(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Standard Techniques For Pelagic Fish Egg And Larva Surveys Fao Fisheries Technical Paper No 175"

PAD Fisheries Technics! Paper No, 175 



STANDARD TECHNIQUES FOR PELAGIC 
FISH EGG AND LARVA .SURVEYS 

by 



Paul E. Smith 

Southwest Fisheries Center 

NMFS, NOAA, US iJept. of Commerce 

La Jolla, California 92038, USA 

and 

Sally L. Richardson 
Department of Oceanography 

Oregon State University 
Corvallis, Oregon 97331, USA 



Rone, December 1977 



~ it - 



The designations employed end the presentation 
of material in this publication do not imply the 
expression of any opinion whatsoever on the 
part of the Food and Agriculture Organization 
of the United Nations concerning the legal 
status of any country, territory, city or area or 
of its authorities, or concerning the delimitation 
of its frontiers or boundaries. 



ism 92-5-100515-x 

The copyright in this book is vested in the Food and Agriculture Orga- 
nization of the United Nations. The book may not be reproduced, in whole 
or in pert, by any method or process, without written permieeion from 
the copyright holder. Applications for such permission, with a statement 
of the purpose end extent of the reproduction desired, should be addressed 
to the Director, Publications Division, Food end Agriculture Organization 
of the United Nations, Via cMto Terme di C*rscetta, 00100 Rome, Italy. 

CFAO 1977 



iii 



PREPARATION OF THIS PAPER 



This publication is one of a series of FAO manuals in fisheries science. 
It has been prepared as part of the programme of the Fishery Resources and 
Environment Division and is largely based on the work of a Working Party on 
Fish Egg and Larval Surveys established following the recommendation of the 
Fourth and Fifth Sessions of the FAO Advisory Committee on Marine Resources 
Research. 

Through a contract awarded to the Southwest Fisheries Center, National 
Marine Fisheries Service, La Jolla, California, the present version of the 
manual was prepared by Paul E. Smith and Sally L. Richardson. 

The manual will be primarily for use in FAO field projects and training 
courses, and by national institutions, especially in developing countries, 
which may be conducting fish egg and larval survey programmes. The final 
edition of the paper in the official languages of FAO will eventually be pub- 
lished incorporating any revisions that may be proposed by the users of the 
manual. . 



Distribution 

FAO Department of Fisheries 

FAO Regional Fishery Officers 

Selector SM 

UNESCO 

IOC 

SCOR 

Author 



Bibliographic reference 

Smith, P.E. & S. Richardson (1977) 
FAO Pish Tech. Pap,. (175):100p* 
Standard techniques for pelagic fish 
egg and larval surveys 

Ichthyoplankton surveys* Methodology. 
Sampling (biological)* Plankton 
collecting devices. Stock assessment. 



PREFACE 

A Working Party on Fish Egg and Larval Surveys was established in 1968 
following the recommendation of the fourth and fifth sessions of the FAO 
Advisory Committee on Marine Resources Research (ACMRR) , whose terms of 
reference included the preparation of a Manual (field guide) for this aspect 
of fishery science. The contributions prepared by the Working Party members 
mainly based on the discussions of the Working Party held in Tunis from 22 
to 27 October 1969 and on further advice provided by ACMRR during its Sixth 
Session, were recently edited by Prof. G. Hempel, and published by FAO: 
Fish Egg and Larval Surveys (Contributions to a Manual) (FAO Fish.Tech.Pap. , 
122). This provisional version has been produced for use as a background 
document during the first International Training Course on Fish Egg and 
Larval Studies, held in La Jolla from 2 to 29 September 1973 (see FAO. Fish. 
Rep. . 144). Based on the experience gained during the course and taking into 
account the recommendation of the ACMRR that the Manual should provide a 
review of the circumstances under which egg and larval surveys are more appro- 
priate and cost-effective than other field surveys, the provisional version was 
edited and supplemented in order to provide the potential new users pf this 
survey technique with a balanced view of its value and limitations as well as 
of cost effectiveness. 

Many people have been generous with their time and thought to make this 
manual possible. The consistent support and encouragement of Dr. Elda Fagetti 
has been the essential energy behind this work and the advice and urging of 
Dr. Gotthilf Hempel and Dr. Elbert Ahlstrom have guided the production of the 
manual. The authors of a precursor to this manual (Hempel, 1973), E.H. 
Ahlstrom, J.M. Colebrook, E. Fagetti, G. Hempel, K. Sherman, S. Tanaka have 
furnished major sections of text within this edition. Much of the procedural 
information in section 2 has been derived from procedural description (Kramer 
al ii, 1972) relative to the CalCOFI surveys. Significant editing was offered 
by Dr. William Richards, Dr. William Aron, Dr. William Lenarz, and Mr. David 
Kramer. Helpful comments were received from Dr. J.D. de Ciechomski, M.E. Diaz, 
Dr. J.A. Gulland, Dr. J. Kinzer, Dr. R> Lasker, Dr. A. Longhurst, R. Marak, 
B. Monsalve, Dr. W. Nellen, M.A. Padilla, A.M.Perez-Franco, Dr. B.J. Rothschild, 
Dr. M. Ruivo, Dr. D. Sahrhage, Dr. H. Santander, Mr. K. Sherman, Dr. A.C. 
Simpson, Dr. M. Vanucci, and K. Venkataramaniyam. Administrative assistance 
was received from Dr. E.F. Akytlz, Mr. Ben Remington, Mr. I. Barrett, Mr. Richard 
Schwartzlose and Ms. Barbara Rowe. Technical assistance was received from 
K. Akutagawa, F. Crowe, S, HuFei, E. Stevens and J. Thrailkill and typing was 
by Janica Scott, Dr. Sally Richardson incorporated the comments and reviews, 
unified the style of the text, and rewrote major sections and she has been 
named assistant editor for this manual. 



Paul E. Smith 



vii 
CONTENTS 



Page 



1. Introduction . . . . * . . 1 

1,1 Background , , . . 1 

1*2 Scope . . . . S 

2. Recommended Procedures for Conducting I ch thy op lank ton Surveys 

for Spawning Bioraass Estimates ...... 6 

2.1 Field Operations . . .... 6 

2.1*1 Cruise Plan and Personnel 6 

2.1.2 Sampling System . . . . 8 

2.1.2.1 The Sampler 

2.1.2.2 The Towing Procedure and Data Records 

2.1.3 Handling the Sample at Sea 16 

2.1.3.1 Preserving the Sample 

2.1.3.2 Labelling 

2.1.3.3 Storage on the Vessel 

2.1.4 Supplementary Data 18 

2.1.4.1 Hydrographic Data 

2.1.4.2 Smaller Zooplankton and Phytoplankton 

2.1.4.3 Neuston 

2.1.4.4 Micronekton 

2.2 Laboratory Procedures 21 

2.2.1 Plankton Volume Determinations 21 

2.2.2 Percentage of the Sample to be Sorted 23 

2.2.3 Sorting Fish Eggs and Larvae from Plankton Samples. .23 

2.2.4 Preliminary Identification, Enumeration and 
Measurement of Fish Eggs and Larvae 26 

2.2.5 Final Identification of Fish Eggs and Larvae . . . . 27 

2.2.6 Bottling, Storing and Curating Identified 
Collections of Fish Eggs and Larvae 30 

2.3 Data Summarization 30 

2.3.1 Flowmeter Calibration 34 

2.3.2 Estimation of Missing Flowmeter Values 34 

2.3.3 Equation of Standardization ..... 36 

2.3.4 Data Forms Used for Standardization 36 

2.4 Census Estimate (Larval Index) 40 

2.5 Estimating the Spawning Biomass 42 

3. Theoretical Considerations 43 

3.1 Statistics 43 

3.1.1 Numbering Systems 43 

3.1.1.1 Normal Distribution 

3.1.1.2 Lognormal Distribution 

3.1.1.3 Negative Binomial Distribution 

3.1.2 Data Comparisons 53 

3.1.3 Advice on Data Transformations and Comparisons . . . 62 
3*2 Volumetric Sampling < . . . 63 

3.2.1 Volume and Distribution of Water Filtered 63 

3.2.1.1 The Oblique Tow 

3.2.1.2 Clogging 

3.2.2 Avoidance 66 

5.2.5 Mesh Retention 69 

3*2*4 Uneven Tow Trajectories 73 

3.3 Sampling to Determine Spawning Area and Time 74 

3.3.1 Geographic Distribution ........ 74 

$.5.2 Vertical Distribution 75 

3.5.3 Seasonal Distribution 77 

3.3.4 Oceanic Transport and Diffusion 80 



Vlll 



Page 

3*4 Sampling to Overcome Problems of Microdistribution .... 81 

3.5 Residence Time (Estimating Mortality) 83 

3.5.1 Temperature Dependent Hatching Time 83 

3.5.2 Temperature Dependent Larval Growth Rate 85 

3.5*3 Food Dependent Growth Rate . 86 

3.5.4 Mortality Estimates 86 

3.6 Spawning Biomass Estimates 86 

3.7 Fecundity 87 

3.8 Alternative Quantitative Ichthyoplankton Samplers .... 88 

3.8.1 Oblique Tow Nets with Bridles 89 

3. ft* 2 Vertical Tow Nets with Bridles 89 

3.8.3 High-Speed Nets * 89 

3.9 Intercalibration of Quantitative Samplers 92 

3.10 Design of Surveys to Minimize Costs 92 

I. References 94 



The purpose of this manual is to describe the techniques necessary for conducting 
quantitative ichthyoplankton surveys. The main focus of the description will be di- 
rected toward the use 6f such Surveys for estimating the spawning bioma*fi o pelagic 
fishes* However , ichthyoplankton surveys have many other uses (Table 1.1) some of which 
will also be taken into account within the framework of the text, especially those 
dealing with detection and Appraisal of fishery resources. 

1.1. Background 

Data on the size of the stock, together with an index of its productivity, are 
essential to understand the dynamics of an exploited fish stock and to evaluate th 
effect of fishing on the stock so as to manage the fishery* The^e indices are usually 
derived from life history data and statistics of total catch. Various methods have 
been developed to produce estimates of absolute or relative siae of stocks. These 
methods include: 

(i) analysis of catch and effort statistics, 

(ii) tagging experiments, 

(iii) observation or direct counting, 

(iv) sonic surveys, 

(v) surveys of eggs and larvae 

Under certain conditions, each of the methods may provide a reliable estimate of the 
size of stocks. There are, however, considerable limitations to the application and 
accuracy of each of the methods. Therefore, whenever possible, more than one inde- 
pendent estimate should be attempted. 

In most cases, much information on stocks is obtained through commercial fisher- 
ies which involve various sources of biae. Hence estimates of stock site made inde- 
pendently of the commercial fisheries are very useful. Among the methods listed 
above, the first two depend much upon commercial fisheries whereas the othfer three are 
largely independent of fisheries. 

Catch and effort data analysis has been applied to many commercially exploited 
fish stocks. Whenever possible, this method should be attempted for stock analysis. 
An advantage of this method is that the basic data for the study arc ofttsn available 
without social surveys or with a small amount of additional woirk. As catch and *f~ 
fort data are needed for many purposes including administrative requirements, surveys 
for them are conducted as government routine in many countries. Schaefer (19S7) and 
Silliman (1967) provide good examples of catch and effort data analysis with a simple 
model of population dynamics. 

In order to apply a more detailed model, such as that proposed by tteverton and 
Holt (1959), age specific data (e.g., growth of fish, and age structure in the catch 
as well as detailed catch and effort statistics classified by areas and seasons) are 
needed. Collecting these data requires some care but generally the work is not quite 
as costly as that required for surveys at sea. ' 

One of the most serious limitations of this method is introduced by rapid changes 
in gear efficiency which are now going on everywhere in the world fisheries. Catch 
per unit ef fbrt will give erroneous patterns of the change in abundance If estimates 
Of major long -term changes in the efficiency of gear are lacking. 

Othet <Hflotilti include eh*nies 1ft availability *&& Vudnertbillty delated to 
fish ecology and ooaanographic conditions. $hese changes *ill striouf ly. ff*ct the 
landings wS catoh per unit effort, ' muitin* in itnpreciin Infortiation ** the sise of 
the stock end Jshe mortality rates. This is particularly true for swiftly moving pela- 
gic fishes. 



Mortality analysis with the catch and effort data is difficult to apply to a 
fish species with a short life span, or to a stock exploited by several types of 
gear each specialized for capturing different developmental stages of fish. 

Furthermore, detailed and reliable statistics on catch and effort are npt always 
available to fishery scientists and the available data do not necessarily represent 
the stock because fisheries are often biased by their economic interest* 

Tagging fish is * very useful method for estimating abundance and has been 
used widely for stock analysis (e*g., IGNAF, 1963). In many cases, this method 
has provided reliable estimates of the size of stock. However, there are many 
limitations. Differential mortality sometimes makes the application of this 
method impossible. This is particularly true for small sized fish such as 
sardine and anchovy or small juveniles. Recovery rate from the catch and re- 
porting rate by fishermen may be considerably lower than unity for the catch 
treated in mass. Tag shedding and non-random distribution of tagged fish in the 
stock are among many factors which introduce error and bias in the estimates. 
Cost of fish for release is sometimes exceedingly high, if a large enough number 
are to be released for a reliable estimate. 

Direct counting by eye can be applied only for large sized fish. The number 
of schools of fish such as sardine could be counted but this would be performed 
much better by acoustic instruments. 



^Table 1.1. On the use of ichthyoplankton surveys. (From Hempel, 
1973) 



Detection and appraisal of fishery resources 

Exploring for new resources 

Locating spawning concentrations of important stocks 

Describing relative abundances of. commercially important stocks; 
comparison within and amongst regions 

Monitoring long-term changes in the composition and abundance of 
resources and in spawning times and areas 

Studies in bio logy and sy sterna tics 

Studying the development, growth, behaviour, food requirements and 
mortality of the early stages of economically important fishes as 
related to environmental factors 

Providing a better understanding of oceanic biology, e.g. , zoogeo- 
graphy and ecology of all organisms in the samples 


Clarifying fish systematic* 

Studies In population dynamics of flsttes 

tracing fluctuations in spawning stocks by estimating the abundance 
of their eggs and young larvae 

Forecasting year-class strength on the ft* si* of the abundance of 
older larvae ^ , : ' 

Estiavating abundance of a *tock based 0*1 its pawning production 
Piscriainat ing between stocks of tfce sane species ,' " T '.,. ( 



(QPOQPOOi 



V50QDOO 




10000- 



Figure 1.1. A time series con* 
pa risen of sardine and anchovy 
biomass estimates from 1940 through 
1969. (From Smith, 1972).. Dashed 
lines represent interpolations be- 
tween non- adjacent year*. 

e - sardine biomass from 19.40 to 
1959 calculated from the fish* 
ery by Murphy (1966) 

v 

A - sardine biomass from 1951 to 
1969 derived from a regression 
estimate of the. relationship be* 
tween the Murphy biomass esti- 
mate and the sardine larval 
index (Sections 2.4., 2.5.) 
determined during the refer- 
ence years 

o - anchovy biomass from 1940 to 
1959 derived from the ratio 
of anchovy larvae to sardine 
larvae and the Murphy sardine 
biomass estimate 

A - anchovy biomass from 1951 to 1969 
derived from a regression estimate 
of the relationship between the 
anchovy tonnage calculated from 
the anchovy: sardine larvae ratio, 
and the anchovy larval index. 

Q - Murphy estimate of anchovy 
spawning biomass by 3 -year 
averages (Murphy, 1966) 



Sonic survey is acquiring more and more importance owing to the recent develop- 
ment in acoustic instruments. Although the results obtained to date are not as num- 
erous as those provided by the catch and effort data analysis and tagging experiments, 
the importance of sonic surveys in the future is expected to be extremely high. This 
method is used for locating fish schools and even individual fish. However, the in- 
terpretation of sonic records requires sampling by fishing operations or other means, 
such as photography or submersible television to identify the kind of fiah,~ It is 
difficult to Intercept all parts of the stock if fish are moving rapidly either hori- 
zontally or vertically unless a number of vessels are used simultaneously for a sonic 
survey. Methodological limitations of sonic surveys have been considered by Forbes 
and Nakken (1972). 

There are several marked advantages in using fish eggs and larvae to monitor 
adult populations and estimate biomass. The early .life history stages of fishes are 
restricted, by depth, usually to the upper mixed layers. The passive eggs and feebly 
swimming larvae are quite vulnerable to capture. Many marine fishes have pelagic 
eggs and most have pelagic larvae. Thus, it is easy to quantitatively sample several 
species over broad areas with a simple plankton net. frhis type of gear oan be 
handled by a wide variety of vessels without major installations of equipment. The 
increase in cost for adding species to be studied for biomass estimates is much less 
when eggs and larvae are used than when adults are used. The index of larval abun- 
dance obtained from ich thy op lank ton surveys has been shown to provide a reliable esti- 
mate of biomass, e.g., for sardine and anchovy (Fig. 1.1) and Pacific mackerel 
(Fig, 1.2). 



as 



X30NI 

T I 



a 



I ............ Till T I T I T ! 

OOOD S8VWOI8 



r 7 I i 



\ \ \ I I I 





^,< ; 



i 



100' 







I 




10 15 Eggs 



Figure 1,3. CorreXation between the abun- 
dance of eggs and the adult catch of an- 
chovy in the following fishing season in 
the Pacific waters along Honshu, (r -.91, 
r d -,88). (Data from Hayashi, 



There are additional benefits to be 
obtained from ichthyoplankton survey* pri*" 
marily designed to estimate spawning bio* 
rnaas. The samples contain not only fish 
eggs and larvae, but also part of their 
potential zooplanktonic food and predators. 
Modern ooeanographic techniques Allow sim- 
ultaneous measurements of the physical and 
chemical environment of planktonic commun- 
ities* Trends of water motion can be esti* 
mated. The survey can detect spatial and 
temporal isolation efficiently over wide 
oceanic areas and thereby help define unit 
stocks of fish upon which fishery manege* 
ment normally depends (Gulland, 1969). 
Spawning distributions delimited by sur- 
veys tell when and where the fish will be 
concentrated for efficient capture. Re- 
sults from the survey can be used for mon- 
itoring changes in the species composition 
and diversity of the communities in which 
the fish stocks reside. Data obtained can 
be used for forecasting the stock site in* 
to the next fishing season (Fig. 1.3,) and 
for forecasting the year class strength of 
a species. Ichthyqplankton surveys can 
also contribute information on new stocks 
of fish with commercial potential (Ahl- 
strom, 1968, Gullend, 1970). 



Problems associated with the use of 
ichthyoplankton surveys to estimate spawn- 
ing biomass are of a taxonomic, technical, 
and statistical nature. Problems involved 
with identification stem from the fact 
that early life history stages have been 

described in the literature for relatively few of the species* Also, there are a 
limited number of specialists who are broadly knowledgeable in the identification 
of fish eggs and larvae. These problems are surmountable (Sections 2.2., 4.). 
Technical problems involved with the survey require an adequate staff be assembled 
to handle the collection and processing of samples which may be extensive* Also, 
high standards must be maintained during the sampling process for quantification and 
comparison of data (Sections 2.1.2,, 3,2.). The statistical problems are inherent 
in the oceanic sampling process and require the ability to interpret difficult dis- 
tributions of sample data (Section 3.1.), 

1,2. Scope 

This manual Describes recommended standard procedures for conducting quantita- 
tive ichthyPp^anHton surveys for spawning bioniaas estimates, This description (Sec- 
tion 2) cpverp survey planning fW* field operation* , laboratory procedures, d*ta 
suinmariiation,, making census estimates, and estimating the spawning biojwss. Theoret- 
ical considerations of the problems and biases involved with the recommended proce- 
dures are discussed in Section 3* 



ft l*rge elected bibliography on ighthyoplenfctan fw?vi (?AQ Fisheries 
706) i* toeing prepared to complement this manual. In addition to papers on survey 
method* *n4 result*, the bibliography includes references on spteitf identification, 
whioj* is cwtsWe the soppe gf thia nwmal, and taWtf o**e*9P inferences of the 
paper* an survey ?epults and specie* identification toy gapgraphio area and teicun&mic 
group , In PHHPi where thfrp are no .pubUihed works and fish must be reared for identi- 



fication purpose*, the bibliography of May (1971) should to consulted for references 
on appropriate culture techniques* 



2. Hftoomiended Procedures for Conducting Ichthyoplankton Surveys for Spawning 
Biomars Estimate's ' 



10 



The chief barriers to the successful execution of surveys and delivery of survey 
results are I) under estimation of the necessary technical effort, and 2) imprecise 
definition of survey objectives. Most hew ichthyoplarikton surveys will encounter a 
mixture of new problems, specific to the survey area, and problems which have been 
well-defined and solved in older surveys. This section of the manual will be directed 
toward the standard methods recommended on the basis of survey experience so that the 
newer surveys may best encounter and solve the problems specific to the new survey en- 
vironment. 

The planning stage is likely to be the most important part of the survey for this 
r . is where the objectives of the survey are 

* compared with monetary and personnel re- 

sources. The commitment of money and time 
to the surveys is an irreversible step. 
To minimize risk, it is necessary to set 
phased re-evaluations of objectives and 
survey performance. This process should 
be incorporated into the planning so that 
the tactics of the survey may evolve to- 
ward the most effective delivery of survey 
results. 

2.1. Field Operations 

In this example, it is considered 
that the field operations will be con- 
ducted from ships 15 to 100 meters in 
length which are equipped to make plankton 
tows and hydrographic observations. 

2.1.1. Cruise Plan and Personnel 

Normally the cruise plan (Section 
3.3.) is prepared by the chief scientist 
and the chief marine technician. After 
the cruise track is prepared, it is dis- 
cussed with the ship's captain* Figure 
2.1 shows a two- ship cruise plan to mea- 
sure the extent and intensity ot spawning 
in the outlined area. 

Log sheets ("Captain's sheets") which 
list the desired stations and their posi- 
tions, the order in which the stations are 
to be occupied , the captain and the navi- 
gator for the cruise ship, the month and 
year of the cruise, and the niqiie of the 
ship (Fig. 2.2) are filled out by the 
chief scientist and the chief marine tech- 
nician beflore each cruise. During the 
cruise, Information is added to these 
sheets including station position actually 
occtipied, the day of the month, the time 
of arrival to and departure from the sta- 
tion, and the method tot locating the 
station p<>sition. Omitted stations arte 
noted and the order in which the stations 
were actually occupied is corrected accortt- 
ingly. 




270 2K> 150 SO 90 



Figure 2*1. An example of a cen- 
. trie systematic area sample grid 
(Milne, 1959) superimposed on a 
hypothetical coastal spawning 
area. She solid dots repre*n 
complete biological oceanographic 
sampling stations and the "X f s* 
represent supplementary stations 
for detailing coastal gradients. 
The lines connecting the stations 
represent a possible method of oc- 
cupying each station with two ships. 
diamond-enclosed 



The diamond-enclosed **jp*. represents 
* test station occupied by both 
ships on each cruise to cimdk the 
' completeness artd operability of 
the ship's survey equipment and to 
compare the performance of each 
ship's measuring equipment. 




c 

H 


t 

<0*W 
O O 



05 

-H C 
4J H 
H 
ID CO 



8 

H (0 



a-H 

3 rH 

u o 

8 






tM 
O 
.p 
(0 

a 



ts 



STATION *tQUffiCMNT SHttT 




w^wWpnii 



Figure 2.3* Station requirement 
*heet. (Courtesy University of 
California, Scripps Institution 
0? Oceanography*} 

2.1.2. Sampling System 



The chief scientist and the chief 
marine technician also fill out a station 
requirement sheet, (fig. 2*3.) which lists 
the stations the same as the "Captain's 
sheets 11 but includes additional information 
of the number and type of ablfc^vttiona to 
be carried out on each station. Copies 
on this sheet are provided to eacm cruise 
leader (one of whom may be tJfte ahief ma- 
rine technician). Following the descrip- 
tion of the cruise on the station require- 
ment sheet, the chief marine technician 
and the cruise leader select the. marine 
technicians and watch leaders required for 
each cruise. Following the selection of 
the scientific staff for the cruise, a 
cruise announcement is iasued which pro- 
vides the beginning date and the proposed 
final date of the cruise, the objectives 
and procedures for the cruise and the per- 
sonnel who are to conduct the cruise. 

It must be stressed that having an 
experienced, responsible marine technician 
(who may also serve as cruise leader) is 
essential to the success of the survey* 
Duties. assigned to the chief marine techni- 
cian go far beyond those involved with 
setting up the cruise plan. He is respon- 
sible for seeing that the station proce- 
dures (Sections 2.1.2., 2.1.3., and 2.1*4) 
are carried out correctly for each station 
that is occupied. 



The recofflroended sampling system requires that the ship be equipped with a hydro- 
graphic winch with more than 400 meters of standard hydrographic wire (i.e., 0.48 
cm in diameter) , a meter block or a metering system on the winch to measure the 
amount of wire out end an angle indicator (inclinometer) to measure the angle at 
which the wire enters the water. 

2.1.SM. The 



the Bongo net {Figs. 2.4 and 2.5) towed at slow speeds, is recommended as the 
best type of simple gear fished from hydrographic winches for ichthyoplankton surveys. 
It provides a minimum of variation in the biases caused by uneven filtration per unit 
depth, avoidance of the net, and escapement or extrusion of organisms through the 
meshes, this recommended "slow* Bongo equipment is derived from bridle-free plankton 
nets (McGowan **td Brown, 1X6). It incorporate* simplification* from the original 
opening-closing design suggested by Posgay, Marak, and Hennenrath (1908) as well ai 
modifications made by Smith, Thrailkill f and Vrooatan (1971)* after extent! ve compar- 
ison of the Bfcngo net tod the CalCOFi standard net (Vrooman, 



Bongo towing frame {fig. 2.4) made of anodited aluminum, consists of two cir- 
cular frames, each f >tf ft in diameter, connected by a central yoke to which the towing 



* cruise Report* David 

Pisherie* OettterTEa 

** cruise toport. Da 

Fisheries Center 



~ dffi^jf * 59 * etd April 1971. On file at Southwest 
a, California. 

170* Dated June 1972, on file at Southwest 
ornia. 





wire is attached. Thus there are no 
bridles in front of the mouth of the net. 
A 22 kilogram dead weight depressor is sus- 
pended beneath the frame (Pig. 2.5) for 
making standard oblique tows. 

The towing frame is fitted with two 
cylindrical-conical nets (Figs. 2.4, and 
2.5) made of Nitex or equivalent mbnofila- 
ment netting preferably of a dark color. 
One net, which is the principal ichthyo- 
plankton sample net, has a 0.505 mm mesh. 
The other net, which may be used for p^ank- 
ton biomass studies or fish egg and larva 
escapement and extrusion studies (Lenarz, 
1972), should have a 0.333 mm mesh. The 
use of ^soft" cod ends (Pig. 2.6) is recon- 
mehded primarily as a matter of handling 
ease. Sample spillage back into the net 
is less likely to occur than with a "hard" 
cod end. The cod ends are made of the 
same type netting and mesh size as the 
main part of the net. The nets and the 
cod ends are color-coded for easy recogni- 
tion and match up of mesh size (e.g., red 
for the .505 mm mesh and blue for the .333 
mm mesh) . 



Figure 2.4. The Bongo net recom- 
mended for ichthyoplankton samp- 
liner. 

A flowmeter is mounted in the mouth of 
each net (Fig. 2.5) to provide data on the 
volume of water filtered during each tow. 
This information is essential for quantir 
fication of the data* 

2.1.2.2. The Towing Procedure and Data 

' 



The standard towing procedure is modi- 
fied from Kramer et al. (1972). An .example 
of an on-deck arrangement for making plarrk- 
ton tows is ppbvided in Fig, 2>7. 



The data sheet .- Before the station is 
occupied , m$ following jiunjber'ed items are 
recorded on <th plankton- tow data* sheet, 
a sample of'lJK^'VtoVjLa^^ 2.8. ft&e 
data sheets are made Of a water resistant 
linen which 
handled or 




regular and fine mesh. 



Item 12 Meter miirtber ( flowmeter 
*-"*"' rtiimhef] , **8a 'regtifar and fine 
mesh . 



of 



meter block on research vessel, 
2) flowmeter suspended within 
the mouth of each net, 3) 22 leg 
(50 Ib.) hydroa^aphic weight. 



10 



Item 14. Initial flowroeter reading (carryover from last tow) for regular and 
fine mesh. 

Additional data to be recorded during the tow are mentioned as the towing procedure 
is described, 

The tow - After the gear is assembled and the data sheet for the plankton tow is 
prepare^ las outlined above), the tow is ready to begin. The net tow is made off 
either side of the ship as follows: (Tows off the stern are not recommended because 
of turbulence from the ship's propeller.) 

1. The ship is stopped on station. Bottom depth on station is requested from 
the bridge and recorded in the lower left-hand section of the data sheet. Note: The 
bottom depth will determine the depth to which the nets will be lowered and thus the 
length of wire to be let out. To lower the net to 210m (the standard procedure for 
this type of survey, depth permitting) with a wire angle of 45 requires that 300m of 
wire be let out (wire angle is defined as deviation from the vertical) : 

Length of wire out X cosine 45 -net depth 
300ro X 0.707 - 210m 

If station depth is less than 238m, reference to the "depth of tow 1 * graph (Fig. 2.9) 
will quickly give the proper amount of wire to be let out so that the net will not hit 
the bottom. For shallow tows, the wire is let out and retrieved at the same rate as 
for routine standard tows to avoid relative over-sampling of the surface layers. The 
standard tow filters 1 to 2 cubic meters of water for each meter depth* Alteration 
in the rate at which the wi*e is let out or retrieved would materially affect the 
basic statistical distribution of sample size. 

2, The flowmeter is read and checked against the recorded initial meter reading 
(Item 14 on the tow data sheet). If there had been a previous tow, this should have 
been the final meter reading (Item 13 on the previous tow data sheet) . If the read- 
ing changed between tows, the last recording is crossed out, the new reading is 




Figure 



11 




entered, and an explanation is given in 
the lower right-hand part of the data sheet 
under "Remarks". 

3. The 22 kg weight is lowered below 
the surface of the water. If the ship is 
still slightly underway , the wire (16 in 
Fig, 2.7) is pulled to the side of the 
work-platform (#10 in Fig. 2.7) and fast- 
ened close with a snap hook attached to 
the outside rail. 

4 The inclinometer is fastened to 
the wire above the Bongo net (Fig, 2*7) , 
Enough slack is left on the line to the 
inclinometer so that when the proper angle 
is achieved during the tow, it will not 
ride up on the cable to hit the block. 
(If the survey is for net tows only, the 
inclinometer may be left on the tow wire.) 

5. The winch meter is zeroed. The 
ship is set underway, wind off the bow on 
the side on which the tow is taken, and 
the signal to start the tow is .given from 
the bridge. The blocks or pins, which 
keep the blades of the flowmeters from re- 
volving between tows are removed, and the 
Bongo net is lowered into the water. 
(Some flowmeters have "automatic" blocks 
that release the impeller blades when 
water flows through them.) 



a. 



Figure 2.7. The on-deck layout 
of the ichthyoplahkton towing 
apparatus on the U.S. National 
Marine Fisheries Service research 
vessel David Starr Jordan . 1) 
Winch dbruw which holds the hy- 
drographic wire. 2) Boat deck. 
3) Flood lamp for illuminating 
immediate area of inclinometer 
and overside platform. 4) Meter 
block which is connected to read- 
outs in ship's bridge, dry lab- 
oratory and, and winch console 
and gives an instantaneous read- 
ing of meters of wire out. 5) 
Inclinometer which indicates 
angle of stray of towing wire. 
6) Towing wire which is 0.48cm 
(3/16 in.) in diameter. 7) 
Intercommunications device which 
allows communication between 
bridge and winch operator dur- 
ing tow. 8) Meter block read- 
out on windh console. 9) Winch 
*nd boom tjorftrols at winch con- 
sole* 10) Overside work plat- 
font for hooking up inclinometer 
and, launching and retrieving 
Bongo net. (Adapted and modi- 
fied from Kramer et al., 1972) 



b, 



The nets are allowed to stream 
out briefly before lowering. As 
soon as it is obvious that they 
are not tangled, the wire is im- 
mediately let out the predeter- 
mined amount at a constant rate 
of 50 m/min. 

The stopwatch is started as soon 
as the flowmeters are seen to 
sink below the surface of the 
water. The stopwatch is used to 
record sinking time (Item 8) and 
towing time (Item 9) in seconds. 
The duration of the tow in 
seconds is used for calculation 
of mean velocity of towing. 

The time the net enters the water 
is recorded in nautical time 
(2400 hr) to the nearest 5 min. 
(Item 6, and Item 20 or 20* da- 
pending on whether the tow is 
"Routine* or "Other"). The time 
of day is used for analysis and 
correction of day-night differ- 
ences In the catching power of 
net. 



d. 



When the desired amount of wire 
has been let out, the stopwatch 
is stopped, the sinking time is 
recorded in seconds (Item 8), the 



12 



Jtouriitf* fmt HIT CNTI$ WATER- . AJ-Ji.S 




.,, 
of a sheet mad* out or 



ciojiy 

^^^ j 



13 

stopwatch is zeroed and restarted immediately. <Since the net is "fish- 
ing" on the way down, sinking time is as important as that of retrieval*) 

e. When the stopwatch is restarted, the nets are left at the desired depth 
for 30 sec. (hypothesizing that "falling" nets will straighten out at 
depth in the 30-sec. interval). 

f . At the end of 30 sec. the stopwatch is not stopped, the wire angle is 
recorded for that depth, and retrieval Ti~begun at the rate of 10 m per 
30 sec. for all tows. The wire angle is recorded at every 10 m (Item 

20 to 22 for routine tows with 300 m of wire out or Items 20' to 22 ' for 
other tows). Note: Ship speed, during sinking, during times at depth, 
and during retrieval, is maintained to keep the wire angle at 45. In 
dead calm, it may be necessary to run the ship in circles to maintain 
the wire angle. It is essential to maintain the 45 (3) wire angle to 
assure that the proper amount of water is being filtered from each depth 
(1 to 2 cubic meters per each meter depth) for purposes of data quantifi- 
cation. 

g. The nets are brought directly. out of the water at a steady rate. Note: 
It is important not to allow the nets to fish too long at the surface 
because of the bias that results from over samp ling surface waters (Sec- 
tion 3.2.1.)- When the flowmeters break the water surface, the stop- 
watch is stopped and its reading in seconds is recorded as the towing 
time (Item 9). Note: If the flowmeters are the kind that do not have 
"automatic" blocks 6n the impeller blades, blocks or pins should be 
placed in them to stop the impeller blades from rotating as soon as 
possible after the flowmeters break the water surface. 

6. The nets are washed down from the outside to get all the plankton into the 
cod ends, keeping the net rings at rail height with the cod ends dangling, A salt 
water hose is used for the washing. Water pressure should be enough to loosen the 
plankton adhering to the net mesh but not so much as to damage the plankton organisms. 
This operation usually takes only a few minutes per net. 

7. After all the plankton has been washed into the cod ends, the nets are 
brought aboard. The color-coded cod ends (e.g., red for .505 mm mesh and blue for 
0.333 mm mesh) are removed, keeping the plankton from spilling back into the nets. 
The plankton samples are taken to the ship's wet laboratory and preserved immediately 

(Section 2.1.3,1.) . 

8. After the samples have been preserved, labelled, and stored, the cod ends are 
washed (they may be left everted) and replaced on the nets with matching color codes 
in preaparation for the next tow. 

9. Before leaving the station, the flowmeters are read and recorded as the final 
readings (Item 13) . The initial readings (Item 14) , which were made before the tow 
began, are subtracted from the final readings (Item 13) to get the differences (Item 
15) which are the number of revolutions for the tow fbr each net. Note: It is impor- 
tant that the proper amount of water has been filtered for quality of tow and data 
quantif icmtion. For each flowmeter, the proposed central number of revolutions and 
acceptable limits to this .number of revolutions should be calculated before the cruise 
to serve as a guide for the observer to tell if the tow has been correctly conducted* 
If meter readings are not normal, the net tow may have to be repeated. A very high 
reading may have been caused by too great a ship's speed - check for many high wire 
angles. A low reading may have been due to too slow a ship's speed - check for many 
low wire angles. Another reason for low meter readings may be clogging of the nets. 
This may be cumulative if a net is not rinsed properly or it may occur at a single 
station. If a meter shows a trend toward lower and lower readings, attd it i* not 
malfunctioning, the net should be washed (Section 2.1.2.2., 115). The net tow need 
not be repeated if it is obvious that heavy clogging is the reason for low readings 
(it will only clog again) or if the ship's speed has caused low or high angles. If 
wire angle* me norttial, the net is clean* ttfcd t&f tdwinjfir ***, Option 2 2 :U2.3- til) 
is routine but the meter reading is low, the cause could be that a bit of detritus, a 
fish, or even a large jelly or salp had become entangled in the meter blades for a 
portion of the tow. Under these conditions, the tow should be repeated. If the 



14 



300 



LJ 



200 



U. 
O 



CO 
(T 
LJ 

UJ 



100 




50 100 150 200 
METERS DEPTH 



250 



Flgur* 2.9. 



graph. <Modi*i*d f ron Xriwr *t *1. ,1872). 



IS 

reading ip again very low and it is obvious that the flowroeter is not functioning 
proper ly f replace the meter and repeat the tow. Dp not oil or grease any meter or 
make any repairs that might alter the rotation of the blades. Repairs of this type 
would seriously affect the calibration of the meter. If clogging is apparent, the 
amount is recorded in the appropriate box on the plankton tow data sheet. 

10. Wind, sky, sea condition and sea swell should be requested from a crew mem- 
ber by the observer who is recording the angles while the tow is being made- jThe 
information 1m recorded in the appropriate boxes in the lower section of the plank- 
ton tow data sheet. 

11. Total towing time (Item 16) is recorded as the sum of Items 8 and 9* For a 
300 m tow, total time should be about 21 f 30 tf (6 1 sinking time + IS'30 11 towing time). 
Note: If total time is off by 15 to 20 sec,, it must be explained in the "Remarks* 
section* The most usual variation will be in the sinking time, caused by a slightly 
faster or slower rate in paying out the wire than the recommended 50 m/min. In cer- 
tain conditions, such as poor control of the ship, countercurrents below the sea sur- 
face adversely controlling the net as it falls, etc., the winch operator may have to 
depart from the sinking- time procedures to slow the falling net in order to keep it 
from becoming tangled. Such departures from normal procedures must be recorded in 
the "Remarks" section. 

12. Total towing time is added, in minutes and seconds to Item 6 to record the 
hour, minutes and seconds in Item 7. This is actually the time the net comes out of 
the water. 

13. The actual station position (accepted position) is recorded in Items 16 and 
17. This may be done when the station is occupied, but usually is done at the end of 
the cruise when the captain has compiled a complete list of the positions of all, 
stations (Section 2.1.1.). 

14. Checks are made regarding the sample (Section 2.1.3.) and appropriately re- 
corded on the tow data sheet in the lower left-hand section. 

a. Number of jars per sample. 

b. Centimeters of plankton - this gives approximate volume before water is 
added. 

c. Formalin and sodium borate added - the person who adds the preservative 
and buffer must initial this box for each sample after the Formalin and 
sodium borate are added. 

d. Sample labelled - the person who labels the sample must initial this box 
after the sample is labelled. 

15. Checks are made on the net to see if washing or repair is necessary. The 
appropriate boxes are checked in the lower section of the tow data sheet. If replace- 
ment of a net is necessary, it should be recorded in the "Remarks" section. 

If washing is needed, one of three methods may be used: (1) The net is 
everted, still on its ring, and brushed down with an ordinary sweeping 
broom and running sea water; (2) The rings are stood on edge, the net is 
tightened along its length by tying down the end (without cod end at* 
*&**&} and hosed, down with a high-pressure fire hose. This is very ef- 
' v f etStive provided that plankton has not dried in the meshes; (3) The net 
is detached from the ring and put in a washing machine using , a 30-aiin. 
cycle* war water (not hot) and a non-polluting detergent. , 

b. Hips and holes in the net - the net should be examined aft* r every tow 
to check on needs for repair or replacement, if holes or bears are 
small, they should be sewn before the next tow with nylon thread of A 

t color (e.g., red) that can be easily located for sewing Machine repair 
, later on shore. If the net is torn beyond mending at sea, replace the 

' 



16 , 



16* Recheck the tow data sheet to be sure that all items are filled in. (The . 
occupancy code, Item 2, is filled in onshore at the end of the cruise. This is usu- 
ally one of a series of numbers used by a computer pro^rtajnmer to describe the type oi 
tow or the station occupied.) Notei Some deviations from standard procedure cannot 
be avoided. Circumstances such as strong under-sea currents, high winds,/ heavy seas. 
will cause unavoidable deviations, e.g., odd meter readings, prolonged s'tops at sta- 
tions. Conditions of this kind should be noted in the "Remarks" section. 




17. A new plankton tow data sheet is set up for the next station t Items 1, 3, 
4, 5, 11, 12, and 14 are entered. Item 14 should be the final reading of the pre- 
ceding tow and should be rechecked before starting next tow. 

2.1.3, Handling the Sample at Sea 

Fish eggs and larvae are fragile and 
easily damaged. Proper care is needed in 
all stages of preservation and handling 
plankton samples aboard ship. 

2.1.3.1. Preserving the Sample 

The plankton sample should be preser- 
ved immediately. This is especially criti- 
cal in tropical waters. 

The storage container in which the 
sample is preserved should be of sufficient 
size so that when filled, the preserving 
liquid (5% buffered Formalin is recommend- 
ed) will bccupy at least three times the 
volume of the plankton. No problems are 
posed in having a large ratio of preserv- 
ing liquid to plankton. Glass jars hold- 
ing 1000 ml are recommended. Normally 
they are closed by a screwed-on plastic 
lid with an inside coating to prevent leak- 
age and evaporation. 

The plankton collection is carefully 
poured from the cod end into the container 
in which It will be stored. The cod end is 
then rinsed down to gather the last of the 
plankton at its bottom. When fairly well 
drained, the cod end is everted over and 
into the jar and the remaining plankton 
is washed off carefully. It has to be en- 
sured that no parts of the sample remain on 
the mesh of the cod end. 

At this point, before seawater is add- 
ed, the height of the plankton i^ the jar 
is measured in centimeters. Thi's gives art 
approximate volume to be recorded on the 
plankton t<*w data sheet (Section 2.1.? ,2., 
*14). 

The ^ar containing the plankton is 
then f med t&re^-fburtha full with sea- 
water before adding th% preservative* (full 
strength Formalin) and buffer (sodium bo ir- 
ate ), 3Chi done to avoid '"bttrning* the 
delicate plankton organisms. To obtain 
the recommended 5% solution of Formal in in 
a one -liter jar, 50 ml of concentrated 
commercial Formal in is added , To assure V 
proper buffering, 20 ml of a saturated 



8 .:.-- -- 

Figure 2.10. Ship's wet labora- 
tory setup for preserving and 
labelling plankton samples. 1) 
Plastic carboy (19 liters) con- 
taining concentrated formaldehyde 
and having 1 . 5 m of surgical tub- 
ing. 2) 20 ml disposable syringe 
with cannula for adding buffer to 
plankton samples. 3) Container 
of sodium bo rate in seawater. 4) 
Salt water tap with O.S m of stlr-^ 
gical tubing to rinse cod end, 
concentrate plankton, and fill 
sample jar after fixation and buf- 
fering, 5) Ship's hot and cold 
freshwater taps. 6) 50 ml dis- 
pottable eyringe with automatic 
double valve for measuring and 
dispensing formaldehyde . 7 ) 1 
liter sample jare. 8) Removable 
wooden frame to support sample 
jars im rough weather. The en- 
tire assembly, including sink, 
ean toe fitted Into 1 to & tt&*of 
laboratory bench space. {Adapted 
and modified f row **aer- et al. , 
1972.) ^ 



17 



solution of soditim borate in seawater is added for each liter o preserved 
(9u4T is added to counteract the acidity of plankton In formalin. 
definitely not be used as a buffer. A solution containing too much or too 
buffer is Jiarroful to fish larvae.) The sample jar is then filled aOaosfc *so the' 
with seawater, capped and shaken lightly (including inversion) to ototaim 
uniform preservation of the plankton organisms throughout the sample* 

Data on the number of jars per sample and addition of Formalin and borate to 
the sample must be entered on the plankton tow data aheet (Section 2.1.2.2., fl4>* 
Note* full strength formaldehyde aboard ship is kept in 19 liter polypropylene car- 
boys (Fig* 2*10). With the carboy moored securely above the sink, the preservative 
is drawn by siphon action. A further safety measure is to draw the formaldehyde via 
a teflon tube into a 50-ml" plastic syringe through an automatic double valve. The 
buffer is added with a 20-ml plastic syringe fitted with cannula (a "needle" without 
a point) . 



Depth of 



Horn , <S9l Of Vrrt 



Gar and 



larfc* 



..ff* . 

Collector 



Formalin is the most widely used preservative for plankton collections. It is 
by no means "ideal", but satisfactory when properly handled. Years of experience 
have shown that too strong or too weak a concentration of Formalin can be damaging 
to ichthyoplankton. The recommended concentration of 5% for preserving the sample 

TUTION Of OCEANOGRAPHY and 3% for later storage of eggs and 

ZOOPUNKTON COLLECTION larvae (Section 2,2.3.) have proven to 

2y.ffJr.vT. work well over a long time scale. Other 

sh^ c!* recommendations have been recently pro- 

posed (Steedman, 1976; UNESCO, 1974). 

The concentrated, commercial prepara- 
tion of Formalin is a 40 percent solution 
of formaldehyde gas in water. Formalin 
should be stored in inert containers, 
glass or plastic, not in metal containers; 
Formalin reacts with the latter. One of 
the problems in storage of concentrated 
Formalin is its tendency to polymerize - 
to form para- formaldehyde, a flocculent, 
white solid. 

It should always be kept In mind 
that formaldehyde is a poison that can 
irritate skin and cause serious cases of 
dermatitis . Rubber gloves should be 
worn when preparing formaldehyde solu- 
tion, or when handling specimens pre- 
served in Formalin* Formaldehyde fumes 
can also be irritating to the lungs. 

2.1.3.2. Labelling 

r ' 

It is essential that samples be pro- 
perly labelled, information contained on 
the labels should toe sufficient to*' identi- 
fy the sample with certainty. Each sample 
jar should have two labels* One label is 
placed inside the jar; Because thifc 
label is often diff icu It to, read .unless 
removed, a second. label is placed on *&e 
, lid of the sample jax. Both lab* Is are 
necessary because jar lids can be mixed 
u up* Should t!4 happen, the inside J^abel 

2.11. Labels for field idea- usually assures proper identification of 

label, {Cduftesy Dniver-T * '. < 




tif 1 



ti<m of oc*anografi) 
label on jar lid* 



Outside 



toe madef of heavy weight, cWelcall|r re- 
sistant linen paper which will tiot 7 fall, 



18 



apart in th Fo*m 1 in solution. The label is filled out with a soft carbon pem- 
cil wtoiofc will not fade in the Formalin, This label should contain the following 
information* nama of ship, cruise number, date, time, station designation, depth 
Of tK>w, gear used, meah siie, type of haul, duration of haul, collector** name, 
of jars in the sample (1 of 2, 2 of 2 or 1 of 3, 2 of 3, etc.), 

The outside label (Fig. 2,11) is written on the lid of each sample jar with a 
light-colored waterproof marker or wax pencil. It should contain the following in- 
formations cruise number, station, date, time, gear, mesh sise, nunber of jars in 
the sample (1 of 3, 2 of 3, etc.). This outside label should be color-coded to 
match the net and cod end mesh sizes. This may be done by using an appropriately 
colored waterproof marker or colored tape (e.g., red for ,505 mm mesh and blue for 
.333 mm mesh) 

After the samples are labelled, the person who labels them should initial the 
appropriate box on the plankton tow data sheet (Section 2,1,2.2., 114). 

2.1.3.3, Storage on the Vessel 

Sample jars should be completely filled. This is to prevent agitation of plan] 
ton in the jars during rough weather, as will happen if they are only partially 
filled. Even when filled, some movement of the organisms within a jar is inevitable 
due to vessel motion (rolling, pitching, etc.). Hence, to minimize damage to the d< 
licate organisms during the storage on the vessel, samples should be stored in the 
most stable part of the ship. 

Temperature of storage may be important, particularly in tropical regions. Th: 
factor seldom has been given much attention. Collections from tropical expeditions 
often are of poor quality, even when care has been taken in their preservation and 
handling. Experiments are needed to test the effect of storage temperatures on the 
condition of plankton organisms, especially during long cruises. It may be found 
that air conditioned (temperature-controlled) storage rooms are essential for the 
proper care of plankton collections on tropical expeditions, 

2.1.4. Supplementary Data 

At the critical stages in the life history of young fish, minor changes in the 
environment can cause extensive mortalities and lead to corresponding fluctuations 
in the abundance and availability of fish stocks. It is therefore important to col- 
lect environmental information during the course of ichthyopl*nkton surveys, 

2.1.4.1. Rydroqraphic Data ,; , v .> 

for the standard ichtrhyoplanKton survey, it is ia^ortant to 
scsn physical oceanographic data at each station. Thia usual 3y ir 
temperature taken with a bucket thermometer, an4 
(BT) or expendable bathythermograph (XBT) obse 
ature with depth. A sample data sheet for use 
(If an XBT is used, the "BT instrument number" 

Additional observations cbuld include 
inity, oJcygen, and nutrients and a Secchi disc 

, , t , , " , T , ;, / 

2.1.4.2. Smaller Zooplankton and Phytoplankton 

Information oil the availability and utilization 
obtained by simultaneously sampling for the smaller pli 
th* larvfce. *or studying these organisms, it la recomaer 

2t) etfi ih diameter at the motrth, be u*ed. One net should be fitted' with netting of 
0.25* wi mesh apertures and the other with 0.150 mm aesh f These nets can be fasten* 
to the towing cable immediately above the *t> Cm 36ngb assembly (Fig. 2 t ;3). it is 
also highly desirable to measure photOaynth0tic pigmehti. The ooplinKton, in the 
smaller nets and the photosynthetic pigment <Jata will provide a itaadatd baae for 
^ the rt^atiife pvo*^i*rtty of the areas liNN^tl? 




sa: 



can be 
eaten by 
ed Bongo nets, 



19 



rr 



r* 



If 



itt*if 



/if* 



mei 



.M** 



'* 



M iM 



it* 4*4 



4+ .a 



UfS92 



tar 



"?* 



jo IT 



4*Ji 



^/ 



HIM 



^X 



jn/4**5 



Figure 2.12. Oeeanographic log sheet for bathythermograph observation* . 
tesy University of California, Scripps Institution of Oceanography). 

2.1.4.3. Neuston 



(Cour- 



It is advisable to include a neuston tow for larvae and juveniles for 10 minutes 
during the Bongo net tow. Larvae of several species of commercially* important pela- 
gic fishes are known to occur in concentrations in the surface layer (Klawe, 1963; 
Sund and Richards, 1965; Par in, 1968; Zaitsev, 1970; Richards and Simmons, 1971). 
Their occurrence is mainly ephemeral, varying with stage of development, light, feed- 
ing, and oceanographic conditions (see review by flempel and Weikert, 1972). Neuston 
samples can be collected with relatively small effort and they can provide valuable 
information on a number of kinds of fishes including tunas, billfishes, scombroids, 
and even some flatfishes. A rough check on presence or absence of eggs, larvae, arid 
juveniles in the surface layer can be done with any kind of simple floating net which 
is towed at a speed of two knots. The mouth opening should cut the water surface and 
extend to about 10 to 50 cm deep. The tow should be operated aside of the ship and 
well clear of its bow wave and wake. Meah size should be the same as in other *Ch~ 
thyoplankton sampling gear, i.e., about 0.3 or 0.5 millimeters. In some surveys, 
including the Cooperative Investigation of the Caribbean (CICAR) , simple rectangular 
frame nets of2xlmorlx0.5m mouth aperture have been used. More sophisticated 
gear is now in use at several laboratories with the objective of sampling the near* 
surface layer in a more quantitative manner. The major difficulty in this respect is 
to control the depth of insertion of the net's mouth opening into the water and hence 
the amount of water filtered per meter of tow. The modified David-Neuston net (Hempel 
and Weikert, 1972), which is now recommended for international surveys fluch as the 
Cooperative Investigation of the Eastern Central Atlantic (CINECA) , Sameoto's otter 
surface net (Sameoto and Jaroszynski, 1969), and various gears by Zaitsev (1970) have 
to be mentioned in this context. 

2.1.4.4. Micronekton 

Data oa ichthyoplankton obtained from standard Bongo net samples can be usefully 
supplemented by larger nets with larger mesh for studying the larger, more agile larvae 
and juveniles. The objectives of these larger nets are to reduce the degree of avoidance, 
increase the total volume of water filtered, and reduce the amount of plankton from which 
the larger larvae and juveniles must be sorted, One such net has bean designed f W tfre 
National ItoriW Flrtpri**, Service (NHPS) Mrtfine Resource* ' Iteiiliw^gvA^it**^^^^^ ^' 
Prediction (MARMAP} program. The towing frame is the 6-foot IsaaCs-Kidd midwater trawl 
(IKMT) with 2.B9 m* mouth area, uniform 2 nun knotless mesh, with five times as much 



20 



BATHYKYMOGRAPH 



FLOWMETER 



MONOFILAMtNT 
NYLON SUPPORT 



I.2M V-FIN DEPRESSOR 
( > 3.0 KNOTS) 



(9MM. DIAM 
CHAIN OR CABLE) 




45 KG. DEAD WEIGHT 
DEPRESSOR 

( < S.O KNOTS) 



EYE ft THIMBLE 
2 WIRE STOP 

SHACKLE 
SWIVEL 



EXPLODED VIEW OF 
CABLE ATTACHMENT 
FOft LARGE BONGOS 






21 



iw 3*3 



(Modified fre Kraawr 




Figure 2.14. Plankton volume <tata sheet* 

filtering area as mouth area. The present (1973^ use t> the'n^t is 

from 200 m to the surface filtering 13,000 m3 of water at 4 knots towing . 

duration of the tow is 3& minutes. All samples are take* ****&**& . *,* 

mental work is being conducted in conjunction with a 7S ,** o tt*t .,* 

mesh to discern the uppe* limits of effect!^ sailing * "I* **J 

in the 6- foot I KMT, The MARMAP IKMT is *Qt designed for openinf 

The Tucker net, with square mouth opening,, C*n be modified into 

closing micronekton net (Gavies and Barhaja, 19^9), 

2.2. Laboratory Procedures 

This section describes laboratory ta^hniqws employed to 
samples, collected specif icalj.y for fish mgg*+n& $***" *" 
biomass estimates. D*tirmini*i|F the number *na Iciiid of __ 
in each sample is of primary int**est. A i*t?ond*rjr intere*^$s _ 
constituents as they wlatf to the well-being ** *** young jtag^^l 
as prey (food) or predators. The samples rtay WfP K-fc *^ **i* > fei 
the taxonomy of the vfcribun oonstituints / plpijteton^ 
ganisms, zoogeography, lif* history fCwieS^ etq f ^H 
this manual* 

2*2.1. Plankton 



v'V^'^' 





? which are 

...... ' 



A meamurement of wet plankton voiuame, 
each plankton sample o5i||3fg^^ 

for each oruiw ?^,f JafflraE.K^ 

meaaurement provi^tti '^IPtg^^ 

It also has a prao1btc^* y 1fr^^^ ' * """ *"^r' \. ' ' 

sanples may have to be aliquoted for sorting (Section 2.2.2.) and the eis* of the 

aliquot wil?. often depend on sample sixe. 

Two volumes (Fig. 2.14) are reported for each sample. fb total volume includes 
everything in the sample except adult fishes, juvenile fisHes, large* s^ni^ w 
and ilult pelagic crabs (such as Pleuronoodes ) which are not considered planktonic. 



22 




iH 0) 
OiO 

id 

"0,** 

w 

a s | 

M -p O 
O GU.C 
M 0) Oi 

?r 

4J O ^ ^ 
M K 

^3 

^ Q> W *H 

H -H CD C 
M 0) O 

Si! ? !ii 

P *H H 



0) C 

M ^*H 



0. H 





23 

Th* total volume minus large organisms (Fig. 2.14) is the above volume minus the 
volume of large planktonic organisms such as large jellies or tunicate* whose indivi- 
dual volume exceed 5 ml . ' 



Tfce process of determining *wet" plankton volume by displacement is rather 
simple. The preserving liquid is removed from the plankton by pouring the sample 

' - a tfraihing cone, constructed 6f 0.333 mm Nitex. the planktba i* r^t#i*i*4 in 
th* cone until drainage of liquid from the cone diminishes to eUi < tkwimsi'oti*! ,'iJWp^ ' i : 
Th^ volume of the drained plankton is then determined by displacement in a gc " ' 
cylinder. Bven after drainage, the sainple will #et*ift a 
stltial li*i^ usually between 30 to 40 p*roet of the v 
AKttif and fl&railkill (1963). Usually no alldrfitice is made 
uidl rather the total displacement- volume of the drained *wet* sample is 
The sample is returned to the original preserving liquid, 




Additional information recorded on the plankton-volume data atieet 
cludes station identification, the number of jart containii^ tii aafl^le, th 
tions of fractioned samples, and the initials of the persot* who voluaied the 
Alo, this data sheet is initialed by the sorter when the Sample is che<*ed^iit 
be sorted (Section 2.2.3.). 

2.2*2. Percentage of the Sample to be Sorted ~ 

It is recommended that total samples be sorted for fish eggs and larvae whenever 
possible, and that fractioning of samples be limited to those containing exception- 
ally large numbers of eggs and/or larvae or to samples with exceptionally large 
volumes of plankton material . 

Several considerations should be weighed to determine whether an entire sample 
should be sorted for fish eggs and larvae, or only an aliquot. At times, it is 
desirable to sort only an aliquot for one category, usually eggs, and to sort the 
entire sample for the other, usually larvae. Sample size may or may not be a primary 
consideration for determining whether or not a sample should be aliquoted'or oom- 
pletely sorted. 



Eggs are often more abundant than larvae, and for many species are more '' 
gated so that occasionally, very large collections of eggs of the same species 
obtained. For samples containing large numbers &( eggs, it is usually advisable to 
split the sample into aliquots when sorting for eggs. The Folsom Splitter (Fig. 2415) 
(McEwen, JohnAon and Foisom, 1954) is a standard Apparatus for dividing plankton 
samples into aliquot portions. V , 1 

- < ..... * -I ** ^' ' >,' . i .' -.''., '-., /,\'^,"'' 

There are other considerations in sorting for Inrvae. One 6f the prlmairjr 
reasons far investigating larvae is to obtain information on the success of survival 
during the larval period. Such studies are postulated on the premise that the entire 
size range of larvae can be well sampled. Inasmuch as larger larvae may represent 
only a small percentage of the total number of larvae obtained of a species, it is 
usually necessary to sort entire samples in order to get an adequate. representation 
of the larger- si zed larvae* 

2.2.3. Sorting Fish Eggs and Larvae from Plankton Samples 

A sorter checks out a sample by initialing and dating the plankton-volume data 
sheet (Fig. 2.14). At this time a plankton sorter's work sheet (Fig. 2.16) is. 
started which records the number of eggs and larvae removed from the sample. (Also 
refer to Section 2.2.4.). . .. . . - ^ 

Sorting is probably the step fella L requires the most tiiae, e.g. , one day per sam- 
ple in the CalCOFI program* Before a sample is sorted , the preserving liquid (5% 
buffered Formalin) should be drained off because the formadehyde fumes, when 
breathed, can be irritating, evn injurious to health. The sample can be sorted in a 
medium of f remix water or very weak Formalin solution. Precautions must be taken to 
prevent deterioration of the plankton material. If a sample has not been completely 
sorted during the day it was started, the unsorted plankton should be put back into 
51 buffered Bormalin 4uring (the night. 



24 

The bes^ type of containers to use for sorting are small, e.g. , Syracuse watch 
glasses or divided ,Betri dishe (Fig- 2.1Z). They permit cleaner, ie*tfor* com- 
plete, sorting of fish eggs and larvae than is possible from larger containers* 
Borgorov's counting tray is also widely used (Newell and Newell, 1963). 

Sorting is usually done under a directing microscope (Fig. 2.17) k at a raagnifi- 
o| about p^n feie* >(1 Ox ocular jt Ix objective) . It is inadvisable to try to 

' unaided eye. ,/ /'/v' : ' ". < . / ,"" ' 



h apJLe t0 fee porjbed i* poured into on of the glass 
Its cpntwts are examined clowly under the microscope. All fish eggs and 
^^ a4 fine ^paJLity ftainle^s steel forceps, counted. 



^jC'&jUa** ill a^rcraiately labelled dishes flpjug, |.,174. The degree to which eggs 
and larva* are identified at this tie, Will depend on the objectives of the survey 
arid the training of the sorter (Sectio^* 2.2.4., fe2.5). After the eggs and larvae 
have bn aorted from a dish, its remaining contents are poured into a beaker lab- 

*" ^ ^IS^^tad 11 . thli* process is repeated uftt|,l the entire sample (or aliquot) has 

' * completely* t^e remaining plankton is then replaced in the original jar. 

larvae removed from the sample are recorded on the 

A 




WWL ORI6. va. MINUS 10. OR6. 



OAfE CaLECTH) ^ ~ 
SOUTH*: 

DATE STATED: 
LEFT TIME 



EEICCWEOC WXUME PEWON SORTING 



tNeh at ovrt4 condition^ etc.) 




nm 

IHOLE 

H0 JttCTfrflNgi 1A1L SCCTICWS: 



iBmi3Ni vT 7 ";- tan. SICTIONI 

,ifc 




1 ' ' '^'''j-. MI,''\'^>" ' ; /y. l .^i* ' l^V^vfr , J 



Figure 2:ir idbtiryoplankton sorter's work whaet Plrd WMiiMiB< jHt al. , r 



25 




Figure a, 17. A laboratory sorting table. (From Kraaar t al. , IS72) 



26 



CR 720 333>STA77vW 




plankton sorter's work sheet (Fig. 2.16.)* 
The eggs and larvae are then stored in as- 
parate appropriately labelled (Fig. 2.18.) 
vials. For most samples, 2-dram (about 




j,l,8. . Trfitytl VHMJI 
of .ort*d and itj*tttifd 

' --'y^"''t ' v/f*/ 



preservative added to the larvae if 
cent buffered Formalin in tap water, 
has proved to keep eggs and larvae ia good 
condition indefinitely* The type 
cap reooniwmded for the 2-dr*m gl< 
it ttatto *4$ft * vtny^ i******* tliat 
sel f -seal ing wheia . firmly tightened on 
yial (Fig, t;l) . tht* tor** cap 
preferred to corks r rtibber stoppers, or 
screw cap* lacking the vinyl insert be- 
cause it retards evaporation. 



sorting should be to remove 100% of the fish eggs and larvae. ^ 

._ to establish checks on the completeness of sorting samples. This re- 
lifcher sauplas t **i*cted at random be completely re-sorted by another #*r- 
be devised &MT the routine rechecking by a second sorter of each sam- 
Notes It is firmly recommended that the sorting of a sample for fish , 

! be completed before sorting is commenced on other taxa in the sample, 
the other taxa can be used as a means of rechecking for ichthyop lank ton* 

Preliminary Identification, Enumeration, and Measurement of Fish Eggs 



jjginai 
Larvae 



and Larvae 

technicians who sort fish eggs and larvae from other plankton constituents also 

to recognize and separate eggs and larvae of selected families of 
as clupeids, engratilids, scorabrids, scomber esoc ids, carangids, etc. To 
of same -fishes, the technicians will have to make measurements of dia- 
*nd -.'of -tMtv tftl globules, if present. Technicians, i.e., sorters, 
ow WjftiiM^^ etigrwalid larvae from clupeid larvae; howerver, 

they st*&uld ot be expected to make subtle distinctions among clupeid larvae, for 
exa^le, OT aong engrmilid larvae when several species of either family are repre- 
sented in the collections This is a task for ichthyoplank tologis t s . 





and 



selected species are identified at the sorting level, 
. . recorded (Fig. 2.16) separately with the ren 
in the <dategories OFE (other fish eggs) and OFL (other fish larvae^ 



larvae , of the e- 
lected species may be mMflured 'and the 
lengths recorded on * tabulation sheet 
(Fig. 2.20). The larval length (standard 
length) is estimated to the nearest half 
millimeter. The larvae are measured under 
a microscope by passing each Individual 
over a glass slide which has a transparent 
plastic rule taped beneath it. * 

After the data on preliminary identi- 
fication, enumeration, and measurement 
haVe been recorded on the proper data 
sheets (Figs. 2,16, 2.20), the eggs and 
larvae are bottled for storage (Sect. 
2*2.6.)* An appropriate label (Fig* 2.18) 
is placed on each vial and the sample is 
kept together as a unit. Identifications 
made at the sorting level are rechecked by 
more highly trailed "identifiers" to as- 
sure quality. After all the samples 
frort a cruise have been sorted, a plank- 
ton sorters' master sheet (Fig. 2.21) is 
compiled. 




Figure 2.19. Two-dram (/ ml) 
glass vials and plastic caps used 
for storing sorted and identified 
fish eggs and larvae* (From Kra- 
mer et al., 1*72) 



27 



TPML 



/./2 

tTCa.f. 
>WtX 



Figure 2 
ulating 
larvae. 
1972) 



20. Wbrk sheet for tab* 

surements of fish 
(Prom Kramer, et al., 



2.2.5. Final Identification of Fish Eggs 
and Larva? 

The thoroughness with which collec- 
tions of fish eggs and larvae ate to be 
identified depends on the objectives of a 
survey and on personnel available for the 
work. The subject of identification is 
outside the scope of this manual. A biblio- 
graphy of current research in this field is 
being prepared separately (FAQ Fish. Circ. 706) 

If surveys for fish eggs and larvae 
are designed to deal with only one or a 
few species whose eggs and larvae San be 
readily identified, the identification 
problem discussed below does not exist 
and the task of identification can be 
assigned to easily trained technicians 
who also do the sorting and enumerating 
(Sections 2.2.3. and 2.2.4.). 

Often the problem of identifying fish 
eggs and larvae is not this simple* In 
many groups it is difficult to distinguish 
between larvae of closely related species, 
as for example between the larvae of alba- 
core tuna (Thunnus alalunga) and yellowf in 
tuna (Thunnus albacares ) (Matsumoto et al.. 



1972) or between larvae of clupeid species 
belonging to the same genus, such as Sar- 
dine lla (Fagetti, 1970), or between most 
species of engraulid larvae, or between 
herring (Clupea harengus) and sprat ( Sprat - 

tus sprattus) , etc. The problem is further complicated if all fish eggs and larvae 
are to be identified and enumerated. In most situations, the problem of the initial 
identification of fish eggs and larvae should be entrusted to competent ichthyoplank- 
tologists. 

The development stages of pelagic fish eggs and larvae have yet to be described 
in literature for a majority of marine fishes. The state of our knowledge is not 
as "primitive", however, as the above statement would seem to Indicate. Early life 
history stages have been described for perhaps less than 10 percent of marine fishes, 
but, fortunately, these belong to a large number of marine fish families. By using 
information already available in literature, the majority of fish larvae (say 95 
percent) can be identified to the family level. Scorpaenid larvae, for example, 
can be readily identified to the family level; life history stages have yet to be 
published, however, for roost genera and species of Scorpaenidae. Hence, working Out 
the life history stages of the various genera and species within a family becomes 
the primary task of the trained ichthyoplanktologist. The quality of literature 
dealing with the life history stages of fishes ranges from excellent to miserable, 
with a much higher percentage of poor contributions than good ones. 

It is important that ichthyoplankton scientists be generAiists ^as^rell as spe- 
cialists, i.eT, that they have the training and knowledge to identify fish larvae in 
general. To be a generalist, there is no alternative but to work on identification 
of all constituents (eggs and larvae) from total samples. 

Jhe egg and larval stages of some groups of fishes are particularly 
to identify, uch as those of scombrids (particularly tunas , cluaeids. 
Those of some othet groups are difficult because they contain *Ttt ~wCT 
species, such as scorpaenids, serranids, carangids, !&"' para lep idids .etc. 

- , - . . i needed to work out the life history stages Of .all <Ufi$ 

an In^dif f icult group is facilitated if comaratW tattjip^ 
areas or oceans. There is a great need f or \oc*gft*toiajR Q t 
"a "worldwide basis, in order to prevent duplication of effort, to 



28 




Figur 



ort*r l 



jt ml., 1972) 



exchange of comp*rativ materials, to .tinulate work on neglected woupe, 
and to e*teoupiae ae-wlopm^nt ;of .new 

' ''' 



JciiwJa of fish ems and larvae ar* HtfiCtlt ti iaamblly to onue or 
of fiho f em and lrvl characters CM be nprior 
idnti*l<tion to tke 
vcto|*iia latvke (P*rt%V 
Ahl.tron, 19T2) / bathyla^fl 999* and 
lrvm (Bge , ' 



lv Mir|M of 0oZlection Of fih gg ahd lrv ha* bun 

**^ 2 fea ? lb f' i h i t*y*tlS1bion (SlC0m., 
n f the deciiioh was made at the initiation of th* 



29 



RB* 

i 



*J 

M 



A 



i* 



i -J 



o- ^ 



JL i 



M- 
*T 



to 

rv 






31 






1- t i? ^ j *; ^ \ 

7%w-i* '^?* 

it ^ H r<^* ^^ 

J ^ *>- 

Z -n T5 CJ *i, ^> * 

* ~ x^^X * 4 ix 

f *i O 



jjt ** f~ 



s 



a 

4J 



1 

! 

c 

2 



30 

to identify all constituents to at least the family level , and all common kinds to 
genus or species. When this is done, the person making the identifications records 
the data on a special identification sheet (Fig. 2.22.). The collections so treated 
to date are in. the neighborhood of 30,000, a clear demonstration that this approach 
is feasible. This information has permitted a critical evaluation of the fishery 
resources of the California Current region. It should be noted that some species are 
dealt with in more detail than others. For example, larvae are measured routinely 
for only a few important species, i.e., Pacific sardine, northern anchovy, jack mack- 
erel, Pacific mackerel and recently Pacific hake. Host of the CalCOFI egg and larva 
data have been put oh tape for computer handling. Standardization of data is now 
done by computer, as well as summary tables, distribution charts, statistical tests, 
etc. 

2.2.6. Bottling, Storing and Cur a ting Identified Collections of Fish Eggs and Larvae 

Jtfter a collection of fish eggs and larvae has been identified, it is usually 
inadvisable to try to bottle each category of fish eggs or larvae separately. Hence, 
a decision has to be made as to which categories should be bottled separately, and 
which kept together as a unit. 

X& is advisable to establish a reference collection of identified fish eggs and 
larvae which would include not only eggs and larvae of important species, but eggs 
and larvae of as many kinds of fishes as can be identified with certainty. Specimens 
selected for the reference collection are, of course, bottled separately. The ori- 
ginal identification sheet (Fig. 2.22) should note all specimens of eggs and larvae 
that are separately bottled for placement in the reference collection. Within the 
reference collection, eggs and larvae of fishes belonging to the Same family are 
kept together as a unit, arranged alphabetically by genus and species. A phylogene- 
tic arrangement of families can be utilized, or an alphabetical arrangement of fami- 
lies. The former is the more convenient arrangement for trained ichthyologists, 
the Utter for persons with less scientific background. 

Collections of eggs and of larvae of species under major study should be bottled 
separately for ready retrieval if necessary. The eggs are usually "staged" and/or 
"aged 11 ; larvae are measured. Sometimes food studies are made on gut contents of 
larvae, etc. 

Eggs and larvae of species other than major species that are not separately 
bottled for the reference collection are best kept as units. Identified collections 
should be stored in* an orderly manner, so that any given collection can be retrieved 
easily. For storing the 2-dram vials of eggs and larvae, a covered cardboard box 
having the dimensions length 19 k cm, width 11.5 cm, depth 7 cm is quite ser- 
viceable. 

Samples stored in the above manner must be curated to prevent the samples from 
losing liquid by evaporation and eventually drying out. For ready retrieval of 
collections of fish eggs and larvae, such a storage system is most recommended. 
The eviration time is well worth the effort. An alternative is to store' samples in 
a bath of 3 to 5 percent Formalin. When so stored, the individual vials are usually 
stoppered with cotton* Perhaps the largest collection so curated is the Dana Col- 
lection (Carlesbergfondets at Charlottenlund Slot, Denmark), which utilizes 12 -cm 
glass jars, 13.5 cm tall for storage of cotton-stoppered vials in a 5 percent For- 
malin bath. This is basically a museum storage technique, aftd cannot be recommended 
for collections that have to be consulted frequently. - 

2.3> Data Summarization ;'" 

This section on data summarization considers only the recommended techniques for 
gathering the data collected on the ichthyoplankton survey for the estimation of spawn- 
ing biomass. The analysis of errors in the process of gathering the data will be 
considered in Section 3 r "Theoretic*! Considerations". 

The end result of data summarization is to standardize the number of eggs or 
larvae in each plankton haul to the number under a unit area (10m 2 in this manual) 
of sea surface* This is a necessary prerequisite to estimating spawning biomass. 
Data forms used to assure good data summarization are shown in Fig. 2.23 and 



31 



tf T X 

-Lor A. 



CftlHSC NUMftC* 

VCSSCL 

DATES 



J5L. 

1L_ T 
Jt-L 



wen [ v rr ' n ynri rrr^;] mm mm pro ir 



. 

a^ 
< r^iii>pji> r^mr 



I I I I h-hH ill I 



if T JL 




8CT 



vcstct <S) ilhtKitf 

Mtrt *w.ija J4 

MCIM tuc 



smnw 



DATI 
t* 



^wictifr 



fltr 



^t 



mUL. 



41 



JJUL. 



-tlu 



1972) 



a. 21. Porsui urd for datm uMnricaticMn. (Modified from Kr^a0r 




s/.r 



39-0 



103 
/Of 



5-V3 



3*1 
3Z7 
3.03 



of 



flowtr 



(From KrMr 



33 




1 






2 



s 



*M 

2 



34 

discussed in Section 2.3.4. 
2.3*1* Flowmtter Calibration 

To determine the standard haul factor for each tow requires that a calibrated 
flowweter was placed in the mouth of each net and the flowmeter nunber and the re- 
volutions made during each tow were accurately recorded (Item 12 on the plankton-tow 
data sheet, Fig. 2.8) . 

The calibration factor (f ) is an expression of the number of meters the flowmeter 
/els during each revolution of the iijpeller (- m/rev) . Diffe*eirt flowmeter a are 
likely to differ from each other in thin factor. The factor will not be the same for 
each towing speed for a single flowmeter. The same flowmeter may change its calibra- 
tion factor gradually or may change it suddenly, if dropped, tor example. For these 
reasons, each flownetes to be Used on, a cruise is calibrated before each cruise, its 
performance is monitored during a cruise, and it is calibrated after each cruise, 
Spare flowmeters are taken on each cruise in case of loss or change in performance. 

To calibrate a flowmeter, it is hauled or pushed through a measured distance of 
water at different speeds. The number of revolutions is recorded seperately for each 
test (Fig. 2.24). It is essential that uniform speed be maintained during each test. 
Calibration speeds should include Speeds slow enough to define the friction point of 
each meter, and fast enough to bracket the range of speeds at which the flowmeter 
will be towed at sea* 

The meters per revolution are then plotted against revolutions per second (Fig. 
2.24) and form part of the calibration curve in Fig. 2.25. This graph, based on two 
calibration trials made before and after a cruise, provides an evaluation of the 
function of the flowroeter during the cruise. In this example, the calibration 
factors (f) are: 

f x 0*150 meter* per revolution f 4 0.147 
f 2 - 0.149 f 5 * 0.146 

f 3 0.148 

Where the subscript number refers to individual ranges of speed) 

1 - 2.50 to 2.75 revolutions per second 4 3.50 to 4.19 

2 - 2.76 to 3.09 5 - 4.20 

3 3.10 to 3,49 ' .. . . 

w>i 

The latter value is good for all average 

towing speeds fro* 0.61 to at least 1.07 f 

meters per second . The lowest usable I 

average towing pfd *or **&s meter is .. 

0.3* Mters per sioond. Plotter * 4 * 

are limited in performance at lower J 

speed* due to prop^rfeMwutte increase in f ' 

friction. The speed at which the meter 1 

stops ie called the f pieties* Jpoiflt, tor ~ J 

the type of pl*n***m tow reooiwended in " 

this manual, flowwttera with friction 

point* at or above 0.55 meters per *ec* 

ond should' be repaired ^* discarded. ..,..,*. .... ^. ., . ......... . . . .. -------- *. .... 

' 



2.3.2, Estimat icfq of Kissinf nowmeter T 

Values Figure 2.26. Begreseion line u*ed 

to estimate a missing flowmeter 

If a f lowweter value (number of revo- reading (circled <Jot) for a parti- 

lutioi^*) is missing from the plankton cuiar f lonitar on a partiottlarr 

tow data sheet (Fig. 2.8), or if a re* cruise, given average tangent for 

corded flowweter value is obviously in- the tow in queation and for othar 

correct, it is possible to make an tows made with the sans water. 



35 



Table 2.1. Worksheet for calculation of the average tangent 
from the wire angles recorded each 10 meters of 
wire recovered (See Figure 2.8) 



tan 8 



1 


49 


1.1504 


2 


48 


1.1106 


3 


47 


1.0724 


4 


46 


1.0355 


5 


47 


1.0724 


6 


48 


1.1106 


7 


48 


1.1106 


8 


47 


1.0724 


9 


47 


1.0724 


10 


47 


1.0724 


11 


46 


1.0355 


12 


46 


1.0355 


13 


48 


1.1106 


14 


48 


1.1106 


15 


46 


1.0355 


16 


45 


1.0000 


17 


47 


1.0724 


18 


48 


1.1106 


19 


48 


1.1106 


20 


46 


1.0355 


21 


45 


1.0000 


22 


47 


1.0724 


23 


49 


1.1504 


24 


48 


1.1106 


25 


46 


1.0355 


26 


45 


l.OOOO 


27 


46 


1.0355 


28 


47 


1.0724 


29 


45 


1.0000 . . 


30 


50 


1.191$ 


SUM 




32.2051 


AVERAGE 




1.0734 



estimate of the missing value. The estimate is based on the average of all the Wire 
angle readings made during the tow in question. 

If a ship tows somewhat faster than usual, the wire angle will tend to be higher 
and the revolutions per second will also respond with an increase_in rate. Conse- 
quently, a linear regression (Fig. 2.26) of the average tangent (T) (Table 2.1 and 
Section 2,3.3.) and revolutions per second (u) based on othef tows made by the same 
flowmeter may allow an adequate estimate of the number of revolutions for the tow 
in question* The precision of the estimate will be in part dependent on the number 
of tows taken with the same meter, and also on the technical quality of the tow re- 
cords . 

The least-squares method is used to calculate the regression line* The Indepen- 
dent variable (T) and the dependent variable (u) are listed in pairs (Table 2.2)> 
The regression line, in this example, is described by the equation v 

u * 2,99$ T + 0.874 (r - 0.87 r r 2 - 75%) , / 

If the average tangent of the tow in question is 0.971 (Table 2,2), the estimated 
value of u is 3.79 revolutions per second. '/,',' 



36 



2.3.3. Equation of Standardization 

Tfre basic equation qf standardization is 

C - 10 (a~Vc d) 

"C" is the nufnber of eggs or larvae beneath a unit sea surface area (10 square meters 
in this case); "a" is the area of the mouth of the bongo net in square meters; "b" is 
the length of the tow-path in meters; "c" ifc the number of eggs or larvae in the 
sample; and "d" is the maximum depth of tow in meters* 

The value "a" is derived from the .equation for a circle which is 60 cm in dia- 
meter (0.3 m radius); 2 

a ir r 

- 3.141 (0.3) 2 

0*2827 

The value "b M is derived from the calibrated flowmeter: 

b f r 

"f" is the calibration factor in meters per revolution (m/rev) for a given flow- 
meter at a given number of revolutions per second (Section 2.3.1.); and "r" is the 
number of revolutions Of the flowmeter during the tow. For example, in Fig. 2.23, 
Set I, Line 6 b s 147 ( 5260 ) ' , 

- 773 

The value "d" is determined from the tow data by the equation; 

d = W cos (tan*" 1 f) 

"W" is the maximum length of wire out in meters (m) ; this is usually determined 
by a meter block in the towing winch system .(Section 2.1.2.2.) and is recorded on the 
tow data sheet (Fig. 2*8); "T M is the average tangent of the wire angle taken at 30 
second intervals (Fig. 2.8) during the recovery phase of the plankton tow; 

n 



" 1 ftl 
1 



tan 
n 

and H 9* is the angle at each reading of the wire from the plankton tow data sheet 
(Fig.* 2.8, Table 2.1). For example, (Fig. 2.23, Set I, Line 6) 

d = 300 (0.664) 



for thi* ampl* (Fig. 2.23, Set I, Line 6) "a* is 0.2827; "b" is 773; 
and ** is 199* If 50 larvae (c-60) were taken in th? sample, the solution to 
the ^guAtion would b ' 

c " 10 



455 larvae per 10 m 2 sea surface. 
* Data Forms Used for Standardization 

'-"'--'" , '-"" ' 'J' . '- . ,.i :,.-ui.. "-.-- -r-.----u- .-^ii - .-;.. -f.i_:.- ... mil MI ^ , , | ( ^ _ _ i J, 

The derivation of .the equation of standardization (Section 2.3.3.) is hot neces- 
sarily the same way the data become available or &r needed, for sets of suaanary 
data sheets (Fig. 2.23) are used to assure the completeness and accuracy of organiza- 
tion of , the 4atau -x , .< - - " : - , : ' <; ^ ' 

Set I (Fig. 2.23) summarizes the data from the plankton tow data sheet* Column 
1 (in all four sets) is the station number. Column 2, which gives the order in which 
the station was occupied, comes from the "Captain's sheets 8 (Fig. 2.2) . The order 



Table 2.2. Worksheet for estimation of missing flownuter reading, 
by the method of least squares (regression) . 



TOW 
ORDER 


AVERAGE 
TANGENT 
(T) 


REVOLUTIONS 
PER SECOND 
<u) 


1 


1.243 


4.76 


2 


0.971 


3.79* 


3 


1.003 


4.10 


4 


1.067 


4.01 


5 


1.095 


4.13 


6 


1,021 


3.86 


7 


0.994 


3.98 


8 


1.056 


3.78 


9 


1.199 


4.39 



occupied is used to trace back faults in flowmeters (e.g, malfunction) and nets 
(e.g., clogging) which affect entire sequences of tows. Column 3 is the duration of 
tow (towing time) and is recorded in minutes and seconds on the tow data sheet (Fig. 
2.8). Column 4 changes this value to seconds. The number of revolutions of the 
flowmeter is taken from the original tow data sheet and recorded in column 5. Column 
6 lists the revolutions per second which is calculated by dividing the value in coX* 
umn 5 by the value in column 4. Revolutions per second is used to determine the 
calibration factor for the flowmeter (Section 2.3.1.). Column is also monitored 
for high values to detect a cause of extrusion of small eggs and larvae. Column 7 
is the average tangent derived from the original tow data sheet by converting the 
angle readings to tangents, adding them, and dividing by the number of readings (Ta- 
ble 2.1.). This value is rounded off to three places. It is used to determine the 
maximum depth of tow. The average tangent may also be used to estimate missing read- 
ings for a flowmeter which has been used for a series of tows on the same cruise 
(Sect. 2.3.2). Flowmeter calibration factors (Section 2.3.1) are recorded in column 
8 and are read from the calibration curve (Fig. 2.25) derived for each flowmeter for 
each cruise at the rate of speed in column 6. In column 9 the mouth area of the net 
is multiplied by the calibration factor for use in determining the volume of water 
strained (filtered). In column 10 the volume of water strained is obtained by multi- 
plying column 9, which is in m 3 per revolution, by column 5, which is in revolutions. 
In column 11 the cosine of the angle tan~* T is recorded for use in estimating the 
maximum depth of tow. This value provides a better estimate of the maximum depth of 
tow than the cosine of the angle with maximum wire out. Column 12, wire out is ob- . 
tained from the original tow data sheet and is multiplied by the cosine of the 
angle tan* 1 T (column 11) to estimate the depth of tow (column 13). Column 14, the 
standard haul factor, is obtained by dividing column 13 by column 10 and multiplying 
the answer by 10. The standard haul factor multiplied by the number of eggs or larvae in 
the sample gives the number of eggs or larvae per 10 m^ of sea surface. 



Set II (Fig. 2.23) takes the information in Set I and 'the plankton sample 
placement volume (Section 2.2.1.) and corrects the values to cubic centimeters of 
plankton per 1000m 3 of water strained. The values for volume of plankton may be used 
to estimate the sorting time for the sample. 

Set III (Fig. 2*23) is the summary of plankton volumes in a form which may be 
published. It lists the actual station position from the "Captain's sheets" (Fig. 
2.2) and the mid-time of tow from Set I. Since the normal tow is about 20 minutes 
long, the data from the original tow data sheet (Fig. 2.8) are only recorded to 
within 5 minute accuracy. This information is used primarily for the study and cor- 
rection of the day-night difference in plankton volume and numbers of larvae. Other 



38 



Table 2.3 Basic data table for assembly of regional data to be used in making 
census estimate*. Numbers represent sardine larvae per 10 m 2 sea 
surface (C , Section 2.4) for each station on each cruise, M O* indi- 
cates no larvae of this species were taken. "-0" indicates the station 
was not occupied. 



STATION 


CRUISE 




01 


02 


04 


05 


07 


10 


3.1 


5 


356 


83 


3 


9 





3.2 


-0 


96 


5 





22 





3.3 





14 


36 


6 


21 





3.4 














3 





4.1 














3 





4.2 


-0 





6 


3 


35 


3 


4.3 


-0 





43 


- 


90 


13 


4.4 


-0 


-0 








12 


; 


5*1 


-0 








3 


16 





5.2 


3 


49 


3 


6 





5 


5.3 








4 





3 





5.4 








4 











NUMBER OF STATIONS 
OCCUPIED (N ) 


7 


11 


12 


12 


12 


12 


NUMBER OP STATIONS 

WITH LARVAE (N. ) 

** .* 


2 


4 


8 


5 


10 


3 


NUMBER OF LARVAE (1C) 


8 


515 

* 


184 


21 


214 


21 


rc 2 


34 


138,549 


10,136 


99 


10,758 


203 


tlmC 


1.176 


7.370 


8.267 


2.988 


10.832 


2.290 


X (1J*Q Z 


0.716 


14. 10 


10.820L 


1.894 


13.962 


1.957 



4) > 
> * 



<D 
T3 



5. 
H S 







CM 



CO 6 



m 

4* 
H 10 
-UTS 



-P -H 
10 tt 
-M (0 
CO JQ 



CM 

s 

2 



O 

3 



o 

M 



W 

a 



A so ^ r* r^ en o 

OD vc ^D rH /n in 

CM CO tf> ^ 00 CM 


en o vo CM ^* in 


^H CN rH rH rH rH 

r vo o in oo cn 

m oo r** vo o^\ CM 

_ rH in in rH <* m 

a 

rH 
CO 

00 *^* ro 0^ 00 

in oo o in o 



rH rH rH 

rH 00 * ^- O 

U *^ rH O VD CM CM 
CO 

rH in e^ iH VD m 

m CM CM 

in o o o o 

o r* o CM * o 



CM CM CM 



00 in ** rH ^ rH 
rH 00 CM H CM 
in rH CM 



CM ** oo in o 



!* rH CM CM CM CM 



rH CM <* in !- 
O O H 



<c 
o 





M 
Q) 



<0 
10 



B 



CO 



IU 



z 



^* m u> 
n oo CM 



fM 



oo ro 

Ch IS 



*r u"> m 
CM ^r oo 



r* vo o o> 

m rH CM CM 



m tM CM 



*n 



r- in 

00 rH 



en in * rH 

CM O rH CM 

in CM CM 



ID *> O 

H H 



00 ^ CM CM 
,H CM iH H 



40 

data columns are from Set I and Set II directly. 

, Set IV (Pig. 2.23) is used by the fish egg and larva "identifiers" to record 
sample data on the identification sheet (Fig. 2.22), e.g., standard haul factor, 
mid- time of tow, and percent of sample sorted. 

2.4. Census Estimate (Larval Index) 

To arrive at final census estimates, the spawning area and the survey season are 
divided into units which serve as pooled data areas. This allows a conservative 
estimate of precision and furnishes a concise summary of each year's spawning acti- 
vity within each region. Data from the pooled area may furnish insight into changes 
in the allocation of sampling effort to maximize precision per unit cost for a tar- 
get species (the primary species under study). 

If it has been determined, for example, that the sample survey will be conducted 
over the four quarters of the year (1985) with two cruises each quarter in the 
spawning season and one cruise each quarter out of the normal spawning season, the 
stations close to each other may be grouped regionally and seasonally. If, for this 
example, the region of pooling (region #5) is 120 by 80 nautical miles and contains 
three lines of stations with four stations distributed along each line, the 
data for fish larvae may be assembled as shown in Table 2.3. At the base of each 
column, several sets of numbers have been calculated for analysis and inspection. 
They are the number of stations occupied (N) , the number of stations at which sar- 
dine larvae occur (N, ), the total number of larvae taken in the region during the 
cruise N - 

V 

C 1 

the sums of the squares of the numbers of larvae 

N 



the sums of the logarithms of the numbers of larvae 

N 



I In C * * In C 1 
and the sums of the squares of the logarithms of the numbers of larvae 



rOn 



C) 2 -i (In 



It is essential that these basic data are available for inspection so that questions 
arising in the interpretation can be traced to the original data. 

These basic data (Table 2.3) are then summarized (Table 2.4). The original 
quantities N, N L , and C are retained and the following quantities are calculatedi 

K . 1C 

* *T 

The mean number of larvae per positive station; and 

,; ; Sc- KEC 2 -N^C 2 )/^- 1)] J 

the standard deviation of the number of larvae pet positive station; "IE?" the mean of 
the log of lirvfce fter positive station; "S i n c tf the standard deviation of the logs of 
larvae per positive station;' and "F" the error factor for the positive stations. "P" 
is the proportion of the stations which are positive (an estimate of the proportion of 
the sea surface underlain by larvae); and, "I", the larval index, is the regional census 
estimate of the number of larvae in this time period, which in the given example is 

L 3,29 x 10 9 (P C") 

where the factor 3.29 x 10 9 represents the number of 10 m 2 areas within the pooled region 
which in this case is 120 x 80 nautical mills. 



41 



Tabl 2.6. Total urvy summary (LxlO 9 ) by yar and region*. 



YEAR 


REGION 


TOTAL 


1 


2 


3 


4 


5 


6 


7 


8 


9 


1978 


84 


321 


204 


439 


250 


131 


137 


269 


119 


3193 


1979 


37 


179 


127 


497 


381 


355 


254 


129 





1959, 


1980 


5 


216 


57 


430 


69 


47 


735 


147 





1706,- 


1981 


18 


226 


4 


579 


24 


3 


268 


15 





1137 


1982 


161 


173 


6 


564 


3 


o 


454 


25 


67 


1453 


1983 


128 


90 


4 


288 








252 


160 





922 


1984 


187 


120 


70 


374 


181 


5 


406 


152 


11 


1506 


1985 


36 


35 


6 


99 


189 














Table 2.7. Worksheet for comparing an independent biomass 
estimate with the larval index. 



YEAR 


BIOMASS 
ESTIMATE 
x 10^ Tons 


LARVAL 
INDEX 

x 10* 


1978 


1336 


3193 


1979 


850 


1959 


1980 


586 


1706 


1981 


424 


1137 


1982 


562 


1453 


1983 


380 


922 


1984 


(593)* 


, i ' f ' '* r , ' 

' ' 15i0< - 



Estimate based on larval index 



42 




Many of the factors displayed in 
Table 2.4 are used to assist in the eval- 
uation of data* The cruises In this ex- 
ample dp not represent equal periods of 
time, therei?bre, a furtnet pooling of sta- 
tions in tine must be accomplished for 
the first two pairs of cruises. In Table 
2.5, the two winter and spring cruises 

* pooled and th* regional census 



I 2 

LAfWAL INDCX M 10* 



estimate of the number of larvae in the 
entire ragion in this time period has 
been obtained. This census estimate is 
ah; index of spawning intensity and cover- 
age for this year in this region. 

The larval index for region 15 in 
1985 (total in Table 2.5) has been incor- 
porated into the total survey summary in 
Table 2.6. This larval index can be 
used by itself to monitor changes in 
spawning biomass. The reliability of the 
larval index for such use is exemplified 
in Figure 1.2 when compared with two other 
independent estimates of bi omasa of the 
Pacific mackerel. The larval index can 
also be used in conjunction with other 
data to make instantaneous spawning bio- 
mass estimates. (Sections. 2.5, 3.6.) 



2.5. Estimating the Spawning Biomass 



Figure 2.27. A scatter-plot of 
the relationship between spawn- 
ing biomass and the larval abun- 
dance index. Numbers indicate 
year., e.g., 78-1978, etc. Qne way ^ estimte the 8pawning bio . 

mass of the current population is by using 

the larval index of abundance. Other methods of making spawning biomass estimates 
-from ichthyoplankton surveys are discussed in Section 3.6 along with problems invol- 
ved in using them. 

The method described here involves comparison of the larval index with another 
independent estimate of species abundance. This other estimate could be made, for 
example, by correlating the age coiqposition of the stock with fishing effort and 
thus determining the number of fish in each age class as described by Murphy (196*6) . 
The larval index, wen used in conjunction with Murphy's estimate, can make up for a 
generation of lag time and provide an estimate of the current spawning bi< 



A hypothetical example of this method of estimating biomass is provided in Table 
2.7 and Figure 2,27. In this example, the larval index (total regional census esti- 
mate in Table 2.6.) is available from 1978 through 1984. The other independent 
estimate (Murphy, If 66) of fish biomass (Table 2.7) is available only through 1983. 
A linear relationship exists between these two estimates in the form of the equation: 

B s * 0-436 L - 63.98 (r - .989; r 2 98%) 

"B" is the spawning biomass of sardines, "L" is the larval index, "r" is the correla- 
tion coefficient, mnd "r*" is the first estimate of the degree of dependence of 
spawning biomass on larval index. This equation provides a method of relating pre- 
vi6us quantities ^f* fish biomass with larva abundance in a way that can be used to 
estimate current spawning biomass. Thus by knowing the larval index for 1984 (1506 
K 10), the spawning bioa**s (0.6 x 10* tarfft)**** th*t year can be estlliled. 

Another example is provided in Fi<gu*e Uitv ^'tH* sardine 

fm fi*My;**t* *** * ** *y t*ie Kttrpfcff <l) *t*i<*d 
of abundanoe b*s*d <* ichtfcyoplanktoft survey. w vmil**le fro 

larval il*e*>3^ttili*^ S 

to max. *!. **tto&S*r**t* tho^ tL *t*n*S had 



* larval index 



and 



43 



3* Theoretical Considerations 

Ideally, ichthyoplanton surveys conducted according to the methods presented 
in Section 2 would result in suitable estimates of the spawning biomass of any pela- 
gic spawning fish. Unfortunately, even though the techniques may be standardised, 
the spawning behavior 6f pelagic organisms may vary widely enough to necessitate 
changes in techniques. For example, in individual cases it may be neqessariy to in- 
tensify the icnthyoplankton survey for a particular purpose Or, conversely it may be pos- 
sible to compromise some of the technical quality of the surveys and still obtain 
acceptable estimates of the spawning biomass. While the fundamental reason for 
this section is to provide information to determine individual requirement^ jof stan- 
dard ichthyoplankton surveys, it is hoped that this section will also servkftwo 
other purposes i 1) it should advise on ways in which surveys conducted with non- 
standard sampling techniques can be usefully related to the standard surveys; and 
2) it should provide sufficient information on the sources and the magnitude of 
sampling error so that the survey specialist, with experience and the use of this 
manual and the research papers to which it refers, can make a contribution to the 
solution of continuing sampling problems. * 

3.1. Statistics 

The present manual is not intended to be comprehensive with regard to statisti- 
cal description and analysis of data. However, some subjects will be discussed to 
introduce the reader to particular problems encountered when sampling ichthyoplank- 
ton. Standard statistics texts should be consulted for the general concepts of 
statistical description, testing of hypotheses, and sampling design. For ich 
plankton survey problems Southwood {1966) is recommended. A basic statistical 
description is given by Gulland (1966, 1969). Problems of biases and imprecision 
are treated in the UNESCO (1968) Monograph on Zooplankton Sampling. Sample design 
is treated specifically therein by Cassie (1968). The expense of marine plankton 
sampling and analysis may justify the services of a professional statistician for 
the planning of the data analysis. 

3.1.1. Numbering Systems 

The basic number system includes integers or whole numbers; 

0, 1, 2, 3 f 4, .... infinity (to an unlimited degree). These may also be 
extended to numbers on the negative side of s^iro to infinity. Between any " two whole 
numbers there are fractions in this system which can be infinitely subdivided, e.g.: 

0, 1, 2, 3, 4, .... infinity 
2.1, 2.2, 2.3, 2.4, 
2.21, 2.22, 2.23, 2.24. 

Groups of numbers can also be elaborated to infinity as in the commonly* refer- 
red to "orders of magnitude" system , e.g.: 

0.1, 1, 10, 100, 1000,, 10000, 100000, 1000000. . . 

which in this example extends from one- tenth to one million by orders of magnitude* 
Since this system soon becomes unwieldy, an abbreviated *ysfem qf " exponents" aim, 
be used: t ' _ ' ,,v , . '^ ,^ '..,.,>' -v , 

10* l - o.i io 3 - . lodo . '<* U 
10 - i io 4 10060 

M ID 1 - 10 IO 5 - 10000,0 

ID 2 ' 100 IO 6 - 1000000 ! 



44 



NAUPLW PER SAMPLE 



B 



O 2O 4O 6O 8O (06 i2O I4O ISO ISO 8QO 



NAUPLII PER LITER 



O O 4O 6O 6O tOO I2O I4O I6O ISO ZOO 



NAUPLII PER LITER 



30 35 40 45 50 55 60 65 70 



r ? I i , 



NAUPLII PER LITER 
30 40 50 60 70 



NEW SAMPLES 

. - . m NAUPLII PER SAMPLE 



50 ' too''" ' ' ' 150 ' r ' ' 200 



NAUPLIl PER LITER 



30 40 50 60 TO 






*** " PPUl*tin in 



, in 10 nauplii 

Tha lngth of *ch br i proportional t th * 
number of wunple valtM* within teach intarvml. 



45 



These exponents can also be subdivided like fractions in tfee sy#fce* . 



10 
10 

10 
10 
10 
10 



1.0 



.1-2 . 



.1.4 
1.5 _ 



10 
12.6 

15.8 
20.0 
25.1 
31.6 



10 
10 
10 
10 
10 
10 



1.6 
,1.7 
1.8 
1.9 
2.0 
2.7 



39. ft 

- 50,1 

- 63*1 

- 79.4 

- 100.0 

- SOI 



10 



3.7 



5012 



Approximations of logarithms of numbers and the number approximated by a loga 
rithm may be found in books of mathematical tables. The precision of the approxi- 
mation is controlled by the number of decimal places in the mantissa (the portion 
of the exponent following the decimal point). For example, the text table above m 
be used as one place table of anti-logarithms. The following is a two place table 
of logarithms, showing only the mantissae : 



X 

1 
2 
3 
4 



00 
30 
48 
60 
70 



X 
6 

7 
8 
9 



78 
85 
90 
95 



The characteristic, or order of magnitude must be determined separately , as above. 
(100 i, iol m 10, 10 2 - 100). The characteristic is the power of ten, -and the 
mantissa is zero for all integer powers of ten. The mantissa is used for decimal 
fraction powers of ten. For example, 10- 70 5.012, 10 1 * 70 - 10 x 5.012 50.12, 
10 2.70 m 1Q 2 x 5.012 - 501.2, and 10 3 - 70 - 10 3 x 5.012 5012. Note that the value 
X * has no logarithm. Table 3.5 (below) is an abbreviated table of the mantissae. 
Tables of 4, 5 or 6 place logarithms and anti-logarithms are often available in books, 
and electronic calculators often have the capacity to generate 8 or 10 place equi- 
valents. 

3.1.1.1. Normal Distribution 

Another form of number system is related to the distribution being sampled. For 
example, assume there is a well stirred brood tank of nauplii being reared to feed 
fish. Twenty successive two-liter samples are taken at random from the tank. Numbers 
(listed in descending order) of nauplii counted in each of the 20 samples are: 137, 
128, 122, 118, 114, 111, 108, 105, 102, 100, 99, 97, 94, 91,. 89, 8$, 83, 80, 76, 
69 (see also Fig. 3.1, line A). The ordered counts of nauplii per liter would be 
half the original numbers. On the fractional number scale they would be represented 
as in Fig. 3.1, line B. Only a portion of the fractional number system is necessary 
for this set of samples. On line C (Fig. 3.1), the samples are designated on a number 
scale wiach extends only from 30 to 70 nauglii per liter. Then ail, vmlm* mre t*l< 
to the nearest whole ten, 30, 40, 50, 60, >0, in line D (Pi*- 3,1). C^nt* JMMI 
samples of nauplii from the same tank are first represented by dots on the triple 
number scale (Fig. 3.1, line E) and then classified to "tens" in nauplii per lite* 
(Fig. 3.1, line F) . 



46 



There is a strong resemblance between the 20 samples and the 60 samples with 
regard to the central tendency of both distributions at 50 nauplii per liter and the 
absence of samples at 20 or less and 30 or wore nauplii per liter. If the brood 
tank of nauplii had been completely separated into two liter unite, there is reason 
to believe, on the basis of these two sets of samples, the general form of the dis- 
tribution sample sises in terns of nauplii per saraple or nauplii per liter would be 
the same as the two sample sets illustrated* The only way to determine exactly how 
many nauplii were in the brood chamber would be to empty the chamber and count; each 
nauplius: i.e., take a census. For many purposes the sample information was ade- 
quate and the census would contain little additional information. In many cases, 
like this, a census would be too expensive, unnecessary, or impossible, (For ex- 
ample, the census might kill the nauplii and render them unfit for fish food.) Also, 
during fast growth, the number of nauplii might change during a complete census but 
be stable for most purposes during a short sampling interval. 

The example we have just examined was drawn from a "Normal" distribution. It 
was described by a set of samples which estimated two parameters; one for the posi- 
tion of the center on the number scale and one for the variation of the samples from 
this center. The central position is called the 'arithmetic mean 11 or the "average" 
and the variation of the samples is called the "standard deviation." For the 20- 
sarople series, the arithmetic mean (m) is determined by the equation: 

% X+Y+X4- 4. 

m T 2 L ' *20 

which reads, the. arithmetic mean equals the sum of the first number plus the second 
number plus the third number plus all the numbers up to and including the 20th 
number which then is divided by 20. It is usually abbreviated using the Greek 
letter M E M (Sigma) for the "sum of 11 and the smaller letter "i T? to denote the subscript 
in the following manner: 

20 



which for this sert of data is: 



S0,2jn*uplt1 per liter 

. It is conventional to call this the sample mean "m" and to denote the population 
mean it is intended to estimate by the Greek letter"**" (Mu) . If the brood tank held 
4000 liters, then 2000, 2-liter samples could .be taken and the equation for the pop- 
ulation mean would bet 

2000 



50 naupW per liter 

The second parameter which describes 'the "Normal 11 population is called the stan- 
dard deviation "* M . It tefers to each measurement (x^) to the population M>MI (**) 
The um of the differences between each measure and th* sample mean will be zero; 
thus the standard deviation is calculated fron the square of the values of the 
deviations of individual measurements, or in conventional Shorthand: 




A- t V 



'x 

M 



47 



where s x is the sample estimate of standard deviation of tne variable *x", "n" in 
the saa>le sime, and "m" is the sample estimate of the population wean as auo<W^. - 

,TJift value *n-l" is used as a divisor, because if tf tt fl is used, the sattple stan- 
dard deviation tends to underestimate the population standard deviation. Thi* 
tentenoy A corrected by the use of "n-1" or "degrees of freedom* as the *i**tf&r 
rather -than "n", the number of observations* The value ** 2 is called the sample 
estimate of the variance of "x". ^ ; 

, ' 1 J ft ' 

In the "Normal" population, the "range" (lowest number t?o the highest number) 
is infinite in theory, and sample estimates of the ran^e can only increase with in- 
creasing simple sixes. For this reason, it is usual to seek instead a pair of 
values ifhich encompass a stated proportion (say 95%) of the population* the sample 
standard deviation defined above may be used to estimate pairs of values ifhictt will 
encompass any proportion of the population or sample distribution* Tables fot thi* 

Table 3.1. Normal deviates (Z) and selected probabilities. 





.0 


.1 


.2 


.3 


.4* 


.5 


.6 


,7 


.8 


* 


0. 


.500 


.460 


.421 


.382 


.345 


.309 


.274 


.242 


.212 


.484 


1. 


.159 


.136 


.155 


.097 


.081 


.067 


.055 


.045 


.036 


.029 


2. 


.023 


.018 


.014 


.011 


.008 


.006 


.005 


.004 


.003 


.002 


3. 


.001 





















Table 3.2 Areas under the normal curve. 





.0 


.1 


.2 


.3 


.4 


.5 


.6 


.7 


.8 


.9 


0. 


.000 


.040 


.079 


.118 


.155 


.192 


.226 


.258 


.288 


.316 


1. 


.341 


.364 


.385 


.403 


.419 


.433 


.445 


.455 


.464 


.471 


2. 


.477 


.482 


.486 


.489 


.492 


.494 


.495 


497 


,497 


,499 


3. 


.499 





















purpose are prepared for any situations which require great precision. Abbreviate^ 
tables for learning purposes are given in Tables 3.1 and 3.2 and these may suffice 
for all but the most demanding tests. 

These tables (3.1 and 3.2) are complementary tabulations regarding the "normal" 
curve. !flie basic unit of the "normal" distribution is the difference between; th 
observation and the population mean of all observations divided by the standard 
deviation. JPhe unit is conventionally known by the letter *z* ahd calculated by Jthe 
equation: ' , , '. ' . , , ,.^\ , 






the letter "Xj" represents the variable under consideration, the Greek letiir 
represents the population' mean as above, the Greek letter "cr" (Sigma) is the populs- 
tion ptandaxd deviation, and the letter "Zi* is the nuber of standard deviation uni 
"*!** ii fffP the population mean, I Table 3.1, the <4 Z fl ^afu^of 0.0 re^re^ 
arithmetic mean ox the uistribution. The value at the intercept of the (J. 
the ,0 column is ,500 which is interpreted as mean inf that half tK* '-iaUfiU 
ar* c lbwe J53 "halt the sample values are below the arithmetic *ean. -Thi 
of the 1* .&* ^tdd.thii .5 coiumn if ,097. ?hi i* tdad As 9.71 f -- -- 



48 



are above the aritJwtic ipean j>ius 1,3 tnd*rd deviations and 9*7% of all s*jgle 
values are below the arithmetic mean minus 1.3 standard deviations* In the comple- 
mentary table 3.2, the intercept of the 1, rent and the *3 column la .403. This is 
read that 40-3% of all sample values are between the arithmetic mean and the arith- 
metic ean plus 1*3 standard deviations end 40.3 % of all saaqple values are between 
the arithmetic mean and the arithmetic wean minus 1.1 standard deviations. Note 
that the values of the intercept of 1. row and the .3 column are .09? in Table 3*1 
and .403 in table 3.2 and these values add to .500. All similar intercepts in the 
tiro tabjr*s a*e ooaf>leentary thus either table alone would suffice for tooth types 
of expression. Lastly, asymmetric statements of probability may also be made from 
these tables. For exaaple, the value at the intercept of row 1. and column .6 in 
table 3 .2 is .4^5 and this my be stated* 44,5% of all sample values lie between 
the arithmetic mean and the arithmetic mean plus 1.6 standard deviations. Since we 
know that 50% of all sample values He below the sample mean, this means 94.5% of 
all sample values lie below the arithmetic mean plus 1.6 standard deviations. 

From the nauplii example (Table 3.3) , the mean number of nauplii per liter is 
50.225, the standard deviation is 8.91, the range, 34.5 to 68.5 nauplii per liter, 
and range of deviations extends from 15.725 nauplii per liter below the mean to 
18,273 tiauplii per liter above the mean. Expressed in standard deviation units the 
lowest Value is 1.76 standard deviations below the mean and the highest value is 
2,05 standard deviations above the mean. This is not an unreasonable asymmetry for 
such a small saxrqple. One is justified for suspecting that these 20 samples have not 
explored the entire range of the 2000 samples possible from this tank. One sample 
standard deviation below the mean is 50.225 - 8.91 or 41.315. Sample values x lf 
x 2 and x? are lower than this value. One sample standard deviation above the mean 
is 50.255 + 8.91 or 59.135. Similarly, sample values Xjp, x lg , and x 2Q lie above 
this value. Thus/ 3/20 or 15% of the values lie more than one standard deviation 
below the mean and 3/20 or 15% of the values lie more than one standard deviation 
above the mean. This means that 70% of all samples lie between the one standard 
deviation limits about the mean. 

Table 3.1 at 1.0 2, gives the population value for this as .159 or 15.9% of 
sample values are more than one standard deviation above the mean and 15.9% are more 
than one standard deviation below the mean. This means that 100% - 15.9% or 68. 2% are 
within it one standard deviation unit of the mean. As stated above, Tables 3.1 and 
3.2 are complementary in the sense that one gives the proportion of values likely 
to be outside set standard deviation units, while the other gives the proportion 
of values likely to be within given standard deviation unit boundaries. One should 
become completely familiar with both methods of presentation. Particular attention 
should be paid the value 1.96 standard deviations above and below the mean, because 
this means that 2.5% of all sample values lie outside the upper and lower bounds or 
55% of the values lie within these bounds. This should be conroitted to memory for 
statistical work since estimation, of the chance of being wrong only once in 20 times 
(951) is a coiwoon statistical practice, 

Thu5,. of the number systems mentioned so far, the whole number system, the 
fractional number system, the logarithmic system, and the normal system, the latter 
is the only system which if defined by a population of counts or measurements with 
the central position determined by the arithmetic m*ea and a number in "z fv units of 
standard deviations above and below the arithmetic mean. One property of the normal 
system should be noted: the sample standard deviation is independent of the sample 
mean and thus the two parameters are independent measures of the population. It is 
important to realize that these constraints are not often satisfied in the natural 
distribution of animal populations in the sea or elsewhere, although on occasion 
parts of the sea may approach a state of being mixed somewhat like the tank of nauplii 
in the example. . _, c * , : , , , . ; - - - - ' '' , : 

tapther ejcM^l* o the ncarmal distribution and a simple graphic Method fcr teat- 
ing its utility, may be seen from the data of Prtset (1969) regarding th* degree 
cf penetraticm of ociMUiic plankton into the Worth Sea (Table3.4; Fig, 3.2.)* For 
this example th* ; *tot1& f ea extreme position of the oceanic ^lanktpn was reoor^fl fojr 
32 of the 45 years between 1920 and 1965. These positions were grouped into 
30-mile categories in Table 3*4. At the right margin of the table the individual 



49 



Table 3.3. Ordered set of data from a normal distribution of 
samples of nauplii (an example) . 





NAUPLII 


DEVIATION 


SAMPLE 


PER LITER 


FROM MEAN 


X l 


34.5 


-15.725 . 


X 2 


38 


-12.225 


X 3 


40 


-10.225 


X 4 


41.5 


-8.725 


X 5 


43 


-7.225 


X 6 


44.5 


-5.725 


X 7 


45.5 


-4.725 


X 8 


47 


-3.225 


X 9 


48.5 


-1.725 


v 


49.5 


-0.725 


X 10 






x 


50 


-0.225 




51 


+0.775 


X 12 








52.5 


+2.275 


X 13 








54 


+3.775 


X 14 






V 


55.5 


+5.275 


X 15 








57 


+6.775 


X 16 








59 


+8.775 


X 17 








61 


+10.775 


X 18 








64 


+13.775 


X 19 








68.5 


+18.275 


X 20 








x 1004.05 JE<x 


i-m) n 0.0 




m - 50.225 


md 0.0 




(Xi-m)* - 1*09.2373 



8.91 



so 



a 


* 

M-* 

<O 

CA 



S= 

-*^ 

4^ 
^3 

2*4 

^- 

^ 



r*- csi ir> CM ^ 



vo so 



u-i o u~> <* 



CM 



rsi 



o ro 



<& ^ "0 C9 

** ^^ 3K rS 



at 



51 



group percentages and the cumulative 
group percentages are listed. Special 
graph paper, called probability paper, 
which has the cumulative percentage 
axis adjusted so that a "normal 11 
distribution would be represented as 
a straight line, was then used. The 
axis is numbered from 0,01% to 99.99% 
The cumulative percentages from 17.11 
to 97.1% from Table 3.4 were plotted 
:in position {Fig. 3.2). The fact that 
these points are suitably close to a 
straight line is consistent with a 
hypothesis that the tendency of oceanic 
plankton to penetrate the North Sea is 
a "normally" distributed variable. 
The utility of this graphic method is 
that data sets can be simultaneously 
compared with respect to the 
"normality" of the distributions, the 
similarity of the means, and the 
standard deviations. 

3.1.1.2. Lognormal Pi s tr ibut ion 




9*9 9999 



PROiABIUTY 



Figure 3.2. Penetration of oceanic plankton 
into the North Sea. The data in Table 3.4 
have* been plotted on normal probability 
paper by use of the cumulative percent fre- 
quency of Table 3.4 as the abscissa and the 
nautical miles of penetration recorded l?y 
planktologists as the ordinate. (After 



Fraser, 1969) 

In the lognormal system (Aitchison and Brown, 1957), the sample values vary by 
a series of proportions of the central value rather than by a constant amount from 
the mean as in the Normal distribution. The uses of the central measure and the 
standard deviation are the same: the difference is that the equations are 



log g . log X 1 + log X 2 + log 



log X r 



&-, as before 



log g - IV log X 



'1 



with the difference that the value "g" is the geometric mean of the set of numbers 
rather than the arithmetic mean (m) in the cases above. The sample standard devia- 
tion is similarly determined as above but now refers to a proportion to be multiplied 
or divided by the geometric mean to determine the distribution of values. 



To save the trouble of calculating the &eomeric , 

mean and the ratios of all values and the deviations of the ratios from unity , it is 
conventional to convert all values to logs (any base) and proceed as in the normal 
distribution. Table 3.5 provides a handy method for converting measurements or 
counts to logs. As stated above , if the numbers to be so transformed are between 1 
and A9, f tjie vaiue in the table is used in a base 10 log transformation. If ,the 
numbers are between 10 and 100, the number in the table is preceded by the vaiue 
"l w , and 60 on. for other orders of magnitude. The conventional notation may also 
be used for the sample standard deviation 'of the log of. the geometric mean; 



log X 



(log g - log 



It is im|>ottant to note two limitations on the use of the lognormal distribution: it 
must be demonstrated in each case (as in the normal distribution) that the variance 
is independent of the mean; and, "Lognqrmal theory cannot be applied dirfctly to 
sample whidi contains a zero value. 11 (Aitchison and Brown, 1969, p. 9lK for e* 
ample, in the equation for the mean there is no logarithmic equivalent for x^ - f 



52 



Table 3.5. Three place mantissae for base ten logarithm. 








1 


2 


3 


4 


5 


6 


7 


8 


9 


1 


000 


041 


079 


114 


146 


176 


204 


230 


255 


279 


2 


301 


322 


342 


352 


380 


398 


415 


431 


447 


462 


3 


477 


491 


505 


519 


532 


544 


556 


568 


580 


591 


4 


602 


613 


623 


634 


644 


653 


663 


672 


681 


690 


5 


699 


708 


716 


724 


732 


740 


748 


756 


763 


771 


6 


778 


785 


792 


799 


806 


813 


820 


826 


833 


839 


7 


845 


851 


857 


863 


869 


875 


881 


887 


892 


898 


8 


903 


909 


914 


919 


924 


929 


935 


940 


945 


949 


9 


954 


959 


964 


969 


973 


978 


982 


987 


991 


996 



similarly in the equation for the standard deviation, the expression (log g - log 
is equivalent to division by 'O f when x^ s 'O 1 . 



Users of this manual must assure themselves that in each case where log normal 
analyses are used, sufficient care is taken that the assumptions critical to these 
analyses are examined. Bagenal (1955) lists several misuses of the log normal distri- 
bution in marine plankton sampling and also describes the proper use of the method. 
In particular , the use of logarithmic transformations with the natural base "e w is 
recommended for sample statistics instead of the base '10 1 logarithm because of the 
ease with which unbiased arithmetic means may be derived from the mean and variance 
of In x^. The equation for the mean and variance are as above with the transformed 
variate. 



3.1.1.3. Negative Binomial Distribution 

In the Normal and log normal distributions the variance is independent of 
the mean. In the negative binomial distribution, the variance is related to the 
mean in the following manner t 

S 2 -- 

A theoretical discussion of the properties of the negative binomial is beyond the 
scope of the manual (See Anscombe, 1949; Fisher, 1941). It has been found to be 
useful in the study of trawl catches of fish (Taylor, 1953), 

One goal of statistical analysis, is to minimize the money and effort expended 
to obtain and interpret data* For the purpose stated for this manual, i.e., to 
monitor important changes in the spawning biomaee of fish through ichthyoplankton 
surveys, one fcatfi Identify the level of sampling effort necessary to achieve suitable 
separation of important differences. For example, if one wishes to distinguish 
between years differing by 25%, and if one assumes that the negative binomial is 
the most appropriate distribution model for ichthyoplankton, t$4e following equation 
(Southwood, 1966, p. 20) will allow a first approximation of the number of samples 
needed to achieve the monitoring goal precision (2St): 




53 



"N" is the number of samples required, "m" is the sample mean of the nunrber of e* fi s 
or larvae p*t unit area, "k" is the dispersion parameter of the negative binomial 
distribution, and "D" is the percentage level of difference required to discriminate 
between important and unimportant changes. For example, if we use the value for 
sardine eggs within one day of spawning (Smith, 1973) of about 100 eggs per 10 nTfor 
"", and a k" of 0.014 f and a "D" of 0.25 (25% difference) the number of samples 
required is: . . ^ 



The value of "D" is selected arbitrarily by the researcher to meet the goals of the 
program with respect to detection of important differences. The value of "k" may be 
estimated in various ways (Southwood, 1966) but the simplest estimator ist 

_2 
k 



- 714,386-100 

As above, "m 11 is the arithmetic mean, and H s " is the variance or the square of the 
standard deviation fl s lf . These sample values include the "0" samples, if any, in 
the preliminary samples. 

Table 3.6. Estimations of the number of samples necessary to 
obtain certain levels of precision (D) for varying 
degrees of patchiness (k) . 



k 




D 






.05 


.10 


.25 


.50 


0.01 


40004 


10001 


1600 


400 


0.02 


20004 


5001 


800 


200 


0.05 


8004 


2001 


320 


80 


0.10 


4004 


1001 


160 


40 


0.20 


2004 


501 


80 


20 


0.50 


804 


201 


32 


8 


1.00 


404 


101 


16 


4 


2.00 


204 


51 


8 


2 



Table 3.6 represents a calculation of the number of samples required to dis- 
criminate between samples at levels of difference from 5% (0.05) to 50% (0.50) 
when the value of "k" varies from 0.01 to 2. The mean number of eggs or larvae 
per sample is arbitrarily fixed at 100 for this table. 

3*1.2. Data Comparisons 

In addition to the preparation of data summaries (see Section 2.3.) for practi- 
cal evaluation of spawning biomass estimates, it is often necessary to examine 
c^rtlin statistical properties of the data set. For example, the incidence & a 

Shr*as:'Is^Lsw i 3S SW^TSJ ssa' p s?5ws 
5S^^^^^ ta ^ ta ^=^ssri'gir&%*b. d H!a^' 

by understanding the distribution of spawning products (See Section 3.< 
presenting data in such a way as to expose rare large samples. 



i 7 are three characteristic number distribution* illustrating one, ;'- 
oS and two "lognormal" distributions with different d4**i* of- ' 
of the logs. All three distribution* are centered^ about " * ( 

number of observations exceed 100 as are lo*e* *^>JJ&' ; 

some of the statistical properties of 



&4 



Table 3.7. Three characteristic number distributions, 



Normal Iog Normal *A* Lo# normal *B* 



149 


99 


479 


96 


17785 


100 


143 


98 


380 


93 


9813 


89 


137 


97 


339 


91 


$012 


79 


133 


97 


302 


89 


3981 


71 


130 


96 


269 


85 


2815 


66 


128 


95 


263 


83 


2239 


63 


126 


94 


246 


81 


1778 


56 


123 


93 


229 


79 


1413 


50 


122 


92 


214 


76 


1122 


45 


121 


81 


195 


74 


1097 


40 


119 


90 


191 


72 


891 


35 


118 


89 


178 


69 


708 


32 


117 


89 


166 


63 


631 


28 


115 


88 


162 


62 


562 


25 


114 


87 


159 


58 


501 


22 


113 


86 


148 


56 


447 


20 


112 


85 


145 


55 


398 


18 


111 


84 


141 


52 


355 


14 


no 


83 


138 


51 


316 


13 


109 


82 


135 


50 


282 


11 


108 


81 


132 


47 


269 


10 


107 


80 


129 


45 


251 


9 


106 


78 


123 


43, 


224 


8 


105 


77 


120 


41 


200 


6 


104 


76 


118 


39 


178 


4 


103 


73 


115 


37 


159 


4 


102 


71 


110 


32 


148 


3 


101 


69 


107 


28 


141 


2 


101 


67 


105 


20 


126 


1 


100 


59 


102 


15 


112 


0.7 



m (arithmetic mean) 

B X (standard deviation) 

Coefficient of variation 

In g 

* g (standard deviation) 

g (geometric mean) 



Normal 




Lognormal "A" Lognormal H B* 



123.70 

92.39 

75% 

1.99 

0.31 

96.70 



899.70 
2642.71 

294% 
2.01 
9.98 

102.21 



In Table 3.8 we have divided the original normal number Bet into classes. In 
colunn ""A",, the class boundaries, are 57.5 to 62.5, '62.5 to ft".'S, and so on, tot 19 
classes. ,Tb*S3Sp*esentative va^tfe for ea,ch class,, or jit is 60, 5, TO, 
in fioluaa, s " the cO,a** *#undar'i*s are 55 to 6$, 6 to 
points .41* % TvY *0 t , ' 



145, 



50 



ses, 




jit s 60, 
65 to 75, and 



f e * 

f6r Id 
the el* 
point value ar 

In Column "D w the claii* t>omidariei are 40 to 0, 69 to 80, and 

o 0*1 for 6 cla*eB. We && thai the pajftaitttiir teil**tiii" f row tfeeite climi*ietf ? 
data et differ very li^t]^ * r^m the pto^t^ a4|nf 100. 55 f and the population 
ataii^^^iA^i^ 3to <ac^ CAJi^tfhii \'*mjfrit* Kt thf ciAea Y?*- 3 ??^;:? 1 ^ 

^^ ^alut f oir *Jw dt4^ tJ^.ret^* t^ uinea an4 divl^jbjp t!W* 

" -. ,-_ 4^."4^K ^^ .^^ ;^_ '^*'" oj4|eif i^atiotin KOuld b^i Ansre Jt * 1 



other 61as 



tributi6n 



set 



55 



TabJU 3.8. Sample frequency distributions with wrying > 
boundaries and points for the normal distribution 
in Table 3.7. 



POIttT 


POINT 
2 * 5 


POINT POINT POINT 
5 +10 +10 UNGRDOTED 


SO 






55 




1 


60 


1 


1 3 


65 


1 




70 


2 


4 8 


75 


3 




80 


4 


8 17 


85 


5 




90 


6 


12 22 


95 


6 




100 


6 


12 22 


105 


5 




110 


5 


10 19 


115 


4 




120 


4 


6 14 


125 


2 




130 


2 


4 . 8 


135 


2 




140 





2 4 


145 


1 




150 


1 


1 2 


m 


100.58 


100.50 99.67 100.33 100.55 


8 


19.79 


19.43 20.00 20.99 19.58 



In a similar manner, the logarithmic distributions may be summarized to in- 
spect the frequency distribution. In Table 3.9 one may see the two lognonnal 
number sets summarized by classes. The classes increase by multiples of four 
(any multiple may be used) and the class boundaries are 0.25 to 1, 1 to 4, 4 to 
16 , and *o on for nine classes, and the class points are the gec-metric means of 
the class boundaries, 0.5, 2, 8, 32 r 128, . . . , 3276. One must note that the 
arithmetic means are not particularly precise estimates of the population wean; 
but they are likely no more imprecise than sample estimates of 10 or 20 from 
distributions of this kind. Still the general lognormal appearance of the data is 
emphasised and the lognormal and arithmetic parameters of the rtean and standard 
deviations are unbiased. 

It is Important to become familiar with the 4*i attributions for earah pl*n*toHie 
stage Jtor Wlv species of fish. In this way, errors in judgment may be avoided or 
minimized. 

Some examples from the CalCOPI program may serve to illustrate some kinds of 
statistical distributions of ichthyoplankton. For example, the fundamental sample 
distribution of sardine eggs has not changed markedly, even though the population 



56 



Table 3.9. Sample fr*<jQency distributions with even logarithmic 

cla** boundaries and log-normal freqxiancies (Table* 3*7) 



CLASS 


BOUNDARY 




LOG 


LOG-NORMAL 


LOG-NORMAL 


I<QH8fc 


UPPER 


MID-BQIHT 


MID-POINT 


A 


B 


,25 


1 


0.5 


-0.30 




1 


1 


4 


2 


0.30 




3 


4 


16 


8 


0.90 


1 


9 


16 


64 


32 


1.51 


17 


12 


64 


256 


128 


2.11 


36 


14 


256 


1024 


512 


2.71 


6 


11 


1024 


4096 


2048 


3.31 




7 


4096 


16384 


8192 


3.91 




2 


16384 


65536 


32768 


4.52 




1 








m 


137.20 


1189.57 








8 


132.33 


4425.54 



has decreased 20-fold (Table 3.10). The scale of sample has been different with 
each decade, but the basic distribution of the all-important large samples remains 
proportionate. The lower end of the distribution is a sampling threshold and it 
appears to be essentially lognormal as in 1931-32 but the "unit** sample (one egg 
per sample) appears to be a pool of all areas less than the upper boundary of the 
smallest sample size class. Thus the high-speed sampler of 1950 with a small 
mouth opening, could only represent unit samples for all areas with an average of 
less than 256 eggs per 10 m 2 . One may also note the remarkable stability of the 
standard deviation and the negative binomial "k". 

The sensitivity of the sample frequency distribution is illustrated by the 
spawning cycle of the hake (Table 3.11) and the respective samples of eggs and 
larvae from a typical hake season, December 1954 through April, 1955. This is 
probably the most intense pattern exhibited by ichthyoplankton in the California 
Current. This will likely be explained by the fecundity of the individual females, 
the habit of spawning below the mixed layer of the ocean, and the long persistence 
of eggs and larvae in the colder waters. 

One is often confronted with questions Jkike "Are samples of type N A" comparable 
with samples of type "B"? M Of the statistical tests, few are a widely useful in 
ichthyoplankton sampling as the Kolmogorov-Smirnov Test <Tate and C lei land, 1957, 
p. 62). Two examples are provided here; one in which th* maniples are drawn from 
different distributions ajid one in which the samples are drawn from the sawe 
distribution. 

The f irat *x*pl* is drawn from * test which aks the question "Are anchovy 
egg* the same diameter in August of 1972 a* they wore in February of 1972?" In 
Table 3*12, the midpoints of the frequency classes of the diameters of th* *$ ire 
listed as fractions of a millimeter from 0.5643 to 0.6SSO and the number of eggs mea- 
sured t *aph iz is in the .aeoond *oltfnttw 9h* next Oolwm lists the percent each 
categpry Aft of the total and the lat column it the cumulative sum of th percent* 
for the month of February. 



57 



o 




8 










CD 


















OO 




CO 

















CM 


4 




Qb 


00 oo o ro CM in ^t* 


O 


o 


O 





CD 


SJ 






CM CM ro CM rH 




rH 


in 






3 




OS 




rH 


^ r 


rH 










d en 














c 




U rH 















14 




*o 










CM 


CD 


MH 


CO 


i 


r- o r- rH 


m 

o 

rH 


kO 

CD 
^9* 





CD 


-* 


T3 


a 


CO 
1 






* 
<T5 


(0 






44 


& 


"&o% 









u 








2 


H rH 
















CO 


ac 














CO 


















0) 


|V| 
















rH 


O 


M 
0) 










VD 
Cf\ 




1 . 


i 


i^ 

i ^r 


oo in &\ ^* CD CM 

rH fN rH rH 


OO 


ON 


CO 


rH 




CM 


D> W 










VA 










Q 


"S^. rH 






a 


rH 






CD O 




rH 






cd 








4^ 










o 


(Tj 






8 c 


W 










O 






TJ 5 


s 


^, 










CO 




V4 j^f 


C"4 


0) 










rH 


r-* 


ss 


5 


4J rH 


rH CM vo rH r-* r* ro 


^ 


Cr\ 


^ 





CM 


rH 


i^ 


C ^^ 


CM CM rH CO CM rH 


ro 


VJD 


rH 






tt4 f%| 




1 rH 




rH 


in 


ro 






O 












rH 






C 


















C <D 














VO 




O M 














CO 




-H 0) 




CM 










rH 


Cf> 


"3 *M 




1 


cN^vor-^cNrnincM 


in 


^, 


VO 





CO 


X) -H 






rH rH rH 


VO 


^** 


vo 










Ul CO 






in 


^4 






M 












rH 






-M JC 




3 rH 














CO -P 
















T3 "* 














M- 


frg 

13 


E-i 

g 
2 
^ A 


in 

CM 

rH in 






6 


0) 




O 
(0 

SCO O 
CO -H 


M O 

cu m 


CM 



rH 




> 


rH in rH 
CM 00 










S rH ? 
O* rH 2 O 

CO CQ H 


rH fH 


*V 




in 












f2 



















CU 


1 


(X 

s 


rH VJD in CM CJ\ OO 
CM CD CD CO 












9 


1 


& 


rH 












O 






















5 




in 






- 






* 


S 




CN 

vo in 


















CD CM 












<D 


cr 


P5 


* * ^^J. ^Q 












1 


3 


1 


** rH VD m CM Of 
CM O 
rH ^* 






J 







58 



JQ 

S 



a 
o 



s 

O4 



c 
o 



to 

O) 



SM in 
O i 

4) in 



5 



* 

I 



W 



3 



i 



to 

O 

w 



*<-4 \A m 



CM 



CM <M 



CM 



CM 



CM 



sx> ^ O4 
<M 0* -* 



CM 

CM 



CM r-H VO 



vo 



CM 



in 



in 

CM 
CO 



CO 
CM 



so 



on 



CM 



CM 



CM 

in 



CM cvi to rsi 

-* CM <M r 1 



CM CD so r- so CM 



CM 



in 
tn 



CO 



SO 



CO 
00 
CM 



*v i r- WP o i 



CM 



oo 

00 

CM 



so 
so 



SO 

*-*! 

-l 

CM 
SO 



so 
co 



CM 

m 

o 



CO 
00 



en &n 



so *r o ^ M> 
*~< AO n c m 



VO in 



59 



Table 3.12. Worksheet for Kolmogorov-Smirnov test demonstrating the 

een mthe *y 99 diameters In February and 



DIAMETER 
(mm) 


A 


U G U S T 


F 


E B R U A 


R Y 




FKEQUI 


CNCY FREQUENCY 


CUMULATIVE 
FREQUENCY FREQUENCY 


CUHIHAMVB 

FREQUENCY FREQUENCY 


TOP. 


.5643 
.5814 
.5985 
.6156 


3 
12 
30 
75 


.5 

2.0 
5.0 
12.5 


.5 

2.5 
7.5 
20.5 








.5 
2.5 
7.5 
20.0 


.6327 


87 


14.5 


34.5 








34.5 


.6498 


96 


16.0 


50.5 


6 


1.0 


1.0 


49.5 


.6669 


90 


15.0 


65.5 


18 


3.0 


4.0 


61.5 


.6840 


72 


12.0 


77.5 


84 


14.1 


18.2 


59.3 


.7011 


51 


8.5 


86.0 


75 


12.6 


30.8 


55.2 


.7182 


51 


8.5 


94.5 


135 


22.7 


53.5 


41.0 


.7353 


12 


2.0 


96.5 


96 


16.2 


69.7 


26.8 


.7524 


6 


1.0 


97.5 


69 


11.6 


81.3 


16.2 


.7695 


9 


1.5 


99.0 


48 


8.1 


89.4 


9.6 


.7866 


6 


1.0 


100.0 


30 


5,1 


94.4 


5.6 


.8037 








24 


4.0 


98.5 


1.5 


.8208 








6 


1.0 


99.5 


0.5 


*8379 














0.5 


.8550 








3 


0.5 


100.0 


0.0 


N 


600 






594 








m 


6614 






,7285 








8 


0422 







0368 








D .03 














5.6% 



60 



Table 3.13. Cumulative frequency table for the Kolatogorov-Smirnov 
test demonstrating the similarity between measurements 
of a group of anchovy larvae by two people. 



LENGTH 
(nan) 


CUMULATIVE 
FREQUENCY 


CUMUIATIVE 
FREQUENCY 


DIFFERENCE 


2.0 


0.5 


0.5 


0.0 


2.5 








3.0 


1.6 


1.6 


0.0 


3.5 


2.7 


5.9 


3.2 


4.0 


14.4 


16.2 


1.8 


4.5 


19.8 


20.0 


0.2 


5.0 


28.9 


35.7 


6.8 


5.5 


36.9 


42.2 


5.3 


6.0 


46.0 


49.2 


3.2 


6.5 


51.9 


53.0 


1.1 


7.0 


59.9 


61.6 


1.7 


7.5 


67.4 


70.8 


3.4 


8.0 


73.8 


76.2 


2.4 


8.5 


78.6 


79.5 


0.9 


9.0 


87.7 


88.1 


0.4 


9 -5 


91.4 


90.8 


0.6 


10.0 


94.1 


95.1 


1.0 


10.5 


96.8 


96.2 


0.6 


11.0 


98.4 


97.8 


0.6 


11.5 




98.4 


0.0 


12.0 


98.9 


98.9 


0.0 


12*5 


99.5 


99.5 


0.0 


13.0 








13.5 








14.0 









14.5 








15.0 








15.5 




100.0 


0.5 


16.0 


100.00 




0.0 



'.05 



7.2 



61 



lOOt- 



SO 



Oft5% 




K>0r 



60 



70 
DIAMETER (mm) 




LAftV* LENGTH (mm) 



Figure 3.3. Illustration of the Kolra- 
ogorov-Smirnov comparison of sample 
frequency distributions of anchovy 
egg diameters from samples collected 
February and August of 1972. The 
difference (D) , that is significant 
for the 5% level and 594 samples, 
is shown as a vertical line near the 
right margin. 



Figure 3*4. Illustration of the 
Kolmogorov-Smirnov comparison of 
sample frequency distributions 
of the lengths of anchovy larvae 
as measured by two people. The 
difference (D) that is significant 
for the 5% level and 356 samples 
is shown as a vertical line near 
the right margin. 



The same columns are listed for the August samples. The final column is a dif- 
ference column. For a 5% level of significance, the maximum difference is compared 
to: 




5.6% 

where "D" is the significant difference level and N is the number of sample measure- 
ments (594 in this instance). The fact that the two curves differ by more than 
5.6% is taken as conclusive that the difference is significant and that anchovy 
eggs were smaller in August of 1972 than in February of the same year. The two 
cumulative percent curves are plotted in Figure 3.3. 

In example 2, the question is "do two plankton sorters derive the same length- 
frequency curve from a sample of 356 anchovy larvae?" In Table 3.13, only the 
category values and the cumulative frequencies and differences are listed. In this 
instance, significance at the 5% level is indicated by difference in cumulative per- 
centages greater than 7.2% (D - 136//SW). Since no difference exceeding that 
value occurs, this may be interpreted as supporting the idea tfrat the same 356 
larvae were measured essentially the same by two independent plankton rters. Fig- 
ure 3?4 illustrates the two cumulative percentage curves for this example. The value 
"136" in this example is for a large number of samples (>100) to test a difference 
at the 51 level. 



3.1*3. Advice on Data Transformations and Comparisons 

It is the recommendation of the senior author that no transformation of data 
be done without the documented advice of a professional statistician. The widely 
used log transformation has no explicit or useful method for treating "0" obnerva- 
tions. The "0" observations are necessary for determining the edge of the spawning 
area and are frequent within the spawning area. It is easy to misuse analyses based 
on log transformations (Bagenal, 1955) and analysis of variance and kindred tests 
can sense a significant difference in geometric mean when the arithmetic means are 
identical and detect no difference in geometric means when the arithmetic means are 
different. In some cases, the log transformation generates a set of values in which 
the variance is still a function of the mean, thus stabilization of variance is not 
a guarantee of transformation. 

This manual is aimed at standardized conduct and reporting of fish egg and 
larval surveys, not analysis. In brief , we recommend a generalized distribution which 
1) accommodates contagian or patchiness, 2) has explicit zero terns, 3) approximates 
a poisson at low mean values, and 4) does not assume independence of mean and variance. 
The negative .binomial is our provisional recommendation for data analysis for data of 
the type collected in the CalCOFI surveys. 



G3 



1.2. Volqtoetric Sampling 

The earliest objective of quantitative plankton sampling was to take samples 
from the bdttow to the surface under a defined unit of sea sufctace area. The ear/- 
liest nets did not accept all the water encountered in the column, and factors fur 
flthe oor*iection of this bias were calculated by Hensen (1895) and Birge (1895), 
Reighard CIS*?) suggested the use of a flowraeter to record volume filtered. Kpfc^id 
(1897) pointed out that many smaller organism* were extruded through, the meshes. 
Plartktonic organise were observed to avoid the net (Mackintosh;, 1954) . The survey 
ftarocedures recommended in this manual (Section 2) do not eliminate these biases, * 
but with oare in equipment design and use, these biases and some of th* technical 
sources of variability can be evaluated or minimised. 

3.2.1. Volume and Distribution of Water Filtered 

The estimation of spawning biomass requires quantitative samples of the plank- 
tonic spawning products expressed as numbers under a unit area of sea surface. This 
estimation can best be obtained from a tow designed to filter at a constant rate per 
unit depth from below the deepest occurring egg or larva to the surface. On the 
continental shelf, and in bays and lakes where quantitative sampling began, the s 




I 



TIME IN SECONDS 

20 40 60 80 100 120 
I I I I I I I I I I I I 

DISTANCE IN METERS 
IO 2O 3O 4O 50 60 

I I I I I 



70 
I 




SEA BED 



Figure 3.5. An example of the path of a Hensen net when fished 
vertically with the ship adrift. (From A. C. Simpson, 1959). 

vertical tow was the method of choice. This manual recommends an oblique tow in- 
stead because the vertical tow cannot be conducted with precision on the high seas. 

A.C. Simpson (1959) drew attention to the fact that ".vertical" hauls from a 
drifting ship (Figure 3.5) filtered more water than planned, and he accordingly 
furnished correction factors for different rates of drifting. Unfortunately, a 
second bias occurs in stratified waters which cannot be easily corrected. In a ver- 
tical tow the assumption is that an evenly integrated tow has been taken from the 
miiamui depth to the surface. It is quite clear that this assumption is systematic- 
ally violated in vertical tows from drifting ships or from static ships with under- 
lying current stratification. The effect is that the rate of ascent is rapid in 
portion of the tow, and more and more water is filtered from each unit 

surface is approached <Smith, Counts and Clutter,, 19*8) leading to an un- 
sample of the water colmnn biased in favor o the, .superficial waters. 



64 



"VERTICAL 11 TOWS - INDIAN OCEAN EXPEDITION 



50 r 




5 Z5 45 
WIRE ANGLE C) 



250 380 450 100 
WIRE OUT (m) VOLUME FILTERED (m) 



This effect will not be *riou in 
well mixed wafers. 

Swells and the rolling of the 
ship also cause the vertical tow net 
to pump portions of its sample back 
into the sea. The same motions of 
the ship cause the vertical sanple 
to be accelerated so that portions 
of the sample lying on the mesh are 
likely to be extruded through the 
apertures. 

Some examples of the poor qual- 
ity control of vertical tows are 
available from recent expeditions. 
For example, during the Indian Ocean 
Expedition (Mo tod a, Konno, Kawamura, 
and Osawa, 1963), a 1.13 m diameter 
(1m 2 in cross section) net was towed 
"vertically". Figure 3.6 illustrates 
the variability in the wire angle, 
the amount of wire required to reach 

200 meters, and the volume of water filtered for these tows. Table 3.14 lists some 
of the variations in the wire angle of vertical tows made during the Cooperative 
Study of the Kuroshio (CSK) . Here, too, instead of the intended vertical tow, the 
average tow had a 25 wire angle, the standard deviation (+ 2 s.d.) of the average 
tangent was 0.31, and the actual range of angles was from 9 to 65. Also, the ex- 
tended forward disturbances caused by the tow wire augment the usual avoidance 
problem (Clutter and Anrafcu, 1968). 

Two major faults of the Hensen net are the towing bridle and towing cable ahead 
of the net (see Section 3.3.2. below) and the inability to locate * stable flow 
area in which to place a flowmeter (Figure 3.7). The reduction cone causes accel- 
eration of the water near the periphery of the mouth because of the reduced pressure 

Table 3.14. Recorded angles of tow for "vertical" tows in the 
Cooperative Study of the Kuroshio (CSK) , 



Figure 3.6. A sample of the wire angles, wire 
otit to reach 200 meters depth, and the volume 
of water filtered during each tow using the 
Indian Ocean Standard net, (After Motoda, 
Konno, Kawamura and Osawa, 1963). 



ANGLES OF STRAY 



FREQUENCY 




5 

10 
15 
20 
25 
30 
35 
40 
45 
SO 
55 
60 
65 



16 
35 
58 
52 
59 
33 
59 
37 
38 
11 
11 

3 

o 

2 



65 




* 

T 

H 

* 
II 



i M 

T 



^" 1t I* 



**** 










behind the cone. This effect di- 
minishes toward the center of the 
net behind the bridle apex. If the 
towing speed could be made suitably 
constant, the flow could be calculated, 
but the degree of acceleration is 
quite sensitive and dependent on tow 
speed, therefore the variations experi- 
enced in high- seas towing would pre- 
clude the collection of precise and 
accurate filtration performance data. 

3.2.1.1. The Oblique Tow 

The progress of the oblique tow is 
actively regulated by the , ship when it 
is underway. In areas where currents 
are relatively slow, such as the 
California Current region, it has bean 
found possible to fish plankton nets 
from the standard 210m depth with rea- 
sonable-consistency [e.g., 206m + 16m 
(2 s.d.)]. In areas with strong~under- 
currents, such as the Eastern tropical 
Pacific (Ahlstrom, 1971, 1972) less 
consistency is achieved (e.g., 203m + 
37m) although the variation is still"" 
within acceptable limits. One im- 
portant factor to consider when com- 
paring oblique tows with vertical 'tows 
is that the greatest errors in the oblique tows occur at depth and the error dimin- 
ishes toward the surface (Figure 3.8). The vertical tow path error is cumulative 
toward the surface. Variation in the volume filtered in a series of oblique tows 
is usually much less (e*g., volume filtered - 649m 3 + 86m3 for 165 tows of the Cal- 
COFI net) than it is in vertical tows (e.g., the Indian Ocean Standard net, Fig. 
3.6). 

The tow-by-tow variation in these properties can be measured adequately from 
the surface by monitoring the wire angle closely. Fig. 3.9 demonstrates tjtiis by 
showing the similarity of tow paths for an ideal tow as determined from wire angle 
readings and a bathykymograph . About the best that can actually be accomplished in 
maintaining a straight oblique tow path on a net tow under ideal conditions is illus- 
trated in Figure 3 10. Part "a" of that figure shows the ideal tow path, the tow 
path estimated by a bathykymograph (part w b") , and the time course of the rate of 
ascent. In the ideal standard tow, 20 meters of .wire are retrieved per minute with 
a wire angle of 45. The ideal rate of ascent is 14 meters per minute. In the tow 
in question (Fig. 3.10). the rate of ascent varied + 30% from the ideal rate. An 
example of the maximum acceptable variation from the ideal tow path, for the purposes 
of standardization, is illustrated in Figure 3. XI- * ** iB ;*?*? ** " ta of 
ascent varied + 84% from the ideal. Where the rate of ascent is higher than aver- 
age, less water is filtered per unit depth. The variation in tow path can be eval- 
uated byconverting the wire angle readings, which are made at 30 second intervals 
during the tow, to tangents and calculating the average tangent and tandard devia- 
tion for the tow (Table 3.15). Standards can be established ahead of time for ac- 
cepting or repeating tows. If the standard deviation of the average tangent is 
too great (e?g?; 2 s.d. > 0.1) the tow should probably be repeated. One way to 
assure a oood tow is to maintain the wire between 42" and 48 (45' + 3*). The 
assure a gooa row is TO m exemplified by a series of 204 high-seas tows mad in 
,! h !i 8 P S aiLiSa ?aent was 1.077C+ 2 s.d. - 0.084) with 95% fidu- 



Figure 3.7. A comparison of the accelera- 
tions of water in the mouth and several 
transects ahead of the Hensen net. The 
numbers represent the difference between 
the channel flow with and -without the net 
present in cm/sec. The inset illus- 
trates the entire towing arrangement in 
the test channel. 



a temperate area. The average tangent 
cial limi 



imit;* of 0.91 - 1.24. 






the variation in the rate of ascent is ttnderaoored "by ' ; ';tbf ,^, 

s. 




analysis of variance of egg counts. 
The uneven passage of the net through 
the water column should be carefully 
avoided. Further effects of this un- 
even rate of ascent include the in- 
creased extrusion of flexible organisms 
during the faster part of this retrie- 
val, and the possibility of the flow- 
roeter stopping during the slower part 
of the fluctuation in rate of ascent. 

3.2.1.2. Clogging 

At times, material in the seawater 
may clog the mesh *>f the net thus lower- 
ing the filtering capacity of the net- 
ting. As long as the flowmeter con- 
tinues to function, this deficit in 
water filtered will be monitored. As 
can be seen from Fig. 3.13 however, the 
sample taken gets progressively less 
representative with time. The tendency 
is to undersample the surface layer if 
clogging proceeds throughout the tow. 

In Figure 3.12 the trail of dye streaks through the netting indicates that clogging 
occurs evenly over the entire net surf ace , not just near the cod end. The rapidity 
with which clogging can occur is illustrated in Fig. 3.13. If clogging is routinely 
encountered, the only solution is to increase the filtering area (Smith, Counts and 
Clutter, 1968; Tranter and Smith, 1968) by adding extra panels of mesh in the 
cylindrical part of the net. 

3.^.2. Avoidance 

The use of ichthyoplankton surveys to monitor spawning biomass requires con- 
stancy of larval size distribution in the samples obtained. The size distribution 
of larvae captured in plankton nets is affected by two factors. In particular, as 
larvae grow they appear to undergo heavy mortality while at the same time they 
improve their ability to evade plankton nets. Understanding the relative import- 
ance of mortality and avoidance in the declining catch of fish larvae with increase 
in size should aid in the interpretation and understanding of the role of larvae 
mortality and recruitment of future stocks. 



Figure 3.8. The ideal tow path in the 
CalCOFI oblique tow to 210 meters (solid 
line) and the average depth at each 10 
meters recovered JK 2 standard deviations 
in a test series of 14 tows. 



Research on avoidance was extensively 
TIME (mfetutot) 




Figure 3.9. The ideal tow path of the 
CalCOFI net (solid line), the depth mea- 
sured by bathykyroogr^qph (solid dots) 
an$ th<* toffcfc estimated by the angle of 
stray a*a the wire out (open c4rc!Us) * 



reviewed by Clutter and Anraku (1968). 
Further work on an "ultimate" sampler 

1 by Murphy and Clutter (1972) points out 
the extreme nature of larval avoidance 
(Figure 3.14) and further explains why 
the search for an "ideal" sampler has 
not been very productive. It appears 
likely that the day-night difference in 
avoidance is Only a small part of the 
total avoidance problem, the bulk of 
the avoidance being performed as well 
in the dark as in the light, Barkley 
(1964; 1972) described the relative 
capacities of larvae to avoid when 
given advance warning of the approach 
of the fifcpUt, H0 Ia<ac4tea,fch*t . ,,, 
the response starts three to four mouth 
diameter* Ahead of the net. Larger and 
faster tow h*tf fcay iilter the acuity 
of larvae to avoid very little because 



project 



67 



Figure 3.10. An ideal tow. a) In the 
upper figure the ideal tow path is ill- 
ustrated (solid line) and the attained 
depths f determined by a bathy kymograph, 
are solid dots. The first difference 
in the meter's depth each minute is il- 
lustrated in the lower portion of the 
drawing. This is the critical measure 
of the volume of water filtered for each 
unit depth, b) A bathykymograph trace of 
the same tow. 

VlME (mtoutu) 



TIME (minutes) 
10 15 




KTEOFA90 
<mfti*0 

0088 


* - TTt * 


.. * * * * 



RATEOF A9C 
(m/min) 

o 5 8 g 


" 

. 

* * * 


. 





N 




















litlUllH 


kmMrt 


i '^ 


^ 


> (. 


I).Q o g j 


u 



Figure 9.11. An extremtely variable tow. a) 
In the upper figure the ideal tow path is il- 
lustrated (solid line) and the attained depths/ 
determined by a bathykymograph, are solid 
dots. The first difference in the meter's 
depth each minute is illustrated in the lower 
portion b the drawing. Thi is the 
measure of the volum* of water filtered ~ 
each unit depth, b) A ba thy kymograph t 
of the same tow. 



Table 3*15* An estimation of the variation in towing path 
derived from the average tangent and standard 
deviation of the tangents. 



e 



tan 6 



50 


1.1918 


1 


49 


1.1504 


2 


48 


1.1106 


8 


47 


1.0724 


8 


46 


1.0355 


7 


45 


1.0000 


4 



1.0735 (-47) 



0.0486 



95% Fiducial Limits 



.9782 - 1.1688 ( 44-49) 



Some mechanical cue*, iror avoidance are available both day and night. Smith and 
Clutter (3)65), Mahnken and Jossi (1967), and Tranter and Smith (1968) described 
the accelerations of water in front of the plankton net caused by the legs of the 
towing bridle, the towing bridle apex, and the towing cable. A simulation of a 
thoroughly clogged net (Fig. 3.15) illustrates the fact that the advanced warning 
due to a "bow-wave" effect is probably limited to the region of approximately one 
mouth diameter ahead of the net. The bridles without a net propagate measurable 
disturbances (Fig. 3.16) along the towing line as far as it proceeds ahead of 
the net, and particularly strong disturbances occur at the bridle apex* This is an 
important consideration in vertically towed nets. The absence of these disturbances 
is one of the most important features of the recommended Bongo nets. 

The main reason for understanding day-night differences is to be able to correct 
sample data for size and tiroe-of-day in order to make all the data obtained compar- 
able for use in estimating spawning 
biomass. Lenarz (1973) found that 
the decreased catch per unit larval 
length for a set of CalCOPI tows in- 
creased in the order sardine, ancho- 
vy, hake and jack mackerel. To trans* 
form the data to a comparable basis 
regardless of day or night, or larval 
sire, he fitted the equation: 

J " m A-l (* -* ) * 
%<* ~ *o> + ln 

* N ed" is the nu^" of larwfc of 
dill ** caught during tirae^period 
"d", and "s" takes values o s., 
* 2 .*.s fc where "* " is the modal size 
gfoiip And *s " i8 the largest size 
larvae with significant cttehf "d" 1 is 
the time of day, i.e., day (-1), or 
night (0)*, A is a constant, B is 

**m* variable was incorrectly 

identified as day (0), night (1) 
by Irfmarz (1973). 




ftp tlM fiitwint 
rt indicator of flow. 



69 



toe 
to 
to 



40 

to 
to 

10 




10 tS 10 IB A 4,8 



UK* 



IxK) 9 



KlO 4 



Figure 3.13. The clogging of a 
WP-2 (working party 2) 0.202 mm mesh 
conical net (UNESCO, 1968) under tow 
in Southern California nearshore 
water* 



IxlO 8 



UIO* 



LARVAL NEHU 




I-METER RING NET 



towing speed indicated by a 
telemetering flowmeter out- 
side the net. 

rate of flow through the 
mouth of the net indicated 
by a telemetering flowmeter 
inside the mouth of the net. 

ratio of the inside and outside 
reading* times 100 (% efficien- 
cy) 



4 6 B 10 12 (4 16 I* 20 
LENGTH (mm) 



Figure 3.14. An example of the 
degree of avoidance of larvae (in 
this case larval nehu, Stolephorus 
purpureus) by day from a br idled t 
ring net 'as compared to a plankton 
purse seine. (From Murphy and 
Clutter, 1972) 



^%;ss^nL"n h rtS"^-.j;o"if h "^v-'-;^f s 3;iF" r: "^ 

to be independent of "s" and "d" and to have a log normal distribution. 
The solutions to this equation were: 



Species 

Sardine 
Anchovy 
Hake 

Jack Mackerel 



B 

0.22 
0.42 
0.46 
1.18 



0.20 

0.16 

0.21 

-0.11 



If the neoative value of "C~ for jack mackerel were significant it would mean 



the net a. well at night as during the day. 
3.2.3. Meah Retention 



70 










tt 
II 

t 



IT 
17 

19 

M 
19 
It 

10 
14 
It 

ft 





Figure 3.15. A one meter dia- 
meter solid cone suspended in a 
towing channel. The transects of 
figures represent the difference 
in channel velocity at those 
locations with and without the 
net and are expressed in cm/sec. 
(Prom Tranter and Smith, 1968). 






t 



Figur* J.W.' Aa illustration of the 
erations ahead of the td*M appawtu* of 



pressure and escapement and extrusion 
through the meshes. Lenarz (1972) esti- 
mated the retention ratio of anchovy and 
sardine in nets towed at constant speed 
and found the relationships: 

s a + bL 



Where s is the mesh retention ratio of 
larvae of size "L" (less than 5.75 ram SL) , 
L is the length of larvae , "a" is the usual 
intercept of a linear regression, and "b" 
is the slope. 

For the silk CalCOFI net, the rela- 
tionships were: 

a b 

Sardine -.2382 0.2107 
Anchovy -.1942 0.1945 

and the retention ratio by length was: 
Length (mm SL) Anchovy Sardine 



2.50 
3.75 
4.75 
5.75 



0.292 
0.535 
0.730 
0.924 



0.289 
0.552 
0.763 
0.973 



The static and dynamic relationships 
of mesh retention can be discussed. In 
Fig. 3.17 the width and diagonal measure- 
ments of four different mesh sizes are 
compared for their retention capabilities 
of anchovy and hake larvae. It can be 
seen that larvae are captured because of 
their length but they are retained because 
of their body depth which becomes a criti- 
cal dimension. Another factor to consider 
is the variability in mesh size within 
the netting. For example this variability 
is greater in silk mesh nets than in those 
made of nylon (Fig. 3.18). Mesh vari- 
ability is reviewed by Heron (1968). A 
good review of factors involved with mesh 
selection is given by Vannucci (1968). 

In nets with high filtering rates, 
the complexities of extrusion and larval 
damage are encountered. These rates are 
functions of the square of the velocity 
and the ratio of mesh surface area to 
mouth diameter (Tranter and Smith, 1966). 
As yet, no generalizations can be made on 
the relationship between larval size, mesh 
size, towing speed and variance in towing 
speed. On til this analysis is completed, 
low constant toviitg speeds (1.5-2.0 knots) 
should be maintained. This is the pri- 

**** J* n !"*;XJ2? S^ th ' 
ti<m of th * lov Bon * n * tfl * 



* ***** ** of 

with Increased towing speed 

seen in these preliminary data on percent 



71 



466 



305 



707 



.947 



1.326 



EGG 



2.50 



475 8.75 



1375 



675 10.75 1475 



MESH WIDTH (rt*r) 
MESH DIAGONAL (mm) 



1725 



3 ' 75 5.75 775 9.75 " 75 .3.75 ' 575 



ANCHOVY ( mm SL) 



2OO 



3.75 



I 275 
I 
I 
I 



575 7.75 
475 , 6,75 875 



975 



1075 



HAKE 
(mmSL) 



i i i i i i 1 i i i i 



0.5 1.0 15 

BODY DEPTH AT PECTORAUS 



2.0 



2& 



Flaur 3*17. An exav^le of retention characteristics of various e*h 
The eh width (left bar on upper line) and diagonal (right bar on upper 
line) are compared with standard length and corresponding body depth of an- 
chovy and hake. The dashed line corresponds to the level of absolute reten- 
tion. (Modified from Smith, 1972*) 

*Switii f P.E. 1972. Extrusion. MARMAP, Technical Memorandum 16, June 7 f 
1972. Unpublished manuscript. 



72 





3.18. 
mesh silk 



Above, 



505 mm roonofUw^nt roach (.505 
(extra heaV?) qq (grit, gauze) ] 

, >|,4) v . ' ;,;, r,< -, ^ , 



0.55 



73 



of anchovy larvae retained from Smith (1972)* [based on ideas set forth by Saville 
(1959) and Heron (1968)]: 

Anchovy Standard 

Length (mm) Towing Speed (Knots) 

1.5 ** 2.00 3.15 4.14 

2.50 .292 .478 .154 .063 

3,75 .535 .565 .474 M37 

4.75 .730 .777 .807 .285 

5.75 .924 1 .802 .329 

, 6.75 1 1 -752 .423 

7.75 1 11 -826 

3*2.4. Uneven Tow Trajectories 

Current towing methods do not permit an analysis of the assumption that equal 
volumes of water are filtered at each depth. While these differences can be measured , 
and the effect on a given vertical distribution estimated, each tow in a survey must 
either be accepted or rejected, for no "standardization" or "adjustment" is tractable. 
Until directed research is completed on this problem, one can only say that a poten- 
tial problem exists and that its magnitude is less than the variance associated with 
the normal level of patchiness encountered. 



ftnith. ..!. i?a. l.tru.io., KMWW T.ehnic.1 MonaM 3. 



. 1972), .11 



74 



3.3. Sampling to Determine Spawning Area and Time 

Th* goiraphic f vertical, and seasonal Iwundaries of spawning must be well 
def ijie4<fjtyb the eaf lysi* Of f pa^ni^ fcioroass. 'Sittcse the piriaariy cost* \4tff; 
plankton surveys are related to distance traveled by ship, understanding these 
boundaries may permit substantial reduction of sampling effort between the explora- 
tory phase and the monitoring phase of the sturvey program. Also r the spawning pro- 
ducts diffuse with time and are transported off the spawning grounds, and migrations 
of adults Hay occur during spawning. Both of these factors may affect the seasonal 
placement <# boundaries. 

3.3.1. Geographic Distribution 

The pelagic habitat appears to permit spawning over vast areas* for example, 
the northern anchovy of the California Current regularly spawns over an area of SO 
to 100 t^ousahd square miles. The Pacific saury spawns so widely, beyond the usual 
boundaries of the CalCOFI survey grid, that a satisfactory westerly boundary has 
never been usefully defined (Figure 3.19) and it is possible that the distribution 
is trans-Pec ifi<?. In this figure, the incidence of saury occurrence has been plotted 
at stations sporadically occupied around the periphery of the usual CalCOFI area. 
Were it to become necessary, these peripheral stations could be used to design a 
survey to surround the spawning area of this species. The basic ability to define 
the geographic boundaries of spawning is one criterion for deciding whether the 
ichthyoplankton survey technique is feasible for a particular fish species. 

Figure 3.20 shows the spawning of jack mackerel (Trachurus symraetricus) wad 
undoubtedly occurring to an important eictent outside the usually surveyed area but 
the north-south and the seasonal distribu- 
tions appeared to be adequately enclosed. 
In some instances auch as these, a conserva- 
tive estimate of spawning biomass may be made 
(Ahlstrom, 1966; Gulland, 1970) and plans may 
be modified to enclose the spawning in sub* 
sequent surveys. In this way the monitoring 
surveys for sardine and anchovy may be used 
as the exploratory surveys of other species 
with only small technical costs added to the 
major goal of the survey. 

Colebrook (1973) has examined alterna- 
tive strategies in sampling simulated sea- 
sonal and geographic patterns* His model 
demonstrated that uiider the conditions simu- 
lated there is little dif ference among ran- 

dom, stratified random, grid, and route tran- 
sects over a region at several degrees of 
pattern 4w*ww*sity* However, the grid was 
clearly more expensive than the others in 
terms of the travel distance to occupy sta- 
tions. Colebrook (1973) also emphasised that 
the evenly spaced grid in time and apace has 
advantages for ut if actor ial treatment of 
data. It Should also be kept in mind that 
many ooeaiiptfraphlc techniques requite * sta~ 
tion transect normal to the coast, and great 
economies are available for combined fish- 
eries, biological and oceanographic surveys, 
since chip- time is tike greatest unit cost of 
each ' "' ' " ' 







Switter U967) ha* exatrijied the loss of 
precision resulting f torn the change from an 
even grid to a grid wfcteh cpncentrata* , sta- 
tions along lines and separates the lines by 
an equivalent proportion, for example, the 
error levfitroa a f*f *? v gjid i w^ases only 
10% when a 2*1 rectangular grid is used and 



3^9. 



Raw- 



ing shoeing the 4noidwe of 
sattry eggs in the usual Cal- 
COFI survey a* and *rt*iftd 
the |eri|Aery. (From Smith, 
and Casey, 1970) ' 



75 



JACK HACKC*CL LAKVAC 
1951- 



AJICA oceumto 

< IO% OOCUfHWNOC 

9 Km OCCUMKKNCC 

SHAOCDMICA 
>4% OCCURftEMCE 









AUGUST 




Figure 3.2Q. Average spawning position for the jack mackerel, 
Trachurus symroetricus , over a decade* (After Kramer and Smith, 
1970) 



the distance traveled is reduced 30%. It is recommended that a pilot-study 

survey with square grid be occupied and the alternative results from increasing rec- 

tangularity be analyzed for the optimum grid pattern. 

3.3.2. Vertical Distribution 

The determination of the extent of vertical distribution of a particular 
species is simplified by the observation that most pelagic spawning by fish occurs 
in the upper few hundred meters (Ahlstrom, 1959). Proper survey design is neces- 
sary to assure that the vertical distribution of spawning products and the subse- 
quent movements of the eggs and larvae in the vertical plane will be delimited or 
described. 

There are ecological and physiological aspects of vertical distribution. The 
ecological features, which involve the vertical coincidence of thefish vae and 
their^ common food, competitors, and predators, are outside the scope of this manual. 

;.SMU-^^ 

ture corrections for residence time, 

A method tor determining the proportions of larvae of a * 
tr.ted in Table 3.16 and Figure 3.21 for anchovy *^ hake larvae 







CM vc r- *% nr "o 




in oo in ir* 

Sri -f oo 
rH Csl CM 



trt 

9 * - 

J5 cC in S S 



JtA 



77 



,0 20 40 60 80 100 




ZOO 



300 L 



ANCHOVY! LARVAE 



20 40 60 80 <00 




300 L 



Figure 3.21. The depth distributions of anchovy larvae and hake larvae* 

The solid line is the cumulative percent curve and the individual bars 
at the indicated depths are the percent values of incidence. The arrows in- 
dicate the horizontal sampling levels. (From Ahlstrom, 1959) 

surface . The lowest tow is assumed to represent as much depth below the mid-depth 
as above. The number of larvae per 1000m 3 at each depth is transformed to the 
number of larvae per 10m 2 sea surface area between the limits represented by the 
tow by the following equation: 

r . r r SR ^ 

C " U r ( -rr ) 



Where C is the number per unit sea surface area; C is the number per unit of 
volume (234 anchovy larvae per 1000m 5 )? S is the unit sea surf ace (10 m 2 in this 
example); R is the range of depth for the sample (5 m in this example); V is the 
unit volume (1000 m 3 in this example). Thus anchovy between the surface and 5 mi 



( 10 X 5 v 

( 10DO J 



11.70 



anchovy larva* per 10 m 2 from 0-5 maters depth. In this instance, the depth 
intervals are uneven. For some purposes, it is more convenient to use evenly 
spaced intervals. It now seams likely that stratification is more pronounced than 
earlier thought (Longhurst, 1967) , which implies that much finer depth intervals 
than the above example will be necessary to increase the precision of depth distri- 
bution analysis. It is likely that the above type of analysis will suffice fpr 
determining the maximum depth of tow for species in the upper 300 maters. 

3.3.3. Seasonal Distribution 

Seasonal distribution of sampling will depend on the characteristics of the 
spawning season by species and the residence time of the Spawning products in the 
sea. For example, sampling in polar areas can be expected to center around a brier 
pawning period with an extended residence time of planktonic stages. In tropical 
areas the sharp seasonal cycle may be absent and the residence time of the spawning 
products way be quite short. In .temperate areas these characteristics are, of 
couree, intermediate, wit* tjle added complexity that polar or tropical waters may 
actually be transported into the temperate areas to varying degrees. If a fishery 
on the target p*ci*a exists in an area, gonad samples of the catch may be the beet 
index of when ichtfayopianktm* surveys should be conducted. In the CalCOFl **** 
the fou* months January, April, July and October have been most routinely occupied 
with eupplementml wtiiees *de during the height (February, March, May and June) of 



78 



20- 



% 15- 



10- 



5- 



ANCHOVY SPAWNING 


CYCLE 















1951-60 


1 1 1 I 1 














Illli 



r 



20 



% 15 



10 



. I 



HAKE SPAWNING 

CYCLE 

1951-60 



L 



MONTH 



MONTH 



Figure 3*22. The spawning cycle of the 
northern anchovy, fincrraulip ttprdax 9 
in the California Current region. 



Figure 3.23. The spawning cycle of the 
hake, Merluccius produotus, in the 
California Current region. 



HAKE ANCHOVY 



ONDJFMAMJJASON 




Figure 3.24. The cujtoriativr pro*ntag catch 
cycles of the larvae of HaXe arid anchovy in 
the Calif omia Current region. ^ 

, / ' ' ,-'>' ' ' . ;. ^ , ' 1 ' ' . v ( " 



79 



the spawning season of rnofct dpecies. 






assure a more precise estimate of spawning biomass. Pi*n in g season, 

The same monthly larval index for hake averages 28 and the three 

ved e ivfi b ini? ie8 T,?' I 2 ;," 4 33 ' The twoalternate JoIthSI 
yield larval indices of 24 and 33. Thus, to adequately sample hake, or similarly 
spawaning species which have one or more brief spawning periods ^require! more 

f "' 



, 

hv iK-- * * u cumulatl ve percent frequency curve for anchovy and 
hake (beginning in October, the seasonal minimum for spawning) is shown in Figure 

j * 4 

For some species, the seasonal and geographic distributions of spawning inter- 
act and cause some sampling complexity. These cases are typified by the spawning that 
begins at one end of the range of the adult and spreads across the entire range by 
the end of the spawning season. The statistical difficulty in this case is that 
the survey cruise pattern may at times be conducted in the same direction that the 
spawning is spreading, thus apparently extending the spawning season by region 
Conversely, the cruise may be conducted so that the stations &re taken across the 



AflCA OCCUPIED 

OCCUftftENCE 

9 5 % OCCUftftENCE 




Figure 3*25* The spawning regions and seasons of Pa- 
cific mackerel, 8ot>mber laponicus, in the most com- 

>f the CalCOFI surveys. <From 



sampled months o 
Kramer and Smith, 1970c) . 



area against the trend of spawning appearing to give a shortened season by region. 

An example of incomplete seasonal coverage is shown in Figure 3.25, In this 
case, the Pacific mackerel (Scomber japonicyy) is shown to tys spawning in signifi- 
cant amounts in the region to Ihe south of the survey area* It would, be assumed 
that spawning was also occurring In the missing months of August, September, Novem- 
ber and December. It is possible that the months of August or September could con- 
tain the peak Spawning. Here, as in the jack mackerel example (Figure 3.20) the 
calculated spawning biomass is assumed to be minimal. These data have been collect- 
ed incidental to iehthyoplanktoft surveys designed for the Pacific sardine and may 
be used to modify future surveys to enclose the Pacific mackerel spawning areas 
and times. 

3*3:4. Oceanic Transport and Diffusion 

Monitoring spawning biomass by ichthyop lank ton surveys will normally include 
the sampling of several days' spawning products (Sections 3.4, 3.5). These data 
may be pooled to obtain an egg or larva regional census estimate (Section 2.4). It 
may also be important to estimate the transport of eggs or larvae into and out of 
the region. For example, regions that are transported from will exhibit an unnatur- 
ally high apparent mortality rate because older larvae have more chance of being 
transported away. This causes over- samp ling in the "receiving" area. Figure 
3.26 shows the characteristic lowering of the incidence of young larvae and eleva- 
tion of the relative abundance of older larvae due to increased chances of being 
transported offshore. 

The general .problem of diffusion was treated by Hirano and Fujimoto (1968) and 
some preliminary calculations of the effect of diffusion on the statistical distri- 
bution of sardine egg samples have been made (Smith, 1973). Smith concluded that 
sardine eggs are spawned in a mosaic pattern of fish school dimensions, and the 
eggs at the perimeter subsequently disperse to a condition of randomness at the 
scale of the plankton sample. While an "average" condition leads to the dispersal of 
egg and larva patches with time, there is little information yet on the variability 




Figure 3.26. An example of a lar- 
va catch curve which has been 
influenced by transport of larvae 
away from the coast. (From Smith, 
1972) 

oo catch curve of anchovy 
larvae near the coast 

catch curve of anchovy 
larvae SO miles off the 
coast 

A~*"?A cmtoh curve of anchovy 

larvae 160 miles off the 



81 



of thin rat*. Slight variations in this rate could measurably alter the chances of 
getting or missing the occasional large samples which characterize the samplina of 
pelagic spawning products. Until this property is understood, it would useful 
to me&sure an index of contagion. There are several, but perhaps one of the JSe 
useful ones is the "k" of the negative binomial series. 

3.4. Sampling to Overcome Problems of Microdistribution 

Some understanding of "microdistribution" is useful for interpreting the sample 
data from ichthyoplankton surveys. "Microdistribution" is defined for this manual 
as the distributional characteristics of organisms near the scale of their body 
size or range of movement in a short period. This is distinguished from "hydro- 
graphic distribution 11 which is the distribution associated with particular ocean 
features such as eddies, upwelling areas, or islands. Both are distinguished from 
"physiological distribution" which is the entire geographic and bathymetric area in 
which the organism can survive. These three levels of distribution might be equated 
with the terms "aggregation," environmental preference," and "tolerance." 

Ichthyoplankton survey precision is determined by the scale and intensity of 
microdistribution of the fish eggs and larvae being sampled. These two factors 
are of primary importance in estimating the number of samples required for a unit 
precision of spawning biomass estimate. 

Fish eggs and larvae, as Cassie (1968) noted, " . . . are among the more 
variable in their spatial distribution, probably because they are relatively short- 
lived members of the plankton, and are initially released in a high over-dispersed 
(patchy) pattern." In addition, there appears to be a great difference between 
the dispersal of the spawning products of a demersal fish, the plaice (J.W. Talbot, 
Lowestoft, U.K., personal communication) and a pelagic schooling fish, the sardine 
(Smith, P.E., 1973). It may be that the plaice, in a relatively fixed position, 
discharges its eggs into moving water, while the sardine drifts along with the 
water and spawns into a rather persistent patch. 

Presumably patch sizes will range from the area of spawning of a single female, 

through the entire area of spawn- 
ing activity associated with large 
scale hydrographic features. Thus, 
the spawning activity is the origin 
of the distribution pattern of 
eggs, but subsequent events may 
modify that pattern. The spawning 
products may be transported away 
from the spawning area (Hirano and 
Pujimoto, 1968), they may disperse 
from patch centers (Smith, P.E., 
1973), and local or wide-scale 
water motions may concentrate 
them (Fig. 3.27) in convergences 
(Langmuir, 1938? Owen, 1966) or 
gyrals. Dispersal may also result 
in open spaces between eggs that 
are large enough for samplers to 
pass through even though they are 
inside a "patch". Spaces may also f 
be created by localized sources of 
mortality such as starvation or 
predation. 




Figure 3.27. An example of a pat tern- forming 
feature of the surface layer of the ocean which 
fray interact with organism behavior to yield ' 
f strand lines" at convergences and areas of 
few organisms at divergences (Owen, 1966) . 



As yet there is no reliable 
way to measure the scale and in- 
tensity of pattern in the pelagic 
environment. Patch size can sel- 
dom be determined directly from 
the spawning products because they 
are not usually subject to direct 
observation or continuous measure- 



82 






G 

a> 
u 



O 

O 



c: w r 



O 

a 

tn o a> *TH 

a o 4S 6 

o 4^ o 



CO 



G 'H 

O -M 
/> CO 



1 "" . -. . '- 

*. . 

. I 



:." 



-, ' .. .. : i a 
-;-->: 



U - K-f 4) 

.^1 fO CO G 
M csi CX 

M e a> 

G o% o ao 

o en u co 

u M 

in Ki a> o> 

WH **H o > o 

o pu eo \o 

a> 

^ u G CD 

a> G rt 

r^t 53 ^J- MH 

p^.fH ro to O 

*- CO 

CO t pS = 
K > esi A< 

<u . c r 

a> to o 

*T3 rH *T-I r-l CO 



CO CO U OO CO 

r-H C^ G O> ^G 

SCO IM 

0"H bOG 

H ^3 I- t0 O 

V> 4-* CO CO -* 

> +* 

M *t U OS 

a> G a> -H &e 

4- co -H ^ a> 



G kO 

> do 

*n o co 

a> r u 



Q> G 

jc a> 

4^ U 



4) 

-t 

00 



83 



Two indirect methods have provided a sense of scale of pattern. The- first 
' ^r 01 ^ 11 * ^*Prsal rate, was applied to sardine eggs by P.E. Smith 
(1973 . The second method, involving sample size, has been borrowed from botanical 
techniques (Grieg-Smith, 1952; Pielou, 1969). Pattern has been described statistic- 
ally by reference to the ratio of the variance to the mean (s2/ m) or some derivative 
of it such as the coiowonly used coefficient of variation (100 s/m) . Lloyd's (1967) 
index of "mean crowding" also describes pattern and shows promise of distinguishing 
between various sources of mortality. Colebrook (1969) has shown that variance 
due to density gradients may be confused with patchiness. 

Statistically, the best way, at present, to deal with the problems of micro- 
distribution "is to increase the number of samples using the relationship S.E. - 
S/ /n where 'S.E. 1 is the standard error (standard error of the mean), and 'n 1 is 
the number of samples" (Cassie, 1968). An example is provided by the two computer 
simulated centric aggregations shown in Figure 3.28. These two distribution pat- 
terns differ in intensity but not in scale. For example, the above equation recom- 
mends that 411 samples be taken in order to get 10% accuracy in the mean concentra- 
tion of the pattern in Figure 3.28 a. Only 193 samples need to be taken for the 
same precision in sampling the population in Fig. 3.28k. The aggregation in Fig. 
3.28 a has a mean density of 4.91, the sample variance is 99,23, and the average 
negative binomial H k M is 0.256. The aggregation in Fig. 3.28 b has a mean density 
of 3.76, the sample variance is 27.34 and "k" is 0.600. 

3.5. Residence Time (Estimating Mortality) 

One of the more difficult biases to measure in natural populations of eggs 
and larvae is "residence time" or period and extent of vulnerability to sampling. 
There are two simple techniques for measuring mortality which we will call the 
"cohort birth" method and the "constant birth 11 method. In the "cohort birth" 
method, the spawning products are released during a period of time which is short 
relative to the period of observation. The cohort (which refers to the group of 
eggs that is released at one time and whatever becomes of it) is then sampled in 
close successive time periods, and the mortality rate is estimated between each 
successive measurement. In the "constant birth" method, a single sample is analyzed 
for the abundance of a series of developmental stages of known age. It is beyond 
the scope of this manual to review all the necessary factors that effect these 
estimates such as drift of a cohort away from its point of origin (Smith, 1972) , 
changes in the pattern and rate of dispersal (Smith, P.E. 1973), changes in the 
vulnerability of the larvae to sampling by day, night and size (Barkley, 1972; 
Lenarz, 1972, 1973; Murphy and Clutter, 1972) and the constancy of mortality rates 
among developmental stages (Fager, 1973). We will consider only the temperature- 
dependent hatching time (Ahlstrom, 1943; Lasker, 1964), the temperature-dependent 
larval growth rate (Lasker, 1964), and the food-dependent larval growth rate. 

3.5.1. Temperature-Dependent Hatching Time 

For annual esti nates of spawning biomass, it may be necessary to make correc- 
tions for the temperature-specific hatching time because warm years speed the rate 
of development, reduce the residence times of the eggs, and cause an underestimate 
of the spawning population. Similarly, a colder-than-average year will cause the 
eggs to develop and hatch more slowly, thus increasing the length of time that eggs 
are vulnerable to sampling and causing an overestimate of the spawning bioraass. 
The hatching time at various temperatures can be used to adjust the sample values 
to a central temperature and thereby correct the bias. The following table was t 
developed from data provided by Lasker (1964) for Pacific bardine: 

Temperature Hours to Hatch Resident Time Ratio 
*-7jJ ft) ^ fr) 

130 93.0 .65 

14 78.5 .77 

15 68.1 .88 

16 60.2 1.00 

17 ' 53.7 1.12 
18. 48.4 1.24 
19 42.2 1.39 



84 



8 

O 



3 

O4 



g 







ss 



fi 

^ 



M 



CO 

i 

W 

> 
04 

**3|J2 
tn 



,-H CM GO r-l 



in 



rsi CM GO 



ir> o vo 



CM 



CM 



CM 
CM 



n 



in 



CM 

on 



on 
tn 



CM 

on 



so 

CM 



in 



SO 



CM 
CM 



in 



CM 



* 
0> 




r** m in r-H in 

<M CO CM CM 



CM 



vo in CNJ 



m 
so 



on 
in 



in 
so 



in 



tn 



in 

00 



on 



00 

so 



in 

r-H 
SO 



oo in 



in 
oo 

CM 



CM 



e 



VO 00 

CM 

SO 



CM 



so on 



I 



85 



In this instance, each sample value "x" was multiplied by the residence tine 
ratio "r.j" appropriate to each teraperafdre H j M according to the following equation; 



x i " 



j 



where 
x, is 



- the number of eggs per 10 m in the "i"th sample and its temperature "j, 
16 C - specific equivalent number of eggs per 10 m 2 , r, is the ratio ; 
^16^1' t !6 is the duration of the e 99 stage, from spawning to hitching, at JL6C, and 
t, is j the duration of the egg stage at temperature "j" in the same time units as t njc . 

J 1 w 

3.5.2. Temperature Dependent Larval Growth Rate 

Larval growth is dependent on temperature as well as food. In sardine, the food 
used in early growth comes from the yolk, as opposed to feeding success. Thus, a 
table showing the effect of temperature on growth from hatching to the formation of 
a functional jaw can be made based on work by Lasker (1964): 



Temperature 

(J) 

130 

14 

15 

16 

17 

18 

19 



Hours to Jaw Formation 



(t) 

121.0 
101.5 
81.9 
72.8 
63.3 
55,6 
49.8 



Residence Time Ratio 

(r) 

0.60 

0.78 

9.89 

1.00 

1.15 

1.31 

1.46 



The equation for calculating the 16 - equivalent abundance is the same as for hatch- 
ing time (Section 3.5,1.). After formation of the jaw, the growth rate will likely 



ooo 



900 



IOO 



Lt t t t 

A C MO 
M 




I20 o r EOG 

ABC 3.00 



KXX)- 



800- 



600- 



400 



eoo- 



LARVAE 
4.75 5.76 



675mm 



10 



DAY 



DAY 



Fiure 3.2*. Apparent mortality rate of sardine eggs and larvae 
** logarithmic scale and an arithmetic scale (data from 
1954). 



lstrom, 



86 

be dependent on feeding success as well as temperature, which makes it more diffi- 
cult to measure (Section 3.5.3.). 

3.5.3. Pood*" Dependent Growth Rate 

At present, there are no accurate methods for determining the food-dependent 
growth rate, nor for determining feeding-success in terms of abundance and composi- 
tion of food particles in the field. "Cohbrt birth* methods of determining tis 
value will be biased because of the probability that smaller members of the cohort, 
at the onset of feeding, will probably be more likely to starve than the larger larvae 
Of the same cohort (Kramer and Zweif el, 1970). Also, the more active and larger larvae 
of a given cohort may attract more predation than the smaller more passive larvae. 

3.5.4. Mortality Estimates 

One attempt at deriving crude mortality rates was made by Ahlstrom (1954). These 
values illustrate one method of determining mortality rate: 

Eggs Larvae 

Early Middle Late 3.00 4.75 5.75 6.75mm 
Individuals/lOnT 1044 734 385 395 234 56 22 

Days at Stage 111 2.4 4.8 3.9 3.3 

Individuals per day 1044 734 385 165 49 14 7 

Age at mid-point .5 1.5 2.5 4.2 7.8 12.15 15.75 days 

Figure 3.29 illustrates this crude mortality rate on the logarithmic scale and the 
arithmetic scale. The data are derived from Table 3.17. At interpretation of an 
average line of this kind is that it represents "average" conditions relative to food 
and temperature. Future work on feeding and growth in the field may allow direct 
measurements of the food density and the interpretation of deviations from this 
crude mortality curve as explained by temperature and food density. Another bias 
was pointed out by Isaacs (1965). Since vulnerability to capture generally de- 
creases rapidly in the feeding-larva stages, an explanation for increased catches 
of large larvae may be that those larvae which have starved become more vulnerable 
to capture. Further research is needed in the area of food-specific residence time, 
because growth may be arrested during starvation which leads to another unevaluated 
source of error. 

3.6. Spawning Bi omasa Estimates 

The fundamental relationship between the ichthyoplankton survey and the spawn- 
ing biomass of a given fish stock is: 

B = P 

T 

"B M is the spawning biomass of the fish stock, "C" is the census estimate of the egg 
production, and "P" is the capacity for production of eggs by a unit weight of fish 
stock. The production of eggs is determined by the fecundity of mature females ex- 
pressed as eggs produced per unit time per unit weight and the proportion, by 
weight, of the mature stock made up' of spawning females within each unit time. The 
appropriate census estimate of egg production is obtained by knowing the mean abun- 
dance of eggs in the spawning area, the mean residence tine of the eggs in the water 
column, and the frequency of the egg census cruises. 

The factors on which direct estimations of spawning biomass depend are not yet 
well enough known for most stocks to allow estimates of spawning biomass from ichthyo- 
plankton surveys. The establishment of crude trial values for each parameter of fe- 
cundity and the subsequent refinement of these values to useful levels of accuracy 
and known levels of precision should be Among the primary goals of ichthyoplankton 
research*. Host ot this manual is directed toward obtaining field estimates of egg 
production and analogous larva data from ichthyoplankton surveys. The procedures 
for determining population fecundity are not as well known and include the questions! 

1) How many eggs are spawned by mature females per unit weight per unit time? 

2) What proportion of the mature females are spawning at any given time? 



87 

3) - How frequently does the average female spawn? 

4) Over what period of time does each mature female in the population spawn 
once? 

5) What is the seasonal distribution of spawning? 

6) What is the geographic distribution of spawning by season? 

7) To what degree does spawning rate or absolute fecundity depend on environ- 
mental conditions such as amount and availability of food? 

8) What is the type and extent of variance for the above factors? 

Until answers are found to the questions above, there are ways of obtaining 
preliminary estimates. The simplest (Smith, 1972) is to assume that the larval 
index is proportional to the spawning biomass alone. The equation of proportion* 
ality is then established using an independent estimate of spawning biomass, such as 
obtained by an analysis of the catch from each year class over the time it is in the 
fishery (Murphy, 1966). This simple equation is then used to extend the spawning 
biomass estimates into periods when the entire year class has not yet been captured. 

Another method for estimating spawning biomass, is to determine the fecundity of 
mature females per unit weight and to arbitrarily assign a reasonable number of 
spawnings per year (Smith, 1972) . This has been the primary approach of the ichthyo- 
plankton surveys for the Pacific sardine. Analogous methods have been attempted for 
the northern anchovy. The important advantage of these preliminary attempts at the 
assessment of fish stock size is that they may be done with a minimum of information. 
The basic expense of these estimates is high for the primary species, but the addi- 
tion of new species to the regional census is relatively inexpensive. If the general 
relationship between the larval index, or egg census and spawning biomass is suffici- 
ently robust, the method will continue to provide a suitable check on catch analysis 
and lend supportive data for acoustic and aerial surveys. 

3.7. Fecundity 

The primary linkage between quantitative estimates of spawning intensity and 
coverage and estimates of the size of the spawning stock is fecundity (Saville, 1964). 
For new spawning surveys on relatively unfished stocks, it appears to be necessary 
to begin three phases of fecundity study. Before any fecundity studies have been 
done, the variations in spawning intensity are interpreted as changes of spawning 
biomass, fecundity and sampling error with the latter two assumed to be small 
relative to biomass changes. The first phase of the study of fecundity for the 
purpose of setting realistic limits to spawning biomass estimates is to determine 
the number of "advanced eggs" per gram of spawning female (MacGregor, 1968). The 
second phase is to determine the probability of a female attaining spawning- condi- 
tion by month or season. A third phase might be to understand the variations in 
fecundity from year-to-year, particularly in the number of batches spawned er 
year or alternatively the mean time between adjacent batches during the spawning 
season. 

An example of the use of an egg or a larva census estimate with an independent 
estimate of the spawning biomass of a species was illustrated in Section 2. The 
time series of egg or larva census estimates may also be used alone to indicate time 
series trends in the relative spawning stock size. Estimates of absolute spawning 
stock biomass may be obtained if data exist on the egg production rate of females 
over a well-defined period (for example, a month or a year), the proportion of 
total spawning biomass which is male, the mortality rate of the eggs and larvae, 
and 099 or larva census estimates which extend beyond the spawning region t^rou^h ,, 
the tiie of spawning. 



of the fecundity information may be obtained from the catch through market 
sampling, from the incidental catch through sampling "trash fish" at sea, and simi- 
larly from exploratory surveys. The number of eggs per spawning per gram of fale 
can usually be determined satisfactorily from a small sample (MacGregor, 1968)* The 



88 

felon*** MIX ratio can similarly be estimated from market or sea survey sample*. 
MaoGregor (1968) found the mean and standard error for 19 female anchovies was 
574 + 10% (95% fiducial limit*) eggs per gram of spawning female. This would 
tranSlate to 5.74 x 10 8 eggs per metric ton of female bioroass per spawning. If 
the male biomass is 75% of the female biomass, the population fecundity rate should 
be the same number of eggs per 1.75 metric ton of adult, or 3.28 x 10* egg* per 
metric ton of adult per spawning. If the adults may be expected to spawn two to 
three times per year, the egg* (X-2J*) would be 8.00 x 10 8 eggs per metric ton of 
adult per year. 

Smith (1972) compared three hypothetical models of anchovy spawning to check 
the sensitivity of spawning biomass estimates to those uncertainties about the 
variability in spawning behavior in the context of the increase in spawning bio- 
mass of the northern anchovy. The first model postulated a stable relationship 
between the annual larval census and the spawning biomass. The second model postu- 
lated the single spawning of each female in the winter quarter and does not respond 
to subsequent variation in spawning in the remaining seasons. The third model 
postulated the single spawning of each female in a maximum quarter (winter or spring 
in this instance) . There were no apparent differences in the estimation of spawning 
biomass .but the annual figure was subject to the least variation and the "maximum" 
season was subject to the most variation. Thus, the annual figure was chosen as the 
most representative and stable. 

In a similar evaluation of the Pacific saury (Smith, Ahlstrom and Casey, 1970), 
the bimonthly spawning maximum was chosen because of (1) clear shifts in the season- 
ality of spawning; (2) migration and spawning outside the survey area in some 
months; and (3) evidence from the transpacific stock of this species that the egg 
batches can be matured each two months. (Hatanaka, 1956) 

the Pacific hake spawn only in the CalCOFI grid area in a sharp peak at the 
same time each year. Relative spawning biomass has been calculated by integration 
from December to April, cruises in January-February, and cruises in January, Febru- 
ary and March. Relative to the natural fluctuation in the 1951-60 period, the 
errors of estimate from the less complete surveys appear to be satisfactory. 

Advanced research on the fecundity-spawning season problem should be based on 
the spawning characteristics of the individual species or determined from ichthyo- 
plankton surveys and inspection of adult gonads. The population and individual 
fecundity characteristics to be determined can be expected to vary by major fish 
group and by latitude. For example, more or less continual spawning might be 
expected in the tropic* and highly seasonal spawning could be expected in boreal 
zones. If spawning is essentially continuous, the primary data to be considered 
is the egg production per gram of female per unit time. A single annual spawning 
survey would be sufficient for this spawning behavior. If each spawning female 
spawns a single batch of eggs in a well-defined season, then the biomass estimate 
is derived from a few surveys during the spawning season. The most complete surveys 
and data on individual fecundity and population spawning behavior are required when 
there is multiple spawning at varying times 4uring the year and between years. The 
primary data required is the egg production per gram of female per batch of eggs 
and the minimum time between repetitive spawnings. 

Fundamental research on spawning is required for the direct use of egg and 
larva surveys In tkp direct absolute estimate of spawning biomass of any given 
species. In many species, even indirect estimates of spawning biomass obtained 
from spawning surveys may be the best available for ^in fished or underutilized specie* 



1.8. " tnatlye Quantitative IchtfryolaRfcton Sampler a 



Bon^o net system : ls recognised her* as. the best: type of sampler for 
u*e in quantitative ichthypplankton surveys. However, other systems may be used 
and! produce acceptable results, provided they are used in a rigorously specified 
rdutine. In many instances, results are comparable to those obtained by Bongo* 
(fietii fis-.T^ " 7 , , ..' " ..;; . , " ''.-,..-, ' < '- - ' 

hav* beeii in tt** ' for many years. *hwa* It 



might be advisable to continue using an alternative sampling system, particularly if 
ampling performance ha* been weU^documented over a long time period, until trans- 
fer to the standard equipment seems clearly advantageous. Also, there may be legiti- 
mate concern that the full-scale sampling system recommended here would be beyond 
the scope of the beginning ichthyoplankton program. 

The major spurce of imprecision for each sampler will ba the microdistribution 
pattern (Section 3.4.) of the egg* and larvae. Differences between samplers will 
primarily be related to volumetric sampling (Section 3.2.), i.e., volume and depth 
distribution of water filtered, the rates of avoidance in day and night tows, and 
the rates of escapement and extrusion through the meshes. 

This section is not intended to describe these alternative samplers. (Refer- 
ences are provided in the text.) Rather, the purpose is to review* advantages and 
disadvantages of the major types of alternative sampling Systems, relative to the 
Bongos, which may be wed in quantitative ichthyoplanXton surveys. 

3.8.1. Oblique Tow Nets with Bridles 

Many 1.0 IB and 0.5 m nets have been designed and many are in use to- 
day. Descriptions of some of these are provided by 'Tranter and Smith (1968) . One 
long-used example of this type of net is the CalCOFI net (Kramer et al, 1972) 
which consists of a cylindrical-conical net attached to a one-meter ring. 

The chief advantage of this type of net is the simple and inexpensive bridle 
and ring used to launch and retrieve the net from the water. This process can even 
be done by hand, if necessary. 

The main disadvantage, relative to the Bongo net, is the bridle preceding the 
net which causes directed water disturbances and results in increased avoidance. 
Problems associated with various types of materials used for the netting are dis- 
cussed by Heron, 1968. 

3.8.2. Vertical Tow Nets with Bridles 

The first and mostco*ftOtlly used net which was designed to be towed vertically 
was the Hensen Egg net (Hensen, 1895; Simpson 1959, Santander and de Castillo, 
1969). 

Bridled 1.0m and 0.5m nets also fit into this category when hauled vertically. 
This type of sampling system is generally used in shallow depths. 

The main advantage of vertical tow nets are: (1) simplicity in construction and 
(2) ease with which tow results may be converted to numbers of organisms per unit 
area of sea surface. 



The chief disadvantage is $he common practical limitation of equal sampling per 
unit depth. Other disadvantages include: (1) the sample is taken from water dis- 
turbed by the towing bridle all the way to the ship, thus avoidance may be great; 
(2) the rolling of the ship may alternately pull the net up rapidly and then let it 
fall, thus extruding organisms through the meshes on the ascent and expelling them 
from the mouth of the net on descent; (3) those nets having a small mouth restrict 
the size of the sample that can.be taken. ' 

3.8.3. High-Speed Nets 

High~spe*4 nets, designed to be towed faster than thre^ knots, a*$ still .being 
developed and tested. Some standard forms have evolved <Fig, 3,30), e.g., the 
Gulf JII (Gehringer f 1952, 1962) and modifications qt it,, (Bridger,, 1958* NeUen and 
Bemjtel, 1969; tijlstra, 1970; Harding and Arnold, 1371). The newwr highspeed nets 
appear to be satisfactory for estimating spawning biomass (Bjorke, Dragesuod, and 
Ulltang, 1974; Gjosaeter and Saetre, 1974; Postuma and ZjLjJUtjra, 1574; SavijLle, *xter, 
and ncKay* 1974* Schriaclt, 1974). , r . 



One advantage of the high-speed type net is that it a^, 
to the next station during the tow, thereby saving ship time and money. Other 



SO.3 cm 19cm 
i 




1 > -76 cm 



20 




cm 



B 



Figure 3.30. Scale drawing* of three **&-** 
(International Council for Exploration of the Sea) 
irSgrS in ttie North Sea. TOP) Dutch Gulf III (Getainger, 1952, 
lIj Zijletra, 1970) Thfe net is of 0.4 mm nylon mfft *n* if 
the *** .hown below the body o* the 
' ' 



Survey 



and T 



and 
r 



te, 



1969) The net ia of 0/3 
iy the absence of ;a'; 
at a towing' a$e*l of 

to within five m of the bottom is accomplished by 
the towing eable at ca* 20 meters p^r 
depth reoordet and ftowtocter*. '' 




j^l* ; ' (W 



ing out and recovt 
huve 



91 





-?.! 

1 A P 



U0> fl <H boa 
H O 3 *f4 00 
^HHIUtt.. 

c 



M 4> -H 
O H 



GO 



X >,Cr-4 (0 

i-H O 

fH ft) 10 rH "O V) 

flj tO O ) U 

U > K 0> Ctf 

*H |N Q-C 

*T3 oJ to oo to cn 



o> t> 

C X o> *H p 



M >-. ^ cn 

B O UiH 

o *><+< o> 



> IA 
O O |H-I 
r-1 CO 



0) H 



O f-f M O 
O C*l 



H-l ^>-d 
O <* fl 

B **& 8, 

O M 4^M M 

in 3 4> u * 
H m c * -K <^ 



*> O so N 

^^ a 

W W A ' ^O oO 
^ < IA C H 
4^ 0) 0) 49 CO CU 
UrH p *> 
O 0,0 10 4->*O 

H 3 * 5 te 



O - 

4^ 05 10 4>i-t 

iu 9 



a 

O 



9 

*H 

m 



92 

advantages include: (1) the speeds of tow are frequently high enough to allow 
sea current and wind conditions to be ignored and thus a rigorously controlled tow 
program is possible under severe sampling conditions; (2) larger specimens of the 
species are caught allowing mortality rates to b calculated for a larger portion of 
their planktonic life. 

Disadvantages include* (1) filtration res0ure is often high and variable which 
causes damage to and loss of smaller larvae; <2) the high speed requires a varying 
amount of cable for each increment of depth and this may make a straight tow path 
difficult, which could result in oversampling the surface waters; (3) the initial 
cost of the equipment and the more complicated methods of launch and recovery of the 
nets may be prohibitive for some ship* and programs; (4) the small mouth sizes may 
not allow enough water to be filtered to pick up the larger rarer organisms which 
high-speed samplers are Intended to catch. 

r 4: 
3.9. Intercalibration of Quantit^M aim>lers 



One may wish to in tercalibrat% samplers to allow comparison of (I) older 
data with that currently being doli^cted/ (2) cooperative surveys using different 
samplers, and (3) samplers which collect -some overlapping eiz categories of ichthyo- 
plankton. The standards of comparison should, not differ from the statistical stan- 
dards one applies to the survey results, therefore the estimation of the number of 
samples required can be derived from Tfcbld 3.6 (p, 52). T&& number of f replicate' 
samples should be calculated frpa/a trial stries of samples with both samplers or 
may be estimated from the variance in the existing data from one of the samplers. 

The intercalibration must account for all of the major sources of error: avoid- 
ance, escapement, extrusion, biased sampling of the water column, and different 
scales of sample size. Samplers should be compared under conditions which allow the 
comparison of the mean and variance per unit surface area and particularly the 
ability to detect the boundaries of the geographic distribution For fish larvae, 
a comparison of the size composition of the catch is critical. It is unlikely that 
sample formats which change the scale of sampling radically will yield the same 
sample variance in comparisons and the services of a trained statistician should 
be sought for further tests of this phenomenon. 

The complete results of a comparison of a high speed sampler with a 2.5 cm 
opening, towed in a horizontal series at 10, 20, 30, and 40 meters, and an oblique 
tow with a one-meter ring net towed from 70 meters to the surface is available in 
Ahlstrom, et al (1958). In this teetv conducted over a 72,000 square nautical mile 
area in May of 1950, comparisons were made of total plankton volume, sardine eggs 
and larvae, anchovy larvae, jack-mackerel larvae, lanternfishes and other fish 
larvae* The horizontal distribution 8&ps from this study are jshown in Figure 3.31. 
From these comparisons one can sse that the centers of abun<lancfe' and distributional 
boundaries were wel 1-repre sen ted ty both nets. The high-spe*d net was ultimately 
rejected because it did not catch sufficient numbers of large 'iarvae in its 40 nH 
sample relative to those caught by the larger slower net with a sample 10 times as 
large, and because the high-speed towing apparatus at that time was not capable of 
being fished to the depth required to sample under the distributions of larvae in 
this region* Nonetheless, the results from this comparison can be used to design 
adequate comparisons of standard sampler performance, 

3.10. Design of Surveys to Minimize Costs 

Table 3.18 lists the various cost elements of the CalCOFI surveys. Obviously, 
the increased efficiency of surveys must depend on reducing the major costs without 
substantially reducing the precision. Cursory inspection indicates that the great- 
est cost elements are shiptime, administration and technical personnel. Basic re- 
reductions in ship's costs can be made by: (1) reducing the number of cruises par 
year, (2) reducing the number of lines per cruise, and (3) reducing the number of 
stations per line . the reoi:muuindd procedure is to conduct thorough 

cruise schedules in the exploratory and analytical phases of the egg and larva sur- 
vey, then use tha data fro* these phases to design the most affective and efficient 
monitoring cruises, it is possible that complete surveys need to be made only each 
two or three years with awpplwwmtary data being taken in the interim years. 



93 



Also, it nay be convenient or necessary to merge sea operations of basic ocean- 
ography, exploratory fishing, acoustic surveys, pollution monitoring, or data buoy 
tending with the operations of egg and larva surveys. This can result in consider- 
able avUwr* in we Sea-going phase of the egg and larva survey or alternately mav 
provide extra samples 'vith which to augment the standard survey. One must take pre- 
cautions that the adventitious sampling program is an unbiased and representative 
set of samples. For example, one might need to adjust data from groups of samples 
taken only at night to compare with samples taken at all times of a 24-hour day 
Samples taken over a small area may not be representative of a large area. Provided 
the sampling characteristics of the joint surveys are mutually satisfactory to the 
cooperators, massive savings are possible. For example, the CalCOFl surveys have 
profitted from this joint planning and conduct, and the potential now exists for the 
extensive environmental data to be used with the larval survey data to design ex- 
periments to test the relationships among stock size, larval survival, juvenile 
survival and recruitment. 

Table 3.18. Estimated cost of a decade of ichthyoplankton surveys conducted 
over 300,000 square nautical miles of temperate, eastern boundary 
current off the west coast North America (CalCOFl 1972-base dollars). 



CATEGORIES 



PERSONNEL DAYS/YEAR S/DAY YEARS 



DIRECT SURVEY COSTS 
Biological Aids 
Biological Technicians 
Fishery Biologist 
50 m Research Vessel 



9 

34 
1 
2 



240 
240 
240 
200 



25 

SO 

100 

2500 



10 
10 
10 
10 



$ 540,000 

4,080,000 

240,000 

10,000,000 



SURVEY INTERPRETATION 

Fecundity Estimate 

Biological Technician 1 240 
Fishery Biologist 1 240 

Temperature Specific Hatching and Development Times 
Biological Technician 1 240 
Fishery Biologist 1 240 

Data Processing and Statistical Analysis 
Fishery Biologist 1 240 
Biological Technician 1 240 
Computer Time 240 

Decade Total 

Administrative Overhead (50%) 
TOTAL 



50 
100 

50 
100 

100 

50 

2 



2 
2 

3 
3 

4 
4 
4 



24,000 
48,000 

36,000 

72,000 

i 

96,000 

48,000 

1,920 

15,185,920 

7,592,960 

22,778,880 



4. ' REFERENCES t ( , . 

Ah 1 mtrom, iUH., Studies on the Pacific pilchard or sardine (Sardinop caerulea) 4. 
1943 Influence of temperature' on the rate of development of pilchard egg's in 
nature. Spec . Sc i Rep . DSFWS , (23};26p. 

t Distribution and abundance of egg and larval populations of tbei 
1954 Pacific sardine. Fish Bull USFWS , 56:82-140 



, Vertical distribution of pelagic fish eggs and larvae off California 

T551 ""and Baja California, Fish, Bui l.USFWS, 60*106-46 

, , Co-occurrences of sardine and anchovy larvae in the California Current 

TSITT region off California and Baja California. Rep.CCOFI, (ll):117-35 



f An evaluation of the fishery resources available to California 



1968 fishermen. Univ. Hash. Publ. Fish. (New Ser.) , (4) $65-80 

(ed.), Recommended procedures for measuring the productivity of plankton 

1969 standing stock and related oceanic properties. 5. Sampling zooplankton 
to determine biomass. Washington, D.C., National Academy of Sciences, 
pp. 47-59 

, Kinds and abundance of fish larvae in the eastern tropical Pacific, 

TJ7T based on collections made on BASTROPAC I* Fish. Bui l.MOAA/MUFS, 69(1): 



3-77 

t Kinds and abundance of fish larvae in the eastern tropical Pacific on 
1972 the second multi-vessel EASTROPAC survey, and observations on the annual 
cycle of larva* abundance. Fish. Bull. NOAA/NMFS, 70 (4) :1153-242 

Ahlstrom, E.H. and J*K. Thrailkill, Plankton volume loss with time of preservation. 

1963 Rep.CCOFI, (9) t57-73 

Ahlstrom, E.H. et al., High speed plankton sampler. Fiah.Bull.USFWS, 58:187-214 
1958 

Ahlstrom, E.H. et al.. Sampling zooplankton to determine biomass. In Recommended 
1969 procedures for measuring the productivity of plankton standing stock and 
related oceanic properties, edited by E.H. Ahlstrom. Washington, D.C., 
National Academy of Sciences, 59 p.? 

Aitchison, J. and J.A.C. Brown, The Pognormal distribution. Cambridge, University 
1966 Press 

Anscombe, F.J., 'The statistical analysis of insect counts based on the negative 
1949 binomial distribution. Biometrics, 5:165-73 

Bagenal, M. , A note on the relations of certain parameters following a logarithmic 
1955 transformation. J.Mar.giol.Assoc.U.K. , 34:289-96 

Barkley, R.A., The theoretical effectiveness of towed-net samplers as related to 

1964 sampler size and the swimming speed of organisms, J.Cons.CIEM, 29(2): 
146-57 

Selectivity of towed net samplers. Fish. Bull. NQAA/KMFS, 70:799-820 
1972 

Sever ton, R.J.H. and S.J. Holt, On the dynamics of exploited fish populations. 
19S7 Fish. Invest. Minis t.Aqric.yish. Food G.B. (2 Sea Fish.) , (19)i533 p. 

Beverton, R.J*H, and D*6* Tungate, A multi-purpose plankton sampler, J.Cons.CISM, 
197 



95 



Birge, B.A, , Plankton studiea on Lake Kendota. 1. The vertical distribution of the 
W95 pelmtflc cruetacea during July, 1894. Trans.wis.Aead, Sci. Arts Lett., 10 
421-84 

Bjorke, H., 0. Drageaund and 0. Ulltang, Efficiency test on four high-speed plankton 
1974 samplers. In The early life histoty of fish, edited by J.H.S. 61axter. 
Berlin, Sprlnger-Verlag, pp. 183-200 

Bridget, J.P., On efficiency tests made with a modified Gulf III high-speed tow-net* 
1958 J.Cons.CIEM, 23:357-65 

Cassie, R.M., Sample design. Unesco Monogr . Oceanogr . Methodol . f (2)il05-21 
1968 

Clutter, R.I., The macrodiatribution and social behaviour of some pelagic myeid 
1969 shrimps. J.Exp.Mar.Biol.Ecol. , 3:125-55 

Clutter, R.I. and M. Anraku, Avoidance of samplers. Unesco Monogr . Oceanogr . Methodol . , 
1968 (2):57-76 

Colebrook, J.M., Variability in the plankton. Progr . Oceanogr . , 5:115-25 
1969 

t The design of sampling surveys. FAQ . Fish . Tech . Pap . , (122):52-8 



1973 

Gushing, D.H. and J.G.K. Harris, Stock and recruitment and the problem of density- 
1973 dependence. Rapp.P.-V.R<Sun.CIEM, 164:142-56 

Davies, I.E. and E.G. Barham, The Tucker opening-closing mikronekton net and its 
1969 performance in a study of the deep scattering layer. Mar.Biol. , 2(2): 
127-31 

Ege, V., Sudidae (Paralepis) . Rep . Dan . Oceanogr . Exped . Mediterr . , 2A (13):193 p. 
1930 

, Paralepididae. 2. (Macroparalepsis) . Taxonomy, ontogeny, phylogeny and 



apsis) 

1953 distribution. Dana Rep. , (40) ;184 p. 

_, Paralepididae. 2. (Macroparalepais) . Taxonomy, ontogeny, phylogeny and 

"distribution . Dana Rep., (43):101 p. 



T957 

Fager, E.W., Estimation of mortality coefficients from field samples of zoop lank ton. 
1973 Limnol. Oceanopr., 18 (2) : 297-301 

Fagetti, E., Distribution and relative abundance of Clupeidae and Engraulidae larvae 
1970 in the waters of the continental shelf of Senegal and Gambia during 1969. 
Rapp . Sci . Provis . Cent . Rech . Oceanogr . Dakar-Thiaroye , (1/70) :26 p. 

Fisher, R.A. , The negative binomial distribution. Ann.Eugen. , 11:182-7 
1941 ~ - 

Forbes, S.T. and 0. Nakken, Manual of methods for fisheries resources survey and 

1972 appraisal. Part 2. The use of acoustic instruments for fiah detection and 
abundance estimation. FAQ Man. Fish. Sci. , (5):138p. 

Fraser, J.H., Variability in the oceanic content of the plankton in the Scottish area. 
1$69 Progr.Oceanoqr. , 5:149-59 

Gehringer, J.W. , An all-metal plankton sampler (model Gulf III). Spec . Sci . Rap. USFWS 

TL -- - 



Gehringer, J.W., The Gulf fit and other high-speed plankton samplers. Rapp.PrV.R6un. 

- "^** 



19*2 CIEM, 153*19-22 



96 * 

Gj^saater, 4, ad JU Q+*tx* f The use of data on eggs and larvae for 

1974 spawning stock of iUh populationt wil:h demersal eggs. In Th early Ufa 
history of fieh, edited by J.H.8. fclaxter* Berlin, Springer-Verlag, 
pp. 139-49 

Griag-flmith, P. , Tha uaa of random and oontiguoua quadra ta in tha atudy of structure 
1982 of plant communities. Ann* lot., 16:293-316 

Gulland, J.A. y Manual of sampling and atatiatical mat hod for fishariaa biology. 
1966 Part 1. Sampling methods. FAQ Man.Fiah.8ci. , (3)87 p. (Repr.1971) 

Manual of methoda for fiah stock assessment. Part 1* Fiah population 



1969 analyais. FAQ Man. Fish. Sci. , (4)tl54 p. 

(comp.)/ Tha fiah raaourcaa of tha ocaan. FAQ Fiah Tech. Pap. , (97) i 



1970 425 p. 

Harding, P. , and G,P. Arnold, Fluma axparimant* on tha hydro-dynamic a of tha Jx>watoft 

1971 high-apaad plankton aamplara. J.Cona.CIEM, 34(l):24-36 

Hatanaka, M. , Biological atudiaa on tha population of tha aaury, Cololabia aaira 

1956 (Bravoort) . 1. Raproduction and growth. Tohuku J.Agric.Raa. , f (3) i227-69 

Hayaahi, S., Fiahary biology of tha Japaneaa anchovy, Engraulia japonica (Houttuyn) . 
1961 Bull.Tokai Reg. Fish. Ra. Lab. , (31):145-268 

Hampal, G. (ed.) Fiah eggs and larval atudiaa* (Contributions to a manual). 
1973 FAQ Fi ah. Tech. Pap. , (122) *82 p. 

Hempal/ G. and H. Weikert, The neuaton of tha subtropical and boreal northeastern 

1972 Atlantic Ocean: a review* Mar.Biol., 13(1): 70-88 

Henaen, V., Methodik der Dntersuchungen bei Planktonexpedition. Ergebn . Atl . Ozean . 
1895 Plank tonexped * Hugiboldt-St if f ing , 1:200 p. 

Henaen, V. and C. Apstein, Die Hordsee- Expedition .1895 des Deutschen See fischerei- 
1897 Vereins. Uber die Eimenge der im Winter laichenden Fische. Wise. 
Meeresunters . Helgoland , 2(2)>101 p. 

Heron, A.C., Plankton gauze. Unesco' Monogr.Oceanogr .Methodol. , (2) :19-26 

1968 ; 

Hirano, T and M. Fujimoto, Preliminary results of investigations of the Kuroshio 
1970 functioning as a means of transportation and diffusion of fish eggs and 
larvae. In The Kuroshio, a Symposium on the Japan Current, edited by 
J*C. Marr, Honolulu, East-West Center Press, pp. 4 05-16 

ICNAF, North Atlantic fish marking symposium, Woods Hole, Mass., May 1961. 
1963 Spec. Pub 1. ICNAF. (4):370 p. 

Isaacs, J.D., Night-caught and day-caught larvae of the California sardine. 
1965 Rep.CCOFI, (10) :102-40 

Isaacs^ J.D., A. Fleminger and J.K. Miller, Distributional atlas of zooplankton bio- 
1969 mass in the California Current region : Spring and Fall, 1955-59. 
Atlas CCOFI, (10);252 p. 

Josai, J.W. , Annotated bibliography of zoop lank ton sajnpling devices* SOAA 

^^ 



1970 KMFS/SSRF, (609) : 90 p. 




Klawe/ W.t,,, Cteservations on the spawning of four species of tuna, Neothunnus roacro- 

Katsuwonus pe lamia, Auxia tha sard and Euthynnus ^ 

on the distribution of their 
6 



97 



* rlot 



Kramer, D. and J.R. Zveifel, Growth of anchovy larvae (Engraulis roordax Girard) in 
IffTO thl laboratory as influenced by temperature. ftep.<S(J6fl, <14>:64-7 

Kramer, D. and P.E. Smith, Seasonal and geographic characteristics of fishery 

1971 resources : California Current region. 5. Northern Anchovy Coramer.Fish. 
Rev, , 33(3):33-8 - 

Kramer, D. et al. , Collecting and processing data on fish eggs and larvae in the 

1972 California Current region. NOAA Tech . Rep . NMFS Circ . , (370) :38 p. 

Langmuir, I., Surface motion of water induced by wind. Science, N.Y. , 87*119-23 
1938 

Lasker, R. , An experimental study of the effect of temperature on the incubation 
1964 time, development, and growth of Pacific sardine embryos and larvae. 
Copeia, 1964 (2) :399-405 

Lenarz, W.H., Mesh retention of larvae of Sardinoos caerulea and Enaraulis mordax 
1972 by plankton nets. Fish. Bui l.NOAA/NMFS, 70 (3) ; 839-48 

Dependence of catch rates on size of fish larvae. Rapp.P.-V.R^un. 

CIEM, 164:270-5 



Lloyd, M., "Mean crowding". J.Anim.Ecol. , 36:1-30 
1967 

Longhurst, A.R., Vertical distribution of zooplankton in relation to the eastern 

1967 Pacific oxygen minimum. Deep-Sea Res . , 14:51-63 

MacGregor, J.S., Fecundity of the Pacific sardine (Sardinops caerulea). Fish. Bull. 

1968 USFWS, 57:427-49 

Mackintosh, N.A. , Distribution of the macroplankton in the Atlantic sector of the 
1934 Antarctic. Discovery Rep. , 9:65-160 

Mahnken, C.V.W. and J. Jossi, Flume experiments on the hydrodynamics of plankton 
1967 nets. J, Cons .CIEM, 31(l):38-45 

Mat*umoto, W.M. et al., On the clarification of larval tuna identification parti- 
1972 cularTyTn the genus Thunnus . Fish. Bull. NOAA/NMFS, 70(1):1-12 

May, R.C., Feeding larval marine fishes in the laboratory : a review. Rep.CCOFI, 
1970 (14) t73-83 

Effects of delayed initial feeding on larvae of the grunion 



Leuresthes tenius (Ayres). Fish. Bull. NOAA/NMFS, 69(2):411-25 

McBwen, G.F., M.w. Johnson and T.R. Folsom, A statistical Analysis of the performance 
1954 of the Folsom plankton sample splitter based upon test observations. 
Arch . Meteorol . Geophys . Bioklireatol (A Meteorol Geophy a . ) , 7 $502-27 

J,A. and D.M. Brown, A new opening-closing paired zooplankton net. 
SIO Ref . (Scripps Inst .Oceanogr . ) , (66-23) 

Milne, A. A/, The centric systematic area-sample treated as a random sample. 
Biometrics. 15 (?) 1 270-97 



Moser, E.G., and B*H* Ahlstrom, Development of lanternfishes (family Myctophidae) 
1970 in the California Current. Part 1. Species with narrow-eyed larvae. 
Bull.Ixjs Ana.Ctv.Mus.Nat.HiBt.Sci. , (7) :145 p. 



9 

Moaer, H.G, *nd l.H. Ahlstrom, Development of the lantrnfih (Scogjlopsii 
1172 punctmtu* Brauar ifQff) ; with 4 41fCWf**n 11^ r^lwaitlql 

irTZZ lamily Myctophidae and it* *ole ift "ttf* <tio#rt*& fieeMiasm fr the 
volution of photdphore pattern* in lanternfishes, yjgfa,Biill>liaM/t<MFSr 



., , 
Motoda, s. et al., Proposed method 6f estimation b* tjuaiifelty o^ water filtered by 

1M3 vertical net h*ul, od ita Application on illustrated distribution of zoo- 
plankton biooass in tl$e eaat^rn Indian Ocean, j&t* dA ^tlanktol feg*n r 

' U0) ' ' 




Murphy ,!., Poiwlatlon biiiy ;^f ^^Wt3m^^ 

- - Calif. Acad.aci., 34il-4 > - T "' J! ' : " ' ' -' ' ' ' 

- ' - 



Murphy, G.I. and R.I. Clutter, Sailing aachoyy larvae wit*^a fUnkton purse seine. 
W12 Fish. Bull . MOAA/MiFS , 70lltt-M "<v:r v -v : -xl''//* J. ^v . ., 

Nell en, W. and G. Herapel, Versuche zur Flngigkeit des *Hai" und des roodifizierten 
1969 Gulf-V-Plktan*toplers *!laekt4Bi". ger.)^sai^Bcei.Meeresforsch. , 
20(2)il4l-54 

Nellen, w. and R.C. Newell, Marine plankton : a practical guide. London, Hutchinson 

1963 Educational, 207 p. 

Owen, R.W. Jr., Small-scale horizontal vortices in the surface layer of the sea. 
1966 J. Mar. Res., 24(1): 56-66 

Parin, N.V. , Ichthyo^auna of the eplpelagic *one. Moskva, Izdatel'-stov Nauka, 
1968 186 p. Issued also in English by Israel Program for Scientific Trans- 
lations. Jerusalem, IPST Cat. No. 55 28: 210 p. (1970) 

Parrish, R.H. , Exploitation and recruitment of Pacific mackerel. Scomber japonicus 
1974 in the north eastern Pacific. Rep.CCOPI, (17):136-40 

Perta&va-Ostroujnova, T.A. , Some morphological characteristics of myctophid larvae 

1964 (Myctophidae, Pisces) . In Fishes of the Pacific and Indian Oceans, 
biology and distributionTedited by T.S. Ross, Tr.Inst.Okeanol. , 73: 
79-97 (in Russian) 

Pielou, B.C., An introduction to mathematical ecology. New York, Wiley, 286 p. 
1969 

Posgay, J.A., R.R. Marak and R.C. Hennemuth, Development and test of new zooplankton 
1968 samplers. ICNAF Res, Doc. (85) (mimeo) 

Postuma, K.H. and J.J. Zijlstra, Larval abundance in relation to stock size, 

1974 spawning potential and recruitment in North Sea herring. In The early 
life history of fish, edited by J.H.&* Blaxtar. Berlin, Springer- Ver lag, 
pp. 113-28 

Reighard, J. , Methods of plankton investigation in their relation to practical 
1898 problems, tell. U.S. Fish. Coam. , (17)*l9-tS 

Richards, w.J. and D.C Siimpns^ Distribution of tuna larvae ( Pisces / Scombridae) 

1971 in the Northwestern dulf of Guinea and oft Sierra Leone. Hsh f ull>U8FlfS, 



Saneo to < D < D . 

^^ . T- 1 ' ' 1. ^iv. r"^^ r ^' TT ^^ PT '^"^j*" nF ^^T^f!,u j^ 
W69 J. Fish. ftes. Board Can. , M 



Santander, H. and O.S. de Castillo, 1 dkwoyis 
1969 en los periodos reproductiyos de 1< 



Saville, A. 
1959 




99 



Savilli, A,, I.G. Baxter and D.fr. McKay, Relations between egg production, 

1974 production and spawning stock size in Clyde herring. In the early life 
history of fish, edited by J.H.S. Blaxter. Berlin, SprTnger-Verlag, 

- : - ' * ' *' 



Schaeffer, M.B., A study of the dynamics of the fishery for yellowfin tuna in the 
19S? eastern tropical Pacific Ocean. Bull.I-ATTC, 2(6):247-8S 

Schnack, D. r On the reliability of methods for quantitative surveys of fish larvae. 
1S74 IA *h* early life history of fish, edited by J.H.S. Blaxter. Berlin, 
Springer-Verlag, pp. 201-12 

6ett<fe, O.E. and E.H. Ahlstrom, Estimations of abundance of the eggs of the Pacific 
1948 pilchard (Sardinops caerulea) off southern California during 1940 and 
1941 - J-Mar.Res., 7(3):511-42 

Silliman, R.P., Analog computer models of fish populations. Fish. Bull. USFWS, 66(1) i 

1967 31-46 - 

Simpson, A.C., The spawning of plaice in the North Sea. Fish. Invest. Minist.Agric. 
1959 Fish. Food G.B. (2 Sea Fish.), 22(7);111 p. - - 

Smith, P.E., The increase in spawning biomass of northern anchovy Engraulis mordax* 
1972 Fish. Bull .NOAA/NMFS , 70{3):849-74 -- 

, The mortality and dispersal of sardine eggs and larvae. Rapp.P.-V. 
TST5 R&in.CIEM, 164:282-92 - 

Smith, P.E. and R.I. Clutter, Hydrodynamics of flow and collection in plankton nets. 
1965 In Ocean science and ocean engineering. Transactions of the Joint 

conference and exhibit, 14-17 June 1965, sponsored by the Marine Techno- 
logy Society and the American Society of Limnology and Oceanography. 
Washington, D.C., Marine Technology Society/ASLO, vol. 1:515 (Abstr.) 

Smith, P.E., E.H. Ahlstrom and H.D. Casey, The saury as a latent resource of the 

1970 California Current. Rep.CCOFI, (14) :88-130 

Smith, P.E., R.C. Counts and R.I. Clutter, Changes in filtering efficiency of 

1968 plankton nets due to clogging under tow. J.Cons.CIEM, 32(2):232-48 

Smith, P.E., J.R. Thrailkill and A.N. Vrooman, Cruise report, J-59. La Jolla, 

1971 California, Southwest Fisheries Center, NOAA/NMFS (Unpubl.) 

Southwood, T.R.E., Ecological methods. London, Methuen, 391 p. 
1966 

Squire, J.L., Jr., Apparent abundance of some pelagic marine fishes off the southern 

1972 and central California coast as surveyed by an airborne monitoring 
program. Fish. Bull. NOAA/NMFS, 70 (3) :1005-20 

Steedman, H.R. (ed,), Zooplankton fixation and preservation. Unesco Monogr . Qceanoqr 
1976 Methodol . , (4):350p. 

Sund, P.N. and W.J. Richards, Results of trials with neuston net in the Gulf of 

1965 Guinea. In Ocean science and ocean engineering, Transactions of a Joint 
conference and exhibit, 14-17 June 1965 sponsored by the Marine Techno- 
logy Society and the American Society of Limnology and Oceanography. 
Washington, D.C., Marine Technology Society/ASLO, vol. 1:515-23 

Switaer, P., Reconstructing patterns from sample data. Ann.Math.Stat. , 38*138-54 
1*67 

, N.W. and R.C. Clelland, Non-parametric and short-cut statistics. Danville, 
1957 111., Interstate Publishers 



' Trantir, D, J. Cl. ) , ttooplftnkton sampling. ' Unwo MontxTT.OcaanQgr.tUttoacil. . ' (2) t 
me 174 p. 

"* ,-!! " , , ,, ^ . , ' '. " , / 

D.J. and P*B. Smith r riltration performance* Unasoo 
l>* Mmthodol > , (2):2f^56 



H. , Lo of organism* through the meshes. Uneaco Monogr >Oceanogr ^Hethodol . 

- - - 



Cruiie report. La Jolla, California, Southwest Fisheries Center, 
m? HOIA/KMFS (Dnpubl.) 

Vrooan r A.M. and P.E. Smith, Biomass of the subpopulation of northern anchovy 
1972 (Enoraulifl mordax Girardi Rep, CO OF I f (i5) : 4 9-51 

Saitsev, Yu.P. f Marine neustonology . Kiev, Naukov Duraka, 264 p. (in Russian) 



Siilstra, J.J. f Herring larvae in the central North Sea. Ber . Dtsch . Wise . Koram . 
ft70 Meeresforsch. , 21:92-115 

Zweifel, J.R. f A non-parametric approach to the estimation of relative change in fish 
1973 population size from egg and larval surveys. Rapp.P.-V.R&m.CIEM, 164: 

-