982). Competition for mates and predominant
juvenile male dispersal in mammals. Animal Behaviour,
30, 1183-1192.

Garshelis, D.L. (1983). Ecology of sea otters in Prince
William Sound, Alaska. Unpublished Ph.D. thesis,
University of MInnesota, Minneapolis, MN. 330 pp.

Garshelis, D.L., and Siniff, D.B. (1983). Evaluation of
radio-transmitter attachments for sea otters. Wildlife
Society Bulletin, 11,378-383.

Garshelis, D.L., and Garshelis, J.A. (1984). Movements and
management of sea otters in Alaska. Journal of Wildlife
Management, 48,665-678.

Garshelis, D.L., Johnson, A.M., and Garshelis, J.A. (1984).
Social organization of sea otters in Prince William

Sound, Alaska. Canadian Journal of Zoology,
62,2648-2658.

Garshelis, D.L., Garshelis, J.A., and Kimker, A.T. (1986).

Sea otter time budgets and prey relationships in Alaska.
Journal of Wildlife Management, 50,637-647.

159

Gilmer, D.S., Cowardin, L.M., Duval, R.L., Mechlin, L.M.,
Schaiffer, C.W., and Kuechle, V.B. (1981). Procedures
for the use of aircraft in wildlife biotelemetry studies.
United States Department of the Interior, Fish and
Wildlife Service Research Publication 140. Washington,
D.C. 19 pp.

Green, B.D. (1978). Sexual maturity and senescence of the
male California sea otter (Enhydra lutris). Unpublished
M.A. Thesis, San Jose State University, San Jose, CA.

Greenwood, P.J. (1980). Mating systems, philopatry and
dispersal in birds and mammals. Animal Behaviour,
28,1140-1162.

Greenwood, P.J., and Harvey, P.H. (1982). The natal and
breeding dispersal of birds. Annual Review of Ecology
and Systematics, 13,1-21.

Howard, W.E. (1960). Innate and environmental dispersal of
individual vertebrates. American Midland Naturalist,
63,152-163.

Kenyon, K.W. (1969). The sea otter in the eastern Pacific
Ocean. United States Fish and Wildlife Service. North
American Fauna, No. 68, 352 pp.

Lensink, C.J. (1962). The history and status of sea otters
in Alaska. Unpublished Ph.D. Thesis, Purdue University,
Lafayette, IN 186 pp.

Lidicker, W.Z. Jr. (1975). The role of dispersal in the
demography of small mammals. In: Small Mammals:
Productivity and Dynamics of Populations. (Ed. by K.
Petrusewicz, E.B. Golley, and L. Ryszkowski), pp.
103-128. Cambridge University Press, London.

Loughlin, T.R. (1977). Activity patterns, habitat
partitioning, and grooming behavior of the sea _ otter,
(Enhydra _lutris), in California. Unpublished Ph.D.
thesis, University of California, Los Angeles, CA. 110
Pp.

Monnett, Charles W. 1987. Movement, development and
mortality patterns of sea otters in Alaska. Ph.D.
thesis, University of Minnesota, Minneapolis, MN. 141
PPp-

Odum, E.P., and Kuenzler, E.J. (1965). Measurements of
territory and home range size in birds. Auk, 72,128-137.

160

Peterson, R.S., and Odemar, M.W. (1969). Population growth
of the sea otter in California: results of aerial census
and behavioral studies. Proceedings Annual Conference
on Sonar Diving Mammals, 6,69-72.

Ribic, C.A. (1982). Autumn movement and home range of sea
otters in California. Journal of Wildlife Management,
45,795-801.

Riedman, M.L. (1986). Draft environmental impact statement

of proposed translocation of southern sea otters. Volume
II: Technical support documents, United States Fish and
Wildlife Service and University of California, Santa
Cruz, CA.

Schneider, K.B. (1978). Sex and age segregation of sea
otters. Alaska Department of Fish and Game, Final
Report, Federal Aid Wildlife Restoration Projects W-17-4
to W-17-8.

Siniff, D.B., Williams, T.D., Johnson, A.M., and Garshelis,
D.L. (1982). Experiments on the response of sea otters
Enhydra lutris to oil contamination. Biological
Conservation, 2,261-272.

Vandevere, J.E. (1970). Reproduction in the southern sea

otter. Proceedings of the Annual Conference on the
Biology of Sonar Diving Mammals, 7,221-227.

Williams, T.D., Williams, A.L., and Siniff, D.B. (1981).
Fentanyl and azaperone produced neuroleptananalgesia in
the sea otter (Enhydra _ lutris). Journal of Wildlife

Diseases, 17,337-342.

161

CHAPTER 9

SEX-RELATED PATTERNS IN THE POST-NATAL DEVELOPMENT AND

SURVIVAL OF SEA OTTERS IN PRINCE WILLIAM SOUND, ALASKA

C. MONNETT AND L. ROTTERMAN

NOVEMBER 30, 1988

INTRODUCTION

Pronounced sex differences in morphology, development
(Glucksman, 1974, 1978), and behavior (Clutton-Brock and
Albon, 1982) exist between the males and females of many large
mammals. While the general theoretical explanations for these
differences are fairly well developed, there is still little
known about the proximate causes or consequences of these
differences for most species (Clutton-Brock and Albon, 1982).

According to current evolutionary theory, parental
investment (PI) (Trivers, 1972) should be distributed among
progeny so as to maximize parental inclusive fitness (sensu
Hamilton, 1964). Disproportionate allocation of PI to male
or female offspring could occur in situations where the
production of one sex has a potentially greater effect on
parental inclusive fitness than the production of the other
sex. In polygynous mammals, if an inequality is to occur,
males should be the favored sex for two reasons: intrasexual
competition and male-biased dispersal.

Among polygynous mammals, variation in reproductive
success is likely to be greater among males than among females
(Trivers, 1972; Clutton-Brock, et al., 1977; Clutton-Brock and
Albon, 1982). The allocation of parental resources should
favorably influence an offspring's body condition as an adult
and thus, its relative ability in intrasexual competition with
conspecifics. As a result, if their reproductive success is
affected by their competitive ability, individual offspring
that receive the most PI during dependency should achieve the
highest reproductive success as adults (Trivers and Willard,
1973). Since male mammals generally compete, either directly
or indirectly, for access to females, high quality sons are
likely to leave more offspring than high quality daughters
(Trivers and Willard, 1973). Consequently, females should
tend to invest more resources in individual sons than in
individual daughters (Clutton-Brock and Albon, 1982).

The pattern of PI in individual male and female offspring
could also be influenced by the tendency for one sex to
disperse more than the other sex (e.g. Greenwood, 1980). If
survival during dispersal is influenced by maternal PI before
weaning, females should contribute more PI to progeny of the
dispersing sex (Clutton-Brock and Albon, 1982). In most
mammals those progeny would be males (reviewed in Greenwood
1980).

Given the typical mating system and dispersal patterns
seen in mammals, a unit of PI apparently has a greater
potential to effect the survival and the reproductive success
of male progeny than female progeny. Thus, more PI should
be allocated to dependent males relative to females. This

163

could be accomplished in 2 general ways: First, parents could
manipulate offspring sex at conception (Trivers and Willard,
1973). Second, during the period of dependency, parents could
preferentially allocate more PI to male offspring than to
female offspring (Reiter et al., 1978). For example, males
could be given high quality and/or greater quantities of food
or permitted to have longer dependency periods.

Sea otters (Enhydra lutris) are a sexually dimorphic,
polygynous mustelid that is highly specialized for the marine
environment (Kenyon, 1969). They have a resource-defense
mating system (Greenwood, 1980) with males occupying and
defending breeding territories (e.g., Calkins and Lent, 1975;
Loughlin, 1977; Garshelis and Garshelis, 1984). In Alaska,
males tend to leave the area in which they were reared at the
end of parental care, whereas females do not (Chapter 8). A
radio-telemetry study was conducted from 1984-1987 in order
to investigate differences in juvenile development and
behavior, with specific focus on differences between the
sexes. Comparisons are made of time of birth, growth rates,
dependency periods and mortality patterns. Under the
theoretical arguments reviewed above, male sea otters should
be born earlier in the spring, grow more rapidly and have
longer dependency periods. Differences in post-weaning
behavior between sexes, especially in movement patterns, might
subject juvenile males and females to different risks. Since
male sea otters travel farther from their weaning site, they
should be subject to greater risk and exhibit correspondingly
higher mortality during the first few months following
weaning.

STUDY AREA AND METHODS

Studies of sea otter pups and weanlings were carried out
in two general areas within Prince William Sound (PWS) in
south-central Alaska (Figure 9.1). Observations were made in
the northeastern portion of the Sound during 1984-1986.
Observations were made at Green island, in the south-central
sound, during the summer of 1985. Research activities were
coordinated from cabins in Sheep Bay and on Green Island,
within the Sound, and from a United States Fish and Wildlife
Service warehouse and University of Alaska marine advisory
office in Cordova. The local population of sea otters, its
history, habitat and ecology, have been described by a number
of authors (Calkins and Lent, 1975; Siniff, et al., 1982;
Garshelis and Siniff, 1983; Garshelis and Garshelis, 1984;
Garshelis, et al., 1984; Garshelis, et al., 1986).

164

William Sound, 1984-1987.

\
PRINCE WILLIAM
SOUND

[ill] stuby AREA

: ‘i

is i

Pup capture

Pups (N=157) were captured for routine tagging and data
collection with dip nets and tangle nets. Several types of
nets, including commercially available salmon dip nets, were
tested. The most satisfactory results, and the most pups,
were obtained by using a custom fabricated, aluminum dip net
(Alaska Power Services, Cordova, Ak 99574). This net was
characterized by a long (4m) handle and semicircular "basket"
that attached to the handle at a 90 degree angle. Mother-pup
pairs were pursued in a Boston Whaler skiff (5.5 M) until the
pair surfaced near the bow. The basket of the net was then
dropped in front of the moving animal(s) and drawn, with the
pup, up and back over the bow of the boat, the pup captured
in the net. Large pups were usually not carried by their
mothers and, since they were generally incompetent divers,
were easily netted. Smaller pups were carried by their
mothers as they made repeated dives. Successful captures
usually occurred in 3 circumstances: 1. The mother was caught
as she carried the pup, separated from the pup and released.
2. The pup was separated from the mother by drawing the net
between the pair, over the pup, as they surfaced for air. 3.
The mother released the pup after a few dives and it was
scooped as it floundered on the surface. Exceptions to the
above scenarios included a few mothers (less than 10%) that
abandoned their pups on the surface immediately on the
approach of the boat and a few cases when mothers were on
foraging dives when their pups were captured.

Pup handling

Upon capture, each pup was tagged with numbered, nylon
Temple or button tags (Ames, et al., 1983) in the interdigital
webbing of one, or both, hind flipper(s). The weight, length
and sex were recorded. Weights were taken in pounds because
equipment was available only with those _ scales. The
unevenness of some of the values assumed for calculations
reflects the conversion of those measures to metric scale.
Pups were weighed to the nearest 0.5 pound, with the exception
of the newborns that were weighed to the nearest 0.25 pound.
Pups were held for 5-15 minutes depending upon whether a
blood sample was taken.

Pup release

Pups were released into the water and generally observed
until they reunited with their mothers. If the mother was
not near the boat when a pup was to be released, cassette
playbacks of pup vocalizations (loud cries) were used as an
attractant. Mothers appeared not to discriminate between the
sounds of their own pup and those of others. The recorded
cries of a single pup were used effectively on many different

166

mother-pup pairs. Females were attracted to the boat with
these playbacks from distances of over 1/2 kn. Pups also
responded to the playbacks by swimming toward the boat and,
on some occasions, crying back. They were particularly
indiscriminate and often could be stimulated into lengthy
conversations by crude, human imitations of their own cry.

Recaptures

Forty-one dependent pups were recaptured on at least one
occasion. Previously caught pups were selected for capture
and identified by their flipper tags. They were captured and
handled as described above. Capture activities were usually
spaced to insure that intervals of 30-70 days had passed
between successive captures so that growth rates could be
determined.

Telemetry

Radio-transmitters were surgically implanted in the
peritoneal cavity of 37 dependent pups, during August or
September, 1984 or 1985, by veterinarians, as described in
Chapter 1. Most pups weighed 9-14 kg when implanted (range
7-20 kg).

Upon capture, pups were brought aboard a 5.5 m skiff,
weighed and immobilized with a combination of fentanyl
(generally 0.05 mg/kg) and azaperone (0.20 mg/kg) (Williams,
et al., 1981). Naloxone (0.01 mg/kg), an antagonist to
fentanyl, was injected in all subjects following surgery, but
before release. Pups were released into the water near their
capture site and generally were observed until they reunited
with their mothers. Normally less than 60 minutes elapsed
between an animal's capture and release. Playbacks of pup
cries, as described above, were used to keep mothers attentive
and near the boat during pups' surgeries.

Radio-implanted otters were monitored from small aircraft
or small boats equipped with yagi antennas and 2000 channel,
programmable scanning receivers (Cedar Creek Bioelectronics
Lab). Radios had ranges of 1-5 km and 6-10 km when monitored
from boats and aircraft, respectively. An attempt was made
to observe most pups 1-2 times per week during the fall,
before they became independent.

The transmitters had a maximum life expectancy of about
700 days. One-hundred twenty-five transmitters of the same
design were implanted in sea otters in Alaska and California,
between March, 1984, and September, 1986. To date, only two
are believed to have malfunctioned and many have operated 500
days or longer. The durability of the units is evidenced by
the recovery of 14 operating transmitters from intertidal

167

marine areas at various times following the deaths of the
subjects. Some were found buried under boulders on beaches
and had been subjected to heavy surf. One radio was still
operating after at least 13 months on a gravel beach, in the
intertidal, near Valdez.

Birth date estimates

Sea otter births are seldom, if ever, observed in
natural situations. Consequently, birth dates for pups must
be estimated. Such estimates can be based upon pup weight at
capture, if information is available about normal birth
weights and about pup growth rates (Wendell, et al., 1984).
If relevant data are not available, assumptions must be made
about birth weights and growth rates, based on population
averages.

For estimates given here, birth weight was assumed to be
approximately 4.5 lb (2.04 kg). This value was chosen based
on several types of information available from sea otters in
Alaska. First, 2 kg approximates the average birth weight
observed in this study. Second, based on his observations of
fetuses and newborns in a population near Amchitka, Kenyon
(1969) argued that "normal" birth weight is between 1.87 and
Doss Ikeja | Uslolalagols Schneider (1978b), used the same types of
data to estimate that the mean birth weight in the central
and western Aleutian Islands was 1.8 - 1.9 kg. Te S's
important to note that individual variation in birth weight
has little effect on the accuracy of birth date estimates
since growth rates are fast. An error of several hundred
grams would only change the estimated birth date by a few
days.

Based on data from this study, growth rates are assumed
to be approximately 95 g/day (0.21 1b) for males and 86 g/day
(0.19 lb) for females. No other growth rate data are
available for sea otters in Alaska.

Dependency period estimates

Dependency periods for the radio-implanted pups were
calculated from estimated birth dates and separation dates.
In order to do this, several assumptions were made. As noted
above, birth weight was assumed to be 2.04 kg. Actual growth
rates were used for those pups that had been recaptured and
hence, had been weighed on two or more occasions. If multiple
weights were not available, it was assumed that females gained
86 g/day and that males gained 95 g/day. In these cases,
estimated birth dates were bracketed by estimates made by
assuming growth rates of plus and minus 1 standard deviation
(SD).

168

Survival rates

Survival estimates, based on telemetry data, were
calculated using the method developed by Trent and Rongstad
(1974).

Separate survival estimates were calculated for males
and females during pre-weaning and post-weaning intervals.
Since the exact day of death was rarely known, two
survival-related calculations were made. inp tehe) | rales te
calculation, it was assumed that the animal died the last day
it was seen alive. In the second calculation, it was assumed
that it died the first day it was known to be dead. In some
instances individuals became missing but it was not certain
whether the cause was death, dispersal or radio failure.
Thus, the suggestion of Heisley and Fuller (1985) was followed
and two survival rates were calculated. In the first, it is
assumed that missing individuals were dead; in the second,
that they were alive.

One pup died shortly following surgery as a result of a
veterinary error. This pup is not included in estimates about
survival since it is unlikely that the error will be repeated,
and thus the case is not relevant to understanding normal
survival probabilities or factors influencing survival
schedules.

RESULTS

Birth weight

It was assumed that pups were newborn if pink umbilical
fragments were still attached when they were captured (Kenyon
1969). Three such newborn pups had weights and total body
lengths of 1.7 kg and 48 cm, 1.8 kg and 50 cm and 2.4 kg and
55 cm, respectively. Three other small pups were captured
that had no trace of umbilical fragments. These measured
1.9 kg and 48 cm, 2.5 kg and 57 cm and 2.7 kg and 55 cm.

Growth rates

Twenty-nine pups were recaptured after intervals of 34

days or longer (mean interval = 65.7 days, SD = 11.9, range
(34-98). Male growth rates (mean = 95 g/day, SD = 15 g,
range = 67 - 123, N = 18) were faster than female growth rates
(mean = 83 g/day, SD = 10 g, range = 63 - 88, N= 11) (Fig.

9.2; Mann-Whitney U-test: U = 149,49; N = 18,11; P .03).

Fourteen pups were recaptured after intervals of 11-28

days. As would be expected, rates were much more variable
than those observed for the longer intervals: males mean =
79 g/day, SD = 33 g, range = 41 - 132 g, N = 6; females mean

169

= 72 g/day, SD = 34 g, range = 39 - 132 g, N= 8. A _ female
and a male pup both achieved the maximum growth rate observed
in this study by gaining 3.2 kg in 24 days.

Small pups appeared to grow at approximately the same
rates as large pups (Table 9.1). The five smallest pups
gained on average 92 g/day, whereas, the 6 largest pups
averaged 93 g/day.

TABLE 9.1 - Comparison of growth rates for large vs. small
pups.
SEX WT. (kg) WT. (kg) INTERVAL GROWTH
ist CAPTURE 2nd _ CAPTURE DAYS RATE d
M 1.8 7.7 62 95
SMALL F 2.4 8.4 68 88
PUPS F 3.2 8.4 67 78
M 3.6 Atal dt 65 115
1E 3.6 9.8 ii, 86
MEANS 2.9 9.1 67 92
M BG) ta 58) 63 86
M 5.9 10.9 51 98
LARGE F 5.9 11.8 68 87
PUPS M 6.8 12.0 51 102
M 6.8 14.5 74 104
E 6.8 14.1 88 83
MEANS 6.4 12.4 66 93

Timing of parturition

The modal estimated birth date for Prince William Sound
pups was between May 20 and May 29 (Fig. 9.3). Male and
female pups were born in nearly constant proportions

throughout the spring. The rapid increase in births after
April 30, and decrease after June 28, is a conspicuous
characteristic of the distribution. It is widely accepted

that the timing of the seasonal peak in birth rates varies
seasonally throughout the sea otter's range, but that pups
are born in all seasons. The spring abundance of young sea
otters is readily observable in the Prince William Sound, and
elsewhere. However, since pups were caught only between early
June and late September, it could be argued that the apparent
rapid changes in birth rate were an artifact of the sampling
scheme. A perceived late April increase could result if large
pups, those born in April or earlier, were present in June,
but were too large to be captured in a dip net. In) fact;
capture success does tend to be lower for pups that are larger
than about 10 kg. However, relatively large pups were

170

captured. Seventy-three pups weighing at least 9.1 kg (20 lb)
and 10 pups of at least 13.6 kg (30 1b) were dip-netted. If
pups were commonly born in early April, they should have been
obvious and easily captured since they would have weighed only
8-10 kg by the end of June. However, at that time of year
pups of that size were seldom observed. The pups dip-netted
in June (N = 44) weighed on average 4.9 kg (10.9 1b) with the
largest being only 8.6 kg (19 lb).

Few pups were born during the mid and late summer. Given
the intensity of capture effort after August 28 (N = 50),
quite a few small pups should have been captured if they were
in the study area in the late summer and early fall. Data on
22 pups, dip-netted between 21 and 30 September, 1986, support
the contention that pups were rarely born after June in the
Prince William Sound. The average pup captured during this
interval weighed 10.4 kg (22.8 1b). The smallest pup captured
weighed 5.2 kg, a value heavier than the average weight of
June pups.

Incorrect assumptions about growth rates could cause
errors in calculations. Moreover, such errors would be
greatest for individuals that were not caught until they were
fairly large. In order to illustrate the potential magnitude
of such errors, we calculated the difference between the
estimated birth dates of individuals under two assumptions
about growth rates. That is, we made two separate
calculations: the first using a growth rate of + 1 standard
deviation (SD) and the second using - 1 SD (Table 9.2).
Since it was noted that pups caught before August 1 tended to
be smaller than those caught afterward, data were displayed
accordingly. For example, male pups born before August 1
weighed, on average, 5.7 kg. If a growth rate of 82 kg/day
(-1 SD) was assumed, and the real growth rate was 95 g/day,
the estimated age of one sixth of the male pups, when
captured, would have been at least 6.6 days greater than it
really was. Likewise, if the growth rate was assumed to be
109 g/day (+1 SD) the estimated age of one sixth of the males
would have been at least 4.8 days less than their real age.
Calculations on pups caught after August 1 tended to have a
greater error potential, since the pups were larger.
Conversely, calculations on females had less error potential
since females were smaller and had slower growth rates.

Progeny sex ratio

Sex was determined for 156 dependent pups (Table 9.3).
Total sex ratio favored males, but not significantly so.
Substantial differences in sample sizes and proportions
existed between years so a yearly average was also calculated.
This average favored females, slightly. Additional data on

171

TABLE 9.2 - Error in estimation of birth dates from growth
rate assumptions.

ASSUMED DEVIATION FROM

CAPT. AVE WT GROWTH RATE EST. AGE

DATE (kg) -SD AVE _+SD -SD +SD

EARLY 5.7 +6.6 -4.8
MALES 82 95 109

LATE 9.6 +13.2 -9.8

EARLY 5.6 +4.8 -4.5
FEMALES 77 86 95

LATE 9.0 +9.5 =7 od

22 dependent pups near False Pass, on the Alaska Peninsula,
included 14 males and 10 females. Thus, the observed total
sex ratio, for dependent pups from 2 populations was 95:85.

Dependency period

We assumed that pups were weaned at the time they become
separated from the mothers. This follows from _ the
observations of Schneider (1978b) who found that "...females
with even the largest pup were found to be lactating." and
Payne and Jameson (1984). The peak time of maternal
separation for 21 pups that were instrumented during 1985 was
October 16 - November 15 (Fig. 9.4). The study otters were
not monitored between December 15, 1985 and February 8, 1986.
The fact that three pups were weaned during this interval is
reflected on Fig. 9.4.

The chronology of the dependency periods of radio-
implanted pups is given (Fig. 9.5). Twenty-seven pups are
represented; 6 from 1984 and 21 from 1985. Other pups were
monitored during 1984 but monitoring was inadequate during the
fall and winter to determine relatively accurate weaning
dates. Estimated birth dates for individuals with known
growth rates are displayed as unbracketed open circles.

TABLE 9.3 - Sex of dependent Prince William Sound sea otter
pups.

1984 1985 1986
MALE FEMALE MALE _ FEMALE MALE FEMALE
EASTERN PWS 34 36 19 10 08 14
GREEN ISLAND BO oS 20 15 = =F
TOTAL 34 36 39 25 08 14
GRAND TOTAL 81 75
TOTAL PROPORTIONS -519 ~481

AVERAGE PROPORTIONS 2486 -514

172

FIGURE 9.2 --
otter pups

Growth rate of depende rarer and female sea

ased on two we ee ngs at le 30 days apart.

Males grew faster than female

NUMBER OF INDIVIDUALS
= nm ow p> wo ro)

\Y FEMALES

a MALES

MANN-WHITNEY

P<.03

GON NGS) 7 O75) Ol msSi 9025) 100) 105) 1110) 11115) 1120
GROWTH RATE (GRAMS/DAY)

FIGURE 9.3 -- Estimated birth dates and capture dates of sea
otter pups in Prince William Sound, Alaska. No tendency was
found for one sex to be born- earlier than the other.

EM Mace BintH OATES

Y
KW FEMALE BIRTH DATES
CAPTURE DATES

n°) _W

NUMBER OF PUPS

avid 40 400 5/10 §20 590 69 69 629 79 79 7129 ae sie ame 97 917 9/27
FIRST DAY OF 10 DAY INTERVAL

174

FIGURE 9.4 -- Weaning dates of instrumented sea otter pups in
Prince William Sound, Alaska, 1985-1986. Pups were
considered weaned when they separated from their mothers. No

tendency existed for either sex to be weaned earlier than the
other.

FEMALES

NUMBER OF PUPS

ae a a be
So lo) ie) SR) Om

rr - - wu
Me the Sr Om SO)

WEANING DATES

L75)

RADIO-IMPLANTED PUPS

FIGURE 9.5 -- Chronology of dependency periods of 27 sea
otter pups in Prince William Sound, Alaska, 1984-1986.
Horizontal bars represent individual pups. The main events
during dependency are signified by symbols.

DS eee ee ee
(ot ma ©)
eee
0 SS 5 0S

To) LEGEND

a @aoms BIRTHDATE INTERVAL
sr Fe MIN OE
WEANING INTERVAL
——— = . ——D WEANING DATE
——————— ei ——* WEANED BY INTERFERENCE
—=— CAPTURE DATES

Fiche An ROM iit didi iaiiii Aino S. ah OnseiNae | Dined nb iae Mirae Maidan C3
MONTHS OF THE YEAR

176

When weaning dates were known within plus or minus 4 days they
are also displayed as open circles. When dates were less
certain, they are displayed as intervals; the intervals being
the time between the last sighting before separation and the
first sighting afterwards. Four pups were weaned as a direct
result of research activities. They are not included in the
summary statistics but are considered separately.

The average duration of dependency was 169 days (5.6
months). The range was 76 - 333 days (N = 23). Males did
not differ from females (12 males: 170.1 days, SD = 62, 11
females: 167.6 days, SD = 45.6).

Size at weaning was estimated, assuming that growth rates
remained constant until weaning. Two individuals that were
weaned after 276 and 333 days, respectively, were excluded
from this analysis because growth rates could not remain
constant for that duration. Males were weaned at a slightly
larger size than females (16.2 kg vs. 15.2 kg.). These
estimated weaning weights seem reasonable and are consistent
with field observations, since large dependent pups are not
uncommon. Ten dependent pups were caught in September, or
earlier, that weighed between 13.6 kg and 20.0 kg.

The histories of several individuals are remarkable and
illustrate the ability of young sea otters to survive even if
weaned at very young ages. These are presented as case
histories:

Case 1. A male pup was weaned, apparently under normal
circumstances, at about 76 days of age and weighing
approximately 12 kg. It survived until it drowned in a gill
net 9 months later. Both this pup and his mother were
instrumented. Thus, the individuals' histories were fairly
well known: The mother weaned her previous pup between May
3rd and June 3rd, 1985. She was alone when observed on June
3rd. Although her location was determined several times by
telemetry, she was not observed again until August 18, when
she was observed accompanied by this male pup. The pup was
caught the same day and weighed 5.7 kg. On September 11, 24
days later, it weighed 8.8 kg and was implanted with a
radio-transmitter. It was weaned between October 1 and
October 10. Based on a measured growth rate of 132 g/day,
its birth date was estimated to be 76 days prior to weaning.
The maximum possible dependency period, assuming that the pup
was born the last day the female was seen unaccompanied and
that the pup was weaned the first day it was seen alone, would
be 129 days. However, assuming a birth weight of 2.04 kg,
this would require a growth rate of 48 g/day. The slowest
growth rate observed for a male pup was 67 g/day. If the pup
was actually weaned on the first day it was seen alone, the
pup would have to had traveled about 50 km on the day it was

177

weaned, which was possible but unlikely. Thus, this male pup
was probably weaned at considerably less than 129 days and
possibly at 76 days of age.

Case 2. A female pup was weaned on August 14 when its
mother left the vicinity with a male while the pup was
undergoing surgery to implant a radio-transmitter. When
weaned, it was estimated to be 76 days old. It weighed 9.5
kg. This pup had been previously captured on June 29, when
it weighed 5 kg. Thus, based on a measured growth rate of 95
g/day, it was assumed to have been born on May 30. When it
was released following the surgery, the pup swam several km
across a bay and took up residence in a small, protected cove.
It was known to have remained within a km of that site until
its radio expired, approximately 20 months later.

Cases 3 and 4. Two pups were weaned during the early
fall at under 100 days of age as a result of separation from
their mothers during surgery. They weighed 10.4 kg and 12.0
kg when weaned. One, the 10.4 kg animal, died over 4 months
later of unknown cause. The other was apparently killed as
an incidental take in the local salmon fishery the following
June.

Mortality

Of the 36 pups that were released with radio-implants,
15 are known to have died and contact was lost with 8 during
the life of the study (Fig. 9.6). One of the deaths occurred
during dependency and 14 occurred following weaning.

Deaths were most common during the January-April
interval, but did occur during all seasons of the year. It
was difficult to ascertain the cause of most of the deaths,
since beach-cast carcasses were rapidly consumed by bald
eagles and other scavengers. However, starvation and
predation were both observed and may have been significant
during the fall and winter months.

It was apparent that one female starved to death. Her
carcass was recovered very shortly after death. At that time
she weighed 26% less than she had when implanted, 37 days
earlier.

A recently weaned female was killed by coyotes (Canis
latrans) in Olsen Bay, at the back of a long tidal flat. It
is possible that she became trapped by the falling tide. Her
radio was found in newly fallen snow with fresh blood and
coyote urine markings. The blood was bright red indicating
that it was well oxygenated and thus, that she was killed,
rather than scavenged. Three other weanlings were probably
killed by coyotes in a nearby tidal basin, locally referred

178

FIGURE 9.6 -- Sea otter pups instrumented in Prince William
Sound, Alaska, that died or with which radio contact was
lost. Contact was more frequently lost with males than
females. Females tended to die during the January-April
interval whereas male deaths were distributed year around.

MISSING PUPS
Le)

; \\

SEPT-DEC JAN-APR MAY-AUG

MALES

FEMALES

DEAD PUPS

SEPT-DEC JAN-APR MAY-AUG

179

to as Hell's Hole. All died within a few hundred meters of
each other. The first of these weanlings to die had been seen
hauling-out on a bank that was heavily used by coyotes,
judging from the numerous fresh tracks in the snow cover.
This weanling was dead within a week, leaving no traces except
the radio on the beach within 100 m of the haul out site. The
second weanling's radio was recovered on a nearby beach 2
months later. On the same day, the third weanling was
observed alive a few hundred meters away. It had become
stranded on an exposed mussel bed, by the out going tide,
several hundred meters from deep water. During the next
survey, 3 months later, its radio was recovered at the same
location as the second weanling's. The radios of several
other weanlings were found on large tidal flats. It is
conceivable that those individuals became stranded and were
preyed upon. Other radios were recovered from rocky beaches
in areas subject to wave action. Otters were not observed to
haul out at these locations. Thus, it is probable that they
died elsewhere, remained intact, and floated to those sites.

Entanglement in commercial fishing gear was a significant
problem and a cause of deaths to the sea otters in this study.
Commercial fishing is permitted in the general area from May
until October. All four of the males that were dead or
missing during the May-June interval may have been killed in
the fishery. One male's flipper tags were turned in by a
fisherman who stated that it had drowned in his gill net.
Another weanling was missing a few days after the annual
fishing opener, in May. For the preceding 5 months it had
occupied an area that was heavily fished during the opener.
Presumably, it died and its carcass drifted beyond the study
search area. A third male's radio and remains were recovered
a few km down current from the same area during the middle of
the fishing season. A fourth male became missing near Valdez
following an unusually heavy period of gillnetting in the area
where he was last seen.

Three other study otters, one an adult female, became
entangled in fishing nets but were released unharmed. Three
fishermen reported that otters had snagged their button tags
on the strands of the net. In one case it was thought that
the weanling drowned because it snagged near the bottom of
the net and could not surface. It is not possible to say what
the role of the tags was in the deaths of these animals.
Button tags are no longer used at this study site. It would
seem prudent not to use button tags at any location where
there is a chance that otters may encounter commercial fishing
gear. This includes gill nets, seines, trawls or crab pots.

Radio contact was lost with male weanlings more
frequently than it was lost with female weanlings (7:1). It
is possible that radio contact was lost because those

180

individuals left the study area. In this study male sea otter
weanlings did travel more extensively than their female
counterparts following weaning (Chapter 8). Several of the
missing males were last seen near the edge of the study area
after traveling as far as 123 kn.

Survival and juvenile movements.--The difference in the
fates of individuals was striking depending upon whether they
were travelers or not. Approximately 90% (28 of 31) of the
known or suspected weaning locations were located in three
Bays: Sheep Bay, Simpson Bay or Port Gravina. Of the 12
weanlings that remained in those bays: 2 survived, 1 was
missing and 9 died. Of the 16 weanlings that left those
areas: 9 survived, 4 were missing and 3 died. Even if it is
assumed that the missing individuals died, which biases the
data against a difference in outcome, survival was lower for
weanlings that remained in the weaning areas (Chi square =
3.93, 1 DE Peas 05). If the same comparison is made,
considering only survival until spring, the relationship is
stronger (Chi square = 8.18, 1 DF, p < .05). If it is assumed
that the missing weanlings did not die, the relationship is
stronger since 4 of 5 missing weanlings were travelers. The
argument is also strengthened by the fact that the 3 weanlings
that were known to have been weaned in bays other than Sheep,
Simpson or Port Gravina all survived. Individuals that stayed
in the latter bays appeared to be dying mostly of predation
or starvation during the fall or winter. However, those that
left the weaning areas died mostly in the spring and summer
due to human-related activities.

Survival rates

Two sets of survival-related calculations are given
(Table 9.4). Sea otters that died of all causes are included
in one set. In the second set, deaths caused by human
activities are treated as transmitter failures. That is, the
otter days are included but the death is not. Mortality that
was probably related to the fishery was quite high during this
project. Such deaths may be rare in other areas. It is hoped
that the alternate sets of calculations will make the data
more general.

The only pup to die during dependency was a male pup that
died after being hit by a boat propeller. Thus, the survival
rate for dependent pups is 1.0 when human caused deaths are
not considered but is slightly lower when they are. When the
boat strike was included, the survival rate for dependent
males seemed low at 0.70. It is, perhaps, more insightful to
consider that 20 of 21 (95%) male pups survived the average
of 48 days between being instrumented and being weaned. All
females (N = 15) survived the comparable period.

181

TABLE 9.4 - Survival rates of sea otter pups in Prince William
Sound.

ALL MORTALITY NON-HUMAN MORT.

FEMALES MALES FEMALES MALES
DAY YR DAY YR DAY YR DAY YR
DEPENDENT PUPS
11.0. 109439990; 270 ‘l HOe aie. 0 a! SONU HIS0
(657) (1012) (657) (1012)
INDEPENDENT JUVENILES
ASSUMP. 1 H9969W 2) ne 9982 oS) mec 9969ea320) BOOSOhL eo
(2917) (2720) (2917) (2721)
ASSUMP. 2 .9966 .29 .9923 .20 .9966 .29 .9966 .30
(2917) (2721) (2917) (2721)
ASSUMP. 3 .9974 .38 .9982 .52 .9974 .38 .9989 .68
(3410) (2815) (3410) (2815)
ASSUMP. 4 N9O7Aw 34) 19957 624) 997 es 4ey 99 74e es
(3410) (2815) (3410) (2815)

* Numbers in parentheses are “otter days".

ASSUMPTIONS :
Weaned 1: Died day of last sighting, missing pups are
alive.
Weaned 2: Died day of last sighting, missing pups are
dead.

Weaned 3:
Weaned 4:

Died day carcass found, missing pups are alive.
Died day carcass found, missing pups are dead

The survival rates of weanlings are lower than those of
dependent pups. Female rates are the least variable, ranging
from) O29) .—9 O39 depending upon assumptions made in
calculations. The exclusion of human-caused deaths has no
effect on rates since no females died as a result of the

fishery. As noted, females were caught in fishing gear, but
all were released. Male survival rates are quite variable,
0.21 - 0.70, depending upon the assumptions made in

calculations and whether human-caused deaths are included.
Changes in frequency of monitoring had little effect.
However, potential mistakes in determining weanling survival
(i.e. incorrect assumptions about the status of missing
individuals) could lead to a change in survival rates by a
factor of 2.

DISCUSSION
The results of this study indicated that there exists
substantial variation in the timing of births, growth rates

and dependency periods between individual sea otters. Males

182

grew more rapidly than females, were weaned at slightly
heavier weights and thus, apparently required more parental
resources (i.e. food) than did females. However, male and
female births were timed similarly, and dependency periods
were of approximately equal duration.

Data indicated that young sea otters were capable of
surviving independently long before they reach the typical
age of weaning (i.e. approximately 5-6 months), even when they
were weaned prematurely by human interference. In general,
however, weanling survival rates were low during the first
year of independence. Male weanlings were more likely to
survive their first year of independence than were female
weanlings. Observed differences between male and female
weanlings in rate and timing of mortality existed and appeared
to be strongly related to whether weanlings established their
post-weaning home range in male areas or female areas.

These data offer some support for the body of theory that
suggests that sea otters should make a larger parental
investment in male offspring.

PI and growth rates

The faster growth exhibited by males suggests that
females may allocate more resources to dependent sons than to
dependent daughters. Based on the arguments of Trivers and
Willard (1973), it is possible that only the females that can
best do so, produce males. Such females might be those that
were older, larger or more experienced. This hypothesis
requires the implicit assumption that there exists some
mechanism whereby these individuals would tend to conceive
sons, or possibly abort females (see Clutton-Brock and Albon,
1982, for a discussion of such mechanisms). Substantial
variation in pup sex ratios was observed in this study, but
it is not possible to say whether such variation was related
to maternal factors in any systematic way.

Another way in which a female could allocate more
resources to a son would be trade off future reproductive
potential in order to produce an adequate male, once one is
conceived (reviewed in Clutton-Brock and Albon, 1982). The
faster growth of sea otter males during dependency suggests
that male pups may require greater investment from their
mothers than do female pups. It seems unlikely that males
assimilate nutrients more effectively than females, since
growth rates of male mammals are usually more strongly
affected by food shortages than those of females (Widdowson,
1976). Unless food is unlimited, a male's greater need for
milk or solid food could negatively affect its mother's
health, both by affecting her overall condition and/or by
causing imbalances of critical elements at the time of

183

weaning. These "costs" to the mother could be manifested in
effects on future reproductive potential through, for example,
skipped or aborted pregnancies and reduced life span. It is
possible that such "costs" exist for females that produce
sons, especially if they live in habitats that have been
heavily exploited. For example, higher rates of in utero
mortality and lower overall rates of reproduction have been
observed in sea otter populations that have over-exploited
their food supplies (Schneider 1978b; reviewed in
Simon-Jackson and Rotterman 1987).

Timing of parturition

Other information on sex differences in birth dates is
not available. However, the observed late May peak of pupping
is consistent with observations from other locations within
Alaska. Barbash-Nikiforov, et al. (1978) observed a peak
during May-June in the western Pacific. Schneider (1978b)
reported a May peak in the central and western Aleutian
Islands.

Dependency period

Dependency period lengths were quite variable for pups
in this study and for pups in other studies at different
locations throughout the sea otter's range. Data from research
carried out along the eastern coast of the Soviet Union
(Barabash-Nikiforov, et al., 1978), and in the western and
central Aleutian Islands (Kenyon, 1969; Schneider, 1978b)
suggested that females in some populations nursed pups for
approximately one year. Other studies in Alaska and
California have reported dependency periods averaging 5-7
months (Garshelis, et al., 1984; Wendell, et al., 1984; Payne
and Jameson, 1984). Dependency periods in this study ranged
from 2.5-11 months.

Several variables might contribute to variation in
dependency periods. The underlying food supply could affect
dependency periods both by influencing the development of
young animals directly, and by affecting them. indirectly
through the mother. It is likely that pups must reach a
minimum size and minimum level of experience to be able to
survive, in a given environment, upon independence. Food
abundance should affect growth rates and, hence, the time it
takes to reach that minimum size. If food was abundant,
shorter dependency periods could result. Also, food abundance
might affect the minimum size and experience requirements.
If food was abundant, less precocious weanlings might be able
to feed more effectively, mistakes would be less critical and
thus, pups could be weaned earlier. If females were in good
condition, they might be able to raise a pup more quickly or,

184

conversely, better afford to continue to support one. The
same might be true for older, more experienced, females.

Sex ratio

No data are available from this study that would give any
insight into the reason for the variation in pup sex ratio

that was observed. However, a number of maternal or
environmental factors are believed to be correlated with
biased offspring sex ratios in mammals. These include:

maternal age, parity, reproductive history, dominance status,
size or nutrition and birthdate, litter size or timing of
conception. Recent reviews of these factors, and relevant
theories, are available in Clutton-Brock and Albon (1982) or
Clutton-Brock and Iason (1986).

The only data that are available on the sex ratio of
young sea otters, other than this study, are from Amchitka
during the 1960's (Kenyon 1969). Kenyon (p. 206) reported
that the sex composition of 117 recently "deserted" or
dependent "juveniles" (individuals of less than one year of

age) was 58 males, 58 females, and one unknown. Schneider
(1978b) and Kenyon (1969) reported fetal sex ratios from
studies in the central and western Aleutian Islands.

Combining their data, of 319 fetuses, 171 (57%) were females
and 138 (43% males).

Survival rates and causes of mortality

Parental investment theory suggests that sex differences
in post-weaning behavior, especially movement patterns, could
lead to male and female offspring being subjected to different
risks. Different selection pressures could, in turn, lead to
different mortality patterns. Mortality rates and patterns
were different for male and female sea otter offspring.
Unexpectedly, females (the nondispersing sex) exhibited lower
survival rates, during the first year after weaning, than did
males. The trend toward a higher rate of male post-weaning
survival in this study was similar to the trend in juvenile
survival in California (Chapter 2).

Comprehensive discussions of causes of sea otter
mortality are available elsewhere for California (Riedman,
1986) and Alaska (Kenyon, 1969; Simon-Jackson and Rotterman,
1987). These include: starvation, disease, parasitism,
predation, shark attacks, accidents during research projects,
entanglement in fishing gear, adverse weather conditions
(storms and icing), boat strikes, and injuries received from
conspecifics during fighting or mating.

The results of questionnaires (Simon-Jackson, 1985) and
surveys (Matkin and Fay, 1980; Simon-Jackson, 1986) have

185

indicated that there is a significant mortality of sea otters,
incidental to the salmon fishery in Prince William Sound and
on the nearby Copper River Delta. Incidental take of sea
otters in commercial fisheries is not unique to, but is
especially significant in the Cordova vicinity (Simon-Jackson,
1986; and reviewed in Simon-Jackson and Rotterman, 1987). Sea
otter encounters with fishing gear are frequent and increasing
on the Copper River delta and within the Prince William Sound.
It is known that many untagged otters have died in fishing
gear, either from drowning or from fishermen killing them
directly.

Why is first winter survival high in the male area?

The male area is mostly contained within shallow,
protected channels within estuarine mud flats that contain
abundant shellfish species. Such habitat is scarce in the
weaning areas. Food was considered to be abundant in the
local male area when otters first recolonized it (Garshelis,
et al., 1986) and apparently still is. Otters in the male
area are frequently killed by natives, fishermen, and others.
In such cases, recovered carcasses were examined and found to
have significant subcutaneous fat. Moreover, stomachs usually
contained large quantities of clam and/or crab tissues (pers.
obs.). In addition to abundant food, the local male area
offers young otters other benefits. There are large
aggregations of males of all ages, so opportunities for
learning from adults also exists (Garshelis, et al., 1984).
Moreover, large predator-free, sand bars are exposed at
moderate tides. These are heavily used as haul-out areas,
especially during the winter. The energetic advantages of
such behavior in cold water environments could be substantial.

Why are pups weaned in bays where they are likely to die?

Some weanlings apparently died of starvation during the
first few months after weaning. The apparent maladaptive
practice, of females weaning pups in areas where they have
little chance of over-winter survival, may be a consequence
of changing, conflicting needs during the development of the
pup. On one hand, the deep, protected northern bays offer
shelter and abundant pup food in the blue mussels (Mytilus
edulis) which cling to the steep rock walls and grow in dense
beds in the shallows. Opportunities exist for pups to learn
self feeding in protected channels and lagoons. On the other
hand, these bays tend to be deep and steep-sided so other
larger prey may be relatively unattainable to small, newly
independent otters whose diving abilities are usually limited.
Although it seems probable that small juveniles can survive
on a diet of mussels alone, it is not clear whether large
juveniles can.

Predation appears to be a significant source of mortality
of young sea otters in the weaning areas. Moreover, the rate
of predation could be exacerbated by juveniles' tendency to
haul out when subject to food stress, as would occur under the
preceding argument. Predation on hauled sea otters has not
been noted before this study and may be uncommon. If so, it
would be unlikely that adult sea otters would be prepared to
respond to the possibility when choosing sites for weaning
pups.

Recolonization and future changes

The Prince William Sound sea otter population is still
recovering from near extirpation, which took place at the end
of the 18th century (Lensink, 1962). The bays in which
weaning was observed in this study, Simpson Bay, Sheep Bay
and Port Gravina, were reoccupied by large rafts of males
between 1970-1979 (Garshelis, et al., 1986).

While it is risky to predict the patterns that will
develop for otters in this area when the population approaches
equilibrium, a few speculations can be made. It is likely
that food resources in the three bays have been depressed as
a result of the relatively recent occupancy by large rafts of
males (Garshelis, et al., 1986). However, it is possible that
food levels will eventually rise above current levels, after
the population reaches equilibrium (e.g. Estes, et al., 1978;

Estes, et al., 1982). If food levels do rise, juvenile
starvation may decrease in those bays. Moreover, deaths from
predation may also decrease. Predation can occur when

individuals haul out on beaches. Moribund sea otters usually
haul out (Kenyon 1969), which would be the case if food was
scarce and they were starving. Thus, if food became more
abundant the frequency of hauling-out and, hence, predation
might decline. Conversely, it might be predicted that food
abundance in the male area will decline, at least initially,
after a longer period of sea otter occupancy. If so,
starvation would increase. Predation, however, might remain
unaffected since numerous safe haul out sites are available
on sand bars in that area.

LITERATURE CITED

Ames, J. A.; R. A. Hardy; and F. E. Wendell. 1983. Tagging
materials for sea otters, Enhydra lutris. Calif. Fish
and Game 69:243=-252.

Barabash-Nikiforov, I. I., S. V. Marakov and A. M. Nikolaev.

1968. The kalan or sea otter. Nauka Press, Leningrad.
184 pp.

187

Calkins, D. G. and P. G. Lent. 1975. Territoriality and
mating behavior in Prince William Sound sea otters. J.
Mamm. 26:528-529.

Clutton-Brock, T. H.; P. H. Harvey and B. Rudder. 1977.
Sexual dimorphism, socionomic sex ratio and body weight
in primates. Nature 269:797-799.

Clutton-Brock, T. H. and S. D. Albon. 1982. Parental
investment in male and female offspring in mammals. In:
Current Theories in Sociobiology. (eds. Kings College
Sociobiology Group), Cambridge.

Clutton-Brock, T. H. and G. R. Iason. 1986. Sex ratio
variation in mammals. Quarterly Review of Biology
61:339-374.

Estes, J. A.; N. S. Smith and J. F. Palmisano. 1978. Sea
otter predation and community organization in the western
Aleutian Islands, Alaska. Ecology 59:822-833.

Estes, J. A.; R. L. Jameson and E. B. Rhode. 1982. Activity

and prey selection in the sea otter: influence of
population status on community structure. Am. Nat.
120:242-258.

Garshelis, D. L. 1983. Ecology of sea otters in Prince

William Sound, Alaska. Ph.D. dissertation. University
of Minnesota, Minneapolis, Minnesota. 330 pp.

Garshelic (o9D. Le and’™D. 75.) /Sinatt- 1983. Evaluation of
radio-transmitter attachments for sea otters. Wildlife
Society Bulletin 11:378-383.

Garshelis, D. L. and J. A. Garshelis. 1984. Movements and
management of sea otters in Alaska. J. Wildl. Manage.
48:665-678.

Garshelis, D. L.; A. M. Johnson and J. A. Garshelis. 1984.
Social organization of sea otters in Prince William
Sound, Alaska. Canadian Journal of Zoology 62:
2648-2658.

Garshelis, D. L., J. A. Garshelis and A. T. Kimker. 1986.
Sea otter time budgets and prey relationships in Alaska.
Journal of Wildlife Management 50:637-647.

Greenwood, P. J. 1980. Mating systems, philopatry and
dispersal in birds and mammals. Animal Behavior 28:
1140-1162.

188

Glucksman, A. 1974. Sexual dimorphism in mammals.
Biological Review 49:423-475.

Glucksman, A. 1978. Sex determination and sexual dimorphism
in mammals. London, Wykeham.

Hamilton, W. D. 1964. The genetical evolution of social
behaviour. I. J. Theoret. Biol. 7:1-16.

Kenyon, K. W. 1969. The sea otter in the eastern Pacific
Ocean. USFWS. North American Fauna, No. 68, 352 pp.

Lensink, Cc. J. The history and status of sea otters in
Alaska. Ph.D. dissertation, Purdue University,
Lafayette, Indiana. 186 pp.

Loughlin, T. R. 1977. Activity patterns, habitat
partitioning, and grooming behavior of the sea otter,
(Enhydra lutris), in California. Ph.D. thesis,

University of California, Los Angeles, CA. 110 pp.

Matkin, C. 0. and F. H. Fay. 1980. Marine mammal - fishery
interactions on the Copper River and in Prince William
Sound, Alaska, 1978. U. S. Dept. of Commerce, National
Technical Info. Serv. PB 80-159536. 71 pp.

Payne, S. F. and R. J. Jameson. 1984. Early behavioral
development of the sea otter, (Enhydra lutris). J. Mamn.
65:527-531.

Reiter, J.; N. L. Stinson and B. J. Le Boeuf. 1978. Northern
elephant seal development: the transition from weaning
to nutritional development. Behav. Ecol. Sociob.
3:336-367.

Riedman, M. L. 1986. Draft environmental impact statement of
proposed translocation of southern sea otters. Volume
II: Technical support documents. United States Fish
and Wildlife Service and University of California, Santa
Cruze Ca.

Schneider, K. B. 1978a. Sex and age segregation of sea
otters. Alaska Department of Fish and Game, Final
Report, Federal Aid Wildlife Restoration Projects W-17-4
to W-17-8.

Schneider, K. B. 1978b. Reproduction in the female sea otter

in the Aleutian Islands. Unpublished Report, Alaska
Department of Fish and Game. 44 pp.

189

Simon-Jackson, T. 1985. Fishermen opinions of sea
otter/fisheries issues in Alaska. USFWS, unpubl. rept.

17 pp.
Simon-Jackson, T. 1986. Sea otter survey, Cordova, Alaska
1986, (Orca Inlet to Cape Suckling). USFWS, unpubl.
rept.

Simon-Jackson, T. and L. M. Rotterman 1987. The sea otter in
Alaska (Enhydra lutris): Species account with research
and management recommendations. Prepared for the Marine
Mammal Commission, Washington, D. C.

Siniff, D. B., T. D. Williams, A. M. Johnson and D. L.

Garshelis. 1982. Experiments on the response of sea
otters, Enhydra lutris, to oil contamination.

Biological Conservation 2:261-272.

Trent, T. and T. and O. J. Rongstad. 1974. Home range and
survival of cottontail rabbits in southwestern Wisconsin.
J. Wildl. Manage. 38:459-472.

Trivers, R. L. 1972. Parental investment and sexual
selection. In: Sexual selection and the descent of man,
1871-1971. (ed. B. Campbell), pp. 136-179. Chicago:
Aldine.

Trivers, R. L. and D. E. Willard. 1973. Natural selection
of parental ability to vary the sex ration of offspring.
Science 179:90-91.

Wendell, F.E.; J. A. Ames and R. A. Hardy. 1984. Pup
dependency period and length of reproductive cycle:
estimates from observations of tagged sea otters,

(Enhydra _lutris), in California. Calif. fish and Game
70:89-100.

Williams, T. D.; A. L. Williams and D. B. Siniff. 1981.
Fentanyl and azaperone produced neuroleptananalgesia in

the sea otter (Enhydra lutris). Jie A Walaa) Dilsi
17:337-342.

Widdowson, E. 1976. The response of the sexes to nutritional
stress. Proc. Nutr. Soc. 35:1175-1180.

190

CHAPTER 10

A SIMULATION MODEL FOR ASSESSING THE RISKS OF OIL SPILLS TO
THE CALIFORNIA SEA OTTER POPULATION AND AN ANALYSIS OF THE

HISTORICAL GROWTH OF THE POPULATION

A. BRODY

NOVEMBER 30, 1988

191

INTRODUCTION

The model described in this chapter is designed to
facilitate analysis of the risk of oil spills to the
California sea otter population. Specifically, we provide a
simulation of aspects of otter population biology and behavior
that will likely affect the degree of risk to the population
associated with oil spills. We have not conducted actual risk
analysis with the model.

Ford, et al., (1982) describe two general categories of
oil spill consequences that affect the degree of risk of oil
development activities to a wildlife population: 1) immediate
mortality from a given oil spill, and, 2) long term population
effects. Our model addresses the first category explicitly,
and can be used to address the second category if the long
term effects of oil development are assumed to result only
from mortality due to oil spills. Our model was formulated
to answer two specific questions about a given oil spill: 1)
how many otters will be killed, and 2) how long will it take
for the population to recover.

General approach

There is a good deal of uncertainty surrounding certain
aspects of sea otter ecology in California, and a general lack
of data regarding most aspects of the population dynamics.
There are also ongoing research projects being conducted by
various agencies and organizations aimed at correcting these
situations. A major objective in structuring our model was
that it be flexible enough to explore a variety of conceptual
hypotheses about sea otter ecology, and to incorporate new
data as they become available.

As with any simulation model, some simplifying
assumptions about the system are necessary. Four major ones
will be discussed in this introductory part of this chapter,
others will be mentioned at appropriate places later on.

1) Geometry of sea otter range.-- The sea otter range in
California is essentially linear, consisting of an
approximately 350 km long by 1 km wide band along the coast.
The width of the range is dictated by the slope of the ocean
floor, with otters generally inhabiting only areas shallower
than 18m (USFWS 1986); in some areas of the coast the 18m
depth contour is more than 1 kilometer offshore, and animals
may occasionally be found at locations of deeper depth. In
fact, telemetry data gathered as part of this project
indicates that certain segments of the population may utilize
offshore areas more frequently than previously realized (see
Chapter 3). But in the model we consider the range to be one
dimensional, a line extending up and down the coast.

192

California Department of Fish and Game has traditionally used
the "as the otter swims line" (Ames, Hardy and Wendell,
personal communication), an ordinate system coincident with
the 5 fathom (10m) depth contour, in their census activities.
With an origin at Coyote Point on the San Francisco peninsula,
position along the coast is measured in 500m units south of
the origin. We have adopted this system for the spatial
aspects of our model. This system allows us to incorporate
existing census data, to ignore the shape of an oil spill and
be concerned only with the length of coast affected, and to
facilitate the analysis and modeling of sea otter movement.

2) Density dependence in population dynamics.--A good
deal of controversy surrounds the role of density dependence
and the question of non-equilibrium in the current theories
of population dynamics. The issue is complicated by the range
expansion that has accompanied population growth, differences
in habitat quality at various locations within established and
potential range, and the role of the sea otter as a "keystone
species", able to dramatically affect the quality of its own
habitat (Miller 1980, Estes, et al., 1982, Estes, et al.,
1986, Wendell, et ae 1986). Rather than assume density
dependent or density independent dynamics, we have built in
a flexibility that allows investigation of both. The
population dynamics portion of our model nominally assumes
density dependent growth during recovery from an oil spill.
The equilibrium population size and a parameter governing the
shape of the density dependence function are easily
manipulated at the beginning of a model run, however, and
density-independent population growth can be simulated by
setting parameter values that result in an essentially flat
density dependence function.

3) Range expansion.--Related to range geometry and the
nature of population dynamics is the question of range
expansion. It is an especially critical question given the
purpose of the model, as the offshore areas most likely to be
developed for oil and gas in the future are at the southern
periphery of the existing sea otter range (USFWS, 1986).
Thus, if we are to make reliable predictions about the effects
of oil spills very far into the future we must be able to make
predictions about the extent of sea otter range in the future.
Unfortunately, data collected during this study do not easily
lead to such predictions. We assume in the simulation model
that the extent of the range is static, and that otters
distribute themselves along the coast in the same relative
proportions regardless of total population size. We have,
however, built a separate, small, deterministic model that
will generate predictions of future range length, carrying
capacity, and population size based on historical rates of
population growth and range expansion. This model, OTRANGE,
described elsewhere in this report, can be used to generate

193

range conditions for use in the population model. It only
applies to the existing sea otter range in California and does
not extend to the translocation of otters to San Nicholas
Island.

4) Impact criteria.--In line with the stated purpose of
the model, three criteria are assumed to measure the effect
of an oil spill on the population. The first criterion is
the number of animals and proportion of the population killed
by the spill. The second criterion is the number of years
after the spill that is required for the population to recover
to pre-spill size. The third criterion is the decrease in the
total reproductive value of the population (Wilson and Bossert
1971), providing a measure of how the perturbation in age and
sex structure caused by an oil spill may affect population
dynamics. This is calculated using the survival and
reproduction rates operative in the population just prior to
the spill, before density dependent adjustments to these rates
are effected. In addition to these three criteria the model
can be run in a "control" mode, in which an oil spill is not
introduced, allowing graphic comparison of population dynamics
with and without perturbation by oil spill.

Model structure

We model the effect of an oil spill on the sea otter
population as being determined by the size and location of
the spill in relation to sea otter distribution, the movement
of individual sea otters in the vicinity of the spill, and
sea otter population dynamics.

The model itself consists of four submodels imbedded into
a larger program superstructure (Fig. 10.1). The submodels,
all of which are stochastic, operate in different temporal and
spatial scales. The short term population submodel (OTPOP)
operates with a time step of one month and is spatially
independent. The long term population model (LESLIE) operates
on an annual time scale and is also spatially independent.
The sea otter distribution model (OTDIST) operates on a
Spatial scale of kilometers, and is time independent. The
short term movement and oil response model (OTMOVE) operates
with a time step of days, and on a spatial scale of
kilometers. Each of the submodels operates on a numerical
scale of individual animals.

OTPOP creates a simulated initial population, and
iterates for three simulated years before an oil spill. In
the month of the simulated spill, population vectors,
consisting of the age and reproductive status of each
individual animal in the simulated population, are passed from
OTPOP to OTDIST. OTDIST assigns each individual a position

194

FIGURE 10.1 -- Schematic representation of the interrelation of
the 4 submodels used to predict the potential effects of oil
spills on California sea otter population dynamics.

INITIALIZE LESLIE END
PARAMETERS
AND CONSTRUCT Annual
INITIAL POPULATION Population

Model

population

OTPOP

Monthly
Population
Model

vectors

population
vectors

mortality
vectors

OTMOVE OTDIST

coh a Short—term Rp here tb
HERE —» | Movement ode

Model

along the coast, and passes these vectors to OTMOVE. OTMOVE
introduces an oil spill, moves animals along the coast during
the duration of the spill, and generates a mortality vector.
This mortality vector is then returned to OTPOP, and two years
of population recovery are simulated on a monthly basis. At
the end of the second year after the simulated spill, the
population vectors representing individual animals are
collapsed into age class vectors and passed to LESLIE. LESLIE
then simulates the future course of population growth on an
annual basis for up to 50 years.

While the structure of the model is essentially set, and
can be altered only through reprogramming of the source code,
the parameters used in the model are set each time the model
is run. We have supplied a set of default parameters,
representing our best estimates of the values operating in
the real population, but these may be altered by the user to
investigate the importance or sensitivity of parameters or to
take advantage of revised parameter estimates that may be
available after future research. Following is a detailed
discussion of the structure of each of the submodels, and the
logic that we followed in arriving at the default parameter
estimates.

OTPOP AND LESLIE
Structure

The small size of the population allows a reasonably
efficient consideration of individual animals. The
reproductive biology of sea otters, specifically the fact that
pupping is spread throughout the year rather than concentrated
into a short reproductive season, adds a complexity to the sea
otter model that is not present in population models of most
other large mammals. OTPOP thus iterates on a monthly basis.

At the beginning of each run the user specifies an
initial population size, a carrying capacity for the range,
a maximum population growth rate, and a parameter governing
the shape of the density dependence (population growth rate
vs. population size) curve. The initial growth rate of the
population is calculated on the basis of these parameters and
used to construct an initial population vector with a stable
age distribution.

The age, in months, of each male in the population is
stored in a male age vector. The age, in months, of each
female in the population is stored in a female age vector.
The reproductive status of each female in the female vector
is stored in the
corresponding element of a reproductive vector.

196

During each month of simulation the model loops through
each individual in the population, drawing a random number
that is compared against age- and sex- specific monthly
survival rates to determine whether or not the animal survives
the month. In the female loop the reproductive status of each
surviving female is checked. If a female has a pup the
survival of that pup is determined by drawing a random number
and comparing it to monthly pup survival rates. If a female
has no pup, she becomes pregnant with probability determined
by her age and the month of the year. Inter-uterine mortality
is assumed to be zero so that if a pregnant female survives
a given month her fetus automatically survives also.

The age of any animal that does not survive the month is
flagged. If a pup dies, the mother is assumed to get pregnant
again immediately. If a pup survives until weaning eas
assigned a sex randomly and added to the appropriate age
vector. At the end of each monthly loop through the
population the vectors are reloaded without the animals that
have died, and numbers are totaled and reported.

The model is allowed to run for three simulated years
before an oil spill is introduced, to subject the initial
stable age distribution to stochastic fluctuations. In the
year of the spill, OTDIST and OTMOVE are called and subject
the population vectors to oil spill induced mortality. oOil
spill mortality is considered after the regular loop for that
month has been completed, but after the population vectors
are reloaded. Thus simulated oil spill mortality is strictly
additive to the simulated natural mortality.

OTPOP continues to iterate on a monthly basis for two
simulated years after the spill, with population growth rates
determined by the density dependence function annually. Two
years after the spill the population vectors are collapsed
into age class vectors and passed to LESLIE. Pups and fetuses
are grouped into age class 0, age class 1 contains animals
between seven and 18 months of age, age class 2 contains
animals between 19 and 30 months, etc. The numbers of animals
of each sex in age class 0 are determined by a random draw
from a binomial distribution assuming a 0.5 probability of
being either sex.

LESLIE, iterating on a yearly basis, runs much faster
than OTPOP at the expense of seasonality. Population sizes
are reported once a year, at the end of the month in which
the oil spill occurred. Survival and reproductive rates based
on the density dependence function are determined at each
iteration, just as in OTPOP. Two sources of variation, after
Harris, et al., (1987), are considered explicitly in LESLIE.
The first is a "demographic" stochasticity wherein it is
assumed that all animals of the same age and sex have the same

197

probability of surviving the year, but the number that
actually do survive is determined by a random draw from a
binomial distribution with parameters (n,,S,) where n, is the
number of animals in age class x and s, is the annual survival
rate for age class x. If n, is greater than 30 the normal
approximation to the binomial is used to reduce the computer
time required. The second source of variation is
"environmental" stochasticity that operates simultaneously
across all age classes to the same degree in a given year.
This is introduced to simulate the occurrence of "good years"
and "bad years". The bounds of this stochasticity are
determined by the user at the beginning of the run, and p,
the environmental stochasticity parameter, is assumed to be
uniformly distributed between those bounds. At the beginning
of each simulated year p is determined once by random draw
and the survival rate, s,, of each age class for that year is
modified by adding ps,.

Built in to the structure of LESLIE is the ability to
consider density independent mortality. This is included
primarily because of the possibility that incidental gill-
and trammel-net mortality, which has substantially affected
population growth in recent years (Ames, et al., 1985,
Wendell, et al., 1985, Estes, et al., 1986) operates in a
density independent fashion. The user may set a density
independent mortality rate, applied to all age classes, and
also the degree to which this density independent mortality
compensates for density dependent mortality.

Theoretical framework for parameterization of OTPOP and
LESLIE.

Field data that can be used to infer the dynamics of the
California sea otter population are scarce. Raw data
available to us included aerial and ground censuses conducted
by the CDFG and the U.S. Fish and Wildlife Service (1968-
1985), records from CDFG and USFWS carcass recovery efforts
(1968-present), monthly counts from several CDFG and USFWS
index areas (1976-1982), and our own live-capture and
telemetry data. In addition to this raw data, information on
sea otter population dynamics was gleaned from several
publications and manuscripts.

The information available to us differed greatly in
reliability, in relative quantity, and in the extent to which
it was directly applicable to our purposes. Bringing all of
the information together into a single flexible yet
comprehensive model required the use of a strong theoretical
framework, necessary both for evaluation of the available
data and for estimating model parameters. The theoretical
framework that we used was developed by Eberhardt (1985)

198

following work by Siler (1979), and based on the classical
age structure models of Lotka (1907).

Survivorship.--Siler (1979) and Eberhardt (1985) viewed
survival to any age as a function of three "competing risks".
The first risk is that of "early hazards", risks which are
associated with the early years of life, but become much less
important as the animal matures. In sea otters these early
hazards are most likely to be associated with dispersal from
the natal area after weaning, and with low competitive ability
relative to older, more experienced animals. The early
hazards risk is greatest in the first years of life, but
essentially zero after maturity. The second competing risk
is that of incidental, or "constant hazards", which in sea
otters would include those due to possible predation, severe
weather, infectious disease, and possibly accidental
entanglement in commercial gill and trammel nets. This risk
is seen as constant throughout an animal's life, and is the
only important risk of mortality during the prime adult years.
The last risk is that due to senescence, and in sea otters
would include reduced competitive ability or increased
susceptibility to disease due to old age. This risk is
essentially zero through the prime adult years, but reaches
100% by the maximum age.

The probability of survival to any age, then, is the
probability of surviving all three competing risks at that
age, or the product of the three age specific survival rates
(Fig. 10.2). Estimating the survivorship schedule inside of
this framework allows calculation of survival rate for each
age class without having to estimate each directly from field
data. The number of parameters that have to be estimated is
thus greatly reduced. Sea otters are long-lived; the data
for directly estimating survivorship for perhaps 25 age
classes are simply not available. Using Eberhardt's (1985)
approach we need only to determine the form of the three
competing risk curves.

The equation for survivorship at age x, from Eberhardt
(1985) is:

1, = exp{-a;[1l-exp(-b;x)] - ax - as[exp(b3x) -1] } (1)

where 1, is survivorship, a;, a2, az are the coefficients for
early, incidental, and senescence risks, respectively; and
b,, bz are parameters governing the shape of the early hazards
and senescence curves, respectively. In this formulation the
risk coefficients (a, a2, a3) are taken as -ln(S) where S is
the annual survival rate against early hazards, incidental

199

FIGURE 10.2 -- Hypothetical survivorship curve depicting the
relationships of the 3 competing risks of Siler (1979) and
Eberhardt (1985).

SURVIVORSHIP

200

hazards, and senescence, respectively. Separate survivorship
curves can be generated for males and females using the basic
equation with different parameter values.

Reproduction. --As for survivorship, age specific
reproduction can be modeled as a function of three component
curves: an early reproductive function, a "prime" rate during
adulthood, and a decrease in reproductive output during the

years of senescence (Fig. 10.3). Again, conceptualizing
reproductive rates in this manner greatly reduces the number
of parameters to be estimated. If early reproduction

increases asymptotically to the prime rate a separate term
governing early reproduction is not necessary, and Eberhardt
(1985) gives an equation for the reproductive curve:

m, = A{1l-exp[-b,(x-C) ] }}exp[-azexp (b3x) -1] (2)

where m, is the number of female offspring per year weaned by
each female of age x, A is the maximum reproductive rate (in
number of female offspring weaned per prime aged female), B
is a parameter governing the rate of increase of early
reproductive rate to the prime rate, C is the age before the
age of first reproduction, and az and bz; are the senescence
parameters as in (1).

Population growth rate and density dependence. The per
capita population growth rate is a central parameter in any
population model. Recent investigations suggest that in
marine mammal populations the dependence of population growth
rate on population size is nonlinear, with the growth rate
decreasing more and more rapidly as the population approaches
carrying capacity (Eberhardt and Siniff 1977, DeMaster 1981,
Fowler 1981). We generalized DeMaster's (1981) density-
dependent relationship for survivorship to obtain a simple
non-linear function for population growth rate:

Yr = Ymax{1l-exp[—b(K-N) ] } (3)

where r is the annual per capita growth rate, r,,, is the
maximum annual per capita growth rate, N is the population
size, K is the equilibrium population size, and b governs the
shape of the curve. Because growth rates determined according
to (3) decline unrealistically rapidly when the population
size gets very far above the equilibrium level we impose an
arbitrary floor on the growth rate at -r,,, (Fig. 10.4).

Lett, et al., (1981) and Fowler (1981) point out that
density dependence in age-structured populations may be

201

FIGURE 10.3 -- Hypothetical reproductive curve depicting the
relationship between prime reproductive rate and senescence,
after Eberhardt (1985).

1.0 -—---------___ Sen.

oe prime reproductive rate

FECUNDITY

0.0

202

FIGURE 10.4 -- The effect of the value of b on the non-linearity
of the density dependence function used in OTPOP and LESLIE. XK
is the carrying capacity.

Em
z

b=s0.0025

oO

SEX ISe

Per capita growth rate (r)

Population size (N)

203

mediated through a variety of mechanisms including age at
first reproduction, pregnancy rate, age of weaning, juvenile
survival rate, and/or adult survival rate. Our California
field work indicates that juvenile females are under more
intra-specific competition pressure and suffer higher
mortality rates than adults. Our model incorporates a
hierarchical adjustment mechanism in the calculation of female
parameters to achieve the annual growth rates specified by
(3). Most density dependence is effected only through changes
in the female early risk coefficient (a, in (1)); if the early
survival rate drops below .40 or rises above the incidental
survival rate, the incidental female risk coefficient (a2) is
altered. If the incidental survival rate reaches 0.99 (i.e.,
extremely high values of r), the age at first reproduction (C-
1 in (2)) is reduced. Senescence parameters, and male
survival rates are held constant during a given run of the
model.

In an age structured population with a stable age
distribution and a constant rate of growth the relationship
between the survival and reproductive rates, and the growth
rate of the population is described by Lotka's (1907)
equation:

lme™ = 1 (4)

where x is age, 1, is female survivorship to age x, m, is the
number of female offspring recruited to the population from
each female of age x, and r is the rate of growth. Formulated
in this manner, Lotka's equation holds strictly only for
birth-pulsed populations (where all the reproduction takes
place during a short discrete time period) at a stable age
distribution. California sea otters reproduce throughout the
year, and so are not birth-pulsed; the stochastic nature of
the model precludes attainment of a stable age distribution,
even at equilibrium population size, except by fortuitous
coincidence. But we assume in the model structure and for
purposes of parameter estimation that Lotka's equation
provides an adequate approximation to the structure and
dynamics of the California sea otter population.

At the beginning of each time step in the model r is
calculated according to (3), and then Lotka's equation is
solved numerically by adjusting the female early hazards curve
until the appropriate 1, schedule is found. Because of the
violations in the assumptions of Lotka's equation the
stochastic fluctuation in age structure, the survivorship
schedule determined at the beginning of each time step will
result in a population growth rate that only approximates the
rate calculated in (3).

204

Parameter estimates

Our approach in estimating population parameters was to
begin with independently derived estimates, then vary them
systematically to arrive at a set of parameters that: 1) were
consistent with each other within the structural framework of
the model, and 2) were in reasonable agreement with the
original empirically derived estimates.

Survival. We estimated annual survival rates from our
telemetry data (see Chapter 2) using the method of Heisey and
Fuller (1985). The ages at which these rates applied were
estimated by calculating an average age for known aged
telemetered animals weighted by the number of days each animal
was observed:

x x;d; / a; ;i=1,N (5)

where x is the age at which the calculated annual survival
rate applies, x; is the age of animal i as determined from
tooth annuli (see Chapter 6), and d; is the number of days
animal i was observed. The values in Table 10.1 show the
effect on the estimates of using different age thresholds for
distinguishing between juveniles and adults. We thus obtained
pairs of age-specific annual survival rates for each sex that
could be fitted to survival rates calculated from (1).

Table 10.1. Estimates of annual survival rates of telemetered
sea otters in California, as determined by the method of
Heisey and Fuller (1985), 1983-1986.

Juveniles Adults
Age of Average Estimated Average Estimated
adulthood age! survival age survival
FEMALES :
1 0.61 0.69 5.94 0.91
2 1.41 0.78 7.06 0.92
3 1.88 0.83 7.88 0.92
4 2.16 0.83 8.09 0.91
MALES:
1 0.38 1.00 4.78 0.73
2 1.26 1.00 6.21 0.68
3 1.92 0.87 6.92 0.71
4 1.94 0.76 6.94 0.77

Separate estimates of annual survival can be obtained
from the sample of otter teeth collected from museums and aged

1 ‘5 2
Weighted according to eq. (9) in text.

205

by counting annuli. The method of age determination and the
problems associated with it are discussed in Chapter 6 the
distributions of estimated ages are shown in Fig. 10.5. The
number of animals represented in Fig. 10.5 is smaller than the
total sample because many of the collected animals could not
reliably be assigned a sex. The fact that, in the female
data, the numbers of animals in the even age classes is
consistently higher than those in the odd age classes (except
for age 5) is unexplained. And the fact that there

are apparently fewer animals of both sexes dying at age one
than at age two is questionable, and may be a result of small
carcasses not having the same chance of recovery as larger
ones (due, perhaps to more rapid decomposition or lower
visibility) or that incomplete dentition in some one year olds
may have resulted in selection against them when the teeth to
be aged were extracted from the museum collections.

Despite the uncertainties in the tooth data, we did
calculate annual survival rates using the "segment" method
(Chapman and Robson 1960), which assumes constant annual
survival after a threshold age, to compare to the estimates
obtained from the telemetry data. Assuming a stationary
population:

d, = N(1, - 141) = N(1s"° - Ds FS)
Hs

(6)

where N is the total population size, d, is the number of
animals dying at age x, 1, is survivorship to age x, s is the
constant annual survival rate, c is the threshold age, and H
is the constant represented by: dl,(1-s)s‘s’. Using the
Chapman and Robson (1960) regression to estimate a constant
annual survival rate between ages four and ten (i.e. c =4 in
(6), data for teeth older than 10 were not used) yielded a
female rate of 0.925 (s.e. =0.045) and a male rate of 0.723
(s.e. =0.038). If we assume that actual survivorship is as
in (1), the ages from four to ten represent the segment that
has survived the early hazards but is not yet greatly affected
by senescence. If we further assume that the Chapman-Robson
method estimates average survival during that age segment, we
can calculate the age at which the estimate applies (as we did
for the telemetry data) by calculating weighted average ages
of the samples in the segment. The weighted averages imply
that the female rate applies at age 7.63 and the male rate
applies at age 6.40.

The tooth data can also be used to estimate a,, the early

hazard coefficient in (1). Ignoring the portion of the sample
greater than 10 years old allows us to ignore the senescence

206

tters

la sea oO

-5 -- Distribution of ages of 425 Californ

ted by tooth cementum annuli technique.

FIGURE 10
estima

45

40

35

Mmmm SS CMales

ERR Females

KAA AAK/
60504

wo
N

YdsEWNnn

20

Ti,
YL

ALA
O06,

os
13

+

11. «12

10

ESTIMATED AGE

207

terms in (1), and, assuming that the early hazards are over
by age two, the proportion of animals dying at ages one or two
is:

p = (1-s;s*)/(1-s-s'”) (7)

where s,; is the survival rate against early hazards (i,e.,
-lIn(a,)) and sis the adult survival rate. Using the adult
rates calculated above yields a female early risk coefficient
of 0.051 (s- =0.95). The calculation for males yielded an
illogical value of 0.270 (s;=1.31), indicating that the number
of 1 year olds may indeed be underrepresented in the sample,
and making the female early survival estimate suspect also.

Reproduction.--Since m, and A in (2) are in terms of
offspring weaned, they in turn are functions of pregnancy
rates
and pup survival rates. Otters give birth to a single pup,
twinning is rare enough to be neglected in calculations.
Loughlin et al (1981) suggest a gestation period of four to
six months, a pup dependency period of four to eight months,
and annual reproduction. Wendell et al (1984) report annual
reproduction and a pup dependency period of five to eight
months. The longest period that we observed a radioed female
associating with a pup was about six months. We use constant
gestation and pup dependency periods of six months each in our
model. The pupping interval observed in our telemetered
animals was not significantly different from one year (see
Chapter 2).

Our telemetry data yielded a pup survival (from birth to
6 months) estimate of about 0.50 (see Chapter 2). Assuming
no interuterine mortality, a gestation period of six months,
and a pup dependency period of six months, this rate
translates directly to an annual survival rate of fetuses.

A separate estimate of the ratio of pup survival to adult
survival was derived from unpublished CDFG data collected in
index areas at monthly intervals between 1976 and 1982
(Wendell et al 1986). These data were counts of independent
otters and pups, and the pups were divided by size into a
small stage (here assumed to be from 0-3 months old) anda
large stage (here assumed to be from 3-6 months old) (Fig.
10.6). Average relative pup survival rate was estimated by
contrasting relative numbers of large pups with the peak
number of small pups three months earlier. If s, is the
average survival of independent otters in a given three month
period, and sp is the survival from the small pup stage to the
large pup stage, relative numbers of independents (I), large
pups (L), and small pups (S) can be expressed as:

L/I = (S/T) (So/S1) (8)

208

FIGURE 10.6 -- Relative average number of small pups and large
pups, by month, in the CDFG index areas, 1977-1984.

SMALL PUPS

PUPS PER INDEPENDENT

0.10
0.08 a. LARGE PUPS
0.06 z
0.04 9
0.02

0.00

PUPS PER INDEPENDENT

and, if q is average pup survival relative to adult survival,
q = S/S; = (L/1I)/(S/T) (9)

From the CDFG data q = 0.439/0.607 = 0.723. The observations
were taken three months apart; if we assume that q measures
survival over a three month interval, the relative monthly
survival is the cube root of gq, =0.898. This relative rate
can be used to calculate an absolute estimate of pup survival
once the adult survivorship schedule is determined.

Using the pup survival estimate derived from telemetry,
and assuming a 99% pregnancy rate for prime aged animals and
an even sex ratio at birth, the prime weaning rate (A in (2))
is 0.50*0.99*0.5 = 0.247. In the model pup survival is held
constant over all maternal age classes, and age specific
weaning rates are achieved by varying pregnancy rate with age.

The age of first reproduction for female sea otters in
Alaska appears to be about four years (Schneider 1972),
although limited observation in California suggests that some
otters may reproduce as early as three years. If age at first
reproduction is taken as 4 the parameter C in (2) is set at
3.

Senescence.--In sea otters, which are not subject to
heavy natural predation or killing (gillnet, shooting or
harvest) by humans, and which exhibit low reproductive rates
and long lifespans, the senescence parameters are likely to
be more important than for many other mammalian species.
Siler (1979) relates a; and bz; from (2) and (5) to the modal
age of senescence and the standard deviation around that age:

az = exp(-T/S) (10)

and

b; 1/s (11)
where T is the modal age of senescence and S is the standard
deviation. The senescence parameters can thus be derived from
estimates of T and S. Data from Schneider (1978) suggests
that the modal age of senescence in Alaska otters may be 16
years. Eberhardt (1985) found a significant correlation
between T and S for 10 species of large mammals. A regression
of the data presented by Eberhardt (1985) yielded:

S = 0.161 + 0.144T ;(R°=0.84) (12)
Using this equation and a modal age of senescence of 16, the

standard deviation of age of senescence is 2.46, az = 0.0025
and bz = 0.41.

210

Density-dependence.--Analysis of historical range
expansion and population size data for the OTRANGE model,
discussed elsewhere in this report, suggested values of 0.09,
and 0.035, for Ypx and b, respectively, in (3). Given the
present length of the range and USFWS (1986) estimates of
substrate-specific carrying capacities, the model uses a
default value of 1920 animals for K.

Reconciliation of estimates.--A "spread-sheet" type
program was set up to allow testing of adjustments in the
parameter estimates. Equations (1) and (2) contain a large
number of parameters relative to the number of "data points"
available to use in estimating them; unstructured numerical
solutions to (1) based on annual survival rates at only two
ages undoubtedly would be degenerate. The "Shape" parameters
(b,, bs, ba in (1) and (2)) are particularly difficult because
of their abstractness and their potentially large effect on
the survival and/or reproductive rates at certain ages. We
thus structured our search for the best parameter estimates
as follows:

Siler (1979) describes the time constant at which
maturity is approached (i.e., the rate at which the early risk
becomes asymptotic) as 1/b;. In five mammal populations 1/b,
was always less than one, with the value of b, ranging from
1.06 to 3.84. We conservatively set b, equal to one for both
males and females, giving a relatively slow approach to
maturity, and emphasizing the apparently intense interspecific
competition and reduced survival that young female sea otters
experience. Similarly, b,, which governs the rate at which
the prime reproductive rate is approached, was set equal to
one. The relationship between the senescence parameters, bz
and az, was held initially at the relationship suggested by
Eberhardt (1985), and (12) above, and modal age of senescence
was varied to affect the shape of the survivorship and
reproductive curves through the adult ages. The prime
reproductive rate was set at 0.25 and held constant to start
with, and a;, az, and the modal age of senescence were varied
to find estimates combinations of survivorship and
reproductive schedules that: 1) were in reasonable agreement
with estimated annual survival rates, 2) produced sex ratios
of independent animals in the simulated population in
reasonable agreement with those seen in the wild (i.e.,
female-biased), and, 3) provided reasonable survivorship
schedules at all population growth rates between r,,, and -
IYmxe ANnual survival rates were calculated from constructed
survivorship schedules by :

Sx xa, xt (13)

where s, is annual survival rate at age x and 1, is
survivorship to age x.

211

Following this approach, we arrived at the parameters
given in Table 10.2; these are the default model parameters.
With these parameters exp(-al) (early survival rate) in (1)

Table 10.2. Default parameters used in OTPOP and LESLIE.

Parameter Equation in text Value

Maximum per capita Tey ee Tiea( 3) 0.085
growth rate

Non-linearity of density Voy “altey -(())) 0.020
dependence

Equilibrium population K in (3) 1720
size

Adult female risk a2 in (1) -1n(0.93)
coefficient

Female modal age of ud Ma Bo Wa OD) neS)
senescence

Standard deviation of female Sra meals) 2.46
age of senescence

Female senescence risk az; in (1) -1n(0.9977)
coefficient

"Shape" parameter for female b,; in (1) and (2) ait
early hazards risk

"Shape" parameter for female bz in (1) and (2) 0.41
senescence risk

Prime reproductive rate A in (2) 0.25
"Shape" parameter for approach ba in (2) 1
to prime reproduction

Age before first reproduction Cuan; 1(2))s 3

Adult male risk a2 in (1) 1n(0.87)
coefficient

Male modal age of T in (10) 9
senescence

Standard deviation of male Sian (asl) 3.5
age of senescence

Male senescence risk az in (1) -1n(0.9264)
coefficient

Male early hazards risk a, in (1) 0
coefficient

"Shape" parameter for male bz in (1) and (2) 1

senescence risk

for females varies from 0.96 when the population growth rate
is 0.09 to 0.41 when the population growth rate is -Yyax;
exp(-a2) (incidental survival rate) for females is 0.93,
dropping to 0.91 when r = -Ypax, and increasing to 0.97 when
Y = Ymx (Fig. 10.7). The age of first reproduction is lowered
to three years when r > 0.06 (Fig. 10.8). The modal age of
senescence for females remains constant at 15 years. The fit
of the various estimates of female survival rates to the
default model rates for r=0 is shown in Fig. 10.9.

For males, the early hazards survival rate is set at one
(a, =0, if a male survives weaning there are no additional
hazards associated with youth, the value of b, for males is
thus inconsequential), but the incidental survival rate (exp(-
ao) is 0.87, and the modal age of senescence is nine years.
A modal age of senescence of nine years implies, by (12), a
standard deviation of 1.46 years around that age, values of
az; and bz; calculated as such and combined with the previously
determined rates for a, and ,2 gave a good fit to the two
survival estimates, but led to there being no males over 12
years of age in the population. This is obviously
unrealistic, as four of the 219 male teeth were estimated to
be 12 years or older. Adjusting the standard deviation of
senescence upwards to 3.5 years led to a good fit of the
annual survival estimates and ensured that male otters could
survive until 16 years of age. The low modal age of
senescence combined with a relatively large standard deviation
results in the effects of senescence being manifest in the
male survivorship curve at an early age (Fig. 10.10). MThis
may seem anomalous, but if the concept of "risk of senescence"
in males is stretched to include the risks associated with
holding a breeding territory against younger animals, and if
these risks include a mortality rate that increases with the
years that an animal is not able to hold a territory, the
curve may be biologically justified.

With a population per capita growth rate (r) of zero, the
predicted sex ratio in the independent population under this
parameterization is 0.755 males per female (Fig. 10.11). When
r =0.9 the predicted sex ratio is .572 males per female; when
r=-0.09 the predicted sex ratio is 1.06 males per female.

Using the survivorship schedules described above with

r=0, the average annual survival rate of independent otters
is calculated as:

Sa (1yy/ lyy) ¥S,
=0.844 (14)

taking the 12th root yields an average monthly survival rate
of 0.986. Using the relative pup survival rate calculated

213

FIGURE 10.7 -- Age-specific female survivorship curves and annual
survival rates at different per capita growth rates, under the
default population parameters used in OTPOP and LESLIE.

or
SURVIVORSHIP (¢

ANNUAL SURVIVAL (---)

t 4 J
Oo 1 304 so Geno) 9 10 11 12 13 14 15 16 17 18 19 20 21 22 25

AGE

214

FIGURE 10.8 -- Age-specific reproductive rates under the default
population parameters used in OTPOP and LESLIE.

2
Le)
as

rst eProduction at age 3

0.08

First reproduction at age 4

Fi

WEANED FEMALES PER FEMALE

10 11 12 13 14 15 16 17 18 19 20 21 22 23

AGE

Py By Ch G7 tye)

Oo 1

2515)

FIGURE 10.9 -- Age-specific female California sea otter annual
survival rates calculated under default model parameters and a
per capita growth rate=0 compared to survival rates estimated
from field data. See text for explanation of estimates.

ANNUAL SURVIVAL RATE

0.9 Chapman—Robson estimate
+/— 1 3.e.)
(Applies at ages 4—10)
0.8 -
A Heisey—Fuller estimate

0.7 (adults > 2 years old)
0.6 | Heisey—Fuller estimate

E (adults > 3 years old)
0.5 -
0.4
0.3
0.2 -
0.1

0 12 3 4 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

AGE

216

FIGURE 10.10 -- Age-specific male California sea otter annual
survival rates calculated under default model parameters and
a per capita growth rate=0 compared to survival rates
estimated from field data. See text for explanation of estimates.

Chapmon—Robson estimate
(+/— 1 s.e.
(Applies at ages 4—10)

q A Helsey—Fuller estimate
(adults > 2 years old)

0 Helsey—Fuller estimate
(adults > 3 years old)

ANNUAL SURVIVAL RATE

0123 45 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

AGE

217

FIGURE 10.11 -- Male and female California sea otter
survivorship curves under the default population parameters
used in OTPOP and LESLIE and a per capita growth rate=0.

Female

SURVIVORSHIP -

012 3 4 5 67 8 9 1011 12 13 14 15 16 17 18 19 20 21 22 23

AGE

218

above in (9), the absolute monthly pup survival rate is then
0.986*0.898=0.885, and an annual rate (assuming six months
dependency, six months gestation, and no interuterine
mortality) of 0.482. This is not too distant from the 0.50
derived from the telemetry data when it is realized that the
index areas, being areas of relatively high pup abundance,
certainly contain a higher proportion of females than in the
population as a whole, and the average adult survival rate as
calculated above thus underestimates the average survival of
adults in the index areas. By contrast, the average annual
female survival rate (calculated as in (14) but ignoring
males), is 0.87, and the annual pup survival rate calculated
with that rate is 0.499.

It should be reemphasized at this point that the
parameter estimates supplied above are the model default
values. They may be easily changed by the user to examine the
effects of alternative parameterizations, or if future
research allows refinement of the estimates.

Seasonality.--The CDFG index area counts indicate a
pronounced seasonality in the abundance of pups; this
additional complexity in sea otter reproduction is taken into
account in the model. The abundance of small pups in the
index areas (Fig. 10.6) and the pooling across years (Fig.
2.1, Chapter 2) indicated a peak of pup production in March
and a low point in June or July; the biological basis for
this seasonality is as yet unknown, it may be due to favorable
pup-rearing conditions in the spring and summer leading
directly to a synchronous breeding season, or it may be that
pups born in the fall suffer high mortality and the females
rebreed immediately. We model the seasonality descriptively,
assuming a constant monthly pup mortality, and a constant
monthly base rate of pupping, adding a seasonal pupping rate
to give the observed spring peak.

With a basic year round rate yielding 0.05 pups per
independent as a starting point, and assuming that the small
pup stage lasts three months the basic monthly rate is
0.05/3.0 = 0.0167. An increased seasonal rate for December
to April was then fit to the data in Fig. 10.6 by a numerical
least squares procedure. This added increment was 0.02,
making the monthly pupping rate for the December through April
0.0167+0.020 = 0.0367. Since these rates were based on
relative numbers of pups and independents the ratio of peak
monthly pup production to basic monthly pup production,
0.0367/0.0167 = 2.21 is the important parameter. The model
uses this ratio and the age-specific pregnancy rates to
determine the probability of conception in each month. The
annual pregnancy rate is decomposed into the two relative
pupping rates by numerically solving:

(1-p) = (1-u,)’(1-u,)° (15)

(i.e., the probability of not conceiving during the year is
the product of the monthly probabilities of not conceiving)
where p is the age-specific annual pregnancy rate, u,, is the
basic monthly pupping rate, uz is the peak monthly pupping
rate, and up = 2.21u,;. Intrauterine mortality is again assumed
to be zero, and gestation is again assumed to be six months,
so that the conception rate for November through May is u, and
the conception rate for June through October is up).

Age specific pregnancy rates for use in OTPOP are
calculated according to:

Px = 2m, / {1-(1-s,) —(1-Sp) +(1-S,) (1=So) } (16)

where p, is the annual pregnancy rate for females of age x,
m, is the reproductive rate from (2), s, is the annual survival
rate for females of age x (as in (13)), and Sg is the annual
pup survival rate. This calculation assumes an even sex ratio
at birth and assumes that the number of animals weaned in a
year (2m,) will be the number of pregnancies minus the number
of females that die during the year (1-s,) minus the number of
pups that die during the year (1-S)). Because pup death is
not independent of maternal death (if a mother dies her pup
dies also) the interaction term, (1=-s,) (1-S9), is added. The
age specific pregnancy rates are used to calculate age
specific monthly conception rates according to (15). The
pattern of monthly pup production Generated by OTPOP appears
aliny JaaIe?4 BLOBS MA

In LESLIE maternal deaths are not considered in the
calculation of pregnancy rates because in the life table
formulation using an annual time step only the animals that
survive the year are available to reproduce, so pregnancy
rates are:

Px = 2m,/So (17)

Constructing an initial population

The population growth rate for the first year of
simulation is determined according to the density dependence
function (3) given the initial population size and the
carrying capacity supplied by the user. Survivorship
schedules for each sex are calculated according to (1) using
the population growth rate and the sex-specific risk
parameters and senescence functions. The proportion of the

220

FIGURE 10.12 -- Pattern of monthly pup abundance produced by
OTPOP. Crosses represent the mean of 85 years of simulation.

Total pups

170
150
Zz
Lid
Z
4G 150
Oo
Ld
OQ
Z 110 Small pups
oO
oO
O 30
=

N
om
LJ
Qo 70
io)
oO
2 50 Large pups
N D J F M A M J J A S 0 N
-MONTH

population of each sex in each year class is then determined
by:

Vt Cha lana Camel, ;x=1,W, y=1,2 (18)

where V,, is the proportion of the population of age x and sex
ye

Initial conditions require a distribution of ages in
months within each age class, and an initial distribution of
female reproductive state. Deterministic simulations
convinced us that, given the relative conception rates
described above, the distribution of reproductive status (and
thus of month of birth) converges to a stable distribution
from any initial distribution within a small number of
iterations. This stable distribution depends on both the
basic conception rate (u;) and the pup survival rate. In any
run of OTPOP the pup survival rate is constant, but the
conception rate varies by age according to (2) and (16). An
average basic conception rate, weighted by the initial stable
age distribution, is calculated; beginning with an initial
uniform distribution of reproductive states the distribution
is simulated deterministically for 15 years on a monthly
basis. From the final (assumed stable) distribution of
reproductive status a distribution of month of birth is
extrapolated, and then these distributions are converted to
cumulative density functions.

The age in months of each independent animal in the
initial population is determined by a random draw from the
month of birth function. The number of pups and fetuses in
the initial population is extrapolated from (18) given the
initial population size and assuming a stable age
distribution. The required number of pups and fetuses are
distributed through the female age classes in proportion to
the elements of the l,m,e™ vector, and the age of each pup or
fetus within each female age class is determined by a random
draw from the cumulative density function of reproductive
status. Initial conditions generated in this manner alleviate
the need for long runs of the model prior to introduction of
an oil spill.

OTDIST
Structure

OTDIST distributes the simulated population along the
coast. It differs from the other submodels in the fact that
it is not dynamic. In OTDIST we assume that the position of
an individual otter along the coast is a function of the
animal's sex and reproductive status, the month of the year,

empirically derived population density functions, and
empirically derived local sex ratios.

We have digitized the coast and the associated five
fathom ordinate system for use in the model (Appendix 10.1).
For certain aspects of the model it is useful to consider the
coast in discrete divisions; in these cases we have used the
40 coastal segments delineated by the CDFG and USFWS in their
carcass recovery efforts.

Two density functions are used as input for OTDIST, one
giving the density of otters at any point along the coast in
May, the second giving the density in December. Similarly,
functions giving the sex ratio in each of the 40 coastal
segments in May and December are required input for the model.
The first step in the distribution procedure is to determine
local sex ratios and a density function for the desired month
of the year by linear interpolation between the May and
December functions. Then the density function and local sex
ratios are combined to arrive at density functions for each
sex by multiplying the density at each point along the coast
by the proportion of each sex at that point.

The female density function is then converted to a
cumulative distribution function, and the program loops
through the female age vector, determining each animal's
position by generating a random number and finding a
corresponding location in the cumulative function.

After the females have been distributed OTDIST
distributes territorial males. First potential territories
are set up along the entire coast. Territory size is assumed
to be normally distributed, the location of potential
territories are determined by generating lengths from a normal
distribution with empirically derived parameters, and stacking
them along the coast. Coastal substrates are classified as
either rock or sand, as determined from U.S.G.S. topographic
maps; only rocky areas are allowed to be potential
territories, as no territorial males have been observed in
sandy areas in California (B. Hardy, pers. commn).

Once the locations of potential territories are
determined, the territorial status of each male in the
population is determined. Males six years old and older are
potentially territorial. The probability that a potentially
territorial male will actually be on a territory is viewed as
a function of month of the year, and determined as follows:

P(t) = P(a) (1-P(d)) (19)

where P(t) is the probability of being on a territory, P(a)
if the probability of having arrived on a territory, and P(d)

223

is the probability of having departed a territory. P(a) and
P(d) are assumed normally distributed with empirically derived
parameters. The program then loops through the male
population vector; when a male of 6 years or greater is
encountered a random number is generated and compared to P(t)
to determine territorial status. If the male is territorial
the location of its territory is determined by generating
another random number and finding a corresponding location in
the cumulative distribution function for females. The length
of coast encompassed by this territory is then made
unavailable to non-territorial males by setting the male
density function equal to 0 at all points within the
territory. If a male draws a territory that is already
occupied by a territorial male he is moved to the closest
available unoccupied territory.

The last step in the program is to distribute the non-
territorial males. The male density function, as modified by
territoriality, is converted into a cumulative density
function. Another loop through the male population vector is
executed; if a male does not have a territory a random number
is generated and the corresponding location in the male
cumulative distribution function is found.

Once the positions of all animals in the population have
been determined, the population and location vectors are
passed to OTMOVE, the short term movement and oil response
model. The locations of males occupying territories are
flagged as they are passed to OTMOVE.

Parameterization of OTDIST

Biologically, the spatial distribution of animals ina
population can be viewed as a function of the distribution of
resources and social interactions amongst the individuals in
the population. Our understanding of the actual mechanisms
that produce an observed distribution of animals from the
underlying distribution of resources and social system is very
incomplete, so we must be content with modeling the
distribution in an essentially descriptive manner,
incorporating few mechanisms. OTDIST, the distribution model,
has been structured to utilize what data is available on the
distribution of sea otters within their range in California.

Density functions.--Required inputs for OTDIST include
density vectors for winter and summer, representing the number
of animals in each 500m segment of the five fathom ordinate
system in November and May, respectively. CDFG and USFWS
census data were used to construct these vectors. Dates and
methods of the censuses for which we had access to the raw
data are given in Table 10.3, census methods are described
Wendell, et al., (1986).

Interpretation of the census data suffered from the
questions of range expansion mentioned at the beginning of the
report. For our analysis the locations of animals recorded
on field maps during the censuses were translated to five
fathom line ordinates and summed by census and 500m segment.
Contiguous blocks of 20 500m segments, representing

Table 10.3. CDFG and USFWS censuses used in analysis of
California sea otter distribution.

Date Average
Year Month Total Count Group Size Method

1968 Aug 311 5.27 Air
1968 Nov 659 4.92 Air
1968 Dec 409 3.56 Air
1969 Jan 986 5.94 Air
1969 Feb 685 3.26 Air
1969 Mar 942 4.34 Air
1969 Apr 654 5.03 Air
1969 May 315 4.32 Air
1969 Jun 1013 5.30 Air
1969 Aug 528 3.74 Air
1969 Sep 404 3}, Sal, Air
1969 Oct 485 3.13 Air
1970 May 902 5.43 Air
1970 Sep 607 4.40 Air
1971 Feb 719 3.95 Air
1971 Apr 901 3} oa Air
1971 Jul 957 4.65 Air
1971 Oct 712 4.07 Air
1972 Jan 1064 3.81 Air
1972 Apr 772 2.81 Air
1973 Dec 936 3.38 Air
1974 Mar 956 2.14 Air
1975 Jun 1040 2.30 Air
1976 Jun 1148 2.12 Air
1979 Jun 808 2.32 Air
1982 Nov 1334 akg Ia Ground
1983 Oct 1222 2.06 Ground
1984 Jun 1567 2.32 Ground
1985 May/Jun 1287 2.14 Ground
1985 Oct/Nov 1212 2.20 Ground

10 km of coast each, were grouped together and totaled by
census. The percent of the total count of each census in each
block was then calculated and used in an analysis of variance.
Since sea otter distribution is generally considered to change
on a seasonal basis (USFWS 1986) we grouped the census within

225

each year by season: winter consisting of November through
April, summer consisting of May through October (data were too
sparse to attempt analysis by month). ANOVA (Table 10.4)
showed significant main effects of both year and location, and
of season within year. There was also significant interaction
between year or season and location. The situation is
represented graphically in Figs 10.13 and 10.14.

SS ____eese__......_._.______ Ee

Table 10.4 -- Analysis of variance in CDFG and USFWS
California sea otter census data, 1968-1985. Dependent
variable is the proportion of census total along a 10 km
section of coast.

Source d.f ss F

Year 13 -06065 3.98 0.0001
Season within Year 6 -01985 2.82 0.0100
Location 40 - 41554 8.87 0.0001
Year * Location 362 - 18867 0.45 0.9900
Season & Location 127 »10161 0.68 0.9900
Error 240 ~28108

Total Model 548 - 78632 1.23 0.0300

Model R=0.737

Variation in distribution due to location and season are
easily handled by the structure of OTDIST, but, as mentioned
previously, OTDIST is time-independent, and annual changes in
distribution (aside, of course, from pure stochastic effects)
are not explicitly considered. Distribution of sea otters
throughout the range in California has undoubtedly changed
since the censuses began. We thus decided that the best
parameterization of the density functions would be direct
incorporation of the most recent census data. A separate
analysis showed a highly significant effect of census method
on both total count and on average group size. Ground counts
appear to enumerate a greater proportion of the population,
particularly that part of the population that is solitary on
the day(s) of the census. The relative worth of ground counts
versus ground-truthed aerial counts for estimating population
size has been the subject of some debate, but for present
purposes, that of determining relative distribution, it seems
as if the method that enumerates the higher proportion of the
population will better estimate relative densities. This
contributed to our decision to use only the most recent
censuses, aS coordinated ground counts did not begin until
1982.

Census data from 1984 and 1985 were converted to

probability density functions for each census by dividing each
segment total by the total number of animals recorded in the

226

FIGURE 10.13 -- Contour diagram indicating annual changes in
sea otter density in California from 1968-1985. Y-axis
represents space (5 fathom line ordinate system, increasing,
generally from North to South, in 500 meter increments along
the five fathom line), x-axis represents time. Diagram was
constructed for CDFG and USFWS census data (Table 3).
Individual census counts were totaled by 20km section of coast,
and the proportion of the total count in each section
calculated. These proportions were then averaged by year to
get the values used to produce the diagram. Contour interval
is 3% of the individual census total. Expansion of the range
is evident at the north and south ends of the diagram.

200

LOCATION ALONG 5 FATHOM LINE

YEAR

227)

FIGURE 10.14 -- Contour diagram indicating monthly changes in
sea otter density in California from 1968-1985. Y-axis
represents space (5 fathom line ordinate system, increasing,
generally from North to South, in 500 meter increments along
the five fathom line), x-axis represents time. Diagram was
constructed from CDFG and USFWS census data (Table 3).
Individual census counts were totaled by 20km section of coast,
and the proportion of the total count in each section
calculated. These proportions were then averaged by month to
get the values used to produce the diagram. Contour interval
is 3% of the individual census total. Seasonal contraction of
the range, presumably due to migrations by males, is evident
at the north and south ends of the diagram between February and

July.

LOCATION ALONG 5 FATHOM LINE

228

respective census. These density functions were then averaged
by season to get the empirical distributions that are used in
the model (Fig. 10.15). Approximately 30% of the range is
inaccessible by road, and thus is counted by air, even during
the "ground" counts. Geibel and Miller (1984), Wendell et al
(1986), and Hardy (CDFG, personal communication) report that
aerial observers typically enumerate from 50-80% of the
animals seen by ground truth observers; the density functions
derived from the raw census data are thus further modified by
multiplying the densities in portions of the range that are
counted from the air are multiplied by 1.3.

It is assumed that differences between the two seasonal
distributions are due to seasonal movements of animals,
particularly adult males migrating between male areas in the
winter and breeding territories in the female areas in the
summer. The censuses are timed to reflect the peak of
congregation in the male areas and the peak of territoriality,
respectively; therefore density functions for each month are
obtained by linear interpolation between the summer and winter
functions.

The censuses are an integral part of CDFG's and USFWS's
sea otter research programs, and are scheduled to continue to
be conducted twice yearly in the future. The density
functions are stored in an external file so that they can be
easily updated after each census and incorporated into the
model. USFWS has been supplied with a computer program that
allows rapid digitizing of raw census data and outputs the
data in an appropriate form for the model.

Sex ratios.--The census data provide distributions for
the population as a whole, but provide very little information
about the distribution of each sex throughout the range. Some
local sex ratio information is available from the carcass
recovery data. The location of each recovered carcass is
recorded by recovery area (Ames et al 1985). Each recovery
area is about 12km long. Sex ratios for each recovery area
were calculated, combining all data from December through
April and from May through November (Table 10.5).

The carcass recovery data suffers from small sample size
in many of the areas, from sampling problems discussed by Ames
et al (1984), and, again, from changes in distribution that
have occurred as the range expanded. As an alternative we
used a delphic technique, asking field biologists from CDFG,
USFWS, and other institutions who had been frequent
participants in the semi-annual censuses to estimate present
seasonal sex ratios in each of the recovery areas. Averaging
the responses from the questionnaire gave the sex ratios in
Table 10.5 and Fig. 10.16. Each of the respondents was
familiar with the carcass recovery data, and most said that

229

FIGURE 10.15 Density functions used in OTDIST for the
location of independent sea otters in California in June
(dashed line) and December (solid line). Functions derived
from CDFG and USFWS censuses conducted in 1985. OTDIST works
at a resolution of 50m, but for clarity densities were totaled
by 20km sections to construct this diagram. The numerical
value of the locations along the five fathom line generally
increases as one moves to the south along the California Coast.

ry

0.12 ra
Zz 1\
O i t
~ 53
< 0.10 - i) ! \
! 1
3) 2 \ iN i \
3 Teoog einer
0.08 \ \
ral I | i \ [= : \
Le i { ! ; /\
© 0.06 - es i visleoaty
FL, \ i \ \
Oo Se \
~ { ‘
be 0.04 - | | x
oO A \
QO. 7N > \
\ > S
o 0.02 + /,” . 7 § 2
ra / \i/ E
y \f 2 =
0.00

LOCATION ALONG THE 5 FATHOM LINE

23C.

\

=~
~

220 260 300 340 380 420 460 500 540 580 620 660 700 740 780 820 860 900 940

FIGURE 10.16 -- Local proportion of California sea otters that
are female in June (dashed line) and December (solid line),
used in OTDIST. Data, from questionnaires distributed to
knowledgeable biologits, was collected by CDFG carcass recovery
area and assumed to apply over the entire section of coast
within each recovery area. The numerical value of the
locations along the five fathom line generally increases as one
moves to the south along the California Coast.

0.8

Lif
—! 0.6
=e
=
Ld
Le
Zz
O 0.4
kK
or
O
oO
©
o. 0.2
o
s
o
=
0.0

200 300 400 500 600 700 800 900

LOCATION ALONG THE 5 FATHOM LINE

Zoi

1000

they used that data as a guideline in making their estimates.
We feel that their estimates are more realistic than those
that could have been attained by a purely statistical analysis
of the sparse data, and therefore used their estimates in the
model.

Table 10.5. California sea otter sex ratios of recovered
carcasses (1968-1985) and as subjectively estimated by field
biologists, by season and CDFG carcass recovery area.

CDFG Recovered carcasses Subjective estimates
recovery May-Oct Nov-Apr May-Oct Nov-Apr
area® M F M F M/F M/F.
<7 3 (0) 1 0) 4.5 16.7
7 1 0 3 3 4.5 16.7
8 1 0 5 0 4.5 16.7
9 1 0 0) 3 4.5 16.7
10 2 0) 2 1 4.5 16.7
11 42 8 27 6 4.5 16.7
12 26 24 6 1 1.0 3.0
13 36 24 35 30 0.5 0.9
14 20 20 22 13 0.5 0.9
15 2 6 4 5 0.5 0.9
16 16 11 13 23 0.3 0.3
17 2 5 9 16 0.3 0.3
18 11 30 25 48 0.4 0.4
19 3 17 14 16 0.4 0.4
20 1 7 7 13 0.5 0.4
21 1 5 1 1 0.5 0.4
22 2 4 2 4 0.5 0.4
23 (0) 0 2 5 0.5 0.4
24 (0) 1 1 (0) 0.5 0.4
25 0 1 5 1 0.5 0.4
26 (0) 1 0 0 0.4 0.3
27 3 10 5 6 0.4 0.3
28 16 10 15 17 0.4 0.3
29 18 31 12 30 0.3 0.3
30 10 6 8 2 0.3 0.3
31 16 9 5 7 0.8 0.6
32 70 36 27 18 0.8 0.6
33 25 16 25 12 0.8 0.6
34 2 2 2 2 1.0 3.0
35 5 1 2 0 1.0 3.0
36 22 2 5 3 4.5 16.7
37 33 7 12 5 4.5 16.7
38 4 1 2 0 4.5 16.7
39 0 0 1 0 4.5 16.7
40 2 0 1 2 5.0 17.0
See Appendix 1 for location of CDFG carcass recovery areas.

232

As for the seasonal distributions, the seasonal sex
ratios are assumed to be the result of male migrations to and
from breeding territories, and sex ratios for the each month
are obtained by linear interpolation. Probability density
functions for each sex in a given month are then obtained by
multiplying the population density function for the
appropriate month by the appropriate sex ratio:

Sijk = AjRPijx (20)

where s;;, is the density of sex i, in month j in 500m segment
k, Gj, is the population density in month j in segment k, and
Pijk 1S the proportion of sex i in month j in segment k, and
sex ratio in a recovery area is assumed to apply to all 500m
segments in that recovery area (Fig. 10.17).

Parameters pertaining to territoriality.--The model divides
the coast into potential territories. Territory length is
assumed normally distributed, parameters of the distribution
were obtained from Jameson (1987), who reported mean male
territory length near Piedras Blancas during 1978-1984 at 1.1
km (s.d. =0.43km, N=13), and our telemetry data. Additional
parameters required by the model are mean territory arrival
and departure dates, and associated standard deviations.
Jameson (1987) gives mean arrival date as 22 May (s.d. =33.6
days, N =16) and mean departure date as 21 December (s.d.
=38.1 days, N =18). The proportion of males over the age of
six that are territorial at any given time is assumed to vary
seasonally between 0.25 and 0.75.

The data from Jameson (1987) provides default values for
the male territoriality parameters in OTDIST; but our
telemetry data suggests that the highly seasonal pattern of
male movements observed by Jameson (1987) during 1978-1984 may
not be occurring at the present time or throughout the entire
range (see Chapter 3). The user can change the values of the
territoriality parameters at the beginning of a model run;
setting large standard deviations of arrival and departure
dates and/or small differences between the minimum and maximum
proportions of males that are territorial will reduce the
amount of seasonal variation in male territoriality in the
model.

Expanded sea otter range

As mentioned previously, the size of the sea otter range
and its carrying capacity are fixed for the duration of any
run of the model. Since, however, the peripheral areas of the
current range are the most susceptible to oil spills, and it
is very possible that the range will continue to expand in the

233

FIGURE 10.17a -- Density functions used in OTDIST for the
location of male (dashed line) and female (solid line) sea
otters in California in June. The numerical value of the
locations along the five fathom line generally increases as one
moves to the south along the California Coast.

0.14
Zz 012
2)
K-
yore
=)
au
O “\ \
OL 0.08 7 \ " , I

atee at NV 1\ \ f
O 7\ | \ F \ a \ ! .
=> 9.06 ee F \ ! Ny \\ / _
O J \ I \ 1 \\ / \
f= Sys: Lopdl! sl 7 \
/
004-4, i: Se "
O I 7 . :
QO. U \ > [e} \
O 0 oO :
(v 0.02 - Vig ° a
ij = E a

i. WE 8

0.00

220 260 300 340 380 420 460 500 540 580 620 660 700 740 780 820 860 900 940

LOCATION ALONG THE 5S FATHOM LINE

234

FIGURE 10.17b -- Density functions used in OTDIST for the
location of male (dashed line) and female (solid line) sea
otters in California in December. The numerical value of the
locations along the five fathom line generally increases as one
moves to the south along the California Coast.

0.12

PROPORTION OF POPULATION

>
2
o
ra
:
6
=

0.00 =
220 260 300 340 380 420 460 500 540 580 620 660 700 740 780 820 860 300 3940

LOCATION ALONG THE 5 FATHOM LINE

2335)

near future, we have constructed a simple deterministic model
of sea otter range expansion, OTRANGE, which was described
earlier. The output from OTRANGE includes projected
population sizes, range boundaries, and carrying capacities,
which can be used to initialize OTPOP. Obviously, the census
data required as input for OTDIST will not be available for
the peripheral parts of the simulated range as expanded by
OTRANGE. Thus we provide a sub-routine, EXPRAN, that, in the
event that the north and south ends of the census data used
as input do not coincide with the range boundaries input by
the user, will predict the distribution of otters in the
expanded range.

After examination of the data in Figs. 10.13,10.14,10.15,
and 10.16, we arbitrarily divided the range into a southern
periphery, south of the 840 ordinate on the five fathom line,
a northern periphery, north of the 340 ordinate on the five
fathom line, and a center, between the 340 and 840 ordinates.
Using the density functions constructed by OTDIST from the
census data, EXPRAN calculates the average density of otters
of each sex in the center of the range (Fig. 10.18, point A),
and assuming that the densities at the endpoints of the
censused range are zero, calculates the average slope of the
distribution through the north and south peripheries (Fig.
10.18, slope S). Density is then extrapolated from the end
of the central range into the peripheral range a distance
equal to the length of expansion using the central range
average density combined with censused deviations from the

slope of the peripheral density (Fig. 10.18, line A-B). In
this manner the central range is thus considered to have
extended into existing peripheral range. Densities in the

peripheral part of the range beyond the expanded central part
of the range are calculated by adding the difference in
density implied by the slope of the peripheral density and the
length of expansion (Fig. 10.18, line B=-C). Density in the
expanded part of the range is made cumulatively equal to the
area of CDE in Fig. 10.18, but weighted at any point according
to the type of substrate. USFWS (1986) estimates that rocky
habitats can support 3.1 times as many otters as_ sand
habitats, this ratio is used in the determining the densities
in the expanded range. The heavy line in Fig. 10.18 indicates
the new densities calculated by EXPRAN.

OTMOVE
Structure

OTMOVE simulates the movements of the animals in the
population on a daily basis for up to 30 days, and checks
animal positions against the location of a simulated oil

spill. It iterates on a daily time step, and considers

236

FIGURE 10.18 -- Schematic representation of the algorithm
used in EXPRAN to predict sea otter densities in expanded
range. See text for explanation.

RELATIVE
OTTER DENSITY

Existing: (Expanded Range periphery Range center
Range

a ip

Range periphery Range center

POSITION ALONG COAST

Expanded:

237

position at a spatial resolution of 50 meters. The location
vectors generated by OTDIST are the initial positions of the
animals, the identity of animals dying as a result of contact
with the spill are passed to OTPOP.

As in OTDIST, spatial considerations are simplified by
conceptualizing the system as one dimensional. Otters are
located and move upcoast and downcoast on the five fathom line
ordinate system, and oil spills are 1 dimensional also. At
the beginning of a run the date, duration (in days), and
boundaries of the oil spill are input. Since the purpose of
the movement model is to determine the numbers of animals that
die as a result of the spill, it runs for only as many days
as the duration of the spill. The movements of each animal
are assumed independent of the movements (but not the spatial
distribution) of the other animals in the population. This
allows a structural efficiency of looping days within animals
rather than animals within days; the movements of each animal
are simulated for the duration of the spill and its fate
decided before the next animal is considered.

Each otter in the simulated population is considered to
have a home range (or a territory, for territorial males), the
center of which is the position assigned to the animal in
OTDIST. OTDIST assigns each animal to one of six classes
depending on sex and reproductive status: 1) juvenile male,
2) adult, non-territorial male, 3) adult territorial male, 4)
juvenile female, 5) adult female without pup, 6) adult female
with pup. Juveniles are animals younger than the age of
sexual maturity that is used in the population model, males
over the age of six years are potentially territorial. Three
categories of daily movements are considered in the model: 1)
"routine" movements around the home range or territory center,
2) seasonal migrations by territorial males, and 3) movements
in response to oil spills. The position of an animal at
the end of a simulated day is calculated as:

Xp = Xe-1 + At (21)

where t indexes days, X is the position along the five fathom
line, and d, is the daily movement. Negative values of d,
indicate movement up the coast (i.e., toward the origin of the
five fathom ordinate system), positive values indicate
movement down the coast.

Routine daily movements.--Routine daily movements are
modeled as a function of displacement from the home range or
territory center, and the magnitude and direction of the
previous day's movement:

A, = bydy.y+b2(Xp—-C) +2, (22)

238

where b, is’ the autoregressive parameter, bz is the
displacement parameter, C is the location of the home range
or territory center, and Z is a normally distributed random
error with mean 0. The parameters b, and bz and the standard
deviation of Z vary with class.

Migratory movements.--Migratory movements by adult males
are simulated at appropriate times of the year. Territory
arrival and departure dates are assumed to be normally
distributed, empirically derived means and standard deviations
around those dates are used to calculate the probability of
a male arriving or departing a territory on each day of the
simulation. For territorial males on each day of simulation
a random number is compared with the probability of departing
a territory in that day For non-territorial but potentially
territorial males a random number is compared against the
probability of arriving on a territory; to account for travel
time to the territory, the mean of the probability
distribution is set three days before the actual mean; thus
the distribution gives the probability of non-territorial male
departing for a territory.

Class two (non-territorial adult) males that are determined
to depart their present home range for a territory are
assigned destination territories using the cumulative
distribution of female positions derived from the female
location vector constructed by OTDIST. Since territorial male
density is thought to be negatively correlated with pup
density (Jameson 1987, USFWS 1987), only the locations of
mature females without pups are used to construct the
distribution. The destination is compared to the list of
potential territories and their statuses (occupied or not
occupied) (also generated by OTDIST). If the territory
originally assigned as a destination is occupied, the closest
(to the original destination) available territory becomes the
destination territory. Once a destination territory has been
determined, the male moves according to:

Ap = i* |ViaxtZe | (23)

where Z, is as in (22), Vmx is the maximum daily rate of
movement for a class 2 animal and i = +1 if the destination
is down the coast from present position and -1 if the
destination is up the coast. Once the male has reached the
destination territory it moves routinely according to class
three parameters.

Class three animals that are determined to leave their
territories are assigned destination home ranges by choosing
from a cumulative density function constructed from the
locations of males generated in OTDIST. Simulated male
density in each 500m segment is squared before constructing

239

the cumulative density function in order to accentuate the
aggregation of non-territorial males. Once a destination has
been chosen the male moves according to (23) until he reaches
the territory, at which point he moves routinely according to
(22) with class two parameters.

Movements in response to oil spill.--Each time that a daily
movement can bring an otter into contact with the oil spill
a series of "decisions" on the part of the otter are
simulated, conditional on the spatial relationship between the
animal's home range, its present location, and the oil spill.
If the animal's home range center is inside the spill
boundaries it may elect to abandon its home range and
establish a new range outside of the spill with daily
probability PE. If the animal's home range center is not
within the spill boundaries, or if it is within the spill
boundaries but the animal has elected not to abandon the home
range, it may attempt to avoid the spill with probability PA.
If an animal avoiding a spill with a present location outside
of the spill "bounces" off of the spill boundary it moves a
distance:

Br = -(d,-D,;) (24)

where d,; is the predicted daily movement according to (22),
B, is the distance bounced, and D, is the distance to the oil
spill boundary. If an animal elects to attempt to avoid the
oil after it is already inside of the spill boundaries it
moves according to (23) with the value of V,,, and the standard
deviation of Z appropriate to its class, and the sign of i is
randomly assigned with equal probability. ©

If, at the end of a simulated day, an animal is inside of
the spill boundaries (and thus exposed to oil), it dies with
probability PM.

Assuming for a moment that PM is very low, a number of
behavior patterns in relation to oil on the part of individual
animals may occur depending on the values of PE and PA and the
size of the spill. Animals might depart the spill area
immediately and not return during the life of the spill;
animals might spend a few days in the spill and then depart;
animals might move routinely outside of the spill; avoiding
it by bouncing off when routine movements would ordinarily
bring them inside the spill; animals might continue to move
routinely on the edge of the spill entering it occasionally;
animals might move in a routine manner within the spill;
animals might move long distances up and down the coast inside
the spill in a "panic".

240

Parameterization of OTMOVE

Parameters used in the movement model were derived from our
telemetry data. The daily locations of radioed animals,
recorded in the field on an x,y coordinate system, were moved
to the five fathom line ordinate system to simplify the
analysis of movement patterns and to derive a parameterization
applicable to the single spatial dimension used in the model.
Graphic traces of the movements of each animal along the five
fathom line are presented in Chapter 3. Two of these are
reproduced here for illustrative purposes (Fig. 10.19).

Routine movements.--After original examination of the
traces represented by Fig. 10.19 we attempted to analyze the
movements of each animal as an autoregressive time series,
modeling each day's movement as a function of previous days'
movements and/or correlated error terms. This analysis led
to good predictive equations for the movements of many of the
animals, but the equations for individual animals often
differed dramatically in form and degree. With no biological
basis on which to decide upon the efficacy of one form of
equation over another, incorporation of these equations into
the movement model was unjustifiably complicated.

We thus opted for the simple regression equation (22)
mentioned previously. One autoregressive parameter is
maintained in the equation, but the major factor in the
equation is the displacement term. The horizontal lines in
the traces of Fig. 10.19 mark the mean position over all days
for which data was obtained on each animal. Movements of many
animals were characterized by long periods of time above or
below the overall mean, but localized oscillations around
short term means (Fig. 10.20). Since the movement model is
designed to run for at most 30 days, we ignore patterns that
occur on a longer time scale, and model the oscillation around
short term means. The series of locations for each animal
were divided into arbitrary non-overlapping 30 day segments,
and regression parameters calculated for each segment
(segments wherein an animal was not able to be located for
more than 10 of the 30 days were not included, a total of 383
segments were used in the analysis). This parameterization
was encouraging, as the displacement terms in all equations
were negative, and most were highly significant. We then
restratified the analysis, grouping the segments according to
the six classes of animals described earlier (30 day segments
in which the reproductive status of an adult female changed
or was unknown were discarded). The regression parameters and
the standard deviation of the errors are given in Table 10.6;
these are the default values used in the model.

241

FIGURE 10.19a -- Daily locations of a juvenile female
California sea otter (#35) as determined by telemetry, 1985-
1986. Location is given in 50m units south of San Francisco,
along the 5 fathom line ordinate system. Horizontal line is
the mean position of all daily locations. Julian date 1 is 1
January 1984. The numerical value of the locations along the
five fathom line generally increases as one moves to the south
along the California Coast.

4000

qn
o
c—)
c—}

6000

mam Bomar ns wi NQzZor>zr =zonaTSoor
—
oe
o
eo

8000
400 500 600 700 800 900 1000 1100
JULTAN DATE

242

FIGURE 10.19b -- Daily locations of a juvenile female
California sea otter (#29) as determined by telemetry, 1985-
1986. Location is given in 50m units south of San Francisco,
along the’ 5 fathom line ordinate system. Horizontal line is
the mean position of all daily locations. Julian date 1 is 1
January 1984. The numerical value of the locations along the
five fathom line generally increases as one moves to the south
along the California Coast.

erie
en

ao
wo
[=]
Oo

~~
Ss
oS
—

~
_
ao
o

71200

MSHRr- ZomtHAS sy on Qzor> SOM H SOOT”

7300

7400

7500
600 700 800 900 1000
JULIAN DATE

243

Leora,
i

1100

FIGURE 10.20a -- Daily locations of juvenile female California
sea otter #35, as in Fig. 10.19a. Horizontal lines are the
mean positions during arbitrary 30-day segments. The
numerical value of the locations along the five fathom line
generally increases as one moves to the south along the
California Coast.

6300

~r
_
a
[—J

an
on
(—J
[—]

an
x
c—J
(—J

maser ZrortHere nn] on Qzo~> =zZorH> Oor”
arn
3 3
= =

6900

600 700 800 900 1000
JULTAN DATE

244

1100

FIGURE 10.20b -- Daily locations of juvenile male California
sea otter #29, as in Fig. 10.19b. Horizontal lines are the
mean positions during arbitrary 30-day segments. The
numerical value of the locations along the five fathom line

generally increases as one moves to the south along the
California Coast.

: AY |
Nl \ i"

on
a
o
o
—_
[—
|
a

M=zHR- SOEASTN HW QZFOory BONA SOOM
(2)
w
o
o

6000

600 700 800 900 1000
JULIAN DATE

245

1100

Vmax, the maximum daily rate of travel was estimated for
each class of animal by considering the maximum distance
between locations taken at least 24 hours apart for each
animal, and calculating maximum net daily movement:

Vmax = MAX{ | (dx/dp) |*24 } (25)

where d, is the distance between two successive locations and
d, is the time, in hours, between the locations. If an
animal's class changed during the study periods of different
classes were considered separately. These values were then
grouped and averaged by class to get the values in Table 10.6.
Since the greatest values of V,,, are seen in juveniles
(classes one and four), and it is unlikely that an otter's
swimming speed decreases in adulthood, V,,, for classes one-
three is set equal to 48.6 500m segments/day in the model, and
Vmax f£0r Classes four and five is set equal to 37.5 500m
segments/day. Pups likely restrict female movements, SO Vmax
for class six animals remains at 8.4 500m segments/day in the
model.

ee —_.._._

Table 10.6. Parameters used in short-term otter movement
model. AR and CE are regression parameters discussed in text.
sd is standard deviation of regression errors, R° given for
regressions. Vmax is mean maximum daily movement, derivation
discussed in text.

Class Status AR CE sd R? Vmax
1 Juvenile male -0.045 -0.290 8.56 (0.13) 48.9

2 Adult non-territorial
male 0.105 -0.815 4.64 (0.37) 40.2

3 Adult territorial
male 0.042 -1.044 1.93 (0.43) 36.2

4 Juvenile female
0.367 -0.163 8.09 (0.10) Bi D

5 Adult female w/o
pup -0.025 -0.406 6.39 (0.21) 20.0

6 Adult female
w/pup -0.009 -0.706 2.95 (0.38) 8.4

Migratory movements by adult males.--The probabilities
of migratory movements for males of class two and three are
calculated using the data from Jameson (1987), as in the
distribution model. Probabilities are calculated on a daily
basis assuming a normal distribution of arrival and departure

246

times. Since Jameson (1987) gives mean date of territory
arrival, calculation of the mean date upon which males leave
for their territories requires consideration of the transit
time to the territory. We have arbitrarily set that at three
days, so that the mean date of departure for a territory is
19 May.

The dates given by Jameson are far enough apart that
there is no overlap in simulated territory arrival and
departure. The same caveats about the seasonality in male
migratory movements that were mentioned in the discussion of
OTDIST apply here. The same values of the territorial
parameters that are set for OTDIST are used in OTMOVE;
seasonality less pronounced than that described by Jameson
(1987) and as indicated by our telemetry data, can be
simulated by setting large standard deviations of arrival and
departure dates.

Movements in response to oil spill.--The parameters PM,
PA, and PE are delphic parameters; very little data from which
to estimate their values are available. Costa and Kooyman
(1982) found that otters oiled over 25% of their surface will
die of hypothermia if not cleaned, suggesting that PM could
be very high. Ford and Bonnell (1986) use values of 30%-90%
as most likely mortality rates in their simulations, depending
on the condition of the oil, but allowed the possibility of
mortality varying between 10 and 100%. Siniff, et al., (1982)
found that captive sea otters did not avoid areas of the
holding tanks experimentally contaminated with oil, suggesting
that PA, the probability of localized movements to avoid the
spill, may be very low. It is also likely that PE, the
probability of leaving the spill area to establish a new home
range, is a good deal smaller than PA.

The values of these parameters are set by the user at
runtime, facilitating evaluation of the relative importance
of these parameters in determining the amount of mortality
from a spill within the structure of the model. Additionally,
the values of these parameters are set independently for each
day of the spill, allowing consideration of the effects of
weathering on oil (i.e., PM decreasing with time), or possibly
learning on the part of the animals (i.e., PA increasing with
time), or other scenarios. We anticipate that much of the
model's usefulness will be due to its ability to simulate
different oil spill response scenarios.

Fig. 10.21 traces simulated movements of otters in the
vicinity of an oil spill presumed to occur on December 1, and
lasting 15 days, at the southern end of the Monterey Bay and
eastern side of the Monterey Peninsula. The rangewide
population in these simulations was set at 1600 animals. In
Fig. 10.21la all delphic parameters are set to 0.0, that is,

247

the spill has no effect on the animals' behavior, and the
movements thus reflect a "normal" situation. In Fig 10.21b
PM and PA were set to 0.8 for the duration of the spill, and
PE was set to 0.5 for the duration; in Fig. 10.21c PM was set
to 0.1 for the duration, PA was set to 0.5 for the duration,
and PE was set to 0 for the duration. The same initial
population and distribution, and the same random number seeds,
were used in all 3 simulations.

MODEL OUTPUT

A log file, recording the user-input parameter settings,
is generated each time the model is run. Six files of raw
output data are written as the model runs. One contains the
Simulated population sizes for control runs (runs without
introduction of oil spills), and one contains the simulated
population sizes for runs with oil spills. In each of these
the numbers of males, females, and pups are recorded once a
year, at the end of the month in which the oil spill occurs.
A third file records the numbers of oil spill-caused deaths
by class ( juvenile male, adult non-territorial male, etc.)
and by day of spill. The fourth file records the total number
of deaths of males, females, and pups due to the spill in each
run. The fifth file records the total population size just
prior to the spill, and the number of simulated years that
pass before the population recovers to that size. The sixth
file records the total reproductive value of the population
just before and just after the spill. The reproductive value
of a female is the relative number of female pups she is
expected to wean during the remainder of her life. For a
female of age x, Fisher (1930) and Wilson and Bossert (1971)
give the formula for reproductive value (v,):

Vn = (2/15) 6 am, |; y=x,W (26)

Fig. 10.22 illustrates the reproductive values of females
under default parameter settings and a per capita population
growth rate = 0. The total reproductive value of the
population is:

v7 = nyv, ¢ =X=1,W (27)

where n, is the number of females of age x in the population.
The reduction in total reproductive value may provide a
measure of how the perturbations in age and sex structure of
the population caused by an oil spill effect population
recovery.

248

FIGURE 10.21a -- Simulated movements of sea otters around an oil
spill in Monterey Bay beginning 1 December and lasting 15 days.
Boxed area (heavy dark line) represents spill location, between
Fort Ord (365) and Point Pinos (390). Each trace represents a
different simulated individual otter. Total range wide population
at the time of the spill was set at 1600; OTDIST positioned 56
animals in the area covered by the diagram but for clarity only 25
were chosen, at random, for representation. Movement parameter
settings (see text) were PM=0.0, PA=0.0, PE=0.0, for all days of
the spill, thus simulating no oil-caused mortality or effect on
behavior. The relatively few numbers of animals at the north end
of the diagram reflect the much lower density of otters in the
sandy habitat of Monterey Bay relative to the rocky habitat of the
Monterey Peninsula. The numerical value of the locations along the
five fathom line generally increases as one moves to the south
along the California Coast.

0 1 DharteneS tietares 4 eevee) OmereineeriS Se Oma UA nerlics

345

SDD

365

SYS

385

SEIS)

405

LOCATION ALONG 5 FATHOM LINE

415 <2 aaa ——
OM Si? Mae ae nome er One OL Bi 13) 14

DAY OF SPIRE

249

FIGURE 10.21b -- Simulated movements of sea otters around an
oil spill in Monterey Bay beginning 1 December and lasting 15
days, as in Fig. 10.21a, except that in this simulation
movement parameters were set at PM=0.8, PA=0.8, PE=0.5. A
star (*) indicates an otter death. The numerical value of the
locations along the five fathom line generally increases as
one moves to the south along the California Coast.

0 1 2 OB, VA Seth COE eS ATO RRO IE POTS IS
345

355)

S199)

SS

385

LOCATION ALONG 5 FATHOM LINE

Onn Det Rayne 6 Ai Sh Ake Ova NERA 1s
DANO Or SP iige

14

15

FIGURE 10.21c -- Simulated movements of sea otters around an
oil spill in Monterey Bay beginning 1 December and lasting 15
days, as in Fig. 10.2la, except that in this simulation
movement parameters were set at PM=0.1, PA=0.5, PE=0.0. A
star (*) indicates an otter death. The numerical value of the
locations along the five fathom line generally increases as
one moves to the south along the California Coast.

Cy ee! 2 ot Oph tim TD ee ene: She eal Ojes ylulbet ele al

LOCATION ALONG 5 FATHOM LINE

8 ao en 1D 13
DAY OF SPILL

251

14

14

FIGURE 10.22 -- Age-specific reproductive values of female
California sea otters under default parameter settings and a
per capita growth rate=0.

—_
on

a
[=)

9
ui

REPRODUCTIVE VALUE (pups per female)

0 5 6 7 8 9 1011 12 13 14 15 16 17 18 19 20 21 22 23

AGE

A data summary and analysis program, OTPROC, has been
written to summarize the raw output files and write a short
report on the run. Additionally, OTPROC writes output files
specifically designed to be read by LOTUS123. A LOTUS123
macro worksheet is supplied that will generate simple graphics

illustrating the outcome of the model runs. Detailed
discussion of the processing of model output can be found in
the user's manual (Appendix 10.2). Examples of model output

appear in Fig. 10.23.
OTRANGE

In an effort to be able to make predictions about the
future status of the California sea otter population, and to
aid in the estimation of some of the parameters used in OTPOP
and LESLIE, we undertook an analysis and modeling of the
historical growth of the population. Because of the type of
data used and the speculative nature of the modeling, this
work was conducted separately from the oil spill population
model.

Background

The growth of the California sea otter population
following almost complete elimination by the turn of the 20th
century is described by CDFG (1976), Ralls, et al., (1983),
and USFWS (1986). Information on the size and range of the
population before 1968, when CDFG began intensive study of sea
otters, is scanty and generally anecdotal. A remnant
population of probably less than 100 individuals grew at what
appeared to be a steady rate of approximately 5% per year
until the mid 1970's, expanding the occupied range in the
process. By 1969, when regular counts began, an estimated
1,390 otters inhabited the coast between Seaside and Point
Estero. In 1976 the highest population estimate was recorded,
1,789 animals, and occupied range extended from Rio Del Mar
to Pecho Rock. Population estimates have been relatively
constant since 1976, while the range has continued to expand
slightly.

Reasons for the arrest in population growth have been the
subject of much debate centering around whether density
dependent or density independent factors have been most
influential in bringing about the decline. Miller (1980)
maintained that the population had reached an equilibrium with
habitat carrying capacity, and that starvation was the
principal cause of mortality. Estes et al (1986), based on
comparisons of time budgets with an Alaska population assumed
to be food limited, concluded that the California population
was not at carrying capacity, and that density independent
processes, particularly accidental entanglement in fishing
nets, were the primary factors limiting population growth.

253

FIGURE 10.23a - Log file from a run of the model. A large
oil spill is introduced for 10 days beginning June 15 along
the 50km section of coast between Marina and Yankee Point.
Initial population size is set at 1600, carrying capacity set
at 1600. See user's manual for explanation of parameters.

DATE AND TIME USED TO GENERATE RANDOM NUMBER SEEDS
20,"NUMBER OF YEARS PER RUN"
100,"NUMBER OF RUNS WITH OIL SPILL"
100,"NUMBER OF CONTROL RUNS"
1600,"INITIAL POPULATION SIZE"
6,"MONTH OF SPILL"
15,"DAY OF SPILL"
10,"DURATION OF SPILL"
350,"NORTH BOUNDARY OF SPILL"
450,"SOUTH BOUNDARY OF SPILL"
1600,"EQUILIBRIUM POPULATION SIZE"
201,"NORTH BOUNDARY OF RANGE"
955,"SOUTH BOUNDARY OF RANGE"
0.090,"MAXIMUM PER CAPITA ANNUAL GROWTH RATE"
0.005,"NON-LINEARITY OF DENSITY DEPENDENCE"
0.000,"DENSITY INDEPENDENT MORTALITY RATE"
1.000,"DEGREE OF COMPENSATION"
0.930,"ADULT FEMALE SURVIVAL RATE"
15.000,"MODEL AGE OF FEMALE SENESCENCE"
0.250,"PRIME REPRODUCTIVE RATE"
0.530,"PUP SURVIVAL RATE"
0.870,"ADULT MALE SURVIVAL RATE"
9.000,"MODEL MALE AGE OF SENESCENCE"
5.000,"PERCENT VARIATION IN ADULT SURVIVAL"
5.000,"PERCENT VARIATION IN PUP SURVIVAL"
ww MAR" "CE" "STGMA", "VMAX"
“JUVENILE FEMALES" 0.367-0.163 8.09037.500
“ADULT FEMALES W/PUP"-0.025-0.406 6.39307.500
"ADULT FEMALES W/O PUP"=0.009-0.706 2.950 8.400
"JUVENILE MALES"-0.045-0.290 8.56048.900
“ADULT NON-TERRITORIAL MALES" 0.105-0.815

4.64048.900
“ADULT TERRITORIAL MALES" 0.042-1.044 1.93048.900
6,"AGE AT WHICH POTENTIALLY TERRITORIAL"
60.000,"MAXIMUM % OF POTENTIALS THAT HOLD
TERRITORIES"
20.000,"MINIMUM % OF POTENTIALS THAT HOLD
TERRITORIES"
8.110,"MEAN TERRITORY LENGTH"
0.440,"S.D. OF TERRITORY LENGTH"
5,"MEAN ARRIVAL DATE MONTH"
23,"MEAN ARRIVAL DATE DAY"
11,"S.D. OF ARRIVAL DATE IN DAYS"
12,"MEAN DEPARTURE DATE MONTH"
1,"MEAN DEPARTURE DATE MONTH"
15,"S.D. OF DEPARTURE DATE IN DAYS"
"DAY",

"P (MORTALITY) ","P (AVOIDANCE) ",""P (EMIGRATION) "

N

TABLE 23a. (continued)

1 1.000 0.000 0.000
2 1.000 0.000 0.000
3 1.000 0.000 0.000
4 1.000 0.000 0.000
5 1.000 0.000 0.000
6 1.000 0.000 0.000
7 1.000 0.000 0.000
8 1.000 0.000 0.000
9 1.000 0.000 0.000
10 1.000 0.000 0.000

FIGURE 10.23b - Report file generated by OTPROC after the run in Fig. 10.23a.

CONTROL RUNS:

YEAR FEMALES MALES PUPS
-4 1038.0¢1038-1038, 0.0) 566.0(566-566, 0.0) 195.5¢€175-220, 10.7)
-3 1029.8¢ 997-1064, 13.6) 559.4(529-589, 12.8) 244.8(220-272, 11.8)
-2 1047.1¢1002-1105, 17.7) 579.5(529-615, 18.4) 228.5(193-262, 12.3)
-1 1056.4¢1015-1110, 18.6) 590.6(542-632, 19.3) 215.4¢(170-248, 13.8)
0 1058.1¢1002-1123, 24.7) 597.0(536-652, 20.9) 214.3(¢185-239, 12.6)
1 1035.7¢€ 979-1094, 23.4) 603.6¢560-658, 20.0) 212.5¢177-248, 12.9)
2 1021.3¢ 978-1066, 1938) 606.9(558-650, 18.2) 213.8¢181-249, 14.5)
3 997.1¢€ 916-1049, 22.2) 593.9(526-645, 21.0) 214.7(188-238, 9.5)
4 997.9( 938-1063, 24.7) 582.2(520-640, 24.5) 211.0¢167-240, 11.1)
5 998.8¢( 935-1058, 23.1) 573.3(509-628, 23.8) 208.8(189-235, 9.2)
6 1007.0¢ 947-1071, 23.3) 568.7(506-623, 24.6) 207.6(175-234, 10.9)
7 1010.4¢ 947-1074, 25.0) 563.4(508-627, 22.9) 210.9(¢186-238, 10.7)
8 1018.3¢ 955-1084, 25.5) 559.8(504-619, 23.3) 213.1¢190-245, 11.3)
9 1026.5¢ 969-1085, 24.7) 558.3(507-615, 24.4) 215.2¢183-250, 12.9)
10 1031.2¢ 965-1090, 27.3) 557.3¢499-627, 26.6) 216.7¢179-244, 11.1)
11 1037.4¢ 973-1098, 27.3) 555.8(508-623, 25.8) 219.1(194-242, 11.4)
12 1038.5¢ 980-1103, 28.4) 558.0(484-617, 24.6) 220.9(187-255, 10.3)
13. 1039.4¢ 964-1136, 28.1) 560.9(508-622, 21.1) 219.8(196-241, 9.9)
14 1039.3¢ 965-1115, 26.0) 564.7(494-619, 21.6) 220.2(201-246, 9.4)
15 1039.2¢ 980-1086, 23.1) 565.9(512-616, 21.2) 219.5¢(200-250, 9.9)
16 1036.3¢ 976-1103, 23.3) 565.3¢(516-618, 21.8) 220.9(¢186-254, 12.2)
17 1034.8¢( 972-1091, 23.4) 565.8(503-623, 23.4) 220.3¢188-240, 11.2)
18 1039.0¢ 988-1108, 24.0) 565.9(510-636, 23.8) 218.2¢190-248, 11.1)
19 1034.8¢ 975-1109, 25.6) 564.6(518-642, 22.6) 219.1€¢191-250, 11.0)
20 1032.3¢ 966-1081, 23.8) 564.9(506-616, 23.0) 219.4¢(195-250, 10.6)

256

10.23b (continued)

FEMALES
-4 1038.0 (1038-1038,

0.0)

955-1061,13.0)
983-1090,19.3)
978-1108,26.3)
972-1116,29.8)

713- 860,29.4)
740- 899,31.5)
784- 949,28.7)
836- 984,27.4)

884-1019,26.6)
894-1048,28.0)
933-1085,26.6)
983-1119,26.2)

(1008-1128,23.5)
(1029-1135,22.3)
(1022-1146,26.2)
(1011-1124,24.4)

968-1110,27.7)
963-1107,25.8)
973-1126. 3)
972-1096,27.9)
924-1101,27.8)
896-1132,29.6)
892-1088,27.2)
902-1082,28.3)

FIGURE
OIL SPILL RUNS:
YEAR
-3 1030.4 ¢
-2 1044.0 ¢
-1 1049.3 ¢
-0 1047.4 ¢
+0 785.3 ¢
1 837.5 ¢
2 887.3 ¢
3 923.5 ¢
4 956.7 ¢
5 989.5 ¢
6 1019.8 ¢
7 1045.6 ¢
8 1063.4
9 1073.8
10 1077.9
11 1068.6
12 1060.1 ¢
13° 1047.4 (¢
14 1039.0 ¢
15 1033.9 ¢
16 1027.7 (¢
17 1025.3 ¢
18 1020.4 ¢
19 1018.4 ¢
20 1021.8 ¢

956-1075,24.7)

NUMBER OF DEATHS FROM OIL SPILL:

CLASS MEAN s.D
JUVENILE MALES 13.4 4.0
ADULT MALES 18.7 3.2
JUVENILE FEMALES 75.7 8.3
ADULT FEMALES 186.4 12.1
PUPS 48.4 6.6
TOTAL ANIMALS 342.5 19.0

RECOVERY AFTER OIL SPILL:

REPRODUCTIVE VALUE BEFORE SPILL
REPRODUCTIVE VALUE AFTER SPILL
REDUCTION (4)

YEARS TO RECOVERY

**ON 25 OF

PRE-SPILL SIZE
TIME TO RECOVERY CALCULATED ONLY FOR RUNS THAT DID RECOVER.

100 RUNS (¢ 25.0%) THE POPULATION

MALES
566.0(566-566, 0.0)
559.1(530-589,12.9)
573.5(531-617,18.8)
585.2(531-624,21.2)
590.2(¢541-652,24.2)
558.1(508-612,23.9)
540.3(472-612,26.3)
530.0(468-609,28.6)
515.4(462-575,26.6)
501.0¢440-563,27.2)
496.4(418-566,25.0)
496.0(¢431-571,24.8)
500.7(435-579,25.9)
512.8(¢453-582,24.3)
527.7¢6458-578,25.6)
542.5(490-615,24.2)
556.1(511-608,23.8)
567.4(519-634,23.5)
574.8(519-643,24.6)
578.6(531-638,24.6)
579.1(528-666,27.4)
578.6(532-638,24.5)
577.4(520-642,22.8)
572.3¢(524-619,22.3)
568.3(514-628,23.8)
568.1(511-626,25.1)

PUPS

196.4(165-217,11.1

232.
219.
211.
211.
162.
169.
176.
181.
188.
197.
206.
214.
221.
227.
229.
230.
228.
227.
225.
223.
219.
215.
215.
214.
213.

7¢195-277,16
0(¢188-254,16
5(187-251,13
0¢180-251,12
6(134-194,11
6(140-208,12
0¢151-205,12

3(150-201,10.

4(157-210,11

1¢169-230,10.
11132231, 10+

3(¢186-241,11

0¢200-244,10.
6(204-255,10.

9¢201-259,10

1(196-263,12.
4(189-258,12.

4(202-265,11
5(189-262,11
3(200-256,10
4(196-246,11
2(¢184-252,11
4(166-238,11
0¢165,-241,11
7(188-240,10

PERCENT OF POPULATION

RANGE MEAN S.D.
oc @5) 4.96 1.48
2 74h) 5.86 1.01
- 100 26.88 2.65
- 222 24.35 1.58
SONG, 22.95 2.94
- 383 18.54 1.06

MEAN $.D.
1316.3 39.5
987.7 39.4
24.971 1.52641
9.800 1.71270

257

RA
1.90
3.40

22.00
20.80
15.20
16.20

RAN

NGE
7

GE

-90
8.
37.
0) 4
32.
21.

90
20
40
00
00

1223.2-1409.9

889.9-1073.3
22.271-28.831
7.000-17.000
DID NOT RECOVER TO

-0)
-0)
-3)
5)
-8)
5&2)
at)
3)
-4)
5)
6)
-6)
5)
9)
-5)
3)
5)
-8)
-5)
-7)
oI}
-7)
4)
-0)
-7)

v

FIGURE 10.23c -- Trace of the total simulated population size
for, runs) in Fig. 10.23a-.) Meanvand Gange sot) 1 OO suns Ousl:
spill occurs at year 0.

1800
WY)
4
FE
O high
ue 1600
ZZ
Li
oO =
FZ
Lil \
O 1400 + | mee
OQ
z
LL low
O
1200
oO
LJ
faa)
=
=)
= 1000

Sy <@Q@ © 2 4 6 8 ‘Om Wecatke «AG 18

YEAR OF SIMULATION

FIGURE 10.23d -- Trace of the total simulated population size
for the control (no oil spill) runs in Fig. 10.23a. Mean and
range of 100 runs.

NUMBER OF INDEPENDENT OTTERS

1800

1600

1200

1000

=i

high
mean

low

— 2) 0 2 4 6 8 10 12 14 16 18

YEAR OF SIMULATION

20

FIGURE 10.23e -- Trace comparing the mean values from Figs.
10.23c and 10.23d.

NUMBER OF INDEPENDENT OTTERS

1800

control
1600

1400

oil spill

1200

1000
—4 -—2 0 2 4 6 8 10 12 14

YEAR OF SIMULATION

16

18

20

CUMULATIVE NO. OF DEATHS

FIGURE 10.23f -- Mean cumulative number of otter deaths due
to oiling for the run in Fig. 10.23a.

300
independents

240

180 rn

120
pups

60
0 -
0 2 4 6 8 10
DAY OR TSPIFE

261

Relatively high levels of commercial gill- and trammel-net
fishing along the central coast began in the early 1970's,
coincident with the decline in sea otter population growth.
Wendell, et al., (1985) estimated that approximately 80 sea
otters drowned accidentally in nets each year between 1973
and 1983. Ames, et al., (1985) suggested that density
dependent processes are most important in the central part of
the range, where the population has been established for the
longest period of time and has depressed prey populations
below pre-recolonization levels, while at the range
peripheries, which still have relatively low numbers of
otters, density independent processes, particularly accidental
entanglement and great white shark attacks are the most
important factors limiting the growth of the population. The
time budget data we gathered using telemetry in areas of the
central part of the range tended to support the suggestion
that the juvenile females may be suffering from food
limitations in the central part of the range (see Chapter 4),
but the evidence is not conclusive and does not address the
situation at the range peripheries.

The degree of density dependence in the dynamics of a
particular population is a question that must be addressed by
any attempt to model that population. The uncertainty
surrounding this issue in the California sea otter population
led us to build in to our model the capability of simulating
any degree of density dependence. This approach allows
Simulation of the effect of oil spills under different
conceptual hypotheses about the dynamics of the population,
but does little towards providing a set of parameters most
applicable to a timely and realistic risk analysis. In an
effort to better understand the dynamics and estimate
pertinent parameters we have attempted to simulate the
historical growth of the California population with a model
(OTRANGE) that incorporates a feedback mechanism between
population size and range size. It is hoped that a model that
fits the historical data will be of value in predicting future
population sizes and range extent. Because of the speculative
nature of the mechanisms incorporated in the model, and the
amount of computer time required to fit model parameters to
the historical data, OTRANGE is completely deterministic. It
is intended only as an aid in determining the values of user-
input parameters in the main risk analysis model, particularly
when considering spills in the future.

Structure

The density dependence function used in the population
dynamics portion of OTRANGE is the same as that used in OTPOP
lolol IHSNGIGA ((¥eig (3), Imskef, al@odt, eyerel rely IO) 25) o Range
expansion is incorporated with the assumption that range
expansion is density dependent, and that range expansion

262

results in an increased population-wide carrying capacity
(Fig. 10.24). Biologically, this means that as the population
approaches its carrying capacity it can respond by both
reducing its growth rate, and by increasing its carrying
capacity through an increase in the area of occupied habitat.

The density dependence function for range

size is simply taken as the mirror image of the density
dependence function for population size (Fig. 10.25):

kK = Kpax~Kmax{1-exp{-b (K-N) ]} (27)

where k is the annual rate of change in carrying capacity,
Knax iS the maximum annual rate of change of k, K is the
carrying capacity, N is the population size, and b governs
the shape of the curve. As in the population dynamics portion
of the model, an arbitrary ceiling on the size of k is imposed

at Kmax-

The fact that occupied range has expanded faster to the
south than to the north, and the fact that the range continued
to expand during the period of apparent population decline in
the late 1970,s, led us to modify (27) in the model. First,
range expansion to the north and to the south are considered
separately. Secondly, allowance is made for density
independent range expansion. Biologically, density
independent range expansion may be the result of natural
dispersal of young animals, regardless of the equilibrium
status of the population, superimposed on a limited geographic
range. OTRANGE thus uses

Kg Kgs + Kmax,s7Kmax,s{1-exp{-b (K-N) ] } and
Ky Kan + Knaxin@ Kua, n{ 1-exp{-b(K-N) ] } (28)

where k, and k, are annual range expansion rates to the south
and to the north, respectively, ky, and ky, are density
independent rates of range expansion ‘to the south and to the
north respectively, Kmax,s and Kpax,, are the maximum rates of
range expansion to the south and to the north, respectively,
and b, K, and N are as in (27).

til

Translating k, which is terms of numbers of animals, into
range size requires an estimate of the number of animals that
can be supported in a given area of habitat. Ford and Bonnell
(1986), using USFWS census data, estimated that maximal
densities were 4.7 otters per km? over rocky substrate and 1.3
otters per km, over sandy substrate. By digitizing the coast,
assuming that otter habitat extended from the coast to the 20

263

FIGURE 10.24 -- Schematic representation of OTRANGE. At each
iteration, the quantity K-N determines the next iteration's
population size and carrying capacity.

KN

Density
dependence

264

FIGURE 10.25 -- Density dependence functions used in OTRANGE.
The parameter b affects the shape of the curve (see Fig.
10.4).

Kegs

a LJ
<l =
(eZ <l
(ee
ma
= FE,
= 6
ra Z
O <
ae
: a
(als
<L Lod
O O
on x
Oo Cie
population below e) population above
carrying capacity carrying capacity

N—K

fathom depth contour, coding substrate from USGS maps, and
using Ford and Bonnell's (1986) estimates, we calculated the
carrying capacity of each 0.5 km segment of the 5 fathom line
ordinate system. This allowed calculation of historical
values of K given the length of occupied range at any point
in the past. Using the historical range length data in USFWS
(1986, Table 1.3) we calculated historical carrying capacities
based on Ford and Bonnell's (1986) estimates. The calculated
carrying capacities were well below historical population
estimates, leading us to believe that Ford and Bonnell's
census-based estimates underestimated actual carrying
capacity, and forcing us to include the per hectare carrying
capacities of rock and sand substrates in the group of
parameters to be estimated.

Parameterization

We estimated OTRANGE parameters by means of a 2-stage
numerical search. First, we chose what seemed to be
reasonable bounds on the value of each parameter in the model,
and a testing interval for each parameter between those
bounds. Then the model was run under every possible
combination of parameter values within the bounds, using the
1914 historical estimates of population size and carrying
capacity as initial conditions. Goodness of fit to historical
data was calculated for each run by comparing the modeled
population size and range-wide carrying capacity to the
historical estimates of those values for each year that
historical data were available. Total sum of squares of the
aifference between the modeled and historical values was the
goodness of fit criterion.

A total of over 196,000 combinations of parameter values
were tested in the first stage of estimation. The 20
parameter combinations that gave the best fit to the
historical data by each of the goodness of fit criteria were
saved, new bounds chosen from those combinations, and the
analysis repeated with a smaller testing interval for each
parameter. Over 40,000 parameter combinations were tested in
this second stage of estimation. The parameter combination
that gave the best fit is listed in Table 10.7, and a trace
of the model run under this parameterization is shown in Fig.
10.26.

266

Table 10.7. Parameters giving the best fit of OTRANGE to
historical data. See text for explanation of parameters.

OTRANGE without OTRANGE with
density independent density independent
Parameter mortality mortality
Maximum per capita
growth rate (Ymax) 0.085 0.077
Non-linearity of
density dependence (b) 0.020 0.030

Maximum density dependent
rate of expansion:
North (Kmex,n) * 7.5 8.5
South (Kpax,s) : 20.7 27.5
Density independent rate
of expansion:

North (kg py) : 8.0 2.6
South (Kg ,) : 10.0 15
Per hectare carrying
capacity
Rock substrate: 0.26 6.5
Sand substrate: 0.78 6.75
Density independent
mortality rate (m) -- 0.03

Incorporation of density independent mortality.

The parameterization of the above model that produced the
best fit to the historical data produced oscillations in
population size once the population approached 1600 animals.
This may imply that the stabilization of population size that
occurred in the mid-1970's is a consequence of the internal
dynamics of the population and its habitat, and that
stabilization would have occurred even without gill net
mortality. In other words, it implies that gill-net mortality
completely compensates for natural mortality, and that if
animals would not have been killed in the nets they would have
died from natural causes. Certainly natural mortality is
greater than 5% per year (see Figs. 10.7,10.9, and 10.10), and
compensatory density independent mortality of the magnitude
due to gill-nets is mathematically possible.

We decided to investigate the possibility that gill-net
mortality is completely additive to natural mortality. We
thus modified the population growth equation:

Yr = Ymax{1-exp[-b(K-N) ]}-m (29)
where m is the density independent mortality rate. We then

estimated parameters in the same manner as above, including

267

FIGURE 10.26 -- Fit of OTRANGE output to historical data using
the parameters in Table 10.7 without density independent
mortality. Solid line traces population size, dashed line
traces carrying capacity.

2400
Historical A
population size A
Bolo Historical a
carrying capacity s

1600

1200

800

NUMBERS OF OTTERS

400

1915 1925 19355 1945 1955 1965 1975 1985

YEAR

268

FIGURE 10.27 -- Fit of OTRANGE output to historical data using
the parameters in Table 10.7 incorporating density independent
mortality. Solid line traces population size, dashed line
traces carrying capacity.

2400
Historical
population size
&
fe 2000 Historical a "
carrying capacity oe oe
E <
1
oO 600
Le
O
1200
”Y
or
a
800
=
=)
Fz,

400

1915 3925 1935 1945 1955 1965 1975 1985

YEAR

269

m in the list of parameters that were estimated. Density
independent mortality was incorporated in the model only after
1972, to simulate the effect of gill-net mortality. A trace
of the simulated dynamics that produced the best fit appears
Wg IPSS ALO) 6 BW

Caveats

The model incorporating density independent mortality
produced a better fit to the historical data than did the
basic model (total sum of squares =870,616 vs. 1,121,621).
In the parameterization giving the best fit, m =0.03, less
than the 5% gill net mortality estimated by Wendell et al
(1985), perhaps indicating that gill net mortality is partly
compensatory.

Neither of the models fits the historical data
particularly well, and there is no guarantee that the
mechanisms in the model mimic those in the natural population.
There are two levels on which our approach should be
criticized. First, we made no effort to investigate different
functional forms of density dependent range expansion. As far
as we know no other attempts have been made to model density-
dependent changes in carrying capacity in any situation; we
did not have the advantage of theoretical precedents. We used
the mirror image of the population growth function as a
convenient starting point, but there is no reason to assume
that it is correct. Secondly, of course, parameter estimates
are only as good as the data they are based on. For example,
Geibel and Miller (1980) and Wendell, et al., (1986) described
the problems associated with the estimation of sea otter
population size.

Relationship between OTRANGE and OTPOP

The analysis of OTRANGE under different parameterizations
gave the default parameter values for r,,,, b, and K used in
OTPOP and LESLIE (Table 10.2). Because the proposed oil and
gas developments that are of concern will not occur until the
1990's, at the earliest, and because much of that development
is proposed for what is now peripheral to occupied sea otter
range, we have supplied OTRANGE to MMS in a form that can be
used to predict future population size, range size, and
carrying capacity. These predictions can then be used as
input to the main model to simulate initial conditions in the
future.

LITERATURE CITED

Ames, R. A., F. E. Wendell, and J. J. Geibel. 1985. Sea
otter mortality in California. Draft unpublished

270

report. Marine Resources Branch, California Department
of Fish and Game. 49pp.

California Department of Fish and Game. 1976. A proposal for
sea otter protection and research, and request for return
of management to the state of California. Unpubl.
report, January 1976. 2 Vols.

Chapman, D. G. and D. S. Robson. 1960. The analysis of a
catch curve. Biometrics 16:354-368.

Costa, D. P., and G. L. Kooyman. 1982. Oxygen consumption,
thermoregulation, and the effect of fur oiling and
washing on the sea otter, Enhydra lutris. Can J.
Zoology 60:2761-2767.

DeMaster, D. P. 1981. Incorporation of density dependence
and harvest into a general population model for seals.
Pp. 389-402 in C. W. Fowler and T. D. Smith, eds.,
Dynamics of large mammal populations. J. Wiley and Sons,
New York. 476pp.

Eberhardt, L. L. 1986. Assessing the dynamics of wild
populations. J. Wildl. Manage. 49:997-1012.

Eberhardt, 9b.) L., ands "Ds 9B. Sinver. OMe Population
dynamics of the Pribilof fur seals. Pp 197-220 inc. W.
Fowler and T. D. Smith, eds., Dynamics of large mammal
populations. J. Wiley and Sons, New York. 476pp.

Estes, J. A., R. J. Jameson, and E. B. Rhode. 1982. Activity
and prey selection in the sea otter: influence of
population status on community structure. Am. Nat.
120:242-258.

Estes, J. A., K. E. Underwood, and M. J. Karmann. 1986.
Activity-time budgets of sea otters in California. J.
Wildl. Manage. 50:626-636.

Fisher, R. A. 1930. The genetical theory of natural
selection. Clarendon, Oxford.

Ford, R. G. 1985. A risk analysis model for marine mammals
and seabirds: a Southern California Bight scenario.
Final Report. Minerals Management Service Pacific OCS
Region Contract No. 14-12-0001-30224. Los Angeles.
236pp.

Ford, R. G., J. A. Weins, D. Heinemann and G. L. Hunt. 1982.

Modeling the sensitivity of colonially breeding marine
birds. co) Old Spills: Guilemots and /kittiwake

271

populations on the Pribilof Islands, Bering Sea. J.
Applied Ecology 19:1-31.

Ford, R. G., and M. L. Bonnell. 1986. Analysis of the risk
of oil spills to sea otters-methodology. Technical
Support Document 3, Draft Environmental Impact Statement
for Proposed Translocation of Southern Sea Otters. US
Fish and Wildlife Service.

Fowler, C. W. 1981. Comparative population dynamics in large
mammals. Pp 437-456 in C. W. Fowler and T. D. Smith,
eds., Dynamics of large mammal populations. J. Wiley and
Sons, New York. 476pp.

Garshelis, D. L. 1983. Ecology of sea otters in Prince
William Sound, Alaska. Ph. D. Thesis. Univ. of
Minnesota, Minneapolis. 321pp.

Geibel, J. J., and D. J. Miller. 1984. Estimation of sea
otter, Enhydra lutris, population, with confidence
bounds, from air and ground counts. California Fish and
Game 70:225-233.

Green, B. 1978. Sexual maturity and senescence of the male
California sea otter (Enhydra lutris). M. S. Thesis.
San Jose State University, California.

Harris, R. B., L. A. Maguire, and M. L. Shaffer. 1987. Sample
sizes for minimum viable population estimation. Conserv.
Biol. 1:72-76.

Heisey, D. M., and T. K. Fuller. 1985. Evaluation of
survival and cause-specific mortality rates using
telemetry data. J. Wildl. Manage. 49:668-674.

Jameson, R. J. 1987. Movements, home range, and territories
of male sea otters in central California. Manuscript in
review. 22pp.

ett Push waeRaeks Mohn; gandsDemroEGray., 196i. sensitiv
dependent processes and management strategy for the
northwest Atlantic harp seal populations. Pp. 135-158
in Cc. W. Fowler and T. D. Smith, eds., Dynamics of large
mammal populations. J. Wiley and Sons, New York. 476pp.

Lotka, A. J. 1907. Relation between birth and death rates.
Science, N.S. 26:21-22.

Loughlin, T. R., J. A. Ames, and J. E. Vandevere. 1981.
Annual reproduction, dependency period, and apparent
gestation period in two California sea otters, Enhydra
lutris. Fishery Bull. 79:347-349.

272

Miller, D. J. 1980. The sea otter in California. Pp. 79-81
in CalCOFI Rep. Vol. XXI.

Ralls, K., J. Ballou, and R. L. Brownell. 1983. Genetic
diversity in California sea otters: theoretical
considerations and management implications. Biol.
Conserv. 25:209-232.

Reed, M., D. French, J Calambokidis, and J. Cubbage.
Simulation modeling of the effects of oil spills on
population dynamics of northern fur seals. Final Report.
Minerals Management Service Alaska OCS Region. Contract
NO. 14-12-0001-30145. Anchorage, AK 139pp.

Schneider, K. B. 1972. Reproduction in the female sea otter.
Federal Aid in Wildlife Restoration Project W-17-1.
Project progress report. 26pp.

Schneider, K. B. 1978. Sex and Age segregation in sea
otters. Federal Aid in Wildlife Restoration Project W-
17-4 and W-17-5. Final Report. Alaska Dept. of Fish and
Game. 45pp.

Siler, W. 1979. A competing-risk model for animal mortality.
Ecology 60:750-757.

Siniff, D. B., T. D. Williams, A. M. Johnson, and D. L.
Garshelis. 1982. Experiments on the response of sea
otters, Enhydra lutris, to oil contamination. Biol.
Conserv. 2:261-272.

Wendell, F. E., J. A. Ames, and R. A. Hardy. 1984. Pup
dependency and length of reproductive cycle: estimates
from observations of tagged sea otters, Enhydra lutris,
in California. Calif. Fish and Game 70:89-100.

Wendell, F. E., J. A. Ames, and R. A. Hardy. 1986. A review
of California sea otter, Enhydra lutris, surveys. Marine
resources Technical Report No. 51. 42pp.

Wendell, F. E., R. A. Hardy, and J. A. Ames. 1985.
Assessment of the accidental take of sea otters, Enhydra
lutris, in gill and trammel nets. Draft Report.
California Dept. of Fish and Game. Morro Bay, CA. 30pp.

Wilson, E. O., and W. H. Bossert. 1971. A primer of
population biology. Sinauer Associates,Inc. Sunderland,
MA 192pp.

United States Fish and Wildlife Service. 1986. Summary of
information on the biology of the southern sea otter.

273

Technical Support Document 1, Draft Environmental Impact
Statement for Proposed Translocation of Southern Sea
Otters. 7Opp.

APPENDICES

Appendix 2.1 - Reproductive data on individual females.

SEA OTTER 6

DATE DAYS CUM STATUS
BETWEEN DAYS
05-JUL-84 NO PUP
07-JUL-84 2 2 NO PUP
08-JUL-84 1 3 NO PUP
17-JUL-84 9 12 NO PUP
18-JUL-84 1 13 NO PUP
19-JUL=84 1 14 NO PUP
21-Jul-84 2 16 NO PUP
22-Jul-84 1 17 NO PUP
23-Jul-84 1 18 NO PUP
25-Jul-84 2 20 NO PUP
26-Jul-84 1 21 NO PUP
27-Jul-84 1 22 NO PUP
28-Jul-84 1 23 NO PUP
29-Jul-84 1 24 NO PUP
30-Jul-84 1 25 NO PUP
08-Aug-84 9 34 NO PUP
13-Aug-84 5 39 NO PUP
21-Aug-84 8 47 NO PUP
31-Aug-84 10 57 NO PUP
01-Sep-84 1 58 NO PUP
01-Sep-84 fo) 58 NO PUP
02-Sep-84 1 59 NO PUP
03-Sep-84 1 60 NO PUP
05-Sep-84 2 62 NO PUP
06-Sep-84 1 63 NO PUP
08-Sep-84 2 65 NO PUP
12-Sep-84 4 69 NO PUP
15-Sep-84 3 72 NO PUP
17-Sep-84 2 74 NO PUP
14-Nov-84 58 132 NO PUP
29-Nov-84 15 147 NO PUP
21-Dec-84 22 169 NO PUP
26-Dec-84 5 174 NO PUP
15-Jan-85 20 194 NO PUP
20-Feb-85 36 230 NO PUP
21-Feb-85 1 231 NO PUP
06-Mar-85 13 244 NO PUP
12-Mar-85 6 250 NO PUP
13-Mar-85 1 251 NO PUP
14-Mar-85 1 252 NO PUP
02-Apr-85 19 271 NO PUP
16-Apr-85 14 285 NO PUP
18-Apr-85 ae 287 NO PUP
02-May-85 14 301 NO PUP
11-Jun-85 40 341 NO PUP

21-Jun-85 10 351 NO PUP

SEA OTTER 9

DATE

02-Mar-85
07-Mar-85
08-Mar-85
19-Mar-85
01-Apr-85
02-Apr-85
11-Apr-85
15-Apr-85
20-Apr-85
22-Apr-85
23-Apr-85
04-May-85
05-May-85
07-May-85
15-May-85
16-May-85
17-May-85
28-May-85
10-Jun-85
23-Jun-85
24-Jun-85
25-Jun-85
27-Jun-85
29-Jun-85
02-Jul-85
03-Jul-85
20-Jul-85
29-Jul-85
09-Aug-85
12-Aug-85
15-Aug-85
16-Aug-85
23-Aug-85
24-Aug-85
03-Sep-85
04-Sep-85
06-Sep-85
12-Sep-85
07-Oct-85
23-Oct-85
31-Oct-85
15-Nov-85
12-Dec-85
15-Dec-85
08-Jan-86
10-Jan-86

DAYS
BETWEEN

PRR b
WWPPPONPRPPNULUP

[ed

Pr
DNPRPORPNRPWWRPUOYIRPWNNEF EF

r

N
oO

16

CUM

DAYS

277

SEA OTTER 9 (cont.)

DATES

14-Jan-86
23-Jan-86
05-Feb-86
08-Feb-86
10-Feb-86
13-Feb-86
06-Mar-86
17-Mar-86
29-Mar-86
18-Apr-86
14-May-86
18-Jun-86
05-Jul-86
25-Aug-86

DAYS
BETWEEN

CUM

DAYS

318
327
340
343
345
348
369
380
392
412
438
473
490
541

278

SEA OTTER 11

DATE DAYS CUM STATUS
BETWEEN DAYS
17-May-85 NO PUP
07-Sep-85 113 113 PUP

22-Jul-86 318 431 NO PUP

279

SEA OTTER 14

DATE

18-Mar-85
19-Mar-85
20-Mar-85
27-Mar-85
02-Apr-85
08-Apr-85
14-Apr-85
20-Apr-85
21-Apr-85
22-Apr-85
01-May-85
14-May-85
14-May-85
14-May-85
15-May-85
15-May-85
16-May-85
18-May-85
28-May-85
10-Jun-85
12-Jun-85
23-Jun-85
24-Jun-85
25-Jun-85
29-Jun-85
01-Jul-85
02-Jul-85
03-Jul-85
20-Jul=85
27-Jul-85
29-Jul-85
05-Aug=85
12-Aug-85
15-Aug-85
16-Aug-85
20-Aug-85
22-Aug-85
23-Aug-85
24-Aug=85
27-Aug-85
31-Aug-85
01-Sep=85
04-Sep-85
05-Sep=85
06-Sep-85
10-Sep-85
11-Sep-85

DAYS
BETWEEN

rR r
PNWONFPORPOOWUWOPPANDADYPHE

=)

ra

PPPPWUPPBUPPNAPUNUIYNNYIYGVPPNRPP

DAYS

280

SEA OTTER 14 (cont.)

DATES

19-Sep-85
25-Sep-85
30-Sep-85
23-Oct-85
28-Oct-85
31-Oct-85
15-Nov-85
19-Nov-85
09-Dec-85
10-Dec-85
14-Dec-85
28-Dec-85
18-Jan-86
29-Jan-86
06-Mar-86
09-Mar-86
10-Mar-86
13-Mar-86
17-Mar-86
01-Apr-86
26-Apr-86
05-Jul-86
06-Aug-86
27-Aug-86
30-Aug-86
12-Sep-86
03-Jan-87
14-Feb-87
16-Mar-87

DAYS CUM
BETWEEN DAYS
8 185
6 191
5 196
23 219
5 224
3 227
15 242
4 246
20 266
1 267
4 271
14 285
21 306
11 317
36 353
3 356
1 357
3 360
4 364
15 379
25 404
70 474
32 506
21 527
3 530
13 543
113 656
42 698

281

SEA OTTER 15

DATE

21-Mar-85
22-Mar-85
24-Mar-85
26-Mar-85
27-Mar-85
28-Mar-85
29-Mar-85
30-Mar-85
31-Mar-85
01-Apr-85
02-Apr-85
06-Apr-85
08-Apr-85
11-Apr-85
12-Apr-85
13-Apr-85
14-Apr-85
15-Apr-85
16-Apr-85
17-Apr-85
18-Apr-85
20-Apr-85
22-Apr-85
28-Apr-85
30-Apr-85
02-May-85
03-May-85
04-May-85
05-May-85
06-May-85
07-May-85
07-May-85
11-May-85
12-May-85
13-May-85
15-May-85
16-May-85
17-May-85
23-May-85
24-May-85
25-May-85
26-May-85
28-May-85
02-Jun-85
06-Jun-85
08-Jun-85

DAYS
BETWEEN

NVPUNPRPPOAPPNPPROPPRPRPPNNONNEPRPRPRPEPPPUNAPRPPPPRPERPNNE

CUM

DAYS

282

SEA OTTER 15 (cont.)

DATE

10-Jun-85
11-Jun-85
12-Jun-85
24-Jun-85
25-Jun-85
29-Jun-85
01-Jul-85
03-Jul-85
06-Jul-85
08-Jul-85
25-Jul-85
25-Jul-85
27-Jul-85
29-Jul-85
31-Jul-85
01-Aug-85
02-Aug-85
05-Aug-85
14-Aug-85
16-Aug-85
17-Aug-85
19-Aug-85
20-Aug-85
23-Aug-85
24-Aug-85
27-Aug-85
31-Aug-85
01-Sep-85
04-Sep-85
05-Sep-85
06-Sep-85
07-Sep-85
10-Sep-85
12-Sep-85
13-Sep-85
19-Sep-85
24-Sep-85
25-Sep-85
28-Sep-85
10-Oct-85
14-Oct-85
15-Oct-85
16-Oct-85
18-Oct-85
21-Oct-85
23-Oct-85
26-Oct-85

DAYS
BETWEEN

a

NNWNNPRPNRPEHN

Rr

WNWNPRPENWPUDPNWPPPWUPARWPUWUPNENWOWPPNNNO

SEA OTTER 15 (cont.)

DATES

31-Oct-85
07-Nov-85
11-Nov-85
12-Nov-85
15-Nov-85
19-Nov-85
20-Nov-85
08-Dec=-85
10-Dec-85
13-Dec-85
14-Dec-85
28-Dec-85
03-Jan-86
08-Jan-86
09-Jan-86
13-Jan-86
18-Jan=86
24-Jan-86
25-Jan-86
28-Jan-86
03-Feb=86
07-Feb=86
08=Feb=-86
13-Feb=-86
15-Feb=-86
19-Feb-86
06-Mar-86
10-Mar-86
12-Mar-86
13-Mar-86
24-Mar-86
10-Apr-86
18-Apr-86
29-Apr-86
18-May-86
10-Jun-86
24-Jun-86
05-Jul-86
28-Jul-86
05-Aug-86
08-Aug-86
27-Aug-86
12-Sep-86
03-Oct-86
09-Oct-86
02-Nov-86
08-Nov-86

DAYS
BETWEEN

=)
PPUNORPhWP EY U

Pr
OPNURFPEAWRPAUNPPUOD

Rr

PP
OIP PN >

CUM
DAYS

224
231
235
236
239
243
244
262
264
267
268
282
288
293
294
298
303
309
310
313
319
323
324
329
331
335
350
354
356
357
368
385
393
404
423
446
460
471
494
502
505
524
540
561
567
591
597

284

SEA OTTER 15 (cont.)

DATE

26-Nov-86
13-Dec-86
15-Dec-86
07-Jan-87
06-Mar-87

DAYS
BETWEEN

18
17

2
23
58

CUM STATUS
DAYS

615 NO PUP
632 NO PUP
634 PUP
657 PUP
715 PUP

285

SEA OTTER 16 (cont.)

DATE

08-Apr-85
09-Apr-85
11-Apr-85
13-Apr-85
15-Apr-85
20-Apr-85
25-Apr-85
06-May-85
16-May-85
01-Jul-85
06-Jul-85
16-Aug-85
25-Sep-85
15-Nov-85
11-Dec-85
25-Mar-86
30-Apr-86
12-Jun-86
04-Jul-86
01-Nov-86
23-Nov-86
29-Nov-86

DAYS
BETWEEN

SEA OTTER 19

DATES DAYS CUM STATUS
BETWEEN DAYS

08-Apr-85

09-Apr-85 a 1 PUP
11-Apr-85 2 3 PUP
11-Apr-85 0 3 PUP
22-Apr-85 5 14 NO PUP
25-Apr-85 3 sy) NO PUP
02-May-85 7 24 NO PUP
06-May-85 4 28 NO PUP
16-May-85 10 38 NO PUP
17-Jun-85 32 WO NO PUP
01-Jul-85 14 84 NO PUP
06-Jul-85 6 ve ~89 NO PUP
22-Aug-85 47 136 NO PUP
07-Sep-85 16 152 NO PUP
14-Nov-85 68 220 PUP
15-Nov-85 1 221 PUP
19-Nov-85 4 225 PUP
11-Dec-85 22 247 PUP
03-Jan-86 23 270 PUP
07-Feb-86 35 305 PUP
28-Apr-86 80 385 NO PUP
06-Jul-86 69 454 NO PUP
28-Jul-86 22 476 NO PUP
23-Sep-86 57 533 PUP
28-Sep-86 5 538 PUP
27-Oct-86 29 567 PUP
01-Nov-86 5 572 PUP

06-Nov-86 5 577 NO PUP

SEA OTTER 22

DATES DAYS CUM STATUS
BETWEEN DAYS
12-Apr-85 NO PUP
20-Apr-85 8 8 NO PUP
09-May-85 19 27 NO PUP
10-May-85 1 28 NO PUP
12-May-85 2 30 NO PUP
13-May-85 1 a NO PUP
15-May-85 2 33 NO PUP
16-May-85 1 34 NO PUP
24-May-85 8 42 NO PUP
08-Jun-85 15 57 NO PUP
10-Jun=85 2 59 NO PUP
12-Jun-85 2 61 NO PUP
13-Jun-85 it 62 NO PUP
17-Jun-85 4 66 NO PUP
26-Jun-85 9 75 NO PUP
29-Jun-85 3 78 NO PUP
04-Jul-85 5 83 NO PUP
06-Jul-85 2 85 NO PUP
21-Jul-85 15 100 NO PUP
22-Jul-85 i 101 NO PUP
28-Jul-85 6 107 NO PUP
29-Jul-85 1 108 NO PUP
04-Aug-85 6 114 NO PUP
16-Aug-85 12 126 NO PUP
17-Aug-85 1 27) NO PUP
25-Aug-85 8 135 NO PUP
31-Aug-85 6 141 NO PUP
10-Oct-85 40 181 NO PUP
13-Nov-85 34 215 NO PUP
30-Nov-85 17 232 NO PUP
03=Dec-85 3 235 NO PUP
12-Dec-85 9 244 NO PUP
15-Jan-86 34 278 NO PUP
16-Jan-86 st 279 NO PUP
22-Jan-86 6 285 NO PUP
10-Feb-86 19 304 NO PUP
19-Feb-86 9 313 NO PUP
01-Jun-86 102 415 NO PUP
09-Sep-86 100 515 NO PUP
22-Oct-86 43 558 NO PUP

12-Nov-86 21 579 NO PUP

288

SEA OTTER 25

DATES

20-Apr-85
22-Apr-85
23-Apr-85
28-Apr-85
01-May-85
06-May-85
07-May-85
11-May-85
16-May-85
17-May-85
28-May-85
10-Jun-85
23-Jun-85
25-Jun-85
29-Jun-85
01-Jul-85
02-Jul-85
03-Jul-85
04-Jul-85
0-Jul-85

29-Jul-85
31-Jul-85
01-Aug-85
05-Aug-85
15-Aug-85
16-Aug-85
17-Aug-85
18-Aug-85
19-Aug-85
19-Aug-85
20-Aug-85
21-Aug-85
22-Aug-85
23-Aug-85
24-Aug-85
27-Aug-85
31-Aug-85
01-Sep-85
02-Sep-85
03-Sep-85
04-Sep-85
04-Sep-85
05-Sep-85
06-Sep-85
07-Sep-85
10-Sep-85
12-Sep-85
15-Sep-85
18-Sep-85

DAYS
BETWEEN

r
PRPUPRPUWUPRN

PR
WW

WWNWPPRPOPPPPRUPPPRPRPOPPPRPOBPNUDPPERPNEAN

DAYS

STATUS

NO PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP

SEA OTTER 25 (cont.)

DATES

24-Sep-85
25-Sep-85
28-Sep-85
30-Sep-85
03-Oct-85
08-Oct-85
10-Oct-85
14-Oct-85
15-Oct-85
23-Oct-85
29-Oct-85
30-Oct=85
31-Oct=85
01-Nov-85
07-Nov-85
11-Nov-85
12-Nov-85
15-Nov-85
20-Nov-85
29-Nov-85
04=Dec-85
08=Dec-85
14-Dec=85
15=Dec=85
27=Dec=85
03-Jan-86
o08s=Jan-86
13-Jan-86
15-Jan-86
18-Jan-86
19-Jan-86
23-Jan-86
04-Feb-86
08-Feb-86
11-Feb-86
13-Feb-86
15-Feb-86
19-Feb-86
01-Mar-86
06-Mar-86
10-Mar-86
12-Mar-86
13-Mar-86
17-Mar-86
24-Mar-86
25-Mar-86
10-Apr-86
28-Apr-86
29-Apr-86

POPPRPAOPARNUWNUWE OV

a)

a

kr

PR
POHDPIPPNARUOANNWEN PP WNUNINE RHE UOUWP

CUM

DAYS

157
158
161
163
166
171
173
177
178
186
192
193
194
195
201
205
206
209
214
223
228
232
238
239
251
258
263
268
270
273
274
278
290
294
297
299
301
305
315
320
324
326
327
331
338
339
355
373
374

290

PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP
PUP

SEA OTTER 25 (cont.)

DATES

31-May-86
10-Jun-86
24-Jun-86
05-Jul-86
07-Jul-86
06-Aug-86
08-Aug-86
12-Sep-86
21-Sep-86
03-Oct-86
16-Oct-86

DAYS
BETWEEN

CUM

DAYS

406
416
430
441
443
473
475
510
519
531
544

291

SEA OTTER 27

DATES

04-Oct-85
07-Oct-85
09-Oct-85
11-Oct-85
16-Oct-85
17-Oct-85
23-Oct-85
28-Oct-85
30-Oct-85
31-Oct-85
11-Nov-85
1Li-Nov-85
16-Nov-85
29-Nov-85
10-Dec-85
14-Dec-85
15-Dec-85
20-Dec=85
27-Dec-85
28-Dec-85
30=-Dec-85
03-Jan-86
08-Jan-86
09-Jan-86
15-Jan-86
18-Jan-86
22-Jan-86
23-Jan-86
28-Jan-86
08-Feb-86
11-Feb-86
13-Feb-86
06-Mar=86
10-Mar-86
12-Mar-86
24-Mar-86
02-Apr-86
18-Apr-86
29-Apr-86
05-May-86
23-May-86
18-Jun-86
24-Jun-86
05-Jul-86
28-Jul-86
29-Aug-86
17-Sep-86
27-Sep-86
28-Sep-86

DAYS
BETWEEN

PR
PNPIUP ERP WUOPPNUDPUNN WwW

N Rr
WONNPFPRPNWRUP HE WOAP YU

rR

PR
PO

SEA OTTER 27

DATES

04-Nov-86
07-Nov-86
12-Nov-86
28-Mar-87

(cont. )

DAYS
BETWEEN

37
3

5
136

CUM

DAYS

396
399
404
540

293

STATUS

NO PUP
NO PUP
NO PUP
NO PUP

SEA OTTER 31

DATES

16-Oct-85
21-Oct-85
23-Oct-85
28-Oct-85
30-Oct-85
07-Nov-85
15-Nov-85
16-Nov-85
10-Dec=-85
12-Dec-85
20-Dec-85
28-Dec-85
03-Jan-86
21-Jan-86
23-Jan-86
24-Jan-86
28-Jan-86
05-Feb=-86
08-Feb-86
10-Feb-86
13-Feb-86
15-Feb-86
19-Feb-86
21-Feb-86
01-Mar-86
05-Mar-86
06-Mar-86
10-Mar-86
12-Mar=86
27-Mar-86
01-Apr=86
06-Apr-86
10-Apr-86
18-Apr-86
24-Apr-86
29-Jun-86
08-Aug-86
29-Aug-86
12-Sep-86
09-Oct-86
23-Oct-86
08-Jan-87
05-Feb-87
02-Mar-87
16-Mar-87
18-Mar-87
07-Apr-87
28-Apr-87

DAYS
BETWEEN

iS)

ray
AOR UNUNUNFLRPRPOAONPMENWNWORPNADDANAPHPAODONUN WU

ra

294

SEA OTTER 33

DATES

03-Jul-84
23-Oct-85
29-Oct-85
30-Oct-85
31-Oct-85
02-Nov-85
07-Nov-85
12-Nov-85
14-Nov-85
17-Nov-85
19-Nov-85
20-Nov-85
29-Nov-85
08-Dec-85
12-Dec-85
14-Dec-85
03-Jan-86
10-Jan-86
13-Jan-86
22-Jan-86
23-Jan-86
24-Jan-86
25-Jan-86
28-Jan-86
07-Feb-86
09-Feb-86
11-Feb-86
13-Feb-86
15-Feb-86
19-Feb-86
23-Feb-86
24-Feb-86
01-Mar-86
08-Mar-86
10-Mar-86
12-Mar-86
13-Mar-86
25-Mar-86
24-Apr-86
10-Jun-86
24-Jun-86
06-Aug-86
08-Aug-86
29-Aug-86
03-Oct-86
01-Nov-86
17-Dec-86
23-Mar-87
28-Mar-87

DAYS
BETWEEN

PNNAIUONRPPRPNNNNOWRPRPRPUOWIONRPUOUPRPNWNUUNEFE O

295

STATUS

NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP

NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP
NO PUP

SEA OTTER 33 (cont.)

DATES

05-Apr-87
24-Apr-87

DAYS
BETWEEN

8
19

529
548

296

STATUS

PUP
NO PUP

SEA OTTER 36

DATES DAYS CUM STATUS
BETWEEN DAYS

18-Nov-85 NO PUP
19-Nov-85 1 1 NO PUP
22-Nov-85 3 4 NO PUP
25-Nov-85 3 7 NO PUP
06-Dec-85 11 18 NO PUP
09-Dec-85 3 21 NO PUP
20-Dec-85 11 32 NO PUP
15-Jan-86 26 58 NO PUP
21-Jan-86 6 64 NO PUP
05-Feb-86 15 79 NO PUP
11-Feb-86 6 85 NO PUP
27-Feb-86 16 101 NO PUP
08-Apr-86 40 141 NO PUP
14-Apr-86 6 147 NO PUP
21-May-86 37 184 PUP
22-May-86 1 185 PUP
06-Jun-86 15 200 PUP
11-Jun-86 5 205 PUP
13-Jun-86 2 207 PUP
16-Jun-86 3 210 PUP DEAD
18-Jun-86 2 212 PUP GONE
19-Jun-86 1 213 NO PUP
16-Jul-86 27 240 NO PUP
17-Jul-86 1 241 NO PUP
22-Jul-86 5 246 NO PUP
01-Aug-86 10 256 NO PUP
08-Aug-86 7 263 NO PUP
15-Aug-86 7 270 NO PUP
29-Aug-86 14 284 NO PUP
04-Sep-86 6 290 NO PUP
10-Sep-86 6 296 NO PUP
12-Sep-86 2 298 NO PUP
03-Oct-86 21 319 NO PUP
07-Oct-86 4 323 NO PUP
10-Nov-86 34 357 NO PUP
16-Nov-86 6 363 NO PUP
18-Nov-86 2 365 NO PUP
25-Nov-86 7 372 NO PUP
30-Nov-86 5 377 NO PUP
12-Dec-86 12 389 NO PUP
17-Dec-86 5 394 NO PUP
24-Dec-86 7 401 NO PUP
03-Mar-87 69 470 NO PUP
20-Mar-87 17 487 PUP
25-Mar-87 5 492 NO PUP

Appendix 2.2 Data on instrumented otters used for estimating survival
rates.
OTTER SEX CAPTURE LAST DATE STATUS DAYS LAST DATE
NUMBER DATE TX HEARD TRANSMITTING RECOGNIZED
BY TAGS
1 AM 0O7-MAR-84& 28-MAY-85 HISSING 447
2 AM 16-MAR-84 15-SEP-84 MISSING 183
3 AM 21-MAR-84 28-AUG-85 TX EXPIRED 525
4 AM 21-MAR-84 28-AUG-85 TX EXPIRED 523 26-Nov-86
6 AF 03-JUL-84 23-JUN-85 TX EXPIRED 355
7 AM 15-FEB-85 10-NOV-86 MISSING 633
9 AF 01-MAR-85 27-AUG-86 MISSING 544
10 AM 01-MAR-85 12-NOV-86 MISSING 621
11 AF 15-MAR-85 26-NOV-86 MISSING 621
13 JM 16-MAR-85 O05-JAN-87 MISSING 660 11-Aug-87
14 AF 16-MAR-85 30-MAR-87 MISSING 744
15 AF 20-MAR-85 11-DEC-86 TX EXPIRED 631 06-Mar-87
16 AF O3-APR-85 02-DEC-86 MISSING 608
17 AM 0O3-APR-85 28-MAR-86 MISSING 359
19 AF O3-APR-85 11-NOV-86 MISSING 587
21 AF 10-APR-85 13-APR-85 MISSING 3
22 AF 10-APR-85 16-NOV-86 MISSING 585
23 AM 10-APR-85 27-APR-85 DEAD 17
25 AF 13-APR-85 20-OCT-86 MISSING 555
26 AF O8-MAY-85 O05-JUN-85 DEAD 28
27 AF 04-OCT-85 31-OCT-86 TX EXPIRED 392 28-Mar-87
28 AF 04-G€T-85 04-NOV-85 TX FAILED 31
29 JF 11-OCT-85 14-MAR-87 MISSING 519
30 JM 11-OCT-85 29-JUL-87 MISSING 656
31 AF 11-OCT-85 24-JUL-87 MISSING 651
33 AF 18-OCT-85 26-SEP-87 MISSING 708
34 AM 19-OCT-85 23-SEP-86 MISSING 339
35 JM O8-NOV-85 21-MAR-87 MISSING 498
36 AF O08-NOV-85 07-OCT-87 MISSING 698
37 JF 22-NOV-85 17-OCT-86 MISSING 329
38 JF 22-NOV-85 O02-JAN-86 MISSING 4j
39 JF 22-NOV-85 25-MAR-87 MISSING 488
40 JF 17-DEC-85 03-DEC-87 TRANSMITTING 716
41 JM 17-DEC-85 13-APR-87 DEAD 482
42 JF 17-DEC-85 09-OCT-87 MISSING 661
43 JM 18-DEC-85 10-NOV-87 MISSING 692
44 JF 18-DEC-85 29-JUN-86 DEAD 193
45 JF 18-DEC-85 22-MAR-87 MISSING 459
46 JF 18-DEC-85 25-DEC-87 TRANSMITTING 737
47 JF 30-DEC-85 28-DEC-87 TRANSMITTING 728

298

Appendix 2.3 -- Tag loss information for instrumented sea otters in

California as of

10 July 1987.

OTTER DATE RIGHT TAG LEFT TAG DAYS FROM TAGGING

NUMBER TAGGED LAST SEEN MISSING LAST SEEN MISSING TO DATE LAST SEEN
OR MISSING

RIGHT TAG LEFT TAG

1 O7-MAR-84 28-MAY-85 28-MAY-85 447 447
2 16-MAR-84 15-SEP-84 15-SEP-85 183 183
3 21-MAR-84 28-AUG-85 28-AUG-85 525 525
4 21-MAR-84 26-NOV-86 26-NOV-86 980 980
6 O3-JUL-84 21-FEB-85 06-MAR-85 21-JUN-85 246 353
7 15-FEB-85 06-AUG-86 06-AUG-85 537 537
9 01-MAR-85 07-OCT-85 23-OCT-85 05-FEB-86 29-APR-86 236 424
10 0O1-MAR-85 14-NOV-85 06-APR-86 11-SEP-85 14-NOV-85 401 258
11 15-MAR-85 07-SEP-85 07-SEP-85 176 176
13° 16-MAR-85 31-AUG-85 31-AUG-85 168 168
14° 16-MAR-85 25-SEP-85 15-NOV-85 29-JUN-85 24-JUL-85 244 130
15 20-MAR-85 06-MAR-87 06-MAR- 87 716 716
16 O3-APR-85 16-AUG-85 11-SEP-85 16-AUG-85 11-SEP-85 161 161
17 O3-APR-85 11-NOV-85 11-NOV-85 222 222
19 0O3-APR-85 01-NOV-86 11-SEP-85 11-DEC-85 577 252
21 10-APR-85 13-APR-85 13-APR-85 3 3
22 10-APR-85 22-OCT-86 22-0cT-85 560 560
23. 10-APR-85 27-APR-85 27-APR-85 17 17
25 13-APR-85 21-SEP-86 03-OCT-86 24-JUN-86 06-AUG-86 538 480
26 O08-MAY-85 06-JUN-85 06-JUN-85 29 29
27 04-OCT-85 28-MAR-87 28-MAR-87 540 540
28 04-0CT-85 15-OCT-85 15-OCT-85 11 11
29 11-OCT-85 30-OCT-85 11-FEB-86 30-OCT-85 11-FEB-86 123 123
30 11-OCT-85 30-MAY-87 30-MAY-85 596 596
31 11-OCT-85 18-APR-86 02-MAY-86 16-MAY-87 203 582
33. 18-OCT-85 07-JUL-87 07-JUL-87 627 627
34 19-OCT-85 13-AUG- 86 06-JUL-86 13-AUG-86 298 298
35 O8-NOV-85 03-MAR-87 03-MAR-87 480 480
36 O08-NOV-85 01-JUL-87 06-JUL-87 600 605
37 22-NOV-85 17-JUL-86 17-JUL-86 237 237
38 22-NOV-85 18-DEC-85 18-DEC-85 26 26
39 22-NOV-85 28-MAR-87 28-MAR- 87 491 491
40 17-DEC-85 06-JUN-86 17-JUL-86 09-JUL-87 212 569
41 17-DEC-85 17-OCT-86 17-OCT-86 304 304
42 17-DEC-85 25-JUN-86 17-JUL-86 25-JUN-86 17-JUL-86 212 212
43 18-DEC-85 28-JAN-86 27-FEB-87 11-APR-87 41 479
44 18-DEC-85 04-JUL-86 04-JUL- 86 198 198
45 18-DEC-85 11-JUN-86 17-JUL-86 17-JAN-87 211 395
46 18-DEC-85 09-NOV-86 29-MAR-87 09-NOV-86 29-MAR-87 466 466
47 30-DEC-85 01-JUL-87 01-JUL-87 548 548

APPENDIX 3.1 -- Analysis of variance for the distance

between successive locations of individual instrumented sea
Log transformed data, base 2.

otters.

A. LOCATIONS 18-36 HOURS APART

Adult females
Among individuals
Error

Juvenile females
Among individuals
Error

Adult males
Among individuals
Error

Juvenile males
Among individuals
Error

Age/sex classes
Among classes
Error

df MS F
12 20.1 37.4
3450 0.537
9 8.8 15.8
1918 0.557
7 1.6 4.0
1401 0.398
4 5.0 6.8
964 0.735
3 185.6 342.2
7733 0.542

B. LOCATIONS MORE THAN 36 HOURS APART

Adult females
Among individuals
Error

Juvenile females
Among individuals
Error

Adult males
Among individuals
Error

Juvenile males
Among individuals
Error

Age/sex classes
Among classes
Error

df MS F
12 15.8 18.9
1441 0.837
9 8.1 7.6
913 1.1
7 33.4 22.1
635 1.5
4 1.6 1.0
544 1.6
3 131.7 94.1
3533 1.4

300

Pp
<0.001

<0.001

<0.001

<0.001

<0.001

Pp

<0.001

<0.001

<0.001

ns

<0.001

APPENDIX 3.2 -- Analysis of variance for the minimum convex
polygon daily home ranges. Log transformed data, base 2.

df MS F p
Adult females
Among individuals 4 Seis 5.91 ns
Error 4 1.41
Juvenile females
Among individuals 5 3.04 vA ns
Error 5 1.10
Adult males sample size too small for testing
Juvenile males
Among individuals 3 1.90 0.39 ns
Error 7 3.41 4.87

Age/sex classes after otters
Among classes 2 25.95 28.51 <0.001
Error 16 0.91

301

APPENDIX 3.3 -- Analysis of variance for seasonal
differences in monthly harmonic mean home range size. Log
transformed data, base 2.

af MS F p

Adult females
Among individuals a2. 17.08 7.98 <0.001
Season 1 0.40 0.19 ns
Individualxseason 12 0.94 0.44 ns
Error 227 2.14

Juvenile females
Among individuals 8 2.36 1.78 ns
Season 1 1.50 1.13 ns
Individualxseason 8 0.98 0.73 ns
Error 137 1.33

Adult males
Among individuals 7 9.11 6.70 <0.001
Season 1 0.10 0.07 ns
Individualxseason 7 Nog Bal, 0.89 ns
Error 96 1.36

Juvenile males
Among individuals 4 5.88 2.47 ns
Season 1 5.80 2.43 ns
Individualxseason 4 1.60 0.67 ns
Error 719 2.38

302

APPENDIX 3.4 -- Analysis of variance for differences in
monthly harmonic mean home range size among age/sex classes.
Log transformed data, base 2.

af MS F p
Adult females
Among individuals 12 17.08 8.29 <0.001
Error 240 2.06
Juvenile females
Among individuals 8 2.45 N67) ns
Error 145 0.557
Adult males
Among individuals a 8.47 5.92 <0.001
Error 105 1.43
Juvenile males
Among individuals 4 5.89 2.47 ns
Error 84 2.38
Age/sex classes after otters
Among classes 3 294.87 163.8 <0.001
Error 574 1.80

*This test does not include seasonal effects. The only
age/sex group with significant seasonal effects was the
juvenile females (see appendix 3.4). Variation due to
season is included in the error term; thus, this test is
conservative.

303

APPENDIX 3.5 - Analysis of variance for the distance between
extreme locations of individual instrumented sea otters.
Log transformed data, base 2.

fob a MS F p
Adult females
Among individuals 12 19.8 22.56 <.001
Error 240 0.88
Juvenile females
Among individuals 8 9.81 4.54 <.001
Error 146 2.16
Adult males
Among individuals 7 10.13 3.58 <2 801.
Error ~ 104 2.83
Juvenile males
Among individuals 4 2.67 1.68 ns
Error 84 1.59
Age/Sex classes
Among classes 3 90.63 54.6 <.001
Error 574 1.66

APPENDIX 4.1 - TWENTY-FOUR-HOUR DATA BY OBSERVATION PERIOD.

Fa ee IE ———EEE EE
OTTER AGE/ DATE LENGTH NUMBER OF 10-MIN PERIODS
NUMBER _ SEX HRS REST FEED OTHER
6 AF 19-Jul-84 24 81 27 38
6 AF 25-Jul-84. 48 145 101 62
6 AF 07-Aug-84 24 64 66 20
6 AF 21-Aug-84 14 39 32 14
6 AF 31-Aug-84 24 60 58 33
6 AF 05-Sep-84 17 35 57 15
7 AM 05-Sep-85 24 64 71 12
7 AM 08-Oct-85 24 105 20 20
7 AM 06-Aug-86 24 77 53 13
9 AFP 23-Jul-85 24 49 69 27
9 AF 26-Aug-85 30 73 85 . 22
10 AM 10-Sep-85 23 37 mS 17
11 AF 05-Sep-85 24 64 58 23
11 AF 08-Oct-85 24 61 54 28
13 JM 24-Aug-86 24 65 45 38
14 AFP 24-Jul-85 24 32 86 25
14 AFP 27-Aug-85 18 51 44 12
15 AF 20-May-85 24 79 52 15
15 AF 30-May-85 23 68 61 6
15 AF 03-Jun-85 24 75 45 25
15 AF 18-Jul-85 24 75 45 26
15 AF 12-Aug-85 72 229 128 76
16 AF 19-May-85 24 66 52 26
16 AF 28-May-85 48 147 98 44
16 AF 10-Jul-85 24 75 32 37
16 AF 10-Sep-85 23 66 54 15
16 AF 08-Oct-85 24 75 45 18
16 AFP 04-Nov-86 48 137 104 46
17 AM 02-Jul-85 48 120 105 67
17 AM 19-Jul-85 24 69 69 7
19 AF 19-May-85 23 54 59 24
19 AF 28-May-85 47 130 117 36
19 AF 04-Nov-86 48 173 val 46
19 AF 10-Sep-85 23 62 65 7
19 AF 08-Oct-85 24 94 44 8
22 AF 30-Jul-85 48 166 103 20
22 AF 05-Sep-85 24 77 39 29
22 AF 08-Oct-85 24 71 49 20
25 AFP 07-Aug-85 48 159 152 45
25 AFP 28-Aug-85 11 35 16 ALS)
27 AFP 13-May-85 48 138 97 54
29 JF 25-Jun-86 24 40 69 36
30 JM 23-Apr-86 48 113 98 80
34 AM 30-Jul-86 24 88 24 32
34 AM 13-Aug-86 24 70 60 15
35 JM 25-Feb-86 50 87 116 89
35 JM 16-Sep-86 48 66 109 114

36 AF 13-Mar-86 24 43 85 17

305

APPENDIX 4.1 (cont.)

Se Sun ee eee ee ee ee ee EE EE ee
OTTER AGE/ DATE LENGTH NUMBER OF 10-MIN PERIODS
NUMBER SEX (HRS) REST FEED OTHER

36 AFP 21-May-86 48 172 86 33

37 JF 06-Mar-86 24 55 85 6

39 JF 25-Feb-86 33 69 75 53

39 JF 29-Apr-86 24 48 84 14

40 JF 03-Apr-86 48 103 157 29

40 JF 20-Aug-86 24 90 35 20

41 JM 13-Mar-86 24 40 81 23

41 JM 20-Aug-86 24 70 45 29

41 JM 24-Sep-86 48 73 102 115

42 JF 16-Apr-86 49 128 134 30

43 JM 20-Aug-86 24 50 44 50

44 JF 19-Mar-86 48 124 136 30

45 JF 13-Nov-86 48 87 173 27

46 JF 06-Mar-86 21 76 35 12

46 JF 30-Apr-86 24 61 77 6

47 JF 19-Nov-86 48 74 173 47

306

APPENDIX 4.2

OTTER
NUMBER .

WOUNNNYANAAAAAG

AGE/SEX

DATE

19-Jul-84
25-Jul-84
07-Aug-84
21-Aug-84
31-Aug-84
05-Sep-84
05-Sep-85
08-Oct-85
06-Aug-86
23-Jul-85
26-Aug-85
10-Sep-85
05-Sep-85
08-Oct-85
24-Aug-86
24-Jul-85
27-Aug-85
20-May-85
30-May-85
03-Jun-85
18-Jul-85

DAYLIGHT DATA BY OBSERVATION PERIOD.

NUMBER OF 10-MIN PERIODS

REST

FEED

OTHER

APPENDIX 4.2 (cont.)

OTTER AGE/SEX DATE NUMBER OF 10-MIN PERIODS
NUMBER REST FEED OTHER
36 AF 13-Mar-86 20 48 10
36 AFP 21-May-86 121 43 21
37 JF 06-Mar-86 13 61 4
39 JF 25-Feb-86 46 50 48
39 JF 29-Apr-86 44 38 7
40 JF 03-Apr-86 39 113 13
40 JF 20-Aug-86 61 15 14
41 JM 13-Mar-86 4 53 20
41 JM 20-Aug-86 42 24 21
41 JM 24-Sep-86 34 71 54
42 JF 16-Apr-86 88 53 18
43 JM 20-Aug-86 30 19 38
44 JF 19-Mar-86 37 lil 15
45 JF 13-Nov-86 3 124 11
46 JF 06-Mar-86 41 14 3
46 JF 30-Apr-86 38 43 6
47 JF 19-Nov-86 16 96 27

APPENDIX 4.3

OTTER
NUMBER

OV OV OV OV OV

AGE/SEX

Fy Fey Fey Py ey Fy By Fay Py By ey Be

DATE

02-Jul-85
19-Jul-85
10-Sep-85
18-Jul-85
10-Sep-85
26-Aug-85
10-Sep-85
30-Jul-85
12-Aug-85
24-Jul-85
13-May-85
21-May-86
07-Aug-85
19-May-85
28-May-85
19-May-85
28-May-85
20-May-85
30-May-85
03-Jun-85
04-Nov-86
06-Aug-86
13-Aug-86
04-Nov-86
27-Aug-85
25-Jun-85
20-Aug-86
13-Nov-86
19-Nov-86
06-Mar-86
19-Mar-86
30-Apr-86
29-Apr-86
06-Mar-86
30-Apr-86
16-Apr-86
16-Sep-86
25-Feb-86
23-Apr-86
19-Jul-84
25-Jul-84
07-Aug-84
21-Aug-84
31-Aug-84
05-Sep-84

VISUAL .DATA BY OBSERVATION PERIOD.

NUMBER OF 10-MIN PERIODS

REST FEED
41 39
43 17

7 ie)
21 10
5 5
1 5
10 0)
90 30
110 32
10 38
78 21
115 2
71 28
10 2
56 28
0) 0)
22 6
34 18
38 23
41 14
11 ie)
12 9
ie) 16
8 17
11 0)
4 11
2 fe)
0 30
2 8
te) al
17 42
35 42
33 7
3 27
22 64
63 0
(0) 2
0) 6
43 1
37 5
77 35
11 25
10 14
14 a7
8 7

309

OTHER

2

Pr

rR

N
PONRPNUONNYNFPF RP UORPRPUOONONOKHWVIOOUFENN

APPENDIX 5.1 -- Analysis of variance for the length of dives
made by individual instrumented sea otters in California.
Log-transformed data, base 2. ;

af mean square F p
Adult females
Among individuals 4 130.0 281.4 <0.001
Error 2763 0.46
Adult females with pups
Among individuals 3 94.84 144.33 <0.001
Error 1171 0.66
Juvenile females . |
Among individuals 6 76.92 180.52 <0.001
Error 2129 0.43
Adult males :
Among individuals 6 10.85 14.36 <0.001
Error 1377 0.76
Juvenile males
Among individuals 4 7.39 20.04 <0.001
Error 493 0.37

310

APPENDIX 5.2 -- Analysis of variance for the length of the
surface intervals made by individual instrumented otters.
Log-transformed data, base 2.

df Mean square F p
Adult females
Among individuals 4 20253 179.96 <0.001
Error 2656 1.12
Adult females with pups
Among individuals 3 Sees 25.42 <0.001
Error 1089 1.23
Juvenile females
Among individuals 6 60.75 42.76 <0.001
Error 2114 1.42
Adult males
Among individuals 6 28.85 18.37 <0.001
Error 1310 abo i7/

Juvenile males
Among individuals 4 1.38 1.45 ns
Error 472 0.95

APPENDIX 5.3 - Analysis of variance for the length of dives
made during the day and night by the individual instrumented
otters.

af mean square F Pp

Adult females
Among individuals 4 129.98 302.26 <0.001
Day/night 1 0.10 0.23 ns
Day/night x individual 4 19.55 45.46 <0.001
Error 2758 0.43

Adult females with pups
Among individuals 3 94.80 166.37 <0.001
Day/night 1 43.90 77.01 *<05001
Day/night x individual 3 20.30 35.61 <0.001
Error 1167 0.57

Juvenile males
Among individuals 4 7.40 21.76 <0.001
Day/night 1 0.90 2.65 ns
Day/night x individual 4 4.48 13.16 <0.001
Error 488 0.34

Juvenile females 3
Among individuals 6 76.92 187.60 <0.001
Day/night 1 8.20 20.00 <0.001
Day/night x individual 6 4.50 10.98 <0.001
Error 2122 0.41

312

APPENDIX 5.4 - Analysis of variance for the length of
surface intervals made during the day and night by the

individual instrumented otters.

af
Adult females
Among individuals 4
Day/night 1
Day/night x individual 4
Error 2651
Adult females with pups
Among individuals 3
Day/night al
Day/night x individual 3
Error 1085
Juvenile males
Among individuals 4
Day/night 1
Day/night x individual 4
Error 467
Juvenile females
Among individuals 6
Day/night sl
Day/night x individual 6
Error 2107

mean square

201.33
1.11
5.52
1.11

31.23
1.30
1.67
1.23

1.38
10.80
9.32
0.86

60.75
24.20
8.55
1.39

F

<0.001
<0.001

<0.001
<0.001
<0.001

APPENDIX 10.

314

APPENDIX 10.1A -- Map of central California showing
California Department of Fish and Game mortal ity recovery
areas (after Ames, et al., 1983).

APPENDIX
California coast.

zis. Davenport

10.1B -- Ordinates

for fathom line

along the

t
N

5 0 5 10

—S ee eS

Kilometers

316

Moss

PE)

5+) CARMEL

: 4305
4 :
Pt. Lobos
YS.
soh

b=)

5 0 5
Sas See er ae
Kilometers

317

Ses

Pfeiffer Point

530

Zap

Grimes Point

5 0 5
[== — |
Kilometers

S705) John Little State Park.

318

Gorda

Za!

5 0 ee 10

= Se |
Kilometers

319

av

5 0 5 10
= = ee Se
Kilometers

320

MORRO BAY

Montana de Oro

State Park

Fa,

$2041 GROVER CITY

Santa Maria River

0 5 ~___IJ@
= = ee ae
Kilometers

S2iL

Za,

0 5

Kilometers

10

322

APPENDIX 10.2 -- User's Manual for OTPOP: A Simulation Model
for Assessing the Risks of Oil Spills to the California Sea
Otter Population.

USER'S MANUAL
for

OTPOP

A Simulation Model for Assessing the Risks of
Oil Spills to the California Sea Otter Population

1987

University of Minnesota

323

APPENDIX 10.2 -- User's Manual for OTPOP: A Simulation Model
for Assessing the Risks of Oil Spills to the California Sea
Otter Population.

324

I. INTRODUCTION.

This manual provides information only on running OTPOP,
it does not explain the logical structure of the model nor
the significance of the various parameters. Many problems
will be avoided if the user familiarizes his/her self with
the model documentation volume before attempting to use the
progran.

325

II. TECHNICAL SPECIFICATIONS

Hardware. OTPOP is designed to run on an IBM PC,xXT, or
AT microcomputer. An Intel 8087 or 80287 coprocessor is
required. A hard disk is recommended. A battery operated
clock and associated software are necessary for the random
number generator.

Software. OTPOP is written in FORTRAN and compiled on
the Rand-McFarland IBM Professional FORTRAN Compiler version
1.0. All code is ANSI FORTRAN77 compatible except for 4 IBM
Professional extensions used extensively throughout the
program: 1) all variables and array elements are
automatically set to 0 at the start of the program, 2) some
COMMON and declaration statements are included separately in
program units through the use of INCLUDE statements, 3)
subroutine variables are automatically saved without the use
of SAVE statements, and 4) most integer variables are
declared as 2 byte (INTEGER*2) to save memory. The action of
these extensions must be considered if the program is
transported to a different compiler.

The menu screens used to input parameters that are set
at runtime are generated using K&S Systems Screen Generator

version 4.7. The memory resident portion of the screen
generator must be loaded before running OTPOP (this is
accomplished in the batch file OP.BAT). If OTPOP is

transported to a different compiler a different screen
generator interface must be used, or the data entry portion
of the program rewritten.

Operating environment. DOS version 3.1 or higher is
required as an operating system, and at least 384 kilobytes
of RAM must be available. The following statements must be
included in the "CONFIG.SYS" file available at boot:

DEVICE=ANSI.SYS
BREAK ON
FILES=25
BUFFERS=25

The DOS "ANSI.SYS" file must also reside on the boot disk.

326

III. RUNNING THE PROGRAM

The batch file OT.BAT is supplied to easily load the
resident portion of the screen generator, run OTPOP, process
the raw output, and restore the proper MODE. To run OT.BAT
type "OT" and <enter>. Fig. 1 will briefly appear on the
screen as the screen generator is loaded, and Fig. 2,
introducing the program, will appear as OTPOP is loaded.

The screens pictured in Figs. 3-10 are used to set run
environment and model parameters at runtime. Default values
of all parameters automatically appear when the screens are
presented. To change a default value move the cursor to the
parameter in question and enter the new value. The program
automatically checks for parameter values that are out of
acceptable range or of the wrong type (for instance, entering
a letter when a number is required, or a number with a
decimal point when an integer is required). Move back and
forth between screens using the Fl and F10 function keys as
noted at the bottom of each screen.

Following is a description of the parameters that are
set at runtime using the input screens.

327

Figure 1.

K&S Systems copyright notice for the sceeen n
generator.

C: \OTTERS\INIPOP>sgx

The Screen Generator v4.47

(C) Copyright 1982,83,84,85 K & S Systems

(C) Copyright 1986 The West Chester Group

PO Box 1304, West Chester, PA 19380, (215) 644-4206

328

Figure 2. Introductory screen for OTPOP.

OTPOP

A SIMULATION MODEL FOR THE ANALYSIS
OF THE RISK OF OIL SPILLS TO THE
CALIFORNIA SEA OTTER POPULATION

FOR
USDI MINERALS MANAGEMENT SERVICE

WRITTEN AT THE
UNIVERSITY OF MINNESOTA

6 1987

329

Screen 1: Run parameters (Poet ELA

NUMBER OF YEARS PER RUN. Enter the number of years you want
to simulate after the oil spill.

NUMBER OF RUNS WITH OIL SPILL. Enter the number of separate
runs of the number of years specified above you want
executed.

NUMBER OF CONTROL RUNS. Enter the number of runs you want
conducted without introducing an oil spill. The control runs
run for the same number of years as the runs with oil spills.

INITIAL POPULATION SIZE. Enter the desired number of
independent otters in the simulated population at time of the
spill. Because of the stochasticity built in to the model,
and because the model runs for 3 simulated years before
introducing the spill, the number of animals at the time of
the spill may differ slightly from the inputted value. The
initial population size will also differ between runs.

DATE OF OIL SPILL. Enter the month (1-12) and day (1-31) to
introduce the spill.

DURATION OF SPILL. Enter the number of days (up to 30) that
the spill is to affect the population.

NORTH BOUNDARY OF SPILL. Enter the north boundary of the
simulated spill using the CDFG 5-fathom ordinate (see
Appendix A).

SOUTH BOUNDARY OF SPILL. Enter the south boundary of the
simulated spill, using the CDFG 5-fathom ordinate. This
value must be greater than the value for the northern
boundary entered above.

330

Figure 3. OTPOP Screen #1 -- Run parameters.
Dashed lines indicate location of parameter
edited by user.

SET RUN PARAMETERS:
NUMBER OF YEARS PER RUN
NUMBER OF RUNS WITH OIL SPILL
NUMBER OF CONTROL RUNS
INITIAL POPULATION SIZE
ER a DATE OF OIL SPILL (MONTH/DAY)
DURATION OF SPILL
NORTH BOUNDARY OF SPILL ( <SOUTH )

SOUTH BOUNDARY OF SPILL ( >NORTH )

Fl
F10

PROCEED
PREVIOUS SCREEN

331

values

Screen 2: Population parameters (Fig. 4):

EQUILIBRIUM POPULATION SIZE. Enter the carrying capacity, in
number of independent otters, of the simulated range. The
program OTRANGE may be run separately to determine this
value.

NORTH BOUNDARY OF RANGE. Enter the north boundary of the sea
otter range at the time of the spill using the CDFG 5-fathom
ordinate. Program OTRANGE may be run separately to determine
this value.

SOUTH BOUNDARY OF RANGE. Enter the south boundary of the sea
otter range at the time of the spill using the CDFG 5-fathom
ordinate. This value must be higher than the value entered
above for the north boundary of the range. Program OTRANGE
may be run separately to determine this value.

MAXIMUM PER CAPITA GROWTH RATE. Enter the maximum attainable
per capita annual growth rate of the population, in
animals/animal/year. This is "rpay" from equation (3) in the
documentation volume.

NON-LINEARITY OF DENSITY DEPENDENCE. Enter the value for "b"
in equation (3) in the documentation volume. The higher the
value of "b" the more rectangular the density dependence
function (see Fig. 4 in the documentation volume).

DENSITY INDEPENDENT MORTALITY RATE. Enter the mortality
rate, in animals/animal/year, due to density-independent
factors. This could be used to simulate incidental gill-net
mortality, predation, or harvest ("m" in equation (29) in the
documentation volume). See discussion of density independent
mortality in the OTRANGE section of the documentation volume.

DEGREE OF COMPENSATION. Enter the proportion of the density
independent mortality that will compensate for density
dependent mortality. See discussion of density independent
mortality in the OTRANGE section of the documentation volume.

Cy ke Gd we kd So! EDC ee eee CF TI Ee oD
Note: Density independent growth can be simulated by setting
the equilibrium population size very high relative to the
initial population size (i.e., at least 10 times as high),
and setting the nonlinearity coefficient very high (i.e.,
0.05 or greater).

; i i i, i i i CF

332

Figure 4. OTPOP Screen #2 -- Population parameters.
Lines indicate location of parameter values edited by
user.

SET POPULATION PARAMETERS:

EQUILIBRIUM POPULATION SIZE

NORTH BOUNDARY OF RANGE ( <SOUTH )
SOUTH BOUNDARY OF RANGE ( >NORTH )
MAXIMUM PER CAPITA ANNUAL GROWTH RATE
NON-LINEARITY OF DENSITY DEPENDENCE
DENSITY INDEPENDENT MORTALITY RATE

DEGREE OF COMPENSATION

Fl
F10

PROCEED
PREVIOUS SCREEN

333

Screen 3: Survival and reproductive parameters (Fig. 5):

ADULT FEMALE SURVIVAL RATE. Set the annual rate of survival
of adult females against incidental risks. This is "aj" in
equation (1) in the documentation volume. Express as a
proportion.

MODAL AGE OF FEMALE SENESCENCE. Set the age, in years, of
the modal age of death due to old age for females. MThis is
"T" in equation (10) in the documentation volume.

PRIME REPRODUCTIVE RATE. Set the maximum yearly reproductive
rate, expressed as weaned females per adult female per year.
This is "A" in equation (2) in the documentation volume.

PUP SURVIVAL RATE. Set the proportion of pups that will
survive from birth until weaning. Intrauterine mortality is
not considered.

ADULT MALE SURVIVAL RATE. Set the annual rate of survival of

adult males against incidental risks. This) ds, ag v eid
equation (1) in the documentation volume. Express aS a
proportion.

MODAL AGE OF MALE SENESCENCE. Set the age, in years, of the
modal age of death due to old age for males. This is "T" in
equation (10) in the documentation volume.

% VARIATION IN ADULT SURVIVAL. Set the relative percent by
which annual adult survival rates may vary. This variance is
used only in the recovery phase of the simulation, and the
distribution of annual survival rates is assumed to be
uniform between the specified boundaries. This simulates the
"environmental stochasticity" parameter, "p", described in
the description of the structure of LESLIE in the
documentation volume.

% VARIATION IN PUP SURVIVAL. Set the relative percent by
which annual pup survival rate may vary. This variance is
used only in the recovery phase of the simulation, and the
distribution of annual survival rates is assumed to be
uniform between the specified boundaries. Percent variation
in pup survival may differ from percent variation in adult
survival.

334

Figure 5. OTPOP Screen #3 -- survival and reproductive
parameters.

Lines indicate location of parameter values edited by
user.

SET SURVIVAL & REPRODUCTIVE PARAMETERS:

ADULT FEMALE SURVIVAL RATE
MODEL FEMALE AGE OF SENESCENCE
PRIME REPRODUCTIVE RATE

PUP SURVIVAL RATE

ADULT MALE SURVIVAL RATE

MODEL MALE AGE OF SENESCENCE

I+

PERCENT VARIATION IN ADULT SURVIVAL

I+

PERCENT VARIATION IN PUP SURVIVAL

Fl
F10

PROCEED
PREVIOUS SCREEN

335

Screen 4: Movement parameters (Fig. 6):

Classes of animals are listed down the left side of the
screen, movement parameters along the top. AR is the
autoregressive parameter, CE is the displacement parameter,
SIGMA is the standard deviation of daily distance moved, and
VMAX is the maximum possible daily distance moved. See the
discussion of the structure of OTMOVE and equation (22) in
the documentation volume.

336

Figure 6. OTPOP Screen #4 -- Movement parameters.
Lines indicate location of parameter values edited by
user.

SET OTTER MOVEMENT PARAMETERS:

AR CE SIGMA VMAX
JUVENILE FEMALES
ADULT FEMALES W/PUP
ADULT FEMALE W/O PUP
JUVENILE MALES
ADULT NON-TERRITORIAL
MALES
ADULT TERRITORIAL MALES
Fl PROCEED

F10 PREVIOUS SCREEN

337

Screen 5: Male territorialit arameters (Fig. 7):

AGE AT WHICH POTENTIALLY TERRITORIAL. Enter the age, in
years, at which males may hold breeding territories.

MAXIMUM % OF POTENTIALS THAT HOLD TERRITORIES. Enter the
percent of potentially territorial males that will hold
territories at the height of the breeding season.

MINIMUM % OF POTENTIALS THAT HOLD TERRITORIES. Enter the
minimum percent of potentially territorial males that will
hold territories at any time throughout the year.

MEAN TERRITORY LENGTH. Enter the average length of a male
territory, measured along the 5-fathom line, in 1/2 km units.

S. D. OF TERRITORY LENGTH. Enter the standard deviation
around mean territory length, in 1/2 km units.

MEAN ARRIVAL DATE. Enter the average date (month (1-12) /
day (1-31)) of arrival on a territory.

S. D. OF ARRIVAL DATE. Enter the standard deviation, in
days, of average arrival date.

MEAN DEPARTURE DATE. Enter the average date (month (1-12) /
day (1-31)) of departure from a territory.

S. D. OF DEPARTURE DATE. Enter the standard deviation, in
days, of average departure date.

90 Oo OOD 9 OO OO Oo 8 Oo O to oS
Note: The seasonality of territorial behavior and migrations
can be controlled through the standard deviations of arrival
and departure dates. Large standard deviations lead to less
pronounced seasonality. See discussion in the migratory
movements by adult males section of the documentation volume.

338

Figure 7. OTPOP Screen #5 -- Male territoriality
parameters.

Lines indicate location of parameter values edited by
user.

SET MALE TERRITORIALITY PARAMETERS:

AGE AT WHICH POTENTIALLY TERRITORIAL
MAXIMUM % OF PCTENTIALS THAT HOLD TERRITORIES
MINIMUM % OF POTENTIALS THAT HOLD TERRITORIES
MEAN TERRITORY LENGTH
S.D. OF TERRITORY LENGTH

sory fives MEAN ARRIVAL DATE (MONTH/DAY)

S.D. OF ARRIVAL DATE (IN DAYS)

n/a MEAN DEPARTURE DATE (MONTH/DAY)

S.D. OF DEPARTURE DATE (IN DAYS)

Fl
F10

PROCEED
PREVIOUS SCREEN

339

Screen 6: Oil spill response parameters (Fig. 8):

Columns 1 and 5 show the day of the oil spill, columns 2 and
6 list the probabilities of mortality after contact with the
spill for each day, columns 3 and 7 list the probabilities of
locally avoiding a spill on each day, and columns 4 and 8
list the probabilities of avoiding a spill by shifting the
location of the home range on each day of the spill.
Parameters for days greater than the duration of spill set in
screen 1 are ignored by the program. See discussion of the
structure of OTMOVE in the documentation volume.

340

Figure 8. OTPOP Screen #6 -- Oil spill response
parameters. Lines indicate location of parameter values
edited by user.

SET DAILY PROBABILITIES OF MORTALITY, AVOIDANCE, AND
EMIGRATION DURING EXPOSURE TO OIL SPILL:

DAY P(MORT) P(AVOID) P(EMIG) ) DAY P(MORT) P(AVOID) P(EMIG)
al 16
2 17
3 18
4 19
5 20
6 ial
7 22
8 23
9 24
10 25
int 26
17 27
13 28
14 29
15 30
Fl = PROCEED
F10 = PREVIOUS SCREEN

341

Screen 7: RUNID (Fig. 9):

Enter character string of up to 6 characters that will be
used to identify the output from the current model run. This
identification string will appear at the top of the .LOG file
and on output from PROC. Different RUNID strings should be
used for every production run of the model to ensure that
output\ files from a particular run may be permanently
associated with the appropriate .LOG file.

342

Figure 9. OTPOP Screen #7 -- Set run identification string.
Lines indicate location of parameter values edited by user.

ENTER SIX CHARACTER RUN IDENTIFICATION STRING:

Fl
F10

PROCEED
PREVIOUS SCREEN

343

Screen 8: Set seed for random number generator Fig. 10):

Toggle back and forth between "Use constants" and "Use clock"
with the cursor arrow keys. The same integers will be used
as random number seeds on every run that is initiated by
using constants -- this is supplied as a testing or debugging
aid as it ensures that the same sequence of random numbers
will be used in each run. The clock should be used to set
the random number seeds for production runs.

344

Figure 10. OTPOP Screen #8 -- Set random number generator
seed.

SET SEED FOR PSUEDO-RANDOM NUMBER GENERATOR:

= SELECT |

F1l = PROCEED

345

After all screens have been examined and parameters set
the model will begin execution. Parameter values used in the
current run will be recorded in the OTPOP.LOG file once the
model begins execution. To terminate the program prematurely
at any time press <control> and <break> simultaneously (since
the program will break only during input or output operations
it may be several seconds to a few minutes before the
program terminates after pressing <control><break>).

346

IV. REQUIRED FILES.

Several files are required to reside on the same disk
and subdirectory for the program to function correctly. They
are:

MAINPOP. EXE (OTPOP program execution module)

PROC. EXE (Data processing execution module)

SGX. EXE (Loads memory-resident portion of
screen generator)

MODE. COM (Restores correct mode after
execution)

OP.BAT (Batch file to execute above
programs)

PARASC (Screen generator library)

PROFORT. ERR (Error messages for IBM Professional
FORTRAN)

ZSCORES . DAT

DFLT.DAT

CDIST.DAT

SR.DAT

SBST.DAT

Files with the .DAT extension are data files that are read in
during program execution. ZSCORES.DAT contains values of the
standard normal distribution and should not be disturbed.
The remainder of the .DAT files may be edited by the user
using a standard word processor. If these files are edited,
the edited files must be saved as ASCII text files under
their original names.

DFLT.DAT contains the default values of the parameters
set at runtime. Parameter values are in list format, with 1
space separating each value. The values are listed in the
order they appear on the screens (Fig. 11).

CDST.DAT contains the density functions for summer and
winter used in the distribution algorithms of the model.
They should be derived from the most recent USFWS/CDFG census
data. The file consists of 5 columns in list format, with a
space separating the columns (Fig. 12). The first column is
an integer representing each 1/2 kilometer segment of the 5-
fathom ordinate system. The second column is the proportion
of the population that was observed in that segment of the
range during the most recent spring/summer census. The third
column indicates the method by which that section of the
coast was counted ("0" = from ground, "1" = by air) for that
census. The fourth column is the proportion of the
population that was observed in that segment of the range
during the most recent fall/winter census. The fifth column
indicates the method by which that section of the coast was
counted ("0" = from ground, "1" = by air) for that census.

The digitizing program used by USFWS to enter the census
data will produce files in the correct format that can be

347

Figure 11. Part of the DFLT.DAT file.

aL 0 as es as B= 0K 0 as ea ie Bs 0 a Ee hoy=)
1500 201 955 0.0900 0.0050 0.0000 1.0000
0.9300 15.000 0.2500 0.5300 0.8700 9.0000 5.0 5.0

0.36700
-0.0250
-0.0090
-0.0450
00.1050
00.0420

-0.1630
-0.4060
-0.7060
-0.2900
-0.8150
-1.0440

8.09000
6.39000
2.95000
8.56000
4.64000
1.93000

37.5000
37.5000
08.4000
48.9000
48.9000
48.9000

60 20
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000

8.1100 0.4400 5 23
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000
0.0000 0.0000

2 LS

OBDIHMNFPWNEF OV

348

Figure 12. Part of the CDST.DAT file.

ro
wo

°
ro

AS
or
Pe

PR

1 0.00039 1

(o)
PRR.
°

N
(o}
0
ooo
oooooo0oo0oo0oo0o0o0o0o0°o°coeo°c0o0o090d

(2)
(o}

Ha tan a WeaR ag Cag ay btn ea eae et eae Nea
| ed
°
RP

N
Pp
a
(2)

N

rR

NI
foMo Mo Mo MoM Moone K-)
PRPPPPPPPPPRP-.

{o)

N
N
Nu
fo)
Pr
rh
oO
Pr

IN)
rN)
as
ooo
PRP.

349

copied onto the appropriate disk after each census to keep
the model input data updated.

SR.DAT contains sex ratio data for the sea otter range,
used in the distribution algorithms of the model. Sex ratios

are recorded by CDFG carcass recovery area. The file
consists of 3 columns, in list format, with spaces separating
the columns (Fig. 13). The first column contains an integer

representing the northern boundary of the segment in the 5-
fathom line ordinate system, the second column contains the
number of males per female in the segment during the
spring/summer, the third column contains the number of males
per female in the segment during fall/winter.

SBST.DAT contains codes for the substrate along the
coast. Substrate is either rock or sand. The file consists
of 2 columns, in list format, single spaces separating the
columns (Fig. 14). The first column is an _ integer
representing a point along the 5-fathom line ordinate system
where the substrate changes from one to the other, the second
column represents the substrate of the area south of the
point to the next point where the substrate changes. A "1"
represents rock substrate, a "0" represents sand substrate.
This file was coded from U.S. Geological Survey topographic
maps, and should not have to be changed.

350

Figure 13. Part of the SR.DAT file.

201 4.5 16.7
216 4.5 16.7
229 4.5 16.7
256 4.5 16.7
271 4.5 16.7
321 al 3 ;
366 0.5 0.88
377 0.5 0.88
334 0.5 0.88
390 0.33 0.33
400 0.33 0233
411 0.43 0.43
430 0.43 0.43
449 0.5 0.44
473 0.5 0.44
499 0.5 0.44
524 0.5 0.44
599 0.5 0.44
640 0.41 0.37
668 0.41 0.37
694 0.41 0.37
730 0.33 0.30
753 0.33 0.30
787 0.82 0.61
806 0.82 0.61
825 0.82 0.61
844 lO 51510)
853 1.0 3.0
891 4.5 16.7
916 4.5 16'. 7;
942 4.5 16.7
989 4.5 16.7
999 5. Lee

351

Figure 14. Part of the SBST.DAT file.

199
378
421
426
466
468
489
497
680
682
722
723
730
732
740
742
746
750
815
843
894
896
914
963
966
970
977
978
979
981
989
1007

PoOorPOrPOrOrPOrPOrRPORrRPORPORPOrRPORORPOROROFO

352

V. RAW OUTPUT FILES.

Seven files are generated as output from OTPOP.
OTPOP.LOG is a log file, giving the date and time of
execution, and the values of the parameters set at runtime
(Fig. 15). RUNID.DAT will contain the run identification
string. The other 5 files are output data files, containing
the raw results of the simulations. Program PROC is used to
process and summarize the raw output data.

NTS.DAT contains the simulated population sizes for runs
with an oil spill. There are 5 columns; the first gives the
run number, the second gives the year (the oil spill always
occurs in year 0, so that the years simulated before the
spill are designated -3,-2,-1). Year O will be recorded
twice for each run, once for just before the simulated spill,
once for just after the spill. The third, fourth, and fifth
columns give the number of independent females, number of
independent males, and number of pups, respectively, for that
year during the month of the spill (Fig. 16).

NTC.DAT is exactly the same as NTS.DAT, but contains
data for the control (without oil spill) runs, so year 0 wil
be recorded only once for each run.

DTS.DAT contains the total numbers of animals killed by
the simulated spills. There are 13 columns; the first gives
the run number, the next 6 pairs of columns give the number
of animals and the proportion of the population killed for
each of the 6 classes of animals (see discussion of the
structure of OTMOVE in the documentation volume) respectively
(Pilg 7)

DDS.DAT contains the numbers of animals killed by the
simulated spills on a daily basis. It has 8 columns; the
first gives the run number, the second gives the day of the
spill, and the third through eighth give the numbers of
animals of classes 1-6 (see discussion of the structure of
OTMOVE in the documentation volume) killed on that particular
day (Fig. 18).

RVS.DAT contains the population's reproductive value
(see discussion of model output in documentation volume)
before and after the simulated spill. It has 3 columns, the
first giving the run number, the second giving the population
reproductive value just before the spill, and the third
giving the population reproductive value just after the spill
(Fig. 19).

RCS.DAT contains the recovery times for population. It
has 3 columns. The first gives the run number, the second
gives the population size just prior to the spill, and the
third gives the number of years simulated before that

353

Figure 15. The OTPOP.LOG file.

FOLLOWING ARE RUNTIME INPUTS FOR RUNIL:
EXAMP1

DATE -- 9/ 4/87 TIME -- 22:54

DATE AND TIME USED TO GENERATE RANDOM NUMBER SEEDS
10,"NUMBER OF YEARS PER RUN"
10,"NUMBER OF RUNS WITH OIL SPILL"
10,"NUMBER OF CONTROL RUNS"
1000,"INITIAL POPULATION SIZE"
1,"MONTH OF SPILL"
1,"DAY OF SPILL"
1,"DURATION OF SPILL
345,"NORTH BOUNDARY OF SPILL"
400,"SOUTH BOUNDARY OF SPILL"
1000,"EQUILIBRIUM POPULATION SIZE"
201,"NORTH BOUNDARY OF RANGE"
955,"SOUTH BOUNDARY OF RANGE"
0.090,"MAXIMUM PER CAPITA ANNUAL GROWTH RATE"
0.005, "NON-LINEARITY OF DENSITY DEPENDENCE"
0.000,"DENSITY INDEPENDENT MORTALITY RATE"
1.000,"DEGREE OF COMPENSATION"
0.930,"ADULT FEMALE SURVIVAL RATE"
15.000,"MODAL AGE OF FEMALE SENESCENCE"
0.250,"PRIME REPRODUCTIVE RATE"
0.530,"PUP SURVIVAL RATE"
0.870,"ADULT MALE SURVIVAL RATE"
9.000,"MODAL MALE AGE OF SENESCENCE"
5.000,"PERCENT VARIATION IN ADULT SURVIVAL"
5.000,"PERCENT VARIATION IN PUP SURVIVAL"
woe WARY "CE" "STGMA™, "VMAX"
"JUVENILE FEMALES" 0.367-0.163 8.09037.500
WADULT FEMALES W/PUP"=-0.025-0.406 6.39037.500
"ADULT FEMALES W/O PUP"=-0.009-0.706 2.950 8.400
"JUVENILE MALES"-0.045-0.290 8.56048.900
“ADULT NON-TERRITORIAL MALES" 0.105-0.815 4.64048.900
"ADULT TERRITORIAL MALES" 0.042=-1.044 1.93048.900
6,"AGE AT WHICH POTENTIALLY TERRITORIAL"
60.000,"MAXIMUM % OF POTENTIALS THAT HOLD TERRITORIES"
20.000,"MINIMUM % OF POTENTIALS THAT HOLD TERRITORIES"
8.110,"MEAN TERRITORY LENGTH"
0.440,"S.D. OF TERRITORY LENGTH"
5,"MEAN ARRIVAL DATE MONTH"
23,"MEAN ARRIVAL DATE DAY"
11,"S.D. OF ARRIVAL DATE IN DAYS"
12,"MEAN DEPARTURE DATE MONTH"
1,"MEAN DEPARTURE DATE DAY"
15,"S.D. OF DEPARTURE DATE IN DAYS"
"DAY", "P (MORTALITY) ","P(AVOIDANCE) ","P(EMIGRATION) "
1 1.000 0.000 0.000

354

Figure 16. Part of the NTS.DAT file.

1 1 593 361 130
1 2 591 367 126
1 3 590 368 127
1 4 617 358 124
1 5 641 376 138
1 6 622 368 120
1 7 599 345 122
al 8 597 347 118
1 9 578 335 121
1 10 579 327 125
2 -4 647 352 101
2 =3 630 343 128
2 -2 630 331 137
2 -1 632 348 131
2 0 644 337 117
2 0) 597 324 108
2 1 607 310 137
2 2 623 344 130
2 3 624 340 114
2 4 621 322 137
2 5 650 346 121
2 6 673 331 142
2 7 676 350 143
2 8 666 366 132
2 9 656 351 138
2 10 664 355 140
3 -4 647 352 105
3 -3 659 355 130
3 =2 666 366 129
3 cal 658 371 121
3 0) 662 350 118
3 fo) 621 330 111
3 1 635 332 119
3 2 645 340 142
3 3 623 337 132
3 4 626 345 123
3 5 624 322 132
3 6 630 337 123
3 7 645 321 121
3 8 637 327 130
3 9 654 325 114
3 10 654 319 138
4 -4 647 352 122
4 -3 653 336 148
4 -2 669 318 137
4 al, 668 325 133

355

Figure 17.

rR

OUWUDNAHAUPWNER

(oe

h
WWONNWE OO W

0.020
0.040
0.043
0.073
0.019
0.044
0.012
0.061
0.017
0.057

Part of the DTS.DAT file.

0.045
0.037
0.066
0.064
0.031
0.044
0.020
0.050
0.057
0.052

0.093
0.071
0.079
0.070
0.080
0.083
0.105
0.101
0.039
0.068

356

0.083
0.074
0.055
0.065
0.083
0.090
0.070
0.098

0.071

0.067

0.122
0.077
0.059
0.089
0.065
0.102
0.060
0.102
0.083
0.074

0.073
0.063
0.060
0.070
0.062
0.076
0.057
0.084
0.057
0.064

Figure 18. Part of the DDS.DAT file.

1 1 3 7 3 ab7/ 23 16
2 1 6 5 2 12 26 9
3 1 6 13 1 15 19 7
4 1 11 10 1 13 21 11
5 1 3 4 3 15 33 8
6 1 7 8 1 15 31 12
7 1 2 1 3 18 26 7
8 1 10 9 2 16 33 11
9 1 3 10 2 6 23 11
10 1 9 9 2 13 22 10

357

Figure 19. Part of the RVS.DAT file.

1 820.5 749.6
2 819.9 761.1
3 829.4 781.5
4 861.8 807.8
5 856.5 788.0
6 829.5 753.2
7 800.4 739.0
8 762.9 691.9
9 807.1 755.7
10 859.6 802.3

358

population size was reached again (Fig. 20). For simulations
in which the population never attained pre-spill size the
time to recovery is recorded as a negative number.

359

Figure 20. Part of the RCS.DAT file.

1 1158 -10
2 1098 5
3 1130 -10
4 1121 -10
5 1191 -10
6 1139 8
7 1131 -10
8 1099 9
9 1152 -10
10 1176 -10

360

VI. PROCESSING RAW OUTPUT DATA

Raw output data are summarized using the FORTRAN program
PROC and LOTUS123. PROC is run automatically by the batch
file OP.BAT after MAINPOP finishes execution. It produces a
summary of the run (Fig. 21) in the file "OTPOP.RPT". At the
top of OTPOP.RPT the RUNID identification string is given.
Following is a summary of the control runs listing the year
and the number of independent females, independent males, and
dependent pups in the population. In parentheses after each
of these are given the range and standard deviations of the
population sizes in each year.

Following the summary of the control runs is a summary
of the oil spill runs, in the same format. Year "-0" is just
prior to the spill, year "+0" is just after the spill.

Following that is a summary of the total number of
deaths from the oil spili, and a summary of the recovery
after the spill.

Besides OTPOP.RPT, PROC generates 3 files,
"NO SPILL.DAT", "DEATHS.DAT", and "SPILL.DAT", that can be
used by LOTUS123 to produce rough graphs of the model
output. A LOTUS macro, "0OT123.WK1" is supplied. teas
expected that you have a general understanding of the
LOTUS123 package (including PrintGraph) in order to use the
macro.

Files "SPILL.PIC", "NO _SPILL.PIC", "COMPARE.PIC", and
"DEATHS.PIC" are supplied with "0T123.WK1" and must be
present in order for the macro to run. To run the macro, run
Lotus 123 and then retrieve [/FR] the worksheet 0T123.WK1 in
the usual manner (see your LOTUS123 manual for details). At
this point you may invoke the macro by keying [alt]G. As the
macro runs, 4 graphs will be displayed on the screen. The
macro will pause while displaying each graph and you will be
required to depress the space bar in order to proceed with
the execution of the macro.

After the macro has been executed successfully the 4
above mentioned picture files (*.PIC) will be created. In
addition to these newly created files, the graphs will be
named in the current worksheet. If you wish to keep this
information, save the current worksheet OT123.WK1 under a new
name. From this point, you may wish to modify the graphs
using the LOTUS123 /Graph menus and save modified graphs in
new .PIC files. For information on how to create or modify
graphs consult your LOTUS123 manual.

The first graph (NO SPILL.PIC) produced by the macro

traces the mean total independent population size through
each year of the control simulations (Fig. 22). The lines on

361

OUTPUT FOR RUNID EARP!

CONTROL RUNS:

YEAR FEMALES

HALES

PUPS

ee on ow ow on oo nn ono nn ws een nw coe wn come wn ce ewn ones ececese sc coen nese coeooe

647.0( 647- 647,
836.61 622- 447,
640.44 624- 451,
&44.81 622- 667,
0 640.81 S97- 468,
639.51 S99- 659,
632.81 b01- d61,
624.91 592- bbl,
625.0( 602- 472,
830.51 604- 447,
637.3( 602- 676,
837.41 602-63,
640.91 608- 688,
645.41 625- 478,
648.81 631- 669,

ee a

OIL SPILL RUNS:

0.0)

6.3)

8.3)
13.4)
21.8)
19.0)
20.9)
19.5)
16.1)
16.4)
21.2)
18.8)
22.8)
15.3)
13.9)

352.0( 352- 352,
349.5( 333- 341,
352.1 331- 372,
353.7( 345- 377,
354.8( 317- 376,
345.2 328- 385,
363.41 327- 390, 20.2)
358.7( 330- 383, 18.5)
356.51 320- 393, 2101)
348.94 32t- 387, 26.01
341.3 315- 376, 18.0)
345.20 31S- 371, 18.6)
345.8 313- 376, 18.2)
348.6¢ 304- 401, 25.5)
352.71 31S 385, 19.0)

0.0)
8.2)
14.0)
9.5)
18.4)
16.8)

$6.4)
7.3)
9.7)
7.3)
11.8)
7.2)
10.2)
8.7)
7.2)
3.6)
6.1)
6.4)
12.3)
10.4)
5.1)

$20.1
135.01
131.01
126.51
127.3
126.51
126.44
126.71
130.9
126.1
127.16
127.7
130.3
134.86
135.41

10b- 129,
123- 152,
115-149,
117-139,
102- 142,
112-135,
112-148,
113- 141,
119- 141,
110-138,
\N7= 138,
116-139,
IN1- 152,
121- 153,
123- 142,

847.04 G47- 847,
842.8 G21- 459,
846.71 B10- 69,
850.9 G1S- 668,
852.21 608- 6&0,
602.31 548- 629,
605.7 SSI- 835,
6°.5( ST7- 85l,
o1b.8t Sal- 654,
624.34 607-443,
830.21 595- 450,
&35.41 16-873,
639.31 579- 476,
842.5 S97- 873,
b49.1( 578- 678,
$50.9 579- 681,

owm vo Ue WA

18.2)
17.8)
21.4)
24.9)
28.0)
26.2)

352.0( 352 352, 0.0)
348,1( 335- 364, 8.6)
354.8( 316- 380, 18.8)
340.8( 325- 382, 17.7)
364.9 323- 387, %.5)
349.30 Pre- 377, 22-91
347.6( 310- 379, 22.4)
348.5 324- 369, 14.9)
345.41 329- 372, 15.5)
338.6 322- 362, 13.91
343.2 305- 387, 23.3)
338.0( 299- 388, 19.21
334.8 314- 352, 13.8)
330.7( 298 366, 21.2)
330.8( 305- 351, 14.3).
333.5( 297- 355, 18.9)

WUMBER OF DEATHS FROM DIL SPILL:

10t- 128,
125- 134,
M7- 137.
{7- *s
108-
q-
104-

1272.4
112.26
118.46
121.0
120.71
126.81
126.4
123.7
127.5
127.4
129.26
134.71

113-137,
12i- 138, 5.2)
120- 142, 6.3)
S18- 143,
118- 141,
118-141,
123- 147,

PERCENT OF POPULATION

CLASS WEAN S.D. RANGE REAR S.D.. RANGE
JUVENILE HALES 6.0 63.1 2- it 3.86 1.99 1.20 7.30
ADULT MALES 9.6 2.8 4- 14 4.65 1.37 2.00 6.60
JUVENILE FEMALES 14.0 3.2 6- 18 7.89 1.80 3.70 10.50
ADULT FEMALES 35.9 5.4 26 - 44 7.56 1.22 5.50 9.80
PUPS 10.2 2.6 T- 16 8.33 1.96 5.90 12.20
TOTAL ANIMALS 13.7 9.0 b4- 92 6.66 0.84 5.70 8.40
RECOVERY AFTER OIL SPILL:

MEAN $.D. RANGE
REPRODUCTIVE VALUE BEFORE SPILL 824.8 29.0 762.9- 9861.8
REPRODUCTIVE YALUE AFTER SPILL 763.0 32.4 671.9- 907.8
REPRODUCTIVE VALUE REDUCTION (2) 7.506 1.19617 S.775- 9.307
YEARS TO RECOVERY 7.333 1.89967 3.000- 9.000

$8 ON 67 OF 10 RUNS ( 70.02) THE POPULATION

DID MOT RECOVER TO PRE-SPILL SIZE

TIME TO RECOVERY CALCULATED ONLY FOR RUNS THAT DID RECOVER.
Figure 21. The OTPOP.RPT file generated by PROC.

362

Figure 22. Population sizes through each year of control
simulations.

NO SPILL @ YEAR ZERO

EXAMP14
1.03
1.02
4.01
o 1
nn
\
+@
-3 0.99
us
oO
23 c.98
FE
= yw
rf 0.97
z
0.96
0.95
0.94 7
ag = oO 2 4 6 8 10
YEAR

363

either side of the mean trace the range of population sizes
that occurred during the simulations. The second graph
(SPILL.PIC) traces the oil spill simulations in the same
format as NO SPILL.PIC (Fig. 23). The third graph
(COMPARE.PIC) plots the means from both the control and the
oil spill runs on the same graph, allowing visual comparison
of the population trajectories (Fig. 24). The last graph
(DEATH.PIC) traces the mean cumulative number of deaths due
to oil spill on each day of the spill (Fig. 25).

If you wish to modify the macro or just want to see the
macro depress the [End] then the [Home] key and you will move
to the far end of the worksheet, where the macro is located.
If you wish to modify the macro we suggest that you first
make a copy of it in case you need to refer to the original
while editing.

364

Figure 23.

MEAN g INGEP. +—S.D.
(CThousande)

Population sizes during oil spill simulations.

SPIRE TG YEAR ZER©

SAMP 1

365

MEAN # W/SPILL & W/O SPILL
(Thousands)

Figure 24. Plot of control population and oil
population for simulations.

COMPARE

EXAMP1
1,02 7

1.01

099
0.98 >
0.97
0.96

0.98

366

10

CUMM. LOSS INOEP. & PUPS

Figure 25. Cumulative deaths due to oil spill.

DEATHS

EXAMP 4

DAY

VII. ERRORS.

Errors in model runs can occur at 3 levels. Screen
generator errors occur when values out of range or of the
wrong type are input. The program will warn you of the error
and it can be corrected by simply typing in an acceptable
value. A screen generator error will also occur if the SGX
interface has not been loaded prior to running the model, or
if the screen library PARASC is not available.

IBM Professional FORTRAN runtime errors can occur when
the program is unable to execute a program statement. For
instance, if one of the input .DAT files is missing,
incomplete, or in the wrong format. The error messages are
generated from the PROFORT.ERR file, and will be noted on the
screen. These messages may often be cryptic. If a PROFORT
message indicates a unit or input error, check that all input
files are present and in the correct format. Other
situations that may lead to PROFORT runtime errors occur when
impossible mathematical operations are‘ attempted, such as
division by 0 or taking the log of a negative number.
Usually ‘this will occur when an unrealistic combination of
parameter values have been input. The error may or may not
be fatal to program execution, but even if the program
continues to run the output will be suspect.

Program error trapping is the third level of possible
error. Because there are so many possible combinations of
parameter values that can be set, it has been impossible to
build in complete logical error-trapping, but the program
does check for many impossible combinations, such as setting
survival rates too low to achieve specified growth rates, or
setting pup survival rate too low to achieve the specified
reproductive rate. Should an error occur on this level the
program will terminate and you should correct the parameter
values in the next run.

The program is extremely complex, and it is very likely
that errors not described herein may occur. No program of
this size can ever be guaranteed to be bug-free. Should you
not be able to correct an error, be sure to save the log file
and all input files, and any output files generated during
the run. Notify Allan Brody or Don Siniff and send copies of
the saved files.

368

SMITHSONIAN INSTITUTION LIBRARIES

wi

10 2245) put il

e7/93 canst

DAN
S>

4

Aas

‘NOSE

NVINOSHLINS S31Y¥V

a

e [sj

ie sy
fon 0c Y

ON,

1st

NSTITUTION

\e
“
& NY

ny,

INSTITL

NOILNLILSNI_ NVINOSHLIWS

SMITHSONIAN
NVINOSHLINS S3IYV

3J1YVYGIT LIBRARIES SMITHSONIAN

iS)

NOILNLILSNI

NSTITUTION NOILNLILSNI NVINOSHLINS

SMITHSONIAN

INSTITUTION NOILNLILSNI

INSTITUTION

LIBRARIES

NVINOSHLINS S3IYVYUSITLIBRARIE

SMITHSONIAN

STITUTION NOILAL NVINOSHLIWS

= F}
e ¢
x

YY,
K S

eC;
on ne. Y

S3Iuvuaii_LiB

INSTITUTION

Saluvudi7 LIBRARIES

INSTITUTION

saruvuagi

INSTITU

SAIYVYSIT LIBRARIES SMITHSONIAN

NVINOSHLIWS

LIBRARIES SMITHSONIAN

X
nase

RARIES

NOILNLILSNI

S SA1uv

NVINOSHLIW

INSTITUTION NOILNLILSNI

SMITHSONIAN

SMITHSONIAN

< 207,

()
A,
SINOSA

NVINOSHLINS SJ1av:

nis

INSTITUTION
WS
AN

s31uvugit

INSTITU

NOILNLILSNI

Gy

LIBRARIES

N

SMITHSONIA
SMITHSONIAN

NVINOSHLINS s

Ss

vine in

LIBRARIES
NOILNLILSNI

N

SMITHSONIAN _INSTITUTION NOILNLILSNI N

o

S

SAIYVUSITLIBRARIE

Ssalyvagit

INSTITUTION

NVINOSHLINS Sit

uvugi7_LIBRARIES

NVINOSHLIWS
SMITHSONIAN
NVINOSHLINS

INSTITUTION NOILNL N

=a
<
za
ro)
wo
ae
E
a
wo 2 Mi Zz 4 i 2 @ aq
is) re) Ww , Lu ‘
z a E 4 = Oy, % = 4
ss e yx = < Often < aa
= = a S oc 5 = =i
3 2 z 2 = 2 = 2
— =) od }
JIYVUGIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI NVINOSHLIWS Salevia LIBRARIES §
& Zz - z c z =
= S = S Str CE = S
g = 2 = SON 2 =
> eis > = > MG 2 > 3
= n = n = eo ie m ca
m z mn z i z a Z
on = wn = n Ses wn =.
NSTITUTIO NOILNLILSNI NVINOSHLINS S3IYVYalT_LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI_N
<8 Z g 2 ee 2 gE z
XK < _< VY .- = =< NURS => J
Wo = is = = KS SG 5 =)
Wee Sr fe) 35 DG KK = SD Wes = oe
WE 2 : g 3 ER 3 ZUR 2 a
\ \. = = z E XY Ze = SS -
> = 5 = > = =
‘. 7a (2) i Zz w : a ?))
31uvagia SMITHSONIAN _ INSTITUTION, NOILNLILSNI_NVINOSHLIWS’ S31¥VHEI1_ LIBRARIES. S
= wy 2 ws 2 Ww a wl
= =. ih fp 3 z = ez = z
si i Yip = < = < = <}
5 a “fora = A 4 a =| a
= mS = st =
re) = p 3° = ro) = fe =
2 ; 2 =| z = z =
CST UTONR NOLLNMLILSNI~NVINOSHLIWS _ S3 fuvugd iy tl BRARI ESe PH INSTITUTIONS NOILALILSNI_N
~ &
ie) S ° = ° = ° o
= x S 2 E 2 = 2
= > = > = > = =
= ie = 2 a 12 E a
n 2) n* m m
z Chee ue o z o = pee
ZLIBRARIES, SMITHSONIAN INSTITUTION NOILMLILSNI_ NVINOSHLINS, S31uvua17 LIBRARIES,
=." = < = wb, = = ff, 5 =
z a = g = z
oO # Go (eo) a=
2 NOs 5 MWe ig z 8 3
= NS 9 = ENS Q = 2 E zZ
Sh aN = SS NES = > = SS >
(7) “aS > ose > w = w P=!
NSTITUTION p NOMLALILSNI_NVINOSHLINS ” S31 uUVYsIT LIBRARI ES SMITHSONIAN_INSTITUTION on
a 2 z - z Ww <Gul)
JAN us STITUS > uw <p> Sa us y, a ONS uw VON n

2 Sa = Ray 2 SQ = Gap 2 apy = 2 Rea) = yyy) 2
m Quoc Z Cents m CR assy m Sonne 2 m None ” CoE 5
pT IEVUAITULIBRARIES SMITHSONIAN — yy NOLLNLILSNI_ NVINOSHLINS, Sa1uvu a (ou BRARIES,
= < Va = < = <= = = NS z
4 z \\ = 2 S z Sj Wis Xs =
<= DD Qe Sz ro) aw {o} Bea 0 RYN 5
g ZN 3 2 g : B 2 Ne :
2 i eli = 2 E = 2 \. :
: Br ae EN Ral tena = fe en Se
_INSTITUTION NOILNLILSNI NVINOSHLINS S3IYVYSIT_LIBRARIES SMITHSONIAN _ NOILALILSNI
z “ S <8 2) = 7) S “” =
B = a es a Uh ery tee w é
AM %, =
=] iB = Ss = Oe, wy = e :
c c = Ee <= YY c < =
= o = z = : “Vg =
ro) = re) a fe) a TG fe) = C
z a z Set a a =a =) >
_SalaVu¥di7 LIBRARIES SMITHSONIAN_INSTITUTION NOILNLILSNI_NVINOSHLINS S3IYVYSIT LIBRARIES
a Sie = 2 Eas : e
= o = o = o \ = ive) -
5 Ee 5 a = rs NN oS D t
= > = > = > \WSE > :
= 2 =. a iE 2 Xs E 2 :
| INSTITUTION NOILMLILSNI NVINOSHLINS S31NVYSIT_LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI
Zs 2) z a) 5 Z 2 et n g
< = < = x = < = :
za ar z 4 z a =z = ate
8 en AB o a WX. F o ~NN & cf
aT : Oo = ° 3e SS fo} ac SENSO) :
Ee z S Z, = AS = E Wr 2 :
= Se 3 z Flue ee = ane :
> SSIMYVUSIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI_NVINOSHLINS S3IYVYGIT LIBRARIES
a w =, g (ap) a n , art wn
a z= i Gps. =a fa el fea =I ;
< = < Gils A < 2a YS < 3] :
o S 02 Gt S co = WL oc 5 P
a 5 a. 3 es Soe S 3
a z a z =) z a) a
| INSTITUTION NOILNLILSNI NVINOSHLINS S31YVYGIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI
a z Ur $6 Z & S = = :
wo 2 wo Ne ° o 2 se) 2 C
2 = a WS 2 5 a 5 :
> = > MG 2 > rE ; > E :
r 5 fe = = 7 =
ae = a = ae re a ie
E :
e Z B be o = @ z |
> SSJYVYGIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSN! SAlYVYEIT_ LIBRARIES
on z ; 7) z= : 7p) z a) z an <
= eS = < Ww = = = i 4 is
4 z 4 z GX =) z Sl ag “ :
aoe o = fo) Ve SM Is 9 5 YLY Ors Se
ny n 7) rp) Na S\ wo n » GY 2) (
2 : : ESQ : ag! |
5 : = > = > = = a Sage
Za 7) ae a . S o Z 7)
1 INSTITUTION NOILQLILSNI SAIMVHYGIT LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI
s a) = “n 5 on z a
Oo us 7) a on iS OB =
| = = Set See Asa =H
= ee = ce = ed) a s
e = E s E a Uity § e
3 » z © 2 a “Gf 3 s
z x z i zE a ¢ 72 -
3_SSIYVYGIT LIBRARIES SMITHSONIAN_INSTITUTION NOILOLILSNI_NVINOSHLINS_Saluvualt
3 = ee 3 Se :
— [ov] — ow — oO ay ee)
= 0 i= a) 5 2 5 a
E D> re i > ps > = >
= Bd =. ae = x ie 2
Zz ny B° nm Z o = o |
\_ INSTITUTION NOILALILSNI_ NVINOSHLINS SaluvugI7_LIBRARIES SMITHSONIAN INSTITUTION NOILNLILSNI
a “e w =z w p = 2) Zz x 2) )
< = < = < = < = |
BO = 2 = Zz = 53. = |
8 NN: - 5 g 3 s WA:
= MR AS GO x= fe) <= = =ZE Q'S |
E WW 2 S = = = EF WM’ z }
Sa) = = > = = ihe tes :
S$ ,94!8VNGIT_LIBRARIES SMITHSONIAN INSTITUTION SEE ees eee Tiew Tt ae
" LGR 2 GER" Gin,7 aR» * e = O gore |
ake a) 5 (AM) «= ly, | (Ma) = S E S |
oe we] Clee se) < (fpr e Be) < eS ao \eK 45) S |

1 * if oe
a t x ‘Penee ib steyaste

farbea emia Dy DAG Si ben hy
DIMLaN Ha Woh
SUCH ak hast ehube ata ¢
peepee 8,
deh ‘an Udit

Loree tees ' : ayiaale
RLS AUA ENT Ai i PCa Cee ee ane
ter cory Bee te wr

hae iec ks 8 teh
See bins readin Sen! ie
PR ROCHE TES Thhe a Myre ie alata

ebwlih
Dyk aes DD pra

Sieh tee Ue alstgain

nurde onan
x isk 3 Saat

ns nets TEP
achueteiardnate ceestaree sey? faye ssbuit a. (abl
i Digan nn ashe}

teiree
vacate fader

ie yee E

Ta yrubed pub es bre,
ednderd ent

BETTS ce yte fe

Ey

fat one eet si
Peake:

Predhangeasay

prec

iG
eee Tuy Ty ath M
Saernriete’

nese art
Wgearesyaus
eae kl gy
yearn?
arery
SPIN
Rhruaes Fippeaeieeu ne
SiNpA ae atyouee sts ee yan:
be vatept ela tary ans Mb e ere ¥

Bebra

Seaside
concn wlye dao r
payee is

" pee ta wasias
Hee eT (Choo EEE Sn
Viviyre Ne

syrat

eh

Ee
ith