U.S. Department of Commerce Volume 111 Number 1 January 2013 Fishery Bulletin U.S. Department of Commerce Rebecca M. Blank Acting Secretary of Commerce National Oceanic and Atmospheric Administration Jane Lubchenco, Ph.D. Under Secretary for Oceans and Atmosphere National Marine Fisheries Service Samuel D. Rauch III Acting Assistant Administrator for Fisheries SrATES 0* * The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE, Seattle, WA 98115-0070. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $32.00 domestic and $44.80 foreign. Cost per single issue: $19.00 domestic and $26.60 foreign. See back for order form. Scientific Editor Bruce C. Mundy Associate Editor Kathryn Dennis National Marine Fisheries Service Pacific Islands Fisheries Science Center Aiea Heights Research Facility 99-193 Aiea Heights Drive, Suite 417 Aiea, Hawaii 96701-3911 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE Seattle, Washington 98115-0070 Editorial Committee Richard Brodeur John Carlson Kevin Craig Jeff Leis Rich McBride Rick Methot Adam Moles Frank Parrish Dave Somerton Ed Trippel Mary Yoklavich National Marine Fisheries Service, Newport, Oregon National Marine Fisheries Service, Panama City, Florida National Marine Fisheries Service, Beaufort, North Carolina Australian Museum, Sydney, New South Wales, Australia National Marine Fisheries Service, Woods Hole, Massachusetts National Marine Fisheries Service, Seattle, Washington National Marine Fisheries Service, Auke Bay, Alaska National Marine Fisheries Service, Honolulu, Hawaii National Marine Fisheries Service, Seattle, Washington Department of Fisheries and Oceans, St. Andrews, New Brunswick, Canada National Marine Fisheries Service, Santa Cruz, California Fishery Bulletin web site: www.fisherybulletin.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 111 Number t January 2013 Fishery Bulletin Contents Articles 1 —12 Cowen, Robert K., Adam T. Greer, Cedric M. Guigand, Jonathan A. Hare, David E. Richardson, and Harvey J. Walsh Evaluation of the In Situ Ichthyoplankton Imaging System (ISIIS): comparison with the traditional (bongo net) sampler Companion articles The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprie- tary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, rec- ommends, or endorses any propri- etary product or proprietary mate- rial mentioned herein, or which has as its purpose an intent to cause directly or indirectly the advertised product to be used or purchased be- cause of this NMFS publication. The NMFS Scientific Publications Office is not responsible for the contents of the articles or for the standard of English used in them. 13-26 Bromaghin, Jeffrey F., Monique M. Lance, Elizabeth W. Elliott, Steven J. Jeffries, Alejandro Acevedo-Gutierrez, and John M. Kennish New insights into the diets of harbor seals ( Phoca vituhna) in the Salish Sea revealed by analysis of fatty acid signatures 27-41 Howard, Sarah M. S., Monique M. Lance, Steven J. Jeffries, and Alejandro Acevedo-Gutierrez Fish consumption by harbor seals ( Phoca vituhna ) in the San Juan Islands, Washington 42-53 Rose, Craig S., Carwyn F. Hammond, Allan W. Stoner, J. Eric Munk, and John R. Gauvin Quantification and reduction of unobserved mortality rates for snow, southern Tanner, and red king crabs ( Chionoecetes opilio, C. bairdi, and Parahthodes camtschaticus) after encounters with trawls on the seafloor 54-67 Laidig, Thomas E., Lisa M. Krigsman, and Mary M. Yoklavich Reactions of fishes to two underwater survey tools, a manned submersible and a remotely operated vehicle Fishery Bulletin 111(1) 68-77 Weber, Thomas C., Christopher Rooper, John Butler, Darin Jones, and Chris Wilson Seabed classification for trawlability determined with a multibeam echo sounder on Snakehead Bank in the Gulf of Alaska 78-89 Staaf, Danna J., Jessica V. Redfern, William F. Gilly, William Watson, and Lisa T. Ballance Distribution of ommastrephid paralarvae in the eastern Tropical Pacific 90-106 Burchard, Katie A., Francis Juanes, Rodney A. Rountree, and William A. Roumillat Staging ovaries of Haddock (Melanogrammus aeglefinus): implications for maturity indices and field sampling practices 107 Errata 108-109 Guidelines for authors 1 Evaluation of the In Situ Ichthyoplankton Imaging System (ISIIS): comparison with the traditional (bongo net) sampler Email address for contact author rcowen@rsmas.miami edu 1 Rosenstiel School of Marine and Atmospheric Science University of Miami 4600 Rickenbacker Causeway Miami, Florida 33149 Abstract — Plankton and larval fish sampling programs often are limited by a balance between sampling fre- quency (for precision) and costs. Ad- vancements in sampling techniques hold the potential to add consider- able efficiency and, therefore, add sampling frequency to improve preci- sion. We compare a newly developed plankton imaging system, In Situ Ichthyoplankton Imaging System (ISIIS), with a bongo sampler, which is a traditional plankton sampling gear developed in the 1960s. Com- parative sampling was conducted along 2 transects -30-40 km long. Over 2 days, we completed 36 ISIIS tow-yo undulations and 11 bongo oblique tows, each from the surface to within 10 m of the seafloor. Over- all, the 2 gears detected comparable numbers of larval fishes, represent- ing similar taxonomic compositions, although larvae captured with the bongo were capable of being identi- fied to lower taxonomic levels, espe- cially larvae in the small (<5 mm), preflexion stages. Size distributions of the sampled larval fishes differed considerably between these 2 sam- pling methods, with the size range and mean size of larval fishes larger with ISIIS than with the bongo sam- pler. The high frequency and fine spatial scale of ISIIS allow it to add considerable sampling precision (i.e., more vertical sections) to plankton surveys. Improvements in the ISIIS technology (including greater depth of field and image resolution) should also increase taxonomic resolution and decrease processing time. When coupled with appropriate net sam- pling (for the purpose of collecting and verifying the identification of biological samples), the use of ISIIS could improve overall survey design and simultaneously provide detailed, process-oriented information for fish- eries scientists and oceanographers. Manuscript submitted 8 December 2011. Manuscript accepted 21 September 2012. Fish. Bull. 111(1): 1—12 (2013). doi:10.7755/FB. 11 1.1.1 The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessar- ily reflect the position of the National Marine Fisheries Service, NOAA. Robert K. Cowen (contact author)' Adam T. Greer' Cedric M. Guigand' Jonathan A. Hare2 David E. Richardson2 Harvey J. Walsh2 2 Northeast Fisheries Science Center National Marine Fisheries Service Narragansett Laboratory 28 Tarzwell Drive Narragansett, Rhode Island 02882 Regular surveys of early life stages of fishes provide a wealth of informa- tion for fisheries managers and fish- ery oceanographers. Indices of larval abundance are used quantitatively as fishery-independent measures of population abundance in stock as- sessments (Scott et al., 1993; Gledhill and Lyczkowski-Shultz, 2000; Sim- monds, 2009). Larval fish abundance also is used qualitatively, as evidence for change in stock status (Smith and Morse, 1993; Lo et al., 2010; Richardson et al., 2010). Spawning areas and times are inferred from early-life-stage abundance and dis- tribution, and they contribute to the definition of essential fish habitat (Brodziak, 2005; Levin and Stunz, 2005) and stock identification (Begg et al., 1999; Hare, 2005). Larval fish surveys combined with process- oriented research also help forecast- ing capability of year-class strength (e.g., Megrey et al., 1996; Lough and O’Brien, 2012). Although larval fish studies make substantial contributions to the as- sessment of fish stocks, 3 factors currently limit their applicability. First, larval fishes are relatively rare within the plankton and estimates of variance in larval abundance can be large, limiting the power of sta- tistical comparisons of abundance between years or locations (Cyr et al., 1992). Second, larval fishes are patchily distributed (e.g., Davis et al., 1990; Cowen et al., 1993; Pe- pin, 2004) but not randomly distrib- uted; patches often are associated with fronts, thermoclines, or specific water masses (Cowen et al., 1993; Kingsford and Suthers, 1994). Most larval surveys, however, are conduct- ed along fixed grids or as random stratified designs; significant differ- ences in larval abundance between sampling times may simply reflect a varying intersection of sampling with dynamic larval habitat. Third, the cost of ichthyoplankton surveys is an important consideration and most programs are cost-limited in terms of ship time or the number of samples that can be processed (Tanaka, 1973; Lo et al., 2001; Simmonds, 2009). In the United States, there are numerous federally supported ich- thyoplankton programs that provide 2 Fishery Bulletin 1 1 1 (1) data for fisheries management. All these efforts are limited by the 3 factors described above: rarity, patchi- ness, and cost. The In Situ Ichthyoplankton Imaging System (ISIIS; Cowen and Guigand, 2008) has the po- tential to minimize all 3 limitations, and, if successful, would provide the stock assessment toolbox with robust and timely fishery-independent measures of spawning distribution and stock size based on early-life-stage in- formation. The overall goal of this study, therefore, was to evaluate the effectiveness of ISIIS for quantifying fish larvae and thus show the potential benefits of its integration into larval surveys, with the ultimate goal of improving stock assessments. Specifically, we compare ISIIS with a traditional bongo sampler, which is composed of a frame support- ing paired nets with mouth openings on either side of and in front of the towing wire (Posgay and Marak, 1980). The bongo has been used in ichthyoplankton programs throughout the United States since its de- velopment in the late 1960s: in the shelf ecosystem of the northeastern United States since 1971 (Richardson et al., 2010), in the Gulf of Mexico since 1982 (Lycz- kowski-Shultz and Hanisko, 2007), and in the north- east Pacific Ocean since 1972 (Matarese et al., 2003). Here we present a comparison of larval fish abundance and size distribution based on results from the ISIIS and bongo sampler. Methods This study was conducted 54 km south of Woods Hole, Massachusetts, (Fig. 1), on 23-24 October, 2008, on the NOAA Ship Delaware II. The cruise immediately followed the passage of a low-pressure system, which brought strong winds to the study area; these winds diminished throughout the duration of the cruise. Sam- pling was completed along 2 parallel transects, which were 41.4 and 27.7 km in length and separated by -6 km. To complete the comparison, the prototype ISIIS- 1 (herein referred to as ISIIS) was towed along a tran- sect; then the ship returned to the beginning of the transect, and net samples were made with the bongo over the same transect. Sampling along each transect encompassed both day and night periods, but no at- tempt was made to compare day and night differences in larval abundance or vertical distribution. Morse (1989) compared daymight catches in the region and found no significant differences for most of the taxa captured in this study. He did find some daymight bias at larger transect lengths, but, in our study, both the bongo net and ISIIS sampled during day and night, and therefore we assume this length bias was random- ly distributed between the gears. Sampling gear The imaging output from ISIIS is unique in that it pro- vides a continual image for the entire tow duration, with a pixel resolution of -68 pm. Such fine resolu- tion enables detection of particles as small as a 100 pm (e.g., diatoms), although the ability to clearly resolve particles is typically in the range of 700 pm (i.e., small copepods and larvaceans) and larger sizes (e.g., larval fishes, chaetognaths, and ctenophores). One distinctive feature of ISIIS is its large depth of field (—30 cm for Figure 1 Eight-day average (20-27 October 2008) sea-surface temperature (SST, °C) of northeastern U.S. continental shelf from Cape Hatteras, North Carolina, to Nova Scotia, Canada. (A) The sampling location offshore of Martha’s Vineyard, Massachusetts. (B) The inset shows the 2 In Situ Ichthyoplankton Imaging System (ISIIS) transects and the bongo collection locations marked by black dots along the same transects. Note the change in SST scale between the 2 panels. Cowen et at: Evaluation of the In Situ Ichthyoplankton Imaging System and comparison with the bongo-net sampler 3 mesozooplankton), which enables the concentration of even relatively rare mesoplankters, such as larval fish- es and gelatinous zooplankton, to be quantified (Cowen and Guigand, 2008; McClatchie et ah, 2012). Using the image analysis software that we have developed (Tsechpenakis et al., 2007, 2008), we could essential- ly quantify the plankton field for every centimeter of our tow, and we could match these data centimeter by centimeter with the corresponding environmental data collected by the onboard sensors (pressure [depth], temperature, salinity, and fluorometry). Consequently, ISIIS can evaluate from very fine-scale (centimeters) to submesoscale features. ISIIS sensors for this study were those for temperature (SBE 31 Sea-Bird Electron- ics, Inc., Bellevue, WA) and conductivity (SBE 4) and a fluorometer (ECO FLRT, WET Labs, Philomath, OR). A 61-cm bongo sampler was used and fitted with 505- and 333-pm mesh nets (Posgay and Marak, 1980). A flowmeter (General Oceanics, Miami, FL) was at- tached in the center of each mouth opening to quantify the volume of water filtered by the net. A conductivity, temperature, depth (CTD) instrument (SeaCAT SBE 19) was attached to the tow wire above the bongo net. The CTD was used in real time to monitor the depth of the bongo net during deployment. Sampling approach For this study, ISIIS was towed at a speed of 2.5 m s-1 in a tow-yo (vertically undulating) fashion between the surface and a target depth of 10 m above the seafloor, thereby following changes in seafloor depth. The ISIIS was towed in an undulating manner by paying cable in and out from the winch, and therefore continual winch operation was required. (Since this study, a self-undu- lating version of ISIIS has been designed and the need for continual winch operation has been eliminated). Each undulation (surface to depth to surface) took ~10 min, resulting in a distance covered of 1.5 km, which also equates to the distance between downcasts (or up- casts). While being towed, ISIIS records environmental data (temperature, salinity, fluorescence) and imagery continually, sending the data up the fiber-optic cable for onboard recording. The continual imagery is parsed into single images of 13x13 cm at a rate of 17.3 images s-1. Thus, ISIIS generates -64,000 images h1, and for this study, an estimated total of -478,000 images over -7.68 h of total recording time. Because the focus of this study was specifically lar- val fishes, processing of images specifically targeted lar- val fishes, thereby eliminating the need to capture and classify all imaged particles (e.g., copepods, larvaceans, medusae, and cfenophores). Consequently, all images were manually reviewed for larval fishes. This process is relatively rapid, although -3 months were required 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. to complete this task because of the large number of images. Future development of ISIIS will include auto- mated image processing; however, the current manual processing requires viewing each image. When a lar- val fish was present, that portion of the image was ex- tracted and saved to a file. All fish images were then reviewed for identification to the lowest taxonomic level possible and measured with ImageJ (National Institute of Health public domain Java-based image- analysis program available at http://rsbweb.nih.gov/ ij/). Environmental data from ISIIS were interpolated across each transect with a cubic interpolation function in Matlab (vers. 7.11.0.584 [R2010b], The MathWorks, Inc., Natick, MA). The depth and environmental vari- ables associated with each fish larva were obtained by matching time stamps from image and environmental data. The bongo tows were conducted in standard fashion by following Jossi and Marak (1983). For each tow, the wire was paid-out at a rate of 50 m min 1 to a depth of 10 m above the seafloor, then the wire was retrieved to the surface obliquely at 20 m min-1, while the ship moved at 0.75-1.0 m s-1. At completion of each tow, the nets were washed down and the contents rinsed onto a 333-pm sieve. The sample was preserved in 5% buffered formalin. Samples were then sorted for larval fishes under a dissecting microscope and identified to the lowest taxonomic level following Fahay (2007). The 333-pm mesh bongo samples were used for compari- sons of the bongo and ISIIS methods since this mesh size is the one that has been used for more than 20 years by the Northeast Fisheries Science Center for ichthyoplankton surveys. To compare larval fish concentrations, each bongo tow and each ISIIS undulation were treated as rep- licates. There are potential statistical problems with this assumption, but to date, the decorrelation length scale in ichthyoplankton distributions in the study re- gion has not been calculated. This assumption will be examined in future studies with ISIIS. The larval fish concentrations were transformed by the natural log, and a Shapiro test was performed to test for normal- ity of larval fish concentrations within each gear type. Where the null hypothesis of normality was accepted, a Welch’s f-test was used to compare larval fish con- centrations between transects within gear and then between gear across both transects. Comparisons were made for total larvae, family-level larvae, and species- level larvae both within and between gears for abun- dance and size differences. In these tests, the nonpara- metric Kruskal-Wallis test was used because concen- trations at the family level were zero-inflated, making transformations to a normal distribution impossible. All counts per tow (or undulation) were standardized to volume sampled (number of fish larvae per cubic meter). All larvae collected in the bongo net were measured to the nearest 0.1 mm for notochord (preflexion) or standard length under a dissecting microscope with 4 Fishery Bulletin 1 1 1 (1) an ocular micrometer. Larvae observed in ISIIS im- ages were measured digitally with Imaged software after each image was calibrated to standard pixel size. Fishes were measured for notochord or standard length (the position of the posterior end of the hypural plate was estimated if the pigmentation on a fish was too dense for the internal caudal fin structure to be vis- ible). A subset (6 out of 409) of the fish images was discarded because orientation of the fish precluded ac- curate measurement. Despite our effort to remove such images from measurement, some fish sizes likely were underestimated when the observer was not able to dis- cern the offset that may have occurred where the orien- tation was not exactly parallel to field of view. Lengths of all larvae were compared between the 2 gears and the 2 transects. To avoid pseudoreplication, the average length of all larvae, family-level larvae, and species- level larvae from a bongo tow or ISIIS undulation was used for comparison. Size distributions were all highly skewed, and therefore a Kruskal-Wallis test was used to compare sizes within and between gear types. Statis- tical analyses were performed in R software, vers. 2.14 (R Development Core Team, 2011) with the package “plyr” (Wickham, 2011) as well as visualization tech- niques with the package “ggplot2” (Wickham, 2009). Results Along 2 transects, we completed 24 and 12 ISIIS un- dulations and 6 and 5 bongo tows, respectively. ISIIS sampled an estimated 297 m3 h 1 (or an average of 63 m3 per tow-yo (i.e., down and up undulation), for a total sampled volume of 2281 m3. The actual volume sam- pled was lower than the maximum possible because of a slight misalignment in the mirrors that occluded 1 b b : 14.5 Transect 1 Transect 2 Latitude (°N) Figure 2 Fluorescence (voltage), temperature (°C), and salinity (ppt) measured from ISIIS along the western (transect 1) and eastern (transect 2) transects during 23—24 October 2008. Dotted lines in the fiuorometry panels represent the undulations of the In the Situ Ichthyoplankton Imaging System (ISIIS). The vertical solid lines represent the ap- proximate tow positions for the bongo sampler which was deployed along the length of the same transect once the ISIIS tow was completed. 40.7 40.8 40 9 40.7 40.8 40.9 14.5 -101 -20: -30 -40 -50: 40.7 40.8 40.9 i J33 | i! |32.5 Cowen et al Evaluation of the In Situ Ichthyoplankton Imaging System and comparison with the bongo-net sampler 5 Figure 3 Example of a full-frame image collected with the In Situ Ichthyoplankton Imaging System (ISIIS). Larval fish (small [~4 mm], Paralichthys dentatus ) and other plankters (especially copepods) are evident throughout. The small circular and elongate particles are diatoms (centric and pinnate) and diatom chains, which can be detected but are too small to clearly resolve. Also seen is a ~1.5-cm ctenophore with tentacles retracted. Several small aggregates (marine snow) are evident in the full-frame image. Overall, the full frame provides a good indication of the plankton field encountered by the observed larval fish. Sur- face is to the top of the image. Select plankters are shown to the right of the full frame in higher magnification (from top to bottom): (A) chaetognath (note that an improved image has been substituted for demonstration purpose only), (B) preflexion stage larval fish, (C) marine snow, (D) small copepod, (E) 2 copepods, (F) diatom chain (rotated to fit figure), and (G) copepod. about 15% of the imaging field (i.e., the image field of view was 11 cm versus 13 cm). In comparison, the typi- cal bongo sampled 137 m3 per oblique tow, for a total volume sampled of 1506 m3. The maximum depth of tows was 49 m for ISIIS tows and 52 m for the bongo tows. The water column along both transects was defined by limited vertical stratification, especially in its upper 35 m (Fig. 2). A slight decrease in chlorophyll concen- tration below a depth of -35 m in the inshore portion of the easterly transect was apparent and also was observed with a change in temperature and salinity; still, the differences were small. In contrast, consider- able horizontal variation (south to north) was observed in hydrography along both transects with tempera- ture lower, salinity lower, and chlorophyll fluorescence higher in the inshore (northern) portions than in the offshore (southern) portions (Fig. 2). The productivity of the water column was evident in ISIIS imagery as a preponderance of diatoms vis- ible throughout most images (Fig 3). Also imaged were a variety of invertebrate plankters, ranging from co- pepods and larvaceans to ctenophores and medusae to invertebrate larval types, such as echinoderm plu- teus. Because most imagery was dominated by the smaller plankton (diatoms, copepods, and larvaceans; 6 Fishery Bulletin 1 1 1 (1) 5 mm 5 mm 5 mm Figure 4 Examples of close-up, in situ images of different lar- val fish taxa imaged with the In Situ Ichthyoplankton Imaging System (ISIIS). (A) Paralichthys dentatus (4 mm); ( B) Gobiidae (8 mm); (C) Gadidae (32 mm); (D) Clupeidae (21 mm); (E) Merluccius spp. (14 mm); (F) unknown (preflexion stage) (3.2 mm). see Fig. 3), and larval fishes were relatively rare, the imagery provided a relative measure of abundance of different plankters. In most cases when fish larvae were encountered, the imagery was sufficient to dis- cern characteristics valuable for identification at the family or genus level (e.g., shape, number and location of fins, overall body shape, fish size, and, in some cases, certain skeletal features; see Fig. 4). The 2 sampling methods allowed us to detect com- parable quantities of larval fishes. ISIIS imaged a total of 409 larvae, and the bongo tows collected a total of 359 larvae. When standardized for the volume of wa- ter actually sampled, ISIIS estimated -0.18 fish larvae (±0.015 standard error of the mean [SE] nr3), a value that was not significantly different from the estimate from the bongo tows (0.24 ±0.037 SE nr3; P=0.074). Similarly, within gears, there were no differences in larval fish concentrations between transects. The estimates of larval abundance, however, were made on the basis of the 2 gears sampling different portions of the water column. The bongo net sampled all depths equally as it was towed from depth to the surface, but ISIIS spent less time at depths >40 m than at depths near the surface (Fig. 5A). This sam- pling effect is evident in the difference in measured fish abundance by depth (Fig. 5B), where the apparent pattern was for a continual increase in fish abundance with depth from the surface down to 40 m and then a decrease in abundance by depth beyond 40 m. This decrease was directly coincident with the drop-off in sampling time with depth by ISIIS. When an adjusted abundance was estimated by computing depth-specific concentrations (Fig. 50, then with the assumption of equal sampling effort per depth as with the bongo tows, an adjusted mean ISIIS fish concentration was 0.22 fish larvae rm3, which is very close to the bongo estimate. The taxonomic diversity collected by each gear also was similar; both collected larval fishes representing the same 7 families (Table 1), although bongo samples were typically identifiable to lower levels (genus and species) than those in ISIIS samples. Images of fish larvae from ISIIS were identifiable to at least the ge- nus level for -35% of larvae (143 out of 409). On the other hand, larvae were unidentifiable in 60 fish im- ages and most of these unidentifiable fishes were in the early preflexion stages (-15%); in contrast, all bongo tow larvae were identified at least to the family level. Comparison of the relative proportions of taxa between the 2 sampling methods indicates that they were simi- lar. There were a few notable exceptions: ISIIS under- estimated paralichthyids and scopthalmids and esti- mated relatively greater proportions of phycids and ophidiids than the bongo sampler. The total number of larvae sampled was similar, but it is not known if the “unknown” category would have evened these dis- crepancies or added further differences among certain taxa. Size distributions of larvae differed considerably be- tween the 2 sampling methods. ISIIS imaged a larger size range and larger mean size of fish larvae than the bongo sampler (Fig. 6, Table 2). This sampling gear pattern was evident across several individual taxa, no- tably the gadiform fishes, Phycidae and Gadidae, with the latter mean size from ISIIS samples being more than 3 times the mean size of this family from bongo samples (Table 2). There was also a significant differ- ence between gear types with respect to size of Para- lichthyidae, although this very small difference (0.103 mm) may not be biologically meaningful and likely was significant only because of the rank nature of the Krus- kal-Wallis test. There was a significant difference in overall larval size between transects for the ISIIS sam- ples, but there was no significant difference in overall larval size for the bongo tows between the 2 transects or for any taxonomic group between transect within gear type (Fig. 6, Table 2). Therefore, most of the dif- ferences in size were attributed to sampling gear. Cowen et al.: Evaluation of the In Situ Ichthyoplankton Imaging System and comparison with the bongo-net sampler 7 A o -10 £ -20 _c Q. 00 ''T Z o o CO co •'tf- Z o o CNJ oo Map of the San Juan Island region, where samples were collected for our investigation of the diet composition of harbor seals ( Phoca vitulina) in the Salish Sea. Harbor seals were captured in the vicinity of Padilla Bay, Bird Rocks, Vendovi Island, and the Belle Chain Islets. Sampling of predator and prey Harbor seals were captured from April 2007 to March 2008 at 3 sites in the San Juan Islands of Washington State and at a fourth site in the adjacent Gulf Islands in British Columbia (Fig. 1). Padilla Bay (48°28.37 N, 122°30.88'W) is characterized by estuarine-mudflat habitat, Vendovi Island (48°67.10'N, 122°61.10'W) con- sists of rocky reef habitat located in close proximity to Bellingham, Samish, and Padilla Bays, and Bird Rocks (48°29.16'N, 122°45.61'W) comprises rocky reef habitat in Rosario Strait. The fourth site was the Belle Chain Islets, a rocky reef in the southeastern Gulf Islands of British Columbia (48°49.67'N, 123°11.56'W) with habi- tat similar to that of Bird Rocks. Forty-nine blubber samples were collected from har- bor seals according to standard techniques (Iverson et al., 1997; Walton et al., 2000; Walton and Pomeroy, 2003) under Marine Mammal Protection Act Research Permit 782-1702-00. Seals were captured in salmon landing nets and physically restrained during process- ing following the method of Jeffries et al. (1993). The sampling location on the left side of the pelvic region was shaved with a razor, rinsed with isopropyl alco- hol, scrubbed with Betadine, and rinsed again with isopropyl alcohol. A complete cross section of blubber from skin to muscle was collected with a sterile, 6-mm biopsy punch. A full cross-section sample provides the most complete information regarding diet because pho- cid blubber is not homogenous throughout its depth and the inner layer responds most quickly to diet shifts (Iverson et al., 1997). The biopsy site was then filled with antiseptic cream and left open to drain. Each sam- ple was placed immediately in chloroform with 0.01% butylated hydroxytoluene to inhibit oxidation in glass vials with Teflon lids, placed on ice while in the field, and subsequently stored frozen at -80°C until analysis. Seal samples were associated with these covariates: sampling location, sex, and season (Table 1). Seasons were defined as spring (March to May), fall (October to November), and winter (December to February). We sampled fish and cephalopod species known to be consumed by harbor seals in the San Juan Islands region on the basis of previous fecal analyses (Lance et al., 2012). Some adult salmon samples were obtained from seafood processors and staff of the NOAA North- west Fisheries Science Center. Other prey were cap- tured from throughout the study area between June 16 Fishery Bulletin 1 1 1 (1) Table 1 Number of harbor seal samples, by location, sex, and season, used in our investigation of diet composition of harbor seals ( Phoca uitulina) in the Salish Sea through quantitative fatty acid signature analysis. Location Female Male Spring Fall Winter Spring Fall Winter Belle Chain 4 0 0 6 0 0 Bird Rocks 1 0 2 5 4 2 Padilla Bay 14 1 0 3 0 0 Vendovi Island 0 2 1 0 4 0 and December, 2008, with a variety of gear, including hook and line, longline, and trawl. Samples were ob- tained from 269 specimens representing these 20 spe- cies: Black (Sebastes melanops), Yellowtail (S. flauidus), Copper, and Puget Sound (S. emphaeus) Rockfish; Chi- nook, Chum ( Oncorhynchus keta), Coho (O. kisutch), Sockeye ( O . nerka ), and Pink (O. gorbuscha ) Salmon; Pacific Herring, Walleye Pollock; Pacific Sand Lance ( Ammodytes hexapterus); Northern Anchovy ( Engrail - lis mordax ); Shiner Perch ( Cymatogaster aggregata ); Plainfin Midshipman ( Porichthys notatus ); Spiny Dog- fish ( Squalus acanthias ); Opalescent Inshore Squid ( Loligo opalescens)-, Kelp Greenling ( Hexagrammos decagrammus ); Pacific Staghorn Sculpin ( Leptocottus armatus ); and Starry Flounder ( Platichthys stellatus). Specimens were identified with Hart (1973) for fish species and Roper et al. (1984) for squid. Because some species were represented by individuals with differenc- es in size and total fat content (for example, immature and mature species of salmon), 27 prey classes were defined (Table 2). Prey specimens were placed in airtight plastic bags and stored at -80°C as soon as possible after collec- tion. In the laboratory, each specimen was given a unique sample number, partially thawed, weighed and measured (standard, fork, and total lengths), and ho- mogenized with a medium or large mechanical blend- er, depending on fish size. The smallest prey animals were homogenized with a mortar and pestle because the blender was ineffective. Stomach contents were not removed from prey specimens, to mimic ingestion by predators (Budge et al., 2002). Approximately 5-10 g of homogenate was placed in labeled scintillation vials with Teflon lids and stored in a -80°C freezer. Samples were express shipped in a cooler on dry ice to the Ap- plied Sciences, Engineering, and Technology (ASET) Laboratory at the University of Alaska Anchorage. Fatty acid extraction and selection All samples were processed at the ASET Laboratory through the use of a method for microscale recovery of total lipids with the Dionex ASE 2001 automated solvent extraction system (Thermo Fisher Scientific, Waltham, MA), which provides lipids for the determi- nation of 80 unique fatty acids (Dodds et al., 2005). The total body mass, percent fat composition, and fat mass of prey specimens were obtained for 27 prey classes (Table 2). Total mass data were not available for ma- ture Chinook, Sockeye, and Pink Salmon obtained from the Northwest Fisheries Science Center; therefore, an approximate mean mass for these prey classes (e.g., Quinn, 2005) was used in calculation of fat mass. Given the large range of mass among prey classes (Table 2), the results were insensitive to our use of these approxi- mate values. Extracted lipids were dissolved in hexane to a con- centration of 100 mg/mL, hydrolyzed by a base-cata- lyzed reaction with potassium hydroxide, and then esterified to form fatty acid methyl esters (FAMEs) by reaction with boron trifluoride in methanol. Each sample was spiked with a C21:0 internal standard (25 pg/mL) and separated on a Hewlett-Packard 5890 gas chromatograph (GC) with a flame ionization detector (FID) (Hewlett-Packard Co., Palo Alto, California) by using a 60-m J&W DB-23 column (Agilent Technolo- gies, Inc., Santa Clara, CA) with a 0.25-mm inside di- ameter and 0.25-pm cyanopropyl polysiloxane film. Sig- nal data were collected and analyzed with Agilent GC Chemstation software. Supelco 37-Component FAME Mix (catalog no. 47885-U; Sigma-Aldrich Co., St. Louis, MI) was used as a continuing calibration verification (CCV) to verify both the retention times and recovery values. This CCV also contained 25 pg/mL of a C21:0 internal standard, which is required to meet a tolerance of no greater than ±20% of actual value. Analyte identity was veri- fied further by mass spectrometry through the use of a Varian CP3800 GC (Agilent Technologies, Inc.) and a Varian Saturn 2200 ion trap mass spectrometer 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the U.S. Government. Bromaghin et al: Diets of Phoca vitulma in the Salish Sea revealed by analysis of fatty acid signatures 17 Table 2 The number of prey animals from which fatty acid signature data were obtained (n) and the prey class (class) into which each prey type was assigned after evaluation of discriminant analysis and mean fat mass in our investigation of the diet composition of harbor seals ( Phoca vitulina ) in the Salish Sea through quantitative fatty acid signature analysis. Prey classes are defined as B&YR (Black [ Sebastes melanops] and Yellowtail [S. flavidus 1 Rockfish), CR (Copper Rockfish [S. caurinus}), PSR (Puget Sound Rockfish IS. emphaeus ]), Chin (mature Chinook Salmon \Oncorhynchus tshawytscha ]), Chum (mature Chum Salmon |0. keta\), Coho (mature Coho Salmon [O. kisutch]), Sock (mature Sockeye salmon [O. nerka ]), Pink (mature pink salmon \0. gorbuscha}), Sal-M (medium-sized Chinook and Coho Salmon), Sal-S (small Chinook, Chum, Sock- eye, and Pink Salmon), Pol (Walleye Pollock [ Theragra chalcogramma ]), Her (Pacific Herring [ Clupea pallasii } at least 2 years old), YH&SL (Pacific Herring less than 2 years old and Pacific Sand Lance [ Ammodytes hexapterus ]), NA (Northern Anchovy [ Engraulis mordax]), SP (Shiner Perch I Cymatogaster aggregata ]), PM (Plainfin Midshipman [Porichthys notatus 1), SD (Spiny Dogfish [Squalus acanthias ]), OIS (Opalescent Inshore Squid [Loligo opalescens]), G&S&F (Kelp Greenling [ Hexa - grammos decagrammus ], Pacific Staghorn Sculpin [ Leptocottus armatus], and Starry Flounder [ Platichthys stellatus 1). For each prey type, the sample size in), mean (mean), and standard deviation (SD) of total mass, percent fat composition, and total fat mass are shown. Mass data were not available for mature Chinook, Sockeye, or Pink Salmon, and an approximate mean mass was used for the computation of fat mass. Mass (g) Percent fat Fat mass (g) Prey type n Class n Mean SD n Mean SD n Mean SD Black Rockfish 5 B&YR 5 293.8 48.3 5 6.5% 0.4% 5 19.3 4.0 Yellowtail Rockfish 5 B&YR 5 152.8 28.2 5 5.7% 1.5% 5 8.8 2.6 Copper Rockfish 12 CR 12 201.3 195.7 12 2.4% 0.4% 12 4.7 4.5 Puget Sound Rockfish 14 PSR 14 53.9 8.9 5 2.2% 0.3% 5 1.1 0.4 Chinook, mature 10 Chin 0 10000.0 NA 10 12.2% 2.3% 10 1218.8 233.3 Chum, mature 10 Chum 10 4955.9 784.6 10 15.1% 7.8% 10 789.7 455.6 Coho, mature 10 Coho 10 3765.4 660.8 10 5.5% 2.8% 10 208.2 125.0 Sockeye, mature 10 Sock 0 2500.0 NA 10 12.4% 1.8% 10 309.4 45.4 Pink, mature 10 Pink 0 2000.0 NA 10 5.3% 2.1% 10 105.6 43.0 Chinook, medium 5 Sal-M 5 133.5 70.3 5 3.0% 1.3% 5 4.8 3.1 Coho, medium 4 Sal-M 4 193.0 28.6 4 2.9% 0.5% 4 5.7 1.7 Chinook, small 11 Sal-S 12 20.9 8.0 12 1.3% 0.3% 12 0.3 0.2 Chum, small 12 Sal-S 12 62.8 24.6 12 2.3% 1.1% 12 1.6 1.5 Sockeye, small 12 Sal-S 12 15.5 2.5 12 1.5% 0.2% 12 0.2 0.1 Pink, small 12 Sal-S 12 47.2 13.6 12 2.4% 0.8% 12 1.2 0.7 Pollock 13 Pol 13 29.4 78.6 13 1.8% 0.4% 13 0.5 1.2 Pacific Herring >2 yr 12 Her 12 37.5 4.2 12 11.7% 3.4% 12 4.4 1.6 Pacific Herring <2 yr 12 YH&SL 12 5.8 0.8 12 3.5% 1.3% 12 0.2 0.1 Pacific Sand Lance 12 YH&SL 12 1.9 0.3 12 3.3% 0.8% 12 0.1 0.0 Northern Anchovy 11 NA 11 18.8 1.7 11 12.2% 3.4% 11 2.3 0.7 Shiner Perch 12 SP 12 21.0 5.8 12 6.9% 2.4% 12 1.5 1.0 Plainfin Midshipman 9 PM 9 61.7 13.4 9 3.4% 0.7% 9 2.1 0.6 Spiny Dogfish 4 SD 4 1712.5 383.8 4 9.0% 3.6% 4 160.5 83.5 Opalescent Inshore Squid 12 OIS 12 7.1 1.9 12 3.0% 0.4% 12 0.2 0.1 Kelp Greenling 7 G&S&F 7 179.7 396.3 7 1.5% 0.4% 7 3.0 6.8 Pacific Staghorn Sculpin 12 G&S&F 12 21.0 10.1 11 1.5% 0.6% 11 3.4 5.7 Starry Flounder 11 G&S&F 11 220.2 410.1 11 1.5% 0.6% 11 3.4 5.7 with a scan range of 50-400 mass-to-charge ratios (m/z). Additionally, a National Institute of Standards and Technology 1946 international standard was used to externally verify the method and the quality of recoveries. The ASET Laboratory implements several protocols to improve data quality that are not routinely imple- mented in analyses of fatty acid data. Rather than normalize the peak data of each sample to C18:0, the laboratory adds an internal standard to all samples, method blanks, and CCVs. This protocol is beneficial because it provides a data point of known quantity to each resulting set, including blanks, allowing the sig- nificance of low-recovery peak data to be verified. In ad- dition, because normalization to a recovered compound incorrectly entails the assumption that all compounds respond equally in the FID, use of an internal stan- dard avoids errors that might otherwise result from that assumption (Dodds et al., 2005). The laboratory also verifies the identity of each peak by using a GC mass spectrometer (GC-MS) — verification that is nec- essary to eliminate misclassification of non-fatty acid 18 Fishery Bulletin 1 11 (1) byproducts from the derivatization process. Finally, the laboratory performs periodic standard calibrations of the spectrometer at varying levels of concentration to determine the limit-of-detection for each compound. Several criteria were used to evaluate the suitability of each fatty acid compound for inclusion in mixture modeling. At a minimum, each compound had to pass GC-MS verification, have a minimal variance for the majority of samples collected (<20% relative standard deviation), and average at least 1% of the total fatty acid contained in each sample. The compounds needed to be predominately from a dietary source, as delin- eated in Iverson et al. (2004). Compounds 18:2n-6 and 18:3n-3 were automatically included as neither com- pound is biosynthesized by seals. These selection crite- ria led to a suite of 22 fatty acid compounds to be used in mixture modeling: C16:2n-6, Cl6:2n-4, C16:4n-1, C18:ln-9, C18:ln-7, C18:2n-6, C18:3n-6, C18:3n-4, Cl8:3n-3, C18:4n-3, C20:ln-ll, C20:ln-9, C2Q:ln-7, C20:2n-6, C20:3n-6, C2Q:4n-6, C20:3n-3, C20:4n-3, C20:5n-3, C22:6n-3, C21:5n-3, and C22:5n-6. Data are available at the Biological and Chemical Oceanography Data Management Office of the National Science Foun- dation (http://osprey.bcodmo.org/project.cfm?flag=viewr &id=224&sortby=project). Estimating diet composition Obtaining unique estimates of diet composition with mixture models requires the number of prey classes to be no greater than the number of fatty acids (e.g., Phillips, 2001). Furthermore, combining prey classes reduces the dimensionality of the parameter space and can increase estimation precision. Linear discriminant functions were used to identify prey classes with po- tential to be merged, with R software, vers. 2.10.1 (R Development Core Team, 2009) and function Ida of package MASS (Venables and Ripley, 2002). The ac- curacy of classifying individual prey into correct prey classes was estimated with discriminant functions and cross validation. Data from each prey specimen were removed temporarily, discriminant functions were es- timated from the remaining data, and the estimated functions were used to classify the excluded specimen to a prey class. Prey classes with the largest misclas- sification rates were candidates to be merged, provided that the mean adipose masses of the 2 classes were similar. Methods of QFASA mixture modeling closely fol- lowed those of Iverson et al. (2004) and Beck et al. (2007), methods that have been applied to the re- search of numerous marine species, including harbor seals (Nordstrom et al., 2008), gray seals ( Halichoerus grypus ; Iverson et al., 2004; Beck et ah, 2007; Tucker et ah, 2008; Lundstrom et ah, 2010), harp seals (Pag- ophilus groenlandicus; Iverson et ah, 2004), northern fur seals (Callorhinus ursinus; Hofmeyr et ah, 2010), Steller sea lions ( Eumetopias jubatus; Hoberecht, 2006), polar bears ( Ursus maritimus; Thiemann et ah, 2008) , and various species of seabirds (Williams et al., 2009) . A mixture model based on the Kulibaek-Liebler (KL) distance measure (Iverson et ah, 2004) was used to estimate the diet composition of each seal. The cali- bration coefficients for harbor seals reported by Nord- strom et al. (2008) were used to convert prey fatty acid signatures (FAS) to the scale of predator FAS, and the distance measure was evaluated on the predator scale; note that Iverson et al. (2004) converted predator FAS to the prey scale. Estimation variance for each seal was estimated with 1000 bootstrap replications of the prey FAS data. The resulting estimates of diet composition (fat unadjusted, the pk of Iverson et ah, 2004), also were transformed to account for adipose mass per prey, expressing diet composition in terms of the number of animals consumed (fat adjusted, the ab of Iverson et ah, 2004). Multivariate analysis of variance (function manova in R; R Development Core Team, 2009) was used to explore diet composition estimates for structure as- sociated with the following covariates: sampling loca- tion, season (spring, fall, winter), and sex. The initial model contained these 3 main effects and all 2-way in- teractions, and nonsignificant terms were sequentially eliminated from the model. A significance level (a) of 0.01 was used for all tests. The mean diet composition for a class of predators (e.g., males or females) was computed as the sample average of their individual diet composition estimates. The variance of mean diet composition was assessed with the estimator of Beck et ah (2007). Mixture proportions and variances were estimated with a custom computer program written in Fortran (Metcalf et ah, 2004) and compiled with the In- tel Visual Fortran Compiler Professional Edition, vers. 11.1 (Intel Corp., Santa Clara, CA). Results Estimating diet composition Given the suite of 22 fatty acid compounds used to form FAS, the 27 original prey classes needed to be reduced to no more than 22 prey classes for mixture model estimates to be unique (Phillips, 2001). Among the 27 original prey types, Black and Yellowtail Rock- fish; medium-size Chinook and Coho Salmon; small Chinook, Chum, Sockeye, and Pink Salmon; young Pa- cific Herring aged 0 to 1 and Pacific Sand Lance; and Kelp Greenling, Pacific Staghorn Sculpin, and Starry Flounder were combined to reduce discriminant analy- sis misclassification among prey classes (Table 2). The resulting prey data set contained 19 prey classes, for which 251 of 269 prey animals (93.3%) were assigned to the correct prey class. The mean diet composition of all 49 seals, both ad- justed and unadjusted for differential fat mass among prey, was estimated with FAS for 22 fatty acid com- pounds and data for 19 prey classes. The species esti- Bromaghin et al: Diets of Phoca vitulma in the Salish Sea revealed by analysis of fatty acid signatures 19 Figure 2 Mean diet composition estimates: (A) adjusted and (B) unadjusted for differential fat mass among prey classes, for all harbor seals (Phoca vitulina) combined in our inves- tigation of the diet composition of harbor seals in the Salish Sea. Error bars are ±1 standard error of the estimate. Prey classes are defined as B&YR (Black [ Sebastes melanops ) and Yellowtail [S. flavidus ] Rockfish), CR (Copper Rockfish [S. caurinus]), PSR (Puget Sound Rockfish [S. emphaeus 1), Chin (mature Chinook Salmon \Oncorhyn- chus tshawytscha]), Chum (mature Chum Salmon 10. keta}), Coho (mature Coho Salmon (O. kisutch ]), Sock (mature Sockeye Salmon [O. nerka J), Pink (mature Pink Salmon 10. gorbuscha]), Sal-M (medium-size Chinook and Coho Salmon), Sal-S (small Chinook, Chum, Sockeye, and Pink Salmon), Pol (Walleye Pollock [ Theragra chalcogramma 1), Her (Pacific Herring [ Clupea pallasii 1 at least 2 years old), YH&SL (Pacific Herring less than 2 years old and Pacific Sand Lance [ AmmocLytes hexapterus]), NA (Northern An- chovy [Engraulis mordax ]), SP (Shiner Perch [ Cymatogaster aggregata]), PM (Plainfin Midshipman | Porichthys notatus ]), SD (Spiny Dogfish [Squalus acanthias]), OIS (Opal- escent Inshore Squid [ Loligo opalescens]), G&S&F (Kelp Greenling [ Hexagrammos decagrammus] , Pacific Staghorn Sculpin [Leptocottus armatus 1, and Starry Flounder [Platichthys stellatus]). 20 Fishery Bulletin 1 1 1 (1) mated to contribute most to harbor seal diets included Black and Yellowtail Rockfish, Chinook Salmon, adult Pacific Herring, and Shiner Perch (Fig. 2). Large differ- ences in fat mass among prey classes led to substantial differences in the 2 estimates. Most noticeably, the high fat content of mature salmon species (Table 2) reduced the contribution of adult Chinook Salmon in the estimates adjusted for fat mass, suggesting that few individual Chinook Salm- on need to be consumed for them to contribute significantly to the fat composition of harbor seals. Multivariate analysis of variance results revealed substantial hetero- geneity among estimated diets of individual seals by sampling loca- tion (PcO.OOl) and sex (P<0.001), although the interaction was not statistically significant (P=0.111). For that reason, the 49 seals were independently stratified by sam- pling location and sex and the mean diet composition, unadjusted for dif- ferential fat mass, was estimated for the seals in each stratum. Sea- son was eliminated from the model because it was not a statistically important covariate (see Discussion section). Seals sampled in the vicin- ity of Belle Chain and Bird Rocks, both of which are characterized by rocky, high-current habitat, had the most diverse diets, with important contributions from Black and Yel- lowtail Rockfish, adult salmon spe- cies, Pacific Herring, Shiner Perch, and Spiny Dogfish (Fig. 3). Con- versely, seals sampled from Padilla Bay, which consists of shallow estu- arine habitat, had diets that were, on average, dominated by Shiner Perch. Harbor seals sampled near Vendovi Island, which has rocky habitat with nearby access to sev- eral bays, appeared to have an in- termediate diet. Male harbor seals were esti- mated to consume larger quanti- ties of Black and Yellowtail Rock- fish, Pacific Herring, and Spiny Dogfish than females, for which Shiner Perch appeared to be more important (Fig. 4). Diet estimates for individual seals reflected ad- ditional between-seal heterogene- ity that was not explained by the covariates. For example, although Black and Yellowtail rockfish were estimated to be more important to males than females overall, males were not consistent in their reliance on rockfish spe- cies. Of the 24 males sampled, 10 had an estimated Prey group Figure 3 Estimates of mean diet composition for harbor seals {Phoca uitulina) in the Salish Sea, unadjusted for differential fat mass among prey classes, by sam- pling location: (A) Belle Chain Islets, (B) Bird Rocks, (C) Padilla Bay, and (D) Vendovi Island. Error bars are ±1 standard error of the estimate. Prey classes are defined as B&YR (Black [Sebastes melanops] and Yellowtail [S. flavidus] Rockfish), CR (Copper Rockfish [S. caurinus]), PSR (Puget Sound Rockfish [S. emphaeus]), Chin (mature Chinook Salmon [Oncorhynchus tshawytscha 3), Chum (mature Chum Salmon [O. keta}), Coho (mature Coho Salmon [O. kisutch]), Sock (mature Sockeye Salmon [O. nerka ]), Pink (mature Pink Salmon [O. gorbuscha]), Sal-M (medium-size Chinook and Coho Salm- on), Sal-S (small Chinook, Chum, Sockeye, and Ppink Salmon), Pol (Walleye Pollock [Theragra chalcogramma]), Her (Pacific Herring [Clupea pallasii ] at least 2 years old), YH&SL (Pacific Herring less than 2 years old and Pacific Sand Lance [Ammodytes hexapterus]), NA (Northern Anchovy [En- grauiis mordax] ), SP (Shiner Perch [Cymatogaster aggregata ]), PM (Plainfm Midshipman [ Porichthys notatus]), SD (Spiny Dogfish [ Squalus acanthias]), OIS (Opalescent Inshore Squid [ Loligo opalescens]), G&S&F (Kelp Greenling [Hexagrammos decagrammus ], Pacific Staghorn Sculpin [Leptocottus arma- tus], and Starry Flounder [Platichthys stellatus}). Bromaghm et al: Diets of Phoca vitulina in the Salish Sea revealed by analysis of fatty acid signatures 21 diet composition of 0.0% for Black and Yellowtail Rock- fish, and estimates for the remaining 14 males ranged from 8.2% to 51.4% and averaged 31.8%. Although fe- males were more consistent in their reliance on Shiner Perch, the estimated contribution of Black and Yellow- tail Rockfish exceeded 25% for 3 individuals. There were no discernible patterns in the capture location or date with respect to the magnitude of rockfish estimates for either males or fe- males, a result that is consistent with the nonsignificant interaction between location and gender in the linear model. One female seal was captured twice, at Padilla Bay in spring 2007 and at Vendovi Island in winter 2008. The diet composi- tion of this female was estimated to be -90% Shiner Perch and -9% Chi- nook Salmon, with negligible contri- butions from other prey classes, on both occasions. Discussion Our findings re-affirm the impor- tance of several commercially impor- tant fish species to harbor seal diets, particularly salmon species, Pacific Herring, and Shiner Perch, reported by prior investigators (Scheffer and Slipp, 1944; Everitt et al., 1981; Brown and Mate, 1983; Olesiuk, 1993; Zamon, 2001; Orr et al., 2004; Wright et al., 2007; Thomas et al., 2011; Lance et al., 2012). However, our results also reveal that rockfish species contribute more substan- tially to harbor seal diets than has been recognized previously, exceed- ing 10% of the average diet of all harbor seals combined. Given that QFASA estimates are thought to describe diets integrated over a pe- riod of weeks to months (Iverson et al., 2004; Budge et al., 2006), esti- mates of this magnitude may reflect substantial periodic (and, perhaps, sustained) predation on species of rockfish. Although quantitative esti- mates of rockfish abundance are un- available, rockfish populations are considered depressed and, given the regional abundance of harbor seals (Jeffries et al., 2003), the predation rates indicated by these findings may be sufficiently high to influ- ence their population dynamics, on a local or, perhaps, regional scale. Consequently, management plans to enhance rockfish abundance may need to give greater consideration to the potential influence of pinniped 06 Tl O 0.4 c o O 03 Q. o 0.2 0) T3 O C o r o a o 06 0.5 0.2 ^ =5? 80% of the total variation in model outputs. Prey groups, such as rockfish, that are targeted for recovery may still be affected by even low levels of predation. This study highlights the importance of salmonids and herring for the seal population and provides a framework for refining consump- tion estimates and their confidence intervals with future data. Manuscript submitted: 4 November 2011. Manuscript accepted 31 October 2012. Fish. Bull. 111:27-41 (2013). doi: 10. 7755/FB. 111.1.3 The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessar- ily reflect the position of the National Marine Fisheries Service, NOAA. Sarah M. S. Howard (contact author)1 Monique M. Lance2 Steven J. Jeffries2 Alejandro Acevedo-Gutierrez1 1 Biology Department Western Washington University 516 High Street Bellingham, Washington 98225 Present address: National Park Service 10 Organ Pipe Drive A|o, Arizona 85321 2 Washington Department of Fish & Wildlife 7801 Phillips Road SW Lakewood, Washington 98498 Overfishing and habitat change have affected fish populations heavily in the inland waters of the Pacific Northwest. Many formerly abundant fish species are now species of con- servation concern, including ground- fish stocks, such as rockfish species ( Sebastes spp.) and Pacific Hake (Merluccius productus), forage fish stocks such as Pacific Herring (Clu- pea pallasii), and several salmonid species ( Oncorhynchus spp.) (Musick et ah, 2000; Mills and Rawson, 2004). Most recently, 3 rockfish species (S. ruberrimus, S. pinniger, S. paucispi- nis) were listed under the Endan- gered Species Act as threatened or endangered in Puget Sound, Wash- ington State (Federal Register, 2010). The decline of all these popula- tions, which perform a critical func- tion in regional food webs (Simenstad et al., 1979; Schindler et al., 2003) and have commercial and recre- ational value, has created a need for recovery strategies at the ecosystem level. Fish recovery efforts currently rely on traditional fisheries manage- ment approaches, such as reduction of fishing pressure and creation of no-take refuges or marine reserves, and on habitat restoration (Allison et al., 1998; Roni et al., 2002). Marine reserves in particular are more like- ly to be successful for species, such as rockfish, that have small home ranges and high site fidelity (Love et al., 2002), and reserves are impor- tant management tools for recovery of rockfish in the Pacific (Murray et al., 1999). More reserves have been proposed recently for the San Juan Islands,1 an island group that is part of the Salish Sea marine ecosystem that spans U.S. and Canadian waters (Fig. 1). For pelagic species, such as salmonids and forage fishes, recovery efforts call for habitat protection and mitigation of water-pollution issues, among other factors, as management tools (Fluharty, 2000; Schindler et al., 2003). The restoration of predators in ma- rine ecosystems can reestablish tro- phic relations and restructure habi- 1 McConnell, M. L., and P. A. Dinnel. 2002. Rocky reef bottomfish recovery in Skagit County. Phase II final report: assessment of eight potential marine reserve sites & final site recommendations. Skagit County Marine Resources Committee, Mount Vernon, WA, 43 p. [Available from http://www.nwstraits.org/Archives/ Library.aspx.] 28 Fishery Bulletin 1 1 1 (1) Figure 1 Map of the study area, the San Juan Islands and eastern bays, where seal scat collections were made for a bioenergetics model to examine the quan- tity of fish consumption by the harbor seal ( Phoca uitulina ) population dur- ing 2007-08. Black circles indicate harbor seal scat collection sites. tat with usually positive results (Shears and Babcock, 2002; Shears et ah, 2006); however, predators also can cause declines in the size distributions and abundance of prey species inside marine reserves (Sala and Zabala, 1996; Fanshawe et ah, 2003). Large-bodied and abundant predators can contribute significantly to fish mortality, especially when prey species are already low in abundance, and may theoretically influence prey population recovery (Mohn and Bowen, 1996; Bundy, 2001; DeMaster et ah, 2001; Fu et al., 2001; Trzcinski et ah, 2006). Therefore, there is a need to under- stand the prey requirements of predators that consume fish species of conservation concern to evaluate if such requirements conflict with regional management goals. In the Salish Sea, the harbor seal ( Phoca vitulina ) is an abundant, generalist marine predator whose population has steadily in- creased since gaining protected status in the 1970s. The harbor seal population in Wash- ington State experienced logistic growth from the 1970s to the 1990s, increased 7- to 10-fold in size in different regions, and now appears to be at carrying capacity (Jeffries et ah, 2003). Estimates of the regional popu- lation in the San Juan Islands and eastern bays in the early 1970s were approximately 1000 animals; currently, there are approxi- mately 8000. 2 The age structure of the har- bor seal population in British Columbia was documented in Bigg (1969), on the basis of seals collected and aged in the 1960s. After exponential population increases, this popu- lation was heavily weighted toward juvenile age classes by the 1980s (Olesiuk, 1993). Given the population in- crease in all regions of the Salish Sea, the current age structure of the harbor seal population in the San Juan Islands is unknown. As with other harbor seal populations in the east- ern Pacific, harbor seals in the San Juan Islands take advantage of the large influx of adult salmonids in late summer and fall and increase the diversity of their diet at other times of the year when salmonids are less available (Hauser et al., 2008; Lance et al., 2012). Salmonids, Pacific Herring, Pacific Sand Lance (Ammodytes hexapterus), Northern Anchovy (Engraulis mordax ), Walleye Pollock (Theragra chalcogramma), and estuarine species, such as Shiner Perch ( Cymato - gaster aggregata), also form significant proportions of their diet in the San Juan Islands and nearby estua- rine ecosystems (Lance et al., 2012). 2 Washington Department of Fish & Wildlife. Unpubl. data. Washington Department of Fish & Wildlife, 7801 Phillips Road SW, Lakewood, WA 98498. To calculate population-level consumption of fish species of conservation concern and other common har- bor seal prey in the San Juan Islands, a bioenergetics model was used to determine energetic requirements. The model incorporated seasonal changes in seal diet and life history parameters during breeding and non- breeding seasons. We also used simulated data and sensitivity analyses to address uncertainty in the over- all model and in 2 specific components that may have a strong influence on predicted consumption of prey: 1) uncertainty in age structure of the harbor seal popu- lation and 2) seasonal changes in energy intake (e.g., fasting during breeding season). Methods Area and timeframe of study The region of the San Juan Islands and eastern bays is an area where many fish species of conservation concern occur and also an area where the majority of the harbor seal population resides in the inland waters Howard et ai.: Fish consumption by harbor seals ( Phoca vitulina ) in the San Juan Islands, Washington 29 of Washington State. The San Juan Islands (48°35'N, 122°55'W) are characterized by tidally influenced rocky reefs and isolated rocks surrounded by deep water where harbor seals often congregate at haul-outs (loca- tions where seals come ashore). The adjacent eastern bays, in contrast, consist of large, soft-bottomed, shal- low bays (48°33'N, 122°30'W) (Fig. 1). The consumption model was constructed for a sin- gle annual cycle for the harbor seal population dur- ing 2007-08. The model included 2 seasons: breeding (15 June-15 September) and nonbreeding (16 Septem- ber-14 June) determined on the basis of seal pupping phenology in the San Juan Islands (Huber et al., 2001; Patterson and Acevedo-Gutierrez, 2008). The 2 sea- sons were delineated to reflect known behavioral shifts (more time spent ashore to nurse pups, shallow-water breeding displays by males) related to pupping and breeding activities and subsequent changes in ener- getic expenditures (Coltman et al., 1998; Bowen et al., 1999). The model was programmed in R software, vers. 2.7.1 (R Development Core Team, 2008) and used re- gional activity, abundance, and diet data, as well as physiological data from the literature. Model param- eters were grouped into 3 categories: bioenergetics, population, and diet (Lavigne et al., 1982; Winship et al., 2002) (Table 1). Model structure Bioenergetics Energetic requirements were calculated with a bioenergetics approach that described the en- ergy budget of an individual seal, which is a function of body size, activity budgets, growth, and reproductive costs. Sex- and age-specific gross energy requirements were calculated with Equation 5 in Boyd (2002): EG, = [l(""‘-(rf9Ai)86400] + ft I“‘ ’ U) where EGt = energy requirements in a particular stage i of the annual cycle; y, = the power (watts) generated under activity /"within stage i of the annual cycle; = proportion of time spent in activity f\ g: - the cost of growth in stage i of the annual cycle; and dl - the digestive efficiencies of food being eaten. The model had 6 sex-and-age classes: 1) adult fe- males (>6 years), 2) adult males (>8 years), 3) subadult females (1-6 years), 4) subadult males (1-8 years), 5) female pups (<1 year), and 6) male pups (<1 year). The subadult to adult division was made at the age(s) har- bor seals reach their predicted maximum weight (ap- proximately 66 kg and 89 kg for females and males, respectively) on the basis of the growth curve in Ole- siuk (1993). Daily growth increments for each sex-and- age class were calculated from the same growth curve. Activity budgets were estimated from free-living har- bor seals tagged with data recorders that recorded 3 behavioral periods: haul-out, diving, and shallow-water activity (Table 1). Population abundance and age structure Aerial popula- tion surveys of harbor seals have been conducted an- nually by the Washington Department of Fish & Wild- life with fixed-wing aircraft to estimate the number of animals hauled-out during the lowest tide of the day since 1978 (Jeffries et al., 2003). Results from these surveys were used to estimate the abundance of harbor seals in the study area in 2007-08. The breeding sea- son (July) correction factor of 1.53 (to account for seals not hauled-out at the time of the survey) was used to estimate the size of the breeding season population (Huber et al., 2001). Age-dependent mortality rates in Olesiuk (1993) were used to estimate the age structure (number of seals in each sex and age class) of the har- bor seal population: ^s(x+l)=Ns(x,e~rt ’ (2) where NS(x) = number of seals in sex class S and age class x; -r = the age-dependent mortality rate; and t = time interval between age classes. The breeding season population vector was adjusted by iteration to sum to the total population estimate from aerial surveys. Seal abundance in the nonbreed- ing season was calculated by estimating the numbers still alive in each sex and age class, by using the same age-dependent mortality rates calculated per day (in- stead of annually) and by multiplying the number of days in the breeding cycle. Population energetic requirements were calculated by multiplying individual requirements by the population abundance vectors to estimate energetic requirements for each sex and age class. Reproductive costs were then calculated for the entire population on the basis of val- ues from the literature for gestation and lactation costs and fertility rates (Bigg, 1969; Olesiuk, 1993). Digestive efficiency Data from the literature were used to translate net energy requirements of the harbor seal population into gross energy requirements and prey consumption by first taking into account assimi- lation efficiency and the heat increment of feeding (the increase in metabolism or heat produced during di- gestion) for harbor seals. We used the minimum and maximum values reported in the literature to account for differences in digestive efficiencies related to pro- tein and fat content of prey (Markussen et al., 1994; Trumble et al., 2003). Table 1 Data sets used in the consumption model in a study of the harbor seal ( Phoca vitulina ) population in the San Juan Islands and eastern bays during breeding and nonbreeding seasons, 2007-08. Model parameter symbols refer to Equations 1-3 in text. All energy units were converted to watts. H=haul-out; d=dive; s=surface. NA= not applicable. 30 Fishery Bulletin 1 1 1 (1) 03 > C 03 03 D c r W g -Q r? cd r2 -Q S-. CLh ~ 03 Oh O 3 3 03 CQ T3 + g CO o 3 • • 03 O r—\ CM GO I I I O CO *-H ■ ” co co o -H ^ rH -Com g a; „ o o 3 ■*“* J-l a> o g ns + J2 t-h 3 CO CM CO I> I I co o T— I 1— ( cm in -C T3 m cd 3 CM CO “ co co l I I o v. ® d o Tf o -C T3 W CO H co CM t> ^ r\ 9^ 03 i-H X no T3 + -C g 3 03 I O CQ m ) ^ bD T3 3 -Q 03 9 I 3 £ O 03 CM II IS H O T3 bD „ ^ b D * co ^ 7 CO 03 T3 I °° bl I k> (N ^ . . CM GO 03 T3 + -C g 3 03 I 03 BS _03 S CM CO I t> O 1 9 l> 9 © lO CO i — i 30 03 C3 3 CU a bD o -a c 03 a; o I m © © 3 no bD 03 00 c^h CO O CO 3 co w oo ; co ^ _ IN W h ^ l»Ht« I> I 00 ^ O'-1 | 00 lO - ^ t~- ID n , tn | ■ ^ s i CM 3 TO s a 3 3 u 3 . -a C a. 3 ! ■£ CO 3 Oh g -o p=i 03 CM -2 =3 03 PP -C -3 -3 m co 3 03 n3 3 3 3 CL, O Ph 3 bD a 03 g g 03 3 ■ o I o , bl i 03 ^ 1 O 1-* q q rH w 1 J3 I - o -a to g | s : 03 ► co ffi : cu r_' cu X g tc .s -ts _C o M W ° o o pq cd cn O' -X3 i 9 O CM 9 ^ i o o oo -x I o ^ © -2 -c - • — a > £ JS ^ -rH CJ ^ -C O o£ bD >> .S cj -a C 03 03 ,03 CD tn a ^ ns 7 « PQ a, *t3 *t3 03 - 03 — ; 03 a Pc3 03 . . 03 CD 03 t— | 3 CM 2 ° *2 ° S « do 'qj ^ lO 2 ^ 03 O -2 o O ° w CM CM*' 03 03 03 r G — « o G jS 03 03 CQ 03 . . 03 CD 03 O ~ 03 _ : 03 a . cd ^ a a eu 03 03 CO O Kd 03 tn 05 TLffi V5 a a, 7 fi r ^ 03 m rc •r 03 CD ^ 03 ^ QJ tT ^ cu a 5 J 5 o . . T— I CM . „ . .. Tt 03 03 CD 03 03 H O O _ cm bp . C ^ "®- Cd T3 ^ a, >i 6h ^5 T3 a *> 03 O bD £ G O ft c m. fl 9 ' 03 co o o CM Cd (N - B ~o q g g 0 ~ N bD 03 bD *“• • l=r 03 m a, j' r- o o M Q Howard et at: Fish consumption by harbor seals (Phoca vitulina) in the San Juan Islands, Washington 31 Diet Collection of scat samples Scat samples were collect- ed at 23 sites that represented regional variation in habitat in the San Juan Islands from 2005 to 2008 as part of a larger harbor seal diet study conducted in the northern Puget Sound (Fig. 1) (Lance et al., 2012). Samples collected during seal breeding and nonbreed- ing seasons in 2007-08 were used in our study. De- tailed scat sample processing, collection information, and analysis of frequency occurrence of prey items in harbor seal diet are summarized in Lance et al. (2012). Briefly, samples for the diet study were collected from harbor seal haul-out locations during daytime low tides, placed in plastic bags, and then frozen until they were processed. Scat samples were processed following Lance et al.3 and Orr et al. (2003). Otoliths were mea- sured and graded according to the methods of Tollit et al. (2007). On otoliths that were graded as good (no or minimal erosion) and fair (small amount of erosion), the width and length were measured with an ocular micrometer. For our study, scat samples were pooled by seal breeding and nonbreeding seasons for further analyses. Reconstruction of wet biomass To choose appropriate input values for diet in the model, a wet biomass re- construction technique (Laake et al., 2002) was used to estimate the proportion by wet weight of prey items in harbor seal diet. This technique focuses on energetic content of seal diet, rather than on frequency of items in diet, by accounting for the number and size of prey consumed in a diet sample. The proportion of wet bio- mass of a prey item (jt() in harbor seal diet was calcu- lated by (Laake et al., 2002): where nt = the corrected number of items of prey item /; and wt = the average weight (in grams) of all prey items i. The corrected number of “items” (n,, number of in- dividuals in the sample) was calculated by applying a species-specific (or closest proxy) correction factor to account for otolith loss during digestion. We used otoliths to enumerate all species except Shiner Perch, for which we used the number of pharyngeal plates to derive a more reliable passage rate. We lacked otolith- loss correction factors for herring (Clupeidae) and Wall- eye Pollock; therefore, we considered the correction fac- tors for Pacific Sardine (Sardinops sagax) and Pacific Hake in Phillips and Harvey (2009), respectively, to Lance, M. M., Orr A. J., Riemer S. D., Weise M. J., and Laake J. L. 2001. Pinniped food habits and prey identification techniques protocol. AFSC Processed Report 2001-04, 41 p. Alaska Fisheries Science Center, Seattle, WA. [Available from http://access.afsc.noaa.gov/pubs/search.cfm.! be reasonable proxies because these species are simi- lar in size and structure (M. M. Lance, personal com- mun.l. We used a Pink Salmon ( Oncorhynchus gorbus- cha) otolith-loss correction factor for all salmonids, a Shortbelly Rockfish ( Sebastes jordani) correction factor for all rockfish species, and species-specific correction factors for Shiner Perch and Pacific Staghorn Sculpin ( Leptocottus armatus) (Harvey, 1989; Phillips and Har- vey, 2009). Length correction factors were applied to measure- ments from otoliths scored as being in good or fair con- dition to account for otolith erosion during digestion. Corrected otolith lengths then were used to calculate the fish size with species-specific length-weight regres- sions (Harvey et al., 2000). When we lacked species- specific correction factors or length-weight regressions, we used estimated body sizes of prey items. Otoliths of juvenile and adult salmonids were distin- guished on the basis of otolith and bone sizes. Otoliths that were graded in good enough condition to measure and reconstruct salmonid size were uncommon in scat samples; therefore, for salmonid adults that were not identified to species, we used an approximate average size (1589 g) for Pink Salmon, the species most com- monly consumed by harbor seals (Lance et al., 2012). An average estimated size of 35 g was used for all sal- monid juveniles. We also lacked otolith-length correc- tion factors for herring and Walleye Pollock; therefore, we used Pacific Sardine and Pacific Hake as proxies. The remaining length correction factors that we used were a Shortbelly Rockfish correction factor for all rockfish species, and species-specific correction factors for Shiner Perch and Pacific Staghorn Sculpin. It should be noted that reconstruction was not pos- sible for all species in the diet samples because of the diversity of harbor seal diet and lack of appropriate correction factors as noted previously and in Table 2. Given the complexity of harbor seal diet and lack of reconstruction techniques for several species, we recon- structed the proportion in the sample only for prey spe- cies of conservation concern or for prey species whose frequency of occurrence was >5.0 in the broader study of harbor seal diet (Lance et al., 2012). Our goal was to set a reasonable range of values for model input in addition to describing diet composition; therefore, we make here a distinction between diet sample results and the parameter values used in the model to calcu- late consumption. When there was great uncertainty in percent contribution by wet weight to harbor seal diet because of the use of proxy correction factors or omission of some species from biomass reconstruction, confidence intervals were increased (see Model uncer- tainty and parameter estimation section). Consumption rates We calculated consumption (as biomass) for 5 key prey species or groups that are species of conserva- tion concern or most common in harbor seal diet: her- Table 2 Wet biomass construction results for the most common (frequency of occurrence >5.0) prey species or groups in diet of harbor seals ( Phoca vitulina ) during breeding and nonbreeding seasons, 2007-08. 1 All prey with frequency of occurrence >5.0 are listed to illustrate which common species or groups were not recon- 32 Fishery Bulletin 1 1 1 (1) C u 02 uG < 2 c _c .2 £ 03 0) l si ^ s T3 cd 02 cO T3 - jg -S 7j .2 G <3 a 3 1 ■° M a cl, g 2 o CO g 2 CL) ’£> Lh a; 02 Q- ^ co co ^ 02 •o ^ CO 'Jj r; G w * ° 1 Vh G 02 a; *-< > 02 03 "cd GG T3 -Q T3 03 O) 0) > '5 3 « 5 o3 co "2 o o3 _Q > O /-H G T3 5 02 2 ^ ^ 02 H B- u CO 02 O % £ G cd *-i QJ VJ3 co 8 £P a -3 .2 U | S £ a CO o G G O C*_H a o 02 CQ W> 3 < is "ibo < " 02 ^ tj g? 1 £ 2.S o £ O £ C1 G 03 02 bJD G -G < 2 ,5P 2 < £ < 2; < 2 cq iq id d < < < zzz oo 'O io oo < < < Z2iZ o o o o TF CD 05 < N1 N IN 2; 03 CM 00 O O [III O O lO CO o o o o d d d d co < d ^ Z * Z 2 ° ° 2 ° < 2; cu 2 S Cu c Si ■S d y c CO CL) D ^ X T3 03 jG CO N 2 e? .2 6 ___ 03 O CQ .§ -g co o 03 C is x a c o Si 3 G .2 c — y o m Q. 22 co co C CO 2 22 S 8 as g co QJ "G ^ o G ^ > 1 £ < .5 co 02 3 'cj co £ 02 02 CJ m 02 CJ a CO CJ 02 G u o G 03 co 02 -G CO yG a CO -G -G &D -a JG CJ dsi CJ co qG 03 G cd S-4 02 O 02 CJ m CO Dh a 02 O QJ CJ tn co GG u eg "o eg CJ 02 G 02 -t-J 03 G 02 > -*-> 3 03 cd lc D Dh CU co CO < % QJ £ g 03 T3 03 -a a O G 02 02 03 O) 02 ^ O 03 ^5 -u o o d a a s O < w P3 co co S T3 G- ;Average and ranges reported are between sampling months, includes all unidentified clupeids. •^No otolith length or otolith loss correction factor was available; these estimates should be treated with caution. Howard et al Fish consumption by harbor seals ( Phoca vituiina ) in the San Juan Islands, Washington 33 ring, salmonids, rockfish, Walleye Pollock, and Shiner Perch. Gross energy requirements were translated to consumption rates by applying the energetic density of prey to the proportion by wet weight of prey items in seal diet (Perez, 1994; Van Pelt et ah, 1997; Paul et ah, 1998; Payne et ah, 1999; Anthony et al., 2000; Roby et al., 2003). After biomass reconstruction, all species of adult and juvenile salmonids were combined into a “salmonid” complex. A “herring” complex represented Clupea pallasii and unidentified clupeid species. There are 2 other clupeid species in the study area, but, be- cause of their rareness, we assumed most species were C. pallasii (M.M. Lance, personal commun.). When prey were placed into broader taxonomic groups, we used the minimum and maximum values for energetic densi- ty reported for all prey sizes and ages in the literature to represent the prey group. Model uncertainty and parameter estimation Model variables described in Table 1 were randomly chosen during 1000 simulations from probability dis- tributions to estimate uncertainty in all model outputs. Where estimation of distribution parameters was not straightforward (e.g., lognormal), a maximum likeli- hood technique with the MASS package in R was used; this technique estimates the joint likelihood for dis- tribution parameter values, given the seal body mass values for each sex-and-age class (Venables and Ripley, 2002). We also made the following changes to diet re- sults to adjust the uniform distribution parameters for percentage by wet weight of prey in diet. If we had set the minimum and maximum values for a uniform distribution for proportion in diet exactly as found in diet samples, it would have been uninformative (i.e., a range of 0-100 often occurred but would imply no prior knowledge of diet composition; Table 2). There- fore, zero values from diet samples were discarded and minimum values for herring and salmonids were set as calculated from the remaining diet samples. For Shiner Perch and Walleye Pollock, zero values also were dis- carded. The minimum possible value was assumed to be 1%, and the maximum value was set near the aver- age calculated from diet samples. Harbor seal diet is diverse; therefore at least 20-30% of harbor seal diet was assumed to be made up of other species, and the maximum value possible for any prey species was set at 70-80% (the maximum value for nonbreeding season was set slightly lower because of increased diversity of diet). All model outputs are reported as means ^stan- dard deviation). Sensitivity analyses also were used to identify pa- rameters with the most influence on model outputs by systematically allowing one parameter at a time to be chosen randomly while other variables were fixed at their mean value(s). In this manner, any variation in the model outputs should be the direct result of varia- tion in the parameter of interest (Shelton et al., 1997; Stenson et al., 1997; Winship et al., 2002). The percent- age of variance explained by a single variable was cal- culated as the variance of model outputs when single random variables were used and divided by the total variance when all variables were randomly chosen. To estimate the effect of age structure on total prey consumption, we used different ratios of adults to sub- adults in 3 alternate model scenarios. We increased the number of adults in the population by 25%, 50%, and 100% and kept the total population size stable. During the breeding season, adult harbor seals fast or reduce consumption (Bowen et al., 1992; Coltman et al., 1998); therefore, there may be a discrepancy between predicted energy requirements and timing of consumption during an annual cycle. Rather than use direct consumption, we addressed the effect of this discrepancy with a correction factor that accounted for energy obtained from burning body fat stores in the breeding season. We estimated the amount of energy consumed, stored as body fat, and later metabolized by adult seals with the same estimates of digestive effi- ciency and energy density of prey that were used in the overall consumption model. Results Fish consumption There were 196 and 361 scat samples collected dur- ing the breeding and nonbreeding seasons, respective- ly. In these samples, 23 and 29 prey taxa were iden- tified during the breeding and nonbreeding seasons. Ten prey taxa were selected for reconstruction in this study; they had a frequency of occurrence >5.0 in the broader harbor seal diet study (Lance et al., 2012) or were species of conservation concern. Of these 10 taxa, 3 prey groups (unidentified gadid, skate species, and American Shad [Alosa sapidissima ]) could not be used because we had insufficient methods (e.g., lack of cor- rection factors) to reconstruct their presence in seal diet. Of the remaining prey, herring comprised the vast majority of reconstructed samples: >80% of wet weight in both breeding and nonbreeding seasons. Salmonids composed 15% and 9% in the breeding and nonbreed- ing seasons, respectively (Table 2). We were not able to identify rockfish otoliths to species in either season. In the breeding season, rockfish frequency of occurrence was 0.5% and therefore was assumed to contribute little in energetic terms to diet and was not further considered for calculation of consumption rates. Mea- surable otoliths were not found for rockfish species in the nonbreeding season; therefore, we were unable to determine species or size. During the nonbreeding sea- son, rockfish frequency of occurrence was 1.4% (Lance et al., 2012); we set a hypothetical range for proportion of wet weight of rockfish in diet at 1. 0-2.0%. Walleye Pollock and Shiner Perch constituted a relatively mi- nor portion (averages 0.5-2. 8%) of reconstructed diet (Table 2). 34 Fishery Bulletin 1 1 1 (1) During the seal breeding season, the average con- sumption for prey species calculated over 1000 simu- lations was 783 (±380) metric tons (t) of salmonids, 646 (±303) t of herring, 50 (±17) t of Walleye Pollock, and 22 (±4) t of Shiner Perch (Fig. 2). Subadult seals of both sexes consumed the greatest proportion of the total biomass (approximately 30-40% each), followed by adult females (27%). Adult males consumed a rela- tively small proportion of total biomass compared with adult females and subadults, and their consumption was only slightly higher than the biomass consumed by pups of both sexes (each <10%). During the nonbreeding season, consumption of herring and salmonids had the widest range of val- ues; rockfish, Shiner Perch, and Walleye Pollock were less variable. The average consumption for prey spe- cies calculated over 1000 simulations was 84 (±26) t of rockfish, 675 (±388) t of salmonids, 2151 (±706) t of herring, 66 (±13) t of Walleye Pollock, and 86 (±22) t of Shiner Perch (Fig. 2). The per capita fish consumption rate predicted by the model was 2.1 kg day-1 seal-1 (annual average 2.9, 2.8, 2.0, 2.2, and 1.0 kg for adult females, adult males, subadult females, subadult males, and pups, respectively). As was evident during the breeding sea- son, subadults (which included pups from the previous breed- ing season) of both sexes con- sumed the greatest proportion of the total biomass (approximately 30-45% each), followed by adult females (19%). Adult female consumption dropped slightly in the nonbreeding season. Adult males consumed the smallest proportion in the population (5%). Sensitivity analyses and assessment of model uncertainty Variation in seal body mass had the largest effect on energy use of the population and account- ed for >80% of model variance in both seasons. Taken togeth- er, all bioenergetics variables (mass, growth rates, and activity) accounted for the majority of the variance in the simulation mod- el. Fertility rates accounted for the next-greatest variance (7.3%) after body mass during the breeding season while pop- ulation size contributed least (1.3%) to overall model variabil- ity (Fig. 3). Consumption estimates of salmonids and herring were most sensitive to estimates of proportion of prey in the diet and energy density of prey. Variation in con- sumption estimates was low when the heat increment of feeding and assimilation efficiency parameters were varied within their estimated ranges. The variance in the nonbreeding season seen in the overall simulation model for both salmonids and herring was not well ex- plained by any single prey variable (Fig. 4). We estimated that adult seals used approximately 1,100,000 MJ of fat stores during the breeding season. Assuming an average prey energy density of 4000 J g1, this use of energy was equivalent to consumption of 300 t or approximately 6% and 21% of annual and breeding-season energy use, respectively. Increasing the number of adult seals in the population led to a positive increase in population energy use, although at a relatively slow rate of increase: even when we dou- bled the number of adults in the population, energy use increased only by 7% (Fig. 5). Howard et al.: Fish consumption by harbor seals ( Phoca vitulma ) in the San Juan Islands, Washington 35 1.4 1.2 1.0 ® O.E 0.6 Breeding -?■ j r • C$3 ^ U -6- $ ^ s > < o : o i i . Nonbreeding 0*0 JSA if & r<0 ,x>° j# x>* ^ 20 L/min. Crabs were assessed daily, and those crabs that died were recorded and removed. Early in the 2009 work, it became apparent that many of the red king crabs with no reflex impairments but apparent injuries were dying. This outcome indi- cated that fatally injured red king crab were not as likely to lose reflexes as were the Chionoecetes crabs and led us to adapt the full RAMP approach so that all red king crab that had either a missing reflex or an apparent injury were held. To ascertain how commonly fatal damage was completely hidden, 367 uninjured crabs displaying all reflexes were held. Our estimator of the probability of mortality for crabs with each reflex score was the proportion of held crabs with that score that died for each species. To estimate overall mortalities, the proportions of crabs in each reflex class were multiplied by the probability mortality of that reflex class and summed (Eq. 1): mc = Sr=0 to 6^r * (1) where mc = the mortality estimate for a species in catch c; mr = the mortality probability from the RAMP for that species for reflex score r ; and prc = the proportion of that species from catch c with reflex score r. For red king crab, this formula was modified to use the actual mortality outcomes of the injured and reflex- impaired crabs, all of which were held for observation (Eq. 2): mc = (mu * puc) + mjNic)*pic), (2) where DIC = the number of impaired or injured crabs that died from catch c; Nic = the number of injured or impaired crabs in catch c; and m and p have the same meaning as in Equation 1, except that i refers to injured or im- paired crabs and u refers to those crabs that were uninjured with all reflexes present. To estimate mortality for crabs that encountered a por- tion of the trawl, mortalities (mc) for all catches from recapture nets installed in that area were averaged and weighted for the number of crabs in each catch. To correct mortality estimates for handling dam- age, we assumed that gear and handling mortalities were independent and sequential. That is, where both processes occurred together in the recapture catches (mg+h)> the Sear mortality (mg) occurred first and only those crabs not killed by the gear (1 - mg) were vulner- able to handling mortality (mh, estimated as the mor- Rose et a! : Mortality rates for Chionoecetes opilio, C. bairdi, and Paralithodes camtschaticus after trawls on the seafloor 47 tality rate from the control net), resulting in Equation 3: mg+h = mg + ((1 - mg) * mh). (3) This equation was solved for mg, resulting in Equa- tion 4: mg = (mg+h _ mh) / (1 - m\)- (4) If the cumulative effects of gear impact and handling caused additional mortalities, this estimator would at- tribute those mortalities to gear effects, resulting in overestimated gear-caused mortalities. To account for variability due to the combination of reflex score assessments, RAMP prediction of mortality, and corrections for handling mortality, a randomization approach was used for hypothesis testing and estima- tion of confidence intervals. A model of the experiment was implemented with the Resampling Stats add-in for Microsoft Excel (Resampling Statistics, Inc., Arlington, VA., http://www.resample.com).2 RAMP estimators were regenerated for each trial by making random binary draws for each reflex score category (Urn procedure) and by using the sample size and mortality probability for that score from the experiment. New probabilities, cal- culated from that draw, were then used in the mortality estimation procedure for that trial. In resampling from the reflex assessments, we used each catch as our sample unit, choosing not to assume that individual crabs within a catch have independent mortality probabilities. To test null hypotheses that 2 groups of catches (e.g., catches from recapture nets at different trawl locations) actually came from the same population, the groups were combined and random draws were made from that combination, without replacement (Shuffle procedure), filling 2 new samples corresponding in number to the samples from the original experiment. A mortality estimate was generated for each trial by using the RAMP and assessment draws. For each test, 5000 trials were generated, and the proportion of those trials with differences greater than the observed esti- mate indicated the probability that our result occurred from a random process in which the mortality rates for both groups were equal. Comparisons were made between catches from each of the 3 gear areas (center footrope, footrope wings and extension, and sweeps) and the control catches to deter- mine whether those trawl encounters caused significant mortality. Subsequent tests were made for differences between the 2 footrope areas and between the sweeps and the combined footrope areas. Confidence intervals were generated by a similar pro- cess, except that samples of the assessment catches for each group, including control catches, were randomly selected with replacement from the actual catches for that group. Handling corrections were applied to mortal- 2 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. ity estimates generated for each gear component, on the basis of the control estimate from each trial. Confidence intervals (95%) were generated by identification of the highest 2.5% and the lowest 2.5% of the estimates from 5000 trials. Effects of sex, size, species, and shell condition on mortality rates were examined with logistic regression after the effects of each gear component were accounted for. Mortality was initially regressed against gear com- ponents, and the effects of these other factors were then tested against the residual variation. Because logistic regression requires binomial outcomes, specific RAMP probabilities of death could not be directly applied. Where direct observations from holding were not avail- able, crab mortality outcomes were scored on the basis of whether RAMP probabilities for their mortality were less or greater than 50%. Significant effects also were tested for interactions of each significant factor with gear area. Results The 159 total tows included 17-21 tows for each spe- cies at each recapture position. Between 154 and 991 crabs from each of the 6 combinations of species and sex were assessed after their capture behind each gear component and in the control position, and a substan- tial range of crab sizes were recorded within each com- bination (Table 1). Augmentation of the Stoner et al. (2008) RAMP re- lationships for the 2 Chionoecetes species by holding additional crabs in 2008 had only minor effects on mor- tality rate estimates (Table 2) other than to reduce un- certainty due to larger sample sizes (Hammond, 2009). For all 3 species, injuries varied widely in affected body part, type of damage, and severity, and were cor- related with both reflex score and mortality rate. Of the red king crab with at least one missing reflex (re- flex scores of 1 to 6), 96% also had observable inju- ries, as opposed to only 5% of those crab with no miss- ing reflexes (reflex score of 0). Crabs of all 3 species never survived removal of their abdomen or carapace, although autotomized legs (dropped off after injury) rarely caused fatalities. Crabs with either leg damage or carapace cracks normally survived, depending on extent, severity, and combination with other injuries. Of the 485 surviving red king crab released at the end of this study, 482 had all reflexes present upon re- lease, including all 14 that initially were missing at least one reflex. The 3 crab that were missing a re- flex upon release were all missing the eye reflex, had been held for 9 or 10 days, and had significant injuries, including carapace cracks. Although 25% (122) of the surviving crab had detectable injuries, their survival through the holding period and vigorous state condi- tion upon release indicated a low likelihood of signifi- cant later mortalities. 48 Fishery Bulletin 11 HI) Table t Number of crabs assessed and size ranges for each species and sex combination after they were captured behind 3 sections of bottom trawl gear, or with a control net. Size ranges, carapace width for snow ( Chionoecetes opilio) and southern Tanner crabs (C. bairdi) and carapace length for red king crab ( Paralithodes camtschaticus ) are given in millimeters. The three gear components were the footrope wings or extensions, the center of the footrope, and the sweep. For red king crab only, a fourth component was added, a sweep raised off of the seafloor (Rose et al., 2010). Snow crab Southern Tanner crab Red king crab Male Female Male Female Male Female No. Size range No. Size range No. Size range No. Size range No. Size range No. Size range Control 467 50-130 154 54-92 567 62-148 157 56-100 448 53-183 433 82-145 Sweep 407 47-126 218 52-93 281 60-147 518 59-98 370 64-188 226 63-150 Raised sweep 321 63-179 278 68-148 Footrope center 991 46-140 353 50-85 677 50-145 756 49-102 753 69-189 393 68-164 Footrope wing 696 48-130 540 50-110 288 51-143 494 52-97 203 61-167 263 70-156 Most southern Tanner and snow crabs captured be- hind the main trawl components had all reflexes pres- ent (76-93% reflex score of 0, Fig. 3), and the next most frequent category was dead crabs (reflex score of 6, no reflexes present) upon capture (2-17%). Similarly, a substantial majority (66-83%) of red king crab cap- tured behind the trawl gear was uninjured and had all reflexes present. Very few of these animals died during holding. Of the red king crab, 6% were dead upon cap- ture, making up 71% of mortalities. Therefore, nearly all of the observed crabs were either extremely likely to survive or moribund; relatively few crabs displayed an intermediate condition where the holding and RAMP results were critical to estimation of their probability of mortality. For both red king and southern Tanner crabs, the control net yielded 97% uninjured crab with all reflexes present and no crabs were dead upon capture. Snow crab had more immediate mortalities in the control net (2%) and only 88% had all reflexes present. Mortality estimates for crabs from the control nets (snow crab 7.1%, southern Tanner crab 8.5%, and red king crab 2.9%) were significantly lower than the estimates for crabs captured behind trawl components. Estimates of the rates of mortality due to contact with the trawl gear, adjusted for capture and handling, were below 16% (Fig. 4), with the exception of red king crab that encountered the wing section of the footrope, for which mortality was estimated at 31%. Overall, estimated mortality rates for all 3 species were sig- Table 2 Number of crabs held to observe delayed mortality and resulting mortality rates by reflex score (number of reflexes missing; 6 reflexes were assessed) and species for snow crab ( Chionoecetes opilio), southern Tanner crab (C. bairdi), and red king crab (Paralithodes camtschaticus). Crabs from Stoner et al. (2008) were included for both Chionoecetes species. Number of reflexes missing None missing None missing + injury * 1 2 3 4 5 All 6 missing Snow crab 500 — 78 70 57 79 67 61 Southern Tanner Crab 375 — 53 35 37 47 38 18 Red king crab 367 145 49 55 60 38 21 1 Mortality rate (%) Snow crab 1.4% — 20.5% 30.0% 43.9% 75.9% 88.1% 100.0% Southern Tanner Crab 7.2% — 32.1% 51.4% 86.5% 91.5% 92.1% 100.0% Red king crab 1.9% 23.4% 81.6% 94.5% 98.3% 100.0% 100.0% 100.0% * This category was used only for red king crab. Rose et al Mortality rates for Ch/onoecetes opiiio, C. bairdi, and Parahthodes camtschaticus after trawls on the seafloor 49 □ No reflexes missing a No reflexes missing + injury (RKC) □ Some reflexes missing (1-5) m All 6 reflexes missing 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Snow crab Control Sweep Footrope center Footrope wing Tanner crab Control Sweep Footrope center Footrope wing Red king crab Control Raised sweep Sweep Footrope center Footrope wing t: .i it n=m Figure 3 Percentage of crabs that displayed a range of reflex states in our study of unobserved mortality rates for snow crab (Chionoecetes opiiio), southern Tanner crab (C. bairdi), and red king crab (Paralithodes camtschaticus) . Reflex states were assigned on the basis of the number of reflexes that were missing; 6 reflexes were assessed. Crabs were captured after they contacted 1 of the 3 components of a bottom trawl representative of the gear used in Bering Sea bottom trawl fisheries — the center of the footrope, the footrope wings or extensions, or the sweep — or, for red king crab only, a sweep raised off of the seafloor (Rose et al., 2010). Crabs were captured with no gear contact during control tows. Injured red king crab with no missing reflexes were categorized separately. RKC = red king crab. nificantly lower for crabs that encountered the sweeps than for those crabs that encountered the footrope and were higher for those crabs that encountered the wing portion of the footrope as opposed to the center foo- trope. Although the mortality rates for the southern Tanner and snow crabs were similar, both had lower mortality rates than did the red king crab for all trawl components. Raising the sweeps with widely spaced disk clusters reduced red king crab mortality from 10% to 4%. Holding only samples of the large numbers of crabs with no missing reflexes (no missing reflexes and unin- jured for red king crab) greatly reduced the number of held crabs and produced minimal effects on precision of the mortality estimates. For example, the confidence interval estimation process was run with a sample size of 2581, representing all such red king crab observed, instead of the 367 crabs actually held. The resulting confidence range (high limit to low limit) for foot- rope wing mortality was reduced only from 14.25% to 13.99% by holding 7 times as many crabs. Confidence ranges for footrope center and sweep mortalities were reduced even less (3.53% to 3.50% and 5.98% to 5.95%, respectively). Logistic regression was used to examine whether mortality rates varied by species, sex, size, and shell condition, after the effects of gear were removed. Near- ly all crabs had either a new hard shell or old shell. For southern Tanner and snow crabs, marginally signif- icant effects were detected between species, sexes, and sizes. When the mean effects across the combinations of those factors were examined, it was apparent that most of those effects were the result of higher mortali- ties for snow crab with carapace widths >95 mm; those large snow crab were nearly all males. Large snow crab were approximately twice as likely to die as smaller 50 Fishery Bulletin 1 1 1 (1) 40% 35% 30% 25% 20% 15% 10% 5% 0% □ Snow crab ■ Tanner crab S Red king crab 0 Red king crab (raised sweep) Footrope wing Footrope center Sweep Figure 4 Estimates and 95% confidence intervals of rates of mortality for snow crab (Chionoecetes opilio), southern Tanner crab (C. bairdi ), and red king crab ( Paralithodes camtschaticus that resulted from contact with 1 of 3 different components of a bottom trawl represen- tative of the gear used bottom trawl fisheries in the Bering Sea — the footrope wings or extensions, the center of the footrope, or the sweep — and, for red king crab only, a sweep raised off of the seafloor (Rose et al., 2010). 1 I snow crab or as any size of southern Tanner crab, and this difference persisted across all gear components and control catches. Large red king crab had higher mortality than smaller king crabs (PcO.OOl), although this effect ex- plained <1% of the variability in mortality, compared with 12% for the difference between gear components. The interaction between crab size and gear component was not statistically significant; therefore, there was no indication that this difference in vulnerability between sizes varied between gear components. Mortality of red king crab did not vary significantly between sexes or between new-hard-shell and old-shell crab. Although the percentage of mortalities was high for soft-shell crab (4 of 5 died) and crab with very old shells (3 of 5 died), those shell types were too rare for a statistical validation of difference. Discussion Our study provides the first reliable estimates of mor- tality rates following noncapture (not bycatch or dis- card) bottom trawl encounters for 3 commercially im- portant crab species. Mortality rates varied by species but depended mainly on that part of the trawl system they encountered. Crabs that passed under the trawl footrope, particu- larly in the wing sections, died at higher rates than those crab struck by the sweeps. Effective herding by sweeps greatly expands the area of seafloor from which flatfishes are captured. Mortality rates were substan- tially lower for crabs that encountered these herding devices in that expanded area than for crabs that en- countered the trawl net itself, specifically the footrope. Therefore, enhancement of fish capture rates through effective herding can also reduce overall crab mor- talities (i.e., capture of equivalent quantities of fishes without herding would expose more crabs to footrope components). The effective reduction of crab mortality through use of sweeps was further augmented for red king crab with modifications to raise sweeps a few cen- timeters above the seafloor (Rose et al., 2010). The lower rates of unobserved crab mortalities from herding devices (i.e., sweeps), compared with mortality rates from trawl footropes, only partially indicate the potential of herding to reduce crab mortalities. Mortali- ties of crabs that encounter the footrope also would in- clude those crabs retained in the net (bycatch). Stevens (1990) found that mortality rates were much higher for both captured red king crab (79%) and captured south- ern Tanner crab (78%) than for escaping crabs. Some herding of crabs is conceivable, but their much slower Rose et al Mortality rates for Chionoecetes opilio, C. bairdi, and Paralithodes camtschaticus after trawls on the seafloor 51 locomotion, compared with that of commercial fish spe- cies, led us to assume that the number of crabs that encountered each part of the trawl system is roughly proportional to the area swept by each part. Red king crab had higher mortalities (6-32%) than 2 species of Chionoecetes, snow and southern Tanner crabs (4-15%) — a result that was expected given the generally smaller size and flatter body shape of Chi- onoecetes crabs. Overall mortality rates, weighted for the approximate relative areas swept by each trawl component for modern Bering Sea flatfish trawls (90% sweeps, 6% footrope wings, 4% footrope center) were 6% for snow crab, 5% for southern Tanner crab, and 11% for red king crab. The raised sweeps reduce mor- tality rate for red king crab to 6%. Such sweep modi- fications were required by the North Pacific Fishery Management Council for Bering Sea flatfish trawlers beginning in January 2011. The trawl gear and methods selected for our experi- ment represented those gear and methods used in the Bering Sea flatfish fisheries. The gear is characterized by long, combination rope sweeps and footropes built with large-diameter, rubber bobbins or disks to keep the net mouth more than 20 cm above the seafloor. This footrope selection by the fleet has been driven partially by pressure to reduce crab bycatch. Decreas- ing bycatch through changes to gear means that more crabs pass under the trawl net. Although other Alas- ka bottom trawl fisheries (e.g., for Pacific Cod [Gadus macrocephalus ]) use similar footropes, they use much shorter sweeps. Therefore, although cod trawls cover less seafloor (and hence contact fewer crabs) per kilo- meter towed than flatfish trawls, a higher proportion of the crabs might die because more of them would encounter the footrope components. The other major trawl fishery that can affect Bering Sea crabs is the fishery for Walleye Pollock { Theragra chalcogramma ). Pollock trawls must meet a number of requirements that allow them to be considered “pelagic” trawls, but this fishery commonly has been fished with substantial seafloor contact. Because regulations disallow any protective bobbins, none of the crab mortality estimates for gear compo- nents examined in our study can be used to estimate mortalities used for the pollock fishery, where chain foo- tropes are used. The differences we found in mortality rates between different gear components indicate that changes in the specific gear configurations could im- prove or worsen crab mortality rates. The rates found here should not be applied to trawls with substantially different ground gear (e.g., chain footropes used in the Bering Sea pollock fishery). Component-specific mortal- ity differences also present an opportunity to reduce crab mortality through identification of less damaging footrope configurations that sustain effective capture of target species. A companion study where an alter- native footrope was tested has been completed (Ham- mond, 2009). Because crabs were held for periods <14 days, our results did not include mortalities delayed over longer periods. The rapid drop of new mortalities after the first few days and the presence of all reflexes at the end of the study suggest that little additional mortality would be expected unless some other mechanism, such as infection or problems with molting, created a pulse of mortalities outside of the time period observed (see also Stoner et ah, 2008). Likewise, holding crabs in on- deck tanks protected them from predation that would have increased delayed mortality if vulnerability to predation was enhanced by injury or stress after trawl exposure. Predators and scavenger species have been observed to move into areas recently swept by bottom trawls (Prena et ah, 1999). Although this potential for additional mortality was not addressed directly in this study, the vast majority of the surviving crabs retained their full suite of assessed reflexes, including mobility of walking legs and defensive reactions. If predators initially focused on the more severely impaired and in- jured crabs that ended up as mortalities in our study, less impaired crabs might have some respite, allowing some time for recovery and reducing any difference be- tween our results and the actual unobserved mortality due to predation. All retained red king crab were held until the end of the study, 4 days after the control tows were complet- ed. Because control crabs were held for only 4-6 days, we examined the proportions of delayed mortalities of crabs held for longer periods. For crabs held more than 10 days, 93% of the mortalities occurred in the first 4 days and 95% in the first 6 days. Because only 9 of the 881 red king crab caught in the control net died, the possibility of missing one additional mortality because of a shortened holding time was not considered to in- troduce a significant potential bias. Short holding time was even less of a concern for southern Tanner and snow crab because all of those crabs were held 7 days or longer and the low proportion of deaths after the first days noted during the pilot project (Stoner et ah, 2008) continued during our 2008 observations. In this study, the RAMP procedure (Stoner et ah, 2008) was successful in prediction of mortality rates for many more crabs than we could have held to ob- serve delayed mortality. Of all crabs assessed, 85% had either all reflexes present ( Chionoecetes spp.) or were uninjured with all reflexes present (red king crab). Holding only one-eighth of these crabs provided gener- ous samples (>350 crabs per species) for estimation of their low mortality probabilities. If we had followed a conventional study method and held all crabs regard- less of reflex state, more than 4 times as many crabs would have been held, with minimal reductions in uncertainty. Although only representing a small proportion of the observed crabs, the RAMP procedure also allowed us to efficiently account for crabs with intermediate reflex assessments (reflex scores of 1 to 5). Because significant mortalities occurred to injured red king crab with all 52 Fishery Bulletin 1 1 1 (1) reflexes present, we held all those crabs, as well as all crabs of any of the 3 species with any missing reflexes. This procedure maintained the primary advantage of our RAMP assessments, accounting for a large group with high survival, and avoided the need to rely on injury assessments to estimate mortality. Both Stevens (1990) and Stoner et al. (2008) applied scoring systems for injuries, but the variety of injury types makes in- jury assessment more subjective and less likely to be repeatable than the reflex assessments. We provided specific estimates of the unobserved mortality rates of crabs swept over by trawl gear com- mon to bottom trawl fisheries in the Bering Sea. How- ever, assessment of the effects of such mortalities on the populations of those crabs will require estimation of the portion of those populations exposed to trawling each year. Although the distribution of trawling effort is well documented by automated position monitoring of vessels and onboard observers, the spatial distribu- tion of crabs throughout the year is not well known. A reliable estimate of the distribution of crabs, including seasonal variability, would be needed to estimate their exposure to trawling and allow for use of our mortality rate estimates in order to estimate resulting mortali- ties to the population. This approach would be subject to error from interannual and seasonal variations in crab distribution — variations that are not well under- stood and would be difficult to monitor. The number of crabs captured in bottom trawls is monitored through catch sampling by onboard ob- servers. Another way to estimate the number of crabs encountering trawls would be to learn the proportion of crabs that are caught in the path of a trawl. Crab bycatch data could then be expanded to estimate the number encountered, a value to which our mortality rates could be applied to estimate overall, unobserved mortality. One significant source of error for this ap- proach is variability or changes in the specific foot- ropes used across the fishery — differences that could substantially alter the proportion of crabs retained by the trawl. Also, should the trawl fishery approach its goal of eliminating crab bycatch, the base bycatch data could become sparse and even more variable. Conclusions Unobserved mortality is an important component of bycatch that is both easily overlooked and difficult to assess. Mortality rates for commercial crab species overrun by bottom trawls used in the Bering Sea var- ied substantially between the different components of trawls, with lower mortality for crabs that encountered sweeps than for crabs that encountered footropes. Re- duction of mortality rates of red king crab from 10% to 4% by raising the sweeps off the seafloor showed that gear modifications can mitigate unobserved mortality. Acknowledgments This study was primarily funded under a grant from the North Pacific Research Board (project 711), with additional support from the National Cooperative Re- search and National Bycatch Reduction Engineering Programs of the National Marine Fisheries Service, NOAA. We gratefully acknowledge the substantial con- tributions of Captain L. Perry and his crew on the FV Pacific Explorer and the invaluable sampling efforts of P. Iseri, S. Walters, D. Evans, and K. Lee, and particu- larly D. Benjamin, who participated during all 3 sum- mers of this study. Literature cited Armstrong, D. A., T. C. Wainwright, G. C. Jensen, P. A. Dinnel, and H. B. Andersen. 1993. Taking refuge from bycatch issues: red king crab (Paralithodes camtschaticus ) and trawl fisheries in the eastern Bering Sea. Can. J. Fish. Aquat. Sci. 50:1993-2000. Broadhurst, M. K., P. Suuronen, and A. Hulme. 2006. Estimating collateral mortality from towed fishing gear. Fish Fish. 7:180-218. Davis M. E., and M. L Ottmar. 2006. Wounding and reflex impairment may be predic- tors for mortality in discarded or escaped fish. Fish. Res. 82:1-6. Dew, C. B., and R. A. McConnaughey. 2005. Did trawling on the brood stock contribute to the collapse of Alaska’s king crab? Ecol. Appl. 15:919-941. Donaldson, W. E., and S. C. Byersdorfer. 2005. Biological field techniques for lithodid crabs. Alas- ka Sea Grant College Program Report AK-SG-05-03, 82 p. Univ. Alaska, Fairbanks, AK. doi:10.4027/bftlc.2005 Jadamec, L. S., W. E. Donaldson, and P. Cullenberg. 1999. Biological field techniques for Chionoecetes crab. Alaska Sea Grant College Program Report AK-SG-99-02, 80 p. Univ. Alaska, Fairbanks, AK. doi:10.4027/ bftcc. 1999 Hammond, C. F. 2009. Using reflex action mortality predictor (RAMP) to investigate if trawl gear modifications reduce unob- served mortality of Chionoecetes sp. M.S. thesis, 52 p. Univ. Washington, Seattle, WA. Murphy, M. C., W. E. Donaldson, and J. Zheng. 1994. Results of a questionnaire on research and man- agement priorities for commercial crab species in Alas- ka. Alaska Fish. Res. Bull. 1:81-96. Otto, R.S. 1990. An overview of eastern Bering Sea king and Tan- ner crab fisheries. In Proceedings of the international symposium on king and Tanner crabs; Anchorage, Alas- ka, 28-30 November, 1989, Lowell Wakefield Fisheries Symposia Series, p. 9-26. Alaska Sea Grant College Program Report 90-04. Univ. Alaska, Fairbanks, AK. Prena, J., P. Schwinghamer, T. W. Rowell, D. C. Gordon, K. D. Gilkinson, W. P. Vass, and D. L. McKeown. 1999. Experimental otter trawling on a sandy bottom ecosystem of the Grand Banks of Newfoundland: analy- Rose et al.: Mortality rates for Chionoecetes opilio, C. bairdi, and Paralithodes camtschaticus after trawls on the seafloor 53 sis of trawl bycatch and effects on epifauna. Mar. Ecol. Prog. Ser. 181:107-124. Rose, C. S. 1995. Behavior of North Pacific groundfish encounter- ing trawls: applications to reduce bycatch. In Solv- ing bycatch: considerations for today and tomorrow, p. 234-242. Alaska Sea Grant College Program Report 96-03. Univ. Alaska, Fairbanks, AK. 1999. Injury rates of red king crab, Paralithodes camts- chaticus, passing under bottom-trawl footropes. Mar. Fish. Rev. 61:72-76. Rose, C. S., J. R. Gauvin, and C. F. Hammond. 2010. Effective herding of flatfish by cables with mini- mal seafloor contact. Fish. Bull. 108:136-144. Stevens, B. G. 1990. Survival of king and Tanner crabs captured by commercial sole trawls. Fish. Bull. 88:731-744. Stoner, A. W., C. S. Rose, J. E. Munk, C. F. Hammond, and M. W. Davis. 2008. An assessment of discard mortality for two Alas- kan crab species, Tanner crab (Chionoecetes bairdi) and snow crab (C. opilio), based on reflex impairment. Fish. Bull. 106:337-347. Thompson, A. 1990. An industry perspective on problems facing the rebuilding of king and Tanner (bairdi) crab stocks of the eastern Bering Sea. In Proceedings of the interna- tional symposium on king and Tanner crabs; Anchorage, Alaska, 28-30 November, 1989, Lowell Wakefield fisher- ies symposia series, p. 533-545. Alaska Sea Grant Col- lege Program Report 90-04. Univ. Alaska, Fairbanks, AK. Witherell, D., C. Pautzke, and D. Fluharty. 2000. An ecosystem-based approach for Alaska ground- fish fisheries. ICES J. Mar. Sci. 57: 771-777. Witherell, D., and C. Pautzke. 1997. A brief history of bycatch management measures for eastern Bering Sea groundfish fisheries. Mar. Fish. Rev. 59:15-22. Witherell, D., and D. Woodby. 2005. Application of marine protected areas for sustain- able production and marine biodiversity off Alaska. Mar. Fish. Rev. 67:1-27. 54 Reactions of fishes to two underwater survey tools, a manned submersible and a remotely operated vehicle Email for address for contact author: tom.laidig@noaa.gov Abstract — We examined the reac- tions of fishes to a manned submers- ible and a remotely operated vehicle (ROV) during surveys conducted in habitats of rock and mud at depths of 30-408 m off central California in 2007. We observed 26 taxa for 10,550 fishes observed from the submersible and for 16,158 fishes observed from the ROV. A reaction was defined as a distinct movement of a fish that, for a benthic or hover- ing individual, was greater than one body length away from its initial po- sition or, for a swimming individual, was a change of course or speed. Of the observed fishes, 57% reacted to the ROV and 11% reacted to the submersible. Aggregating species and those species initially observed off the seafloor reacted most often to both vehicles. Fishes reacted more often to each vehicle when they were >1 m above the seafloor (22% of all fishes >1 m above the seafloor reacted to the submersible and 73% to the ROV) than when they were in contact with the seafloor (2% of all reactions to the submersible and 18% to the ROV). Fishes reacted by swimming away from both vehicles rather than toward them. Consider- ation of these reactions can inform survey designs and selection of sur- vey tools and can, thereby, increase the reliability of fish assemblage metrics (e.g., abundance, density, and biomass) and assessments of fish and habitat associations. Manuscript submitted 16 February 2012. Manuscript accepted 15 November 2012. Fish. Bull. 111:54-67 (2013). doi: 10. 7755/FB. 111.1.5 The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessar- ily reflect the position of the National Marine Fisheries Service, NOAA. Thomas E. Laidig (contact author) Lisa M. Krigsman Mary M. Yoklavich Fisheries Ecology Division Southwest Fisheries Science Center National Marine Fisheries Service, NOAA 110 Shaffer Road Santa Cruz, California 95060 Visual surveys of fishes in deep wa- ter and untrawlable areas have been conducted more frequently in re- cent years than in the past largely because of increased availability of underwater vehicles and the need for nonextractive assessments, par- ticularly in no-take areas. These vehicles have provided researchers with the opportunity to gather valu- able information on species composi- tion, habitat associations, population density, and various behavioral traits that was previously unattainable in these deep (>30 m), structurally complex areas (Carlson and Straty, 1981; Pearcy et ah, 1989; Yoklavich et ah, 2007; Laidig et ah, 2009; Love et ah, 2009). Visual surveys present advantages over traditional sampling methods (e.g., trawling, hook and line, traps) through the use of non- destructive, in situ observations of fishes in their natural habitats. One concern in counting fishes is their reaction to an observer (e.g., in scuba or snorkel surveys) or un- derwater vehicle (e.g., submersibles, remotely operated vehicles [ROVs], and camera sleds; Stoner et ah, 2008). The vehicles, in particular, can produce a number of electronic and mechanical stimuli (e.g., lights, mo- tion, and noise) that could alter be- havior (Krieger, 1993; Uiblein et ah, 2003; Ryer et ah, 2009). Accounting for these reactions is an important aspect of accurate population assess- ments. To this end, Yoklavich et ah (2007) quantified the reactions of fishes to a manned submersible dur- ing a survey of Cowcod ( Sebastes le- uis). Cowcod were found to react very little to the submersible, and that low level of reaction strengthened the ac- curacy of the survey results and as- sociated stock assessment. Other studies have reported fish reactions to both ROVs (Johnson et ah, 2003; Trenkel et ah, 2004a; Lorance and Trenkel, 2006) and manned submers- ibles (Murie et ah, 1994; Krieger and Sigler, 1996; Gibbons et ah, 2002). However, most of these studies were qualitative, simply noting that fishes moved out of the way of the vehicles. More quantitative studies are needed to improve our understanding of the nature and magnitude of reactions of various fish species to a variety of underwater survey vehicles. The goal of our study was to char- acterize the reactions of a wide range of marine fish species to 2 commonly used underwater vehicles (a manned submersible and an ROV) during surveys conducted along the seafloor. We quantified the degree of species- and size-specific reactions of fishes living both on and above the seafloor. Laidig et al Reactions of fishes to a manned submersible and a remotely operated vehicle 55 122’0’0-W Figure t Map of the area inside and outside of 3 marine protected areas (MPAs) off central California that was surveyed in 2007 for our study of the reac- tions of fishes to the manned Delta submersible and a remotely operated vehicle (ROV). Polygon shapes outline the MPAs, and triangles and circles indicate dive locations for the submersible and ROV. Bathymetry is in given meters. Materials and methods Fish surveys were conducted off central California (Fig. 1) inside and outside of 3 recently created marine protected areas (MPAs) — Point Lobos, Portuguese Ledge, and Soquel Canyon — with the 2-person Delta submersible (Delta Oceanographies, Torrance, CA) and a Phantom DS41 ROV (Deep Ocean Engineering, San Jose, CA). The manned submersible surveys occurred during the period of 20 September-5 No- vember, 2007, at depths of 30-365 m, and the ROV surveys were conducted during the period of 18-23 November, 2007, at depths of 71-408 m. All surveys were con- ducted during daylight hours from 0800 to 1700. Each submersible dive comprised 2-6 transects, each of a 10-min duration. The ROV dives were 1-3 h in duration. The ROV surveys were conducted along the same path of only a subset of the sub- mersible transects; in other words, not all submersible transects were paired with an ROV dive (Fig. 1). The Delta submersible (Fig. 2A) was launched from the FV Velero IV and op- erated by experienced pilots from Delta Oceanographies. An experienced scientific observer accompanied the pilot inside the untethered submersible. This yellow- orange submersible was 1.8 m tall, 4.6 m long, and from 0.4 m wide at its forward- most part to 1.1 m wide at mid-vehicle. The submersible was equipped with 2 vid- eo cameras: 1) a forward-facing, low-light, wide-angle, monochrome camera (Super SeaCam 5000, DeepSea Power and Light, San Diego, CA), and 2) a starboard-mount- ed, custom-built, color zoom camera with 400 iines of resolution and an illumina- tion range of 2-100,000 lux (Yoklavich and O’Connell, 2008). The position of the Delta submersible was tracked from the support vessel with WinFrog integrated navigation software (Fugro Pelagos, San Diego, CA) and an ORE Track- point-II ultra-short baseline (USBL) acoustic system (EdgeTech, West Wareham, MA). The distance traveled was estimated with a ring laser gyro and Doppler ve- locity log attached to the outside of the submersible. A single 24-volt propeller provided thrust. During sur- veys, the Delta traveled at an average speed of 0.5 m/s, ~1 m above the seafloor, following a directional heading given to the pilot by scientists aboard the support ves- sel. The submersible was equipped with ten 150-watt 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. halogen bulbs; only 3 of the starboard lights and 1 of the front-mounted forward-facing lights were used to illuminate the transect area. We also used an unmanned Phantom DS4 ROV launched from and tethered to the NOAA Ship Da- vid Starr Jordan and operated by experienced pilots from the National Marine Fisheries Service, South- west Fisheries Science Center, in La Jolla, California (Fig. 2B). The ROV had a yellow body and black frame and was 1 m tall, 2 m long, and 1.4 m wide. The ROV was equipped with a forward-facing, color video cam- era (Sony FCB-IX47C, Sony Corp., Tokyo, Japan) with 470 lines of horizontal resolution and an 18x optical zoom. Like the position of the submersible, the position of the ROV was tracked with WinFrog software and 56 Fishery Bulletin 1 1 1 (1) A Figure 2 (A) Photos of the front and port side of the Delta submersible and (B) views of the front and port side of the Phan- tom DS4 remotely operated vehicle (ROV), the 2 survey tools that were used in 2007 off central California in our study of the reactions of fishes to underwater vehicles. The Delta measures 1.8 m tall, 1.1m wide (tapering to 0.4 m at front port), and 4.6 m long. The ROV is 1.0 m tall, 1.4 m wide, and 2 m long. an ORE Trackpoint-II system. The ROV was propelled by 6 electric thrusters (2 angled and 4 perpendicular to the seafloor). Surveys were conducted at a target speed of 0.5 m/s and a target height of 1 m above the seafloor. Illumination was provided by two 250-watt Multi-SeaLite halogen lights from DeepSea Power and Light. The forward-facing video cameras on each vehicle were used to document fish reactions because these cameras had similar orientations and captured fish reactions in front of both vehicles. Both vehicles also were equipped with lasers to help the observers esti- mate size of fishes and their distance from the vehicle. The Delta submersible had a pair of parallel lasers Laidig et al Reactions of fishes to a manned submersible and a remotely operated vehicle 57 mounted 20 cm apart on either side of the color vid- eo camera and were visible to the observer inside the submersible. For both vehicles, fishes were measured to the nearest 5 cm. Five lasers were mounted on the front of the ROV; these lasers included 2 pairs of par- allel lasers (20 and 60 cm apart) and a single crossing laser used to determine depth of field. The laser spots on the video footage were used in postsurvey analysis to estimate both the size (total length) of fishes and the distance ahead of the ROV at which a fish reaction oc- curred. An effort was made to measure all fishes; how- ever, some fishes were either too far away or partially obscured, and, therefore, they could not be measured. In an important distinction in survey methodology between the 2 vehicles, the scientific observer inside the submersible identified, counted, and estimated length of fishes (as annotated on the audio channel of the video camera), but these tasks were performed only with video footage from the ROV surveys. Video foot- age from both vehicles was reviewed after completion of the surveys. Fishes in both surveys were identified to the lowest possible taxon with taxonomic keys (Love et al. [2002] for rockfishes, and Miller and Lea [1972] and Eschmeyer et al. [1983] for the remaining fishes). All fish reactions were determined solely from video footage of the forward-facing cameras on both vehicles in order for the methods to be similar between survey vehicles. A reaction was defined as a distinct movement of a fish if that movement was greater than one body length away from the initial position of the fish. Some fishes that were hovering off the seafloor would turn and face the vehicle as it passed by, but this movement was not considered a reaction unless a fish actively swam at least one body length in any direction. If a fish was swimming in a particular direction when first observed and continued swimming in the same direc- tion at the same speed during the entire time on video, that fish was considered to have no reaction. However, a reaction was noted if a fish changed course or swim- ming speed. The initial position of a fish was recorded as 1 of 3 categories: resting on the seafloor, <1 m above the seafloor (but not touching the seafloor), or >1 m above the seafloor. Direction of fish reaction was recorded as swimming 1) toward the vehicle, 2) parallel, forward, and away from the vehicle, 3) perpendicular to the left, 4) perpendicular to the right, or 5) down toward the seafloor. No fishes were ever recorded moving upward. We used time and an average vehicle speed of 0.5 m/s to estimate the distance between a reacting fish and the front of the Delta submersible. The distance between a reacting fish at first sighting and the front of the ROV was estimated with the laser array. This distance was binned to <3 m or 3-6 m. A 20-cm fish could be seen at a maximum distance of about 6 m in front of the ROV and about 9 m in front of the submersible (because im- ages could be distinguished farther with the low-light, monochrome camera on the submersible compared with the color camera on the ROV). To ensure that results from the ROV and submersible were comparable, we used data only from fishes that occurred at a distance of at most 6 m from the submersible. Hagfishes ( Eptatretus spp. ), thornyheads ( Sebastolo - bus spp.), and young-of-the-year (YOY) rockfishes ( Se - bastes spp.) were included as taxonomic groups in our analyses. Hagfishes often were seen hiding in holes or under structure and could not be identified to species. However, all the hagfishes that could be identified were Pacific Hagfish (Eptatretus stoutii). The thornyhead group comprised Shortspine Thornyhead ( Sebastolobus alascanus), a few Longspine Thornyhead (S. altivelis ; 1 observed from the submersible and 4 from the ROV), and mostly unidentified thornyheads. YOY rockfishes were a mix of many species, and each was recorded as 5 cm in total length. We determined fish reactions only while the vehi- cles traveled forward in survey mode. No fish reactions were counted if the seafloor, which we used as a sta- tionary reference for fish movement, was not observed in the video footage for >5 s (as when either vehicle transited over narrow ravines or when the ROV was pulled off transect by the ship). A number of species were not considered in our analyses. For instance, pe- lagic schooling fishes, such as Northern Anchovy (En- graulis mordax). Jack Mackerel ( Trachurus symmetri- cus ), and Pacific Chub Mackerel ( Scomber japonicus), swam around the vehicles for extended periods of time (possibly because they were attracted to the vehicles, but this idea was not verified), and these long periods of time increased the possibility that fishes would be double counted. These species also darted in and out of the view of the cameras, making it difficult to assess individual reactions to the vehicles. Only species that accounted for >1% of the total number of fish observed from either vehicle were included in the analyses of reactions to the vehicles. A chi-square test was used to evaluate reactions relative to initial fish position. Results A total of 223 transects (56 h) were surveyed with the Delta submersible in hard (70% rock, boulder, and cobble) and soft (30%) mud and sand) seafloor habitat, and 10,550 fishes were observed (Table 1). A total of 10 ROV dives (21 h) were conducted, and 16,158 fishes were observed. Although the ROV covered only a subset of all submersible dives, the type of habitats surveyed with the ROV (60% hard and 40% soft) were similar to those habitats surveyed with the submersible. Water visibility during submersible dives ranged from 4 to 13 m (as estimated by the submersible pilot during each dive), averaged 8 m, and was greatest at depths >100 m. Observations made from the submersible were lim- ited more by light penetration from the submersible (9 m) than by water visibility. Observations made from the ROV video footage were confined to ~6 m, because Table 1 The number and percentage of reactions of various fish taxa observed during visual surveys conducted from a manned submersible and remotely operated vehicle 58 Fishery Bulletin 1 1 1 (1) TJ a D >. g g O o o CM > O PS > o PS bn a G >H O G CO S a> CJ .2 bo w CO G -G G c/3 Sm cG « "a tf-t a; ° *5 t3 c/3 CL) v- cd ^ o £ “ cj -G G c/3 ci) cG >h S -g G C/3 0) CJ bn w G cd c/3 >h cG cd “ <4-* _G 0) o "is cd "d o 3 CG *- G C/3 ■d ^ CO o lOLOOOOOOOOiOOLOiOOOOiOiOOOiOOuOOLOLO Tff-HCOCDLOt^lOLOlO'^TfCMTfCMCMCOCOCM'^CDCM'^fCOTfLO II I I I I I I I I I I I I I I I I I I I I I I O ifl OOIOIOIOOIOOIOLOOIOIOIOIOIOLOIOIOOOO CM COCOxfHHHHCM CO H rHrHrH^HCMr-H^Hf-HCMCM Oit^COCDCDO^COCDCD CMCMOOCOlOCMHHTfrt HOl^CJOCOCMCJhOiO^OOHO h* b* CO Tp H H CO CD ^ i-h CO lO HCMlOiOCOHiOh00Hioa)CMCOHhC)CDCMTfOOO)O0 03 COCO CM 00 l> CO H H 00 H CM CO H h ifl h O CM t> h i— ( CM CD »— i 03 O CM O i-i O O CO O CM O CM O O ODh0GOCOlChOO^HOO3HOO3CM^HO3COOCOCO^OCO LOCO O3CDHTtrfO30O3COHCDHCMCOCO^ CO W rf CO CO CM lO CO i i i— ( O CM CD CM CM > — i CM tC t-h 03 CO <£ OOOOOOOOOOOiOiOOiOUOOiOLOO I^CDH^iOiO^COCM^CMHCOCOCMTtiOCMCO^ I I I I I I I I I I I I I I I I I I I I lOiOiDiOOOOOuOiOiOiOOOLOOiOOOO CM CM CM CM r- 03 ^ co o m inocoomHcohoo^03 ^tCDCMLOOt^CDO-— i lO i— i h CD H CD OWWHO'tHMN ^mCnO^Ol^CDCM(DCOt-030inCDOOC003HOCOCDOO O^COCDCMinmCMCM^OD in03^Hi0C0CDO^C003hin in CO H CD CM H H l> CM CO CM H in a 55 CO 5 d CJ S 5 d .CJ co co d d CJ 1 CD tuo CJ d ■5 CO d d CJ G d ~d o d s -S .CO s CD co 2 G CD co Ss. O SJD G o CD CD C/3 CJ S o d. CD d ’S, CD £ O CJ co d ^3 CO co d 5P o !3jO co O G G -o 12 CD CO PS co ^ .CD CO CO CO CO Q Q C3 d -o -o CD CD <33 > q . .. » ^ <^3 » .» -a a o, « §* 4 » “ "o 0) P g .g -a N co Hi' C/3 £; S’ * 2 Q CO O I a ^ co CO CO d d -o -o CD CD CO CO CO CO d d -o CD CD co co CO CO d d ~o ~o CD CD CO CO CG o PS -2 CD G CJ O) G G > CD >.V— .wGOOJ-i qpqpppppOOQO O O CD "G -d c c/3 G CG 3 bn -G G G X X cG -d ^ c/3 cj - CG O M PS I S " o >1 J3 £ ■£ bu m >> o CL, PS tn Xl B x ° M O 'S O .S DS OJ -4-0 C/3 O PS co cG -d o C/3 O CG PS fS o PS CD cG Q) d « G Si G G CD O' CO CO cG .G o ° rd -4-0 -G CO H o T3 i; r-1 • •“< cG G cj t? O 3 >H _ O Laidig et al.: Reactions of fishes to a manned submersible and a remotely operated vehicle 59 of the type of lighting and camera (see Materials and methods section). We used 26 taxa of fishes in the analyses of fish reactions to the submersible and ROV (Table 1). Half- banded ( Sebastes semicinctus; 25%), Blue (S. mystinus; 24%), and Pygmy (S. wilsonr, 12%) Rockfishes were the most abundant species observed from the Delta submersible, and Halfbanded (56%) and Pygmy (22%) Rockfishes were most abundant in the ROV survey. In total, observations of 1161 fishes for the Delta sub- mersible and 9206 fishes for the ROV were used in the analyses of directional movements and distance of re- action from each vehicle. Fewer fishes reacted to the manned submersible (11% of all fishes; Table 1) than to the ROV (57% of all fishes). The minimum distance of a fish reaction was 0.5 m from the submersible (96% of reactions were at a distance >1 m) and 1 m from the ROV. The percent reaction varied from 0% for several species to 54% for the Squarespot Rockfish (S. hopkinsi) observed from the submersible and from 0% for some species to 84% for Pink Seaperch, (Z alembius rosaceus) for fishes ob- served from the ROV. Of those taxa observed from the submersible, only Squarespot Rockfish had a reaction rate of at least 50%. Six taxa observed with the ROV had a reaction rate of at least 50%: Pink Seaperch, Pacific Hake ( Merluccius productus), Spotted Ratfish (Hydrolagus colliei), and Yellowtail (S. flauidus), Ca- nary (S. pinniger), and Halfbanded Rockfishes. Cow- cod, Bocaccio (S. paucispinis), and Canary Rockfish are of particular concern to fishery managers and in need of improved assessments (Hilborn et al., 2011; PFMC, 2011). The reaction rate of these 3 species to the sub- mersible ranged from 8% to 19%; their reactions to the ROV varied from 20% to 56%. Thornyheads, YOY rock- fishes, and hagfishes had reaction rates <10% to either vehicle. Fishes of 5 taxa did not react at all to the sub- mersible, and 1 group of taxa (YOY rockfishes) that did not react to the ROV. Fish reactions to both vehicles increased significant- ly as fish distance above the seafloor increased, and this trend in reaction was greater for the ROV than for the submersible (all fishes combined, P<0.001; Table 2; Fig. 3). Only 2% of the fishes observed on the seafloor during submersible surveys (i.e., 27 of 1261 fishes) and 7% observed near the seafloor (i.e., 410 of 6009) reacted to this vehicle. However, 18% of fishes on the seafloor (i.e., 512 of 2895) reacted to the ROV, with Halfbanded Rockfish and Blackeye Goby ( Rhinogobiops nicholsii ) accounting for 71% of these reactions (361 out of 512 fishes that reacted; Table 2). During the ROV surveys, fishes near the seafloor reacted more than fishes in contact with the seafloor (59% versus 18%, respective- ly), with Halfbanded and Pygmy Rockfishes represent- ing 93% of these reactions (3800 out of 4083 fishes that reacted). Fishes in the midwater, a region defined as >1 m above the seafloor, reacted the most to either vehicle (22% to the submersible and 73% to the ROV). Squares- pot and Blue Rockfishes represented 80% of the midwa- ter reactions to the submersible, and Halfbanded and Pygmy Rockfishes accounted for 90% of the reactions of midwater fishes to the ROV. This pattern of greater percentage of reactions with increased height off the seafloor was observed for most individual taxa. Even those species that are primarily demersal, like Cowcod and Greenstriped (S. elongatus) and Greenspotted (S. chlorostictus) Rockfishes, exhibited this pattern in ob- servations from both the submersible and ROV. The fishes that demonstrated any type of reaction to each vehicle primarily swam away rather than to- ward the vehicles (Fig. 4; Table 3, A and B). Only a small percentage (0-8%) of fishes swam toward ei- ther vehicle; most of these fishes were Bocaccio near the seafloor, and 19 of 50 of those Bocaccio reacted by swimming toward the submersible. Most fishes either moved away (forward, ahead of the vehicle) or sideways (to the left or right). However, 37% of all fishes in the midwater reacted by swimming downward when ini- tially encountered by the submersible (Table 3A). This group was dominated by Blue, Widow (S. entomelas), and Splitnose ( S . diploproa) Rockfishes (representing 96% of those midwater fishes that reacted by swim- ming down). Only 13% of all fishes near the seafloor moved downward as the submersible approached; Bo- caccio and Widow Rockfish reacted the most in this category (20% and 25% of all fish that reacted, respec- tively). Only 1% of fishes in the midwater or near the seafloor reacted to the ROV by swimming downward (Table 3B). The distance at which a fish reaction occurred var- ied between vehicles (Table 4; Fig. 5). Blue, Halfband- ed, Widow, Bank (S. rufus), and Splitnose Rockfishes moved at distances >3 m in front of the submersible. These species often were located in the midwater or near the seafloor. However, some species located most often near the seafloor (e.g., Bocaccio and Canary and Squarespot Rockfishes) reacted more often when the vehicle came closer to them (<3 m). Seafloor-dwelling species did not react often to the submersible, and, when they did, there was no clear pattern in reactions related to distance in front of the vehicle. The species that reacted farthest in front of the ROV were Half- banded, Widow, and Yellowtail Rockfishes, all of which were found near the seafloor or in the midwater. Spe- cies that reacted closer (<3 m) to the ROV included fishes living almost entirely on the seafloor (e.g.. Black- eye Goby, Shortspine Combfish [ Zaniolepis frenata], and Greenstriped Rockfish), as well as some near the seafloor and in the midwater (e.g., Bocaccio and Ca- nary, Greenspotted, Rosy [S. rosaceus], Splitnose, and Squarespot Rockfishes). Body length was determined for all fishes that were observed during the submersible surveys (n = 10,550), but only 9177 fishes (57%) of all fishes observed in video footage from the ROV surveys were measured. Total length ranged from 5 to 100 cm for fishes ob- served from the submersible and from 5 to 70 cm for fishes seen during the ROV surveys. Most fishes were Table 2 Number and percentage of fishes that reacted to the manned submersible and the remotely operated vehicle (ROV) that were used in our surveys in 2007 off central California, relative to the initial position of those fishes: on the seafloor, near (<1 m above) the seafloor, or in the midwater (>1 m above the seafloor). YOY=young-of-the-year, s=submersible, r=ROV. Significance levels: *=0.05, **=0.01, ***=0.0001. 60 Fishery Bulletin 111(1) > O OS 05 £ -o S _Q D m > O OS S _Q 3 > O OS 5B 03 co 3 cn O # vs ° JG d ^ ° -g | 1 ° -c ta -2 ^ w 2 o ►2 -5 ° o ^2 CD ° ■« jS tJ ^ cn ^ as ° x: to -2 d -£= 03 ° "2 CO ^ VS ”5 CO (N CM CD h Oi CO T-H CO CO io OO T-H o eg o CD t— i co o i> 05 CM OO t-h CM CD O O t> [> O CM O O O O ID o o 05 IOCO^H05CMCOCMCM 05 LO CD O T-H t-H CO ^ h H co O o CM CO ID 00 LO CD CO 05 05 00 CO TT CM GO CO O O C''“ CD O 05 Tf 05 O CO eg oo o t> ^ O CD co O' e- 05 CD ^ id o t> oo o o t> i> CM CM CMCD00500COCO 00 o o CM CD CD CD 05 i-H ID CM OCMOOOCMCDCMCOO CO t> Tp ID CM CO CM t-H ^ CM O e- t-h CM I> T-H O O CM T-H T-H CM I> O 05 05 CM t> O t-h CM t — i O CD ^OOCOCOCMCO^ l> CD O t~h CM CO e-OOOOCMOCMCDCOLOOOOO’-HOC'-’-HOO lOOOOOt— lOT-Hr-^tT-HOOOOCM O O O CM t-H t-H id ID CM O ID CO CM 05 ^ 00 CM O O O O CM c/3 JD -C o c/5 CD O >> q 05 Q* 03 PS On cj u; S P D CQ 03 03 C/3 cp ^ -p * CJ * o PS >> -a U o 03 CJ a £ 03 O 03 O O -C 03 CJ O o CO 3 *-« a ■3 ^ C/3 Cp cp q o q o DS DS T3 -a 03 a O -c Oh -*-> C/3 C/3 p a 03 03 03 03 V* U 2 Q O O O cp -P CJ o PS ™ " Is aj 12 tg C/3 CQ O cP £

> q cu PS cp jQ a -p ° M o cp P d*3 C/3 o 1 m above the seafloor). The total number of reactions is indicated for each vehicle. Numbers above bars indicate the total number of fishes (i.e., sample size) in each cat- egory for each vehicle. Discussion Although fishes reacted to both sur- vey vehicles, there were proportionally greater numbers of reactions to the ROV than to the submersible. The ROV and submersible traveled at similar speeds and maintained similar heights off the seafloor, yet substantial differ- ences were observed in fish reactions to the 2 vehicles. Possible reasons for these differences in reactions include the presence of a tether that attach- es the ROV to the support ship (the manned submersible is autonomous and untethered), forward lighting on the ROV compared with lighting large- ly on the starboard side of the sub- mersible, differences in vehicle noise, and disparity in vehicle dimensions. Both vehicles were much larger than common predators (e.g., large fishes and pinnipeds) of most of these species, and we, therefore, surmise that size alone was not the factor that caused fishes to re- act. It is possible that the smaller ROV, which was about one-half the height and length of the submers- ible, appeared to be more like a large predator to the fishes than did the submersible, but this idea is dif- ficult to establish. The magnitude of pressure waves generated in front of each vehicle could have differed because the submers- ible was of solid construction and the ROV comprised a frame with attached instruments and a trailing tether. Indeed, pressure waves generated from a deepwater drop-camera system that operated about 130 m above a midwater aggregation of Orange Roughy ( Hoplostethus atlanticus) off Tasmania caused those fishes to disperse rapidly up to 40 m (Koslow et al. 1995). Fish reactions to vehicles can also depend on envi- ronmental conditions (e.g., type of seafloor sediments, relief, ambient light levels, and water currents) and some attributes of the survey itself (e.g., vehicle speed and height off the seafloor). To reduce the effects of some of these conditions, we surveyed only during day- light hours, in similar habitats, during the same time of the same year, at similar speeds, and at similar heights off the seafloor. Whatever the reasons that fishes react to survey ve- hicles, the reaction of the target species must be con- sidered in selection of underwater vehicles to conduct surveys on fish abundance. Population abundance can be either over-or under-estimated if fish reactions to the survey vehicles are not quantified. Once the reac- tion rates are determined, correction factors can be developed to account for species-specific differences in reaction to the survey vehicles and to adjust resultant abundance estimates. Knowledge of fish reactions as- sociated with each survey tool can help ascertain the most appropriate survey method for target species. Clear description and quantification of fish reactions to underwater survey vehicles are not common in the literature. From a review of the literature, fish reac- tions were defined in only 2 of 37 published papers that reported on the reactions of fishes to underwater vehi- cles (see review in Stoner et al. 2008; Davis et a!., 1997; Krieger and Ito, 1999; Else et al., 2002; Moore et al., 2002; Uiblein et al., 2003; Costello et al., 2005; Gartner et al., 2008; Luck and Pietsch, 2008; Benefield et al., 2009; Trenkel and Lorance, 2011; Baker et al., 2012; O’Connell et al.2). A fish reaction was defined in one of these 2 articles as a “disturbed” behavior or “differenc- 2 O’Connell, V., D. Carlile, and C. Brylinsky. 2001. Demersal shelf rockfish stock assessment and fishery evaluation report for 2002. Regional Information Report 1J01-35, 42 p. Alaska Dept. Fish Game, Division of Commercial Fisheries, Juneau, AK. 62 Fishery Bulletin 1 1 1 (1) On seafloor o CD CD .2 c: CD o CD CL Near seafloor ■ Submersible (410 fishes, 8%) □ ROV (4083 fishes, 59%) 50 - 45 - 40 - Toward Away Left Right Down CD CD 03 CD CL Midwater Submersible (724 fishes, 22%) □ ROV (461 1 fishes, 73%) 50 - Toward Away Left Right Down Figure 4 Percentage of fishes that reacted in a particular direction to the manned submersible or the remotely operated vehicle (ROV) that were used in 2007 off central California in our study of the reactions of fishes to underwater vehicles rela- tive to the initial position of those fishes: on the seafloor, near (<1 m above) the seafloor, or in the midwater (>1 m above the seafloor). Total number of fishes that reacted to each vehicle, and the percentage of the total number of fishes in the survey that reacted, are shown in parentheses for each initial position. es in natural behavior” (Lorance and Trenkel, 2006) and as “a marked change in activity level and/or locomotion behavior” in the other article (Uiblein et al., 2003). General categories of re- actions (such as a fish avoided or was attracted to a vehicle, or a fish had no reaction) were used in 6 studies (Adams et al., 1995; Trenkel et al., 2004a; Trenkel et al., 2004b; Costello et al., 2005; Trenkel and Lorance, 2011, Baker et al., 2012), but specific definitions of the reac- tions (in contrast to natural movements) were not reported for these studies. In our surveys, reaction of a nonmoving fish was defined as a distinct movement greater than one body length. We used this proportional measure instead of a specific distance because the total length of observed fishes varied from 5 to 100 cm. The use of our definition of a reac- tion as at least one body length could be prob- lematic, especially for quantification of relative- ly small movements. However, in our study, the minimum distance that any fish traveled was 0.5 m in reaction to the submersible (with 96% of these fishes moving 1 m or greater) and 1.0 m in reaction to the ROV. Therefore, the reac- tions of even the smallest fishes could be read- ily discerned. It can be argued that a fish in motion when first seen in a video footage was already mov- ing in reaction to the survey vehicles (Uiblein et al., 2003; Lorance and Trenkel, 2006). In our study, we surveyed numerous benthopelagic species that were slowly moving when first observed in the video footage. Such movement was not considered a reaction unless a fish ob- viously changed course or speed. Because a fish could not be seen before it came into view on a video footage, it could not be determined if that fish was initially motionless and then reacted as the vehicle approached. This type of behav- ior could be indicated by signs like a dust cloud where a fish had contact with the seafloor, a fish quickly darting into the video footage, or loose aggregations of fishes moving in many dif- ferent directions. In our study, these types of behavior were rarely, if ever, observed. Few quantitative studies have been con- ducted on fish reactions to a submersible or an ROV, and no direct comparisons between the reactions of specific fish species to a submers- ible and ROV have been found in the literature. General reactions to an ROV (fishes moving into and out of a video frame) were quantified during surveys on mud habitats off central Cal- ifornia (Adams et al., 1995). In that study, most fishes that occurred on the seafloor did not re- act to a relatively large working-class ROV, al- though 2 species typically observed off the sea- floor exhibited avoidance behavior: 44% of all Table 3 A Number of fishes that reacted in a particular direction (toward, away, etc.) to a manned submersible in our surveys in 2007 off central California, relative to the initial position of those fishes: on the seafloor, near (<1 m above) the seafloor, or in the midwater (>1 m above the seafloor). YOY=young-of-the-year. Laidig et al Reactions of fishes to a manned submersible and a remotely operated vehicle 63 s O Q 07 O LO O o m o £ o O OOTfOCOOi^OCOtNOOtNCOOCNCOCMCO to CO H r—t r—i O ^ CN t''* CO CO (N W H O LO O CO LO O O O O 00 O O O lo co o lo r-~ r- 03 X O co two X X X o ° <-< rrt co x x X co co X X O X X *3 >> V. w x o ^ ° Vh c a g P 0. £ O G > > CD (D TJ xl X! 03 CD u ^ a - o •£ Dh -b co co X ^5 X o 03 o — j_ X O X X *5 s "S « o _h t(S •£ XI S C Sh ,h 03 CD — > — '-j cu w « r* x X cu ■ — <-j j-i D, D, O' eQPQeqmooQOOO!ilKcL,CL)cu»3qcicnmcom * o 03 O — CD o ■ l— ' X t- « £ a3 £ £ o3 a, tn o o ■ ■— ~ x — ■ a cr X X “ C0 Eh X CD O ^ Total 1261 27 1 6 7 13 0 6009 410 32 123 141 61 53 3280 724 7 317 120 9 271 Percent 2.1 4 22 26 48 0 6.8 8 30 34 15 13 22.1 1 44 17 1 37 Table 3B Number of fishes that reacted in a particular direction to a remotely operated vehicle (ROV) in our surveys in 2007 off central California, relative to the initial position of those fishes: on the seafloor, near (<1 m above) the seafloor, or in the midwater (>1 m above the seafloor). YOY=young-of-the-year. 64 Fishery Bulletin 111(1) s O Q C— CM 00 00 (N CO CM 17-05 CO CO LO » — i 00 *-h O t> Tf (MO CO t-H t— I UO CM CM CO O CO T-H cm c— CM CM CD CM IT- O t-h GO OO^f'^LOOCOCDCOOOCOCMCOLO'^OOCM'tfLO CM CO IT- CM 00 tO CO CO CM CO Tf to O O CM C- CO 05 CD ID CM O ID co -D r* CD! w ^ O ^ cC j- JO C/5 6 * cD -hi cD M o (D bJD cD -hi o c ' o D >> V M -Q a +3 C/5 c/5 D D cD C/) 03 o C/5 « £ « ffl m n d! o o Q ^ ^ ^ u t. t, CB ffl a o o K ffi -D S tb bX) r/l D Qj ro ■ <-i 9-D OCXcr-M-i-H (XPhPhPhPhC/DC/DC/DC/DC/DE-1 " T, ? - £ 03 D cr I £ >h ca cj O o 05 0h CL( Laidig et at. : Reactions of fishes to a manned submersible and a remotely operated vehicle 65 Table 4 Number of fishes reacting within 3 m or >3 m from the front of the manned submersible and the remotely operated vehicle (ROV). “Position” refers to the location where each fish taxon was most frequently observed. S=on the seafloor, N=near the seafloor, M=in midwater (>1 m above seafloor). The total number of fishes (n) observed is indicated for each vehicle. Position Submersible (n = 10,550) ROV ( «= 16, 158 ) Number of reacting fishes Number of reacting fishes <3 m >3 m Total <3 m >3 m Total Bank Rockfish N 8 38 46 14 3 17 Blackeye Goby S 0 74 17 91 Blue Rockfish M 2 179 181 0 2 2 Bocaccio N 57 8 65 30 5 35 Canary Rockfish N 18 2 20 32 3 35 Cowcod S 3 2 5 2 1 3 Dover Sole S 0 10 11 21 Greenblotched Rockfish S 2 1 3 5 0 5 Greenspotted Rockfish S, N 10 6 16 55 32 87 Greenstriped Rockfish s 2 3 5 66 10 76 Hagfishes s 1 0 1 4 2 6 Halfbanded Rockfish N, M 84 134 218 1258 5173 6431 Pacific Hake M 11 6 17 13 2 15 Pink Seaperch N, M 0 211 8 219 Pygmy Rockfish N 0 1201 581 1782 Rosethorn Rockfish M 0 4 4 2 1 3 Rosy Rockfish N 7 2 9 11 0 11 Shortspine Combfish S 1 0 1 23 4 27 Splitnose Rockfish S, N, M 14 40 54 29 10 39 Spotted Ratfish N 10 6 16 4 2 6 Squarespot Rockfish N, M 301 111 412 68 44 112 Stripetail Rockfish S 2 3 5 2 2 4 Thornyheads S 0 4 6 10 Widow Rockfish N, M 26 41 67 4 46 50 Yellowtail Rockfish N, M 12 4 16 37 82 119 YOY Rockfishes S, N 0 0 Total 571 590 1161 3159 6047 9206 Percentage of fish reactions 49 51 34 66 Percentage of all fishes 5 6 20 37 Sablefish ( Anoplopoma fimbria) and 39% of all Pacific Hake. Lorance and Trenkel (2006) observed that all 8 taxa seen in the Bay of Biscay, in habitat types rang- ing from flat to gentle slopes and from fine sediments to boulders, reacted to a large working-class ROV with rates from 10% to 90%. Uiblein et al. (2003), also in the Bay of Biscay, worked with a 3-person submers- ible to study fish behavior and found that most of the 7 more abundant taxa reacted to the vehicle by markedly changing their activity level. Two species observed in both of these studies (Roundnose Grena- dier [ Coryphaenoides rupestris] and Orange Roughy) reacted more often to the ROV than to the submers- ible. In our study, fishes that lived in the midwater above the seafloor reacted to both the Delta submersible and the Phantom ROV at a higher rate than did fishes on the seafloor. Similar results have been reported in other studies. Krieger and Ito (1999) observed that all Shortraker ( Sebastes borealis) and Rougheye (S. aleutianus) Rockfishes that occurred above the sea- floor reacted by swimming toward the seafloor as the Delta submersible approached, but only 5 out of the 531 recorded fishes of these 2 species moved when initially seen on the seafloor. Lorance and Trenkel (2006) examined the reactions of 8 fish taxa in the Bay of Biscay and observed that most species reacted to the working-class ROV; only the seafloor-dwelling, deep-sea Atlantic Thornyhead ( Trachyscorpia cristula- ta echinata) had little reaction to the vehicle. In that study, 2 of the 3 taxa that had the greatest reactions (shark species of the order Squaliformes and the fami- ly Scyliorhinidae) were commonly encountered as they swam high in the water column. Adams et al. (1995) 66 Fishery Bulletin 1 1 1 (1) Distance in front of vehicle Figure 5 Percentage of fishes that reacted at a specific distance in front of the manned submersible and the remotely operated vehicle (ROV). These percentages were used in 2007 off central Califor- nia in our study of the reactions of fishes to underwater vehicles. The total number of reactions in) is indicated for each vehicle. hide for target species and environmental conditions. Through such efforts, researchers will gain a better understanding of the effec- tiveness and limitations of potential survey vehicles. Acknowledgments We thank R. Starr, co-principal investigator of the Delta submersible cruise; J. Butler for the use and operation of the ROV; S. Mau for pi- loting the ROV; Delta Oceanographies; and the crews of the FV Velero IV and the David Starr Jordan. We thank M. Love, M. Nishimoto, T. O’Connell, and D. Watters for help with data collection. D. Watters also created the map of our study site. We also thank C. Rooper, S. Sogard, K. Stierhoff, R. Starr, and L. Wed- ding for their helpful comments on early versions of this manuscript. This study was funded in part by a grant from the California Ocean Protection Council to R. Starr and M. Yoklavich. used a working-class ROV and Starr et al. (1996) used the Delta submersible to estimate fish abundance; both studies determined that these vehicles were not re- liable in assessment of the abundance of fishes well above the seafloor. Conclusions What are the implications of the reaction of a fish to a survey vehicle? If the reaction occurs over a small dis- tance and the fish remains inside the survey transect, then the fish would be counted and its reaction would not affect the outcome of the survey. However, some reactions (both large and small in magnitude) could cause a fish to move out of the survey transect or out of view (e.g., into a hole or behind a rock) — behavior that would, thereby, bias the resultant abundance estimate. Similarly, overestimates of abundance could be made if a fish moves into a transect because of its reaction to a survey vehicle. Reactions of the target species need to be considered in selection of a survey vehicle, and the limitations of vehicles need to be evaluated relevant to the goals of a study. For instance, a comparative study can be undertaken to estimate abundance and reaction rates of fish species with various underwater vehicles (e.g., a submersible, ROV, camera sled, an autonomous underwater vehicle, or drop camera) within a specific survey area or over particular transects. From this type of study, the reaction of fishes and abundance es- timates can be ascertained for each vehicle, thereby aiding in the selection of an appropriate survey ve- Literature cited Adams, P. B., J. L. Butler, C. H. Baxter, T. E. Laidig, K. A. Dahlin, and W. W. Wakefield. 1995. Population estimates of Pacific coast groundfishes from video transects and swept-area trawls. Fish. Bull. 93:446-455. Baker, K. D., R. L. Haedrich, P. V. R. Snelgrove, V. E. Ware- ham, E. N. Edinger, and K. D. Gilkinson. 2012. Small-scale patterns of deep-sea fish distributions and assemblages of the Grand Banks, Newfoundland continental slope. Deep Sea Res. 65:171-188. Benefield, M. C., J. H. Caruso, and K. J. Sulak. 2009. In situ video observations of two manefishes (Per- ciformes: Caristiidae) in the mesopelagic zone of the northern Gulf of Mexico. Copeia 2009:637-641. Carlson, H. R., and R. R. Straty. 1981. Habitat and nursery grounds of Pacific rockfish, Sebastes spp., in rocky coastal areas of Southeastern Alaska. Mar. Fish. Rev. 43:13-19. Costello, M. J., M. McCrea, A. Freiwald, T. Lundalv, L. Jons- son, B. J. Bett, T. C. E. van Weering, H. de Haas, J. M. Rob- erts, and D. Allen. 2005. Role of cold-water Lophelia pertusa coral reefs as fish habitat in the NE Atlantic. In Cold-water corals and ecosystems (A. Freiwald and J. M. Roberts, eds.), p. 771-805. Erlangen Earth Conference Series. Spring- er-Verlag, Berlin. Davis, C. L., L. M. Carla, and D. O. Evans. 1997. Use of a remotely operated vehicle to study habitat and population density of juvenile lake trout. Trans. Am. Fish. Soc. 126:871-875. Else, P, L. Haldorson, and K. Krieger. 2002. Shortspine thornyhead ( Sebastolobus alascanus) abundance and habitat associations in the Gulf of Alas- ka. Fish. Bull. 100:193-199. Laidig et al Reactions of fishes to a manned submersible and a remotely operated vehicle 67 Eschmeyer, W. N, E. S. Herald, and H. Hammann. 1983. A field guide to Pacific Coast fishes of North Amer- ica, 336 p. Houghton Mifflin Co., Boston, MA. Gartner, J. V., K. J. Sulak, S. W. Ross, and A. M. Necaise. 2008. Persistent near-bottom aggregations of mesope- lagic animals along the North Carolina and Virginia continental slopes. Mar. Bio). 153:825-841. Gibbons, M. J., A. J. J. Goosen, and P. A. Wickens. 2002. Habitat use by demersal nekton on the continen- tal shelf in the Benguela ecosystem, southern Africa. Fish. Bull. 100:475-490. Hilborn, R., I. J. Stewart, T. A. Branch, and O. P. Jensen. 2011. Defining trade-offs among conservation, profitabil- ity, and food security in the California Current bottom- trawl fishery. Conserv. Biol. 26:257-266. Johnson, S. W., M. L. Murphy, and D. J. Csepp. 2003. Distribution, habitat, and behavior of rockfishes, Sebastes spp., in nearshore waters of southeastern Alaska: observations from a remotely operated vehicle. Environ. Biol. Fishes 66:259-270. Koslow, J. A., R. Kloser, and C. A. Stanley. 1995. Avoidance of a camera system by a deepwater fish, the orange roughy, ( Hoplostethus atlanticus ). Deep Sea Res. 42:233-244. Krieger, K. J. 1993. Distribution and abundance of rockfish determined from a submersible and by seafloor trawling. Fish. Bull. 91:87-96. Krieger, K. J., and D. H. Ito. 1999. Distribution and abundance of shortraker rockfish, Sebastes borealis , and rougheye rockfish, S. aleutianus, determined from a manned submersible. Fish. Bull. 97:264-272. Krieger, K. J., and M. F. Sigler. 1996. Catchability coefficient for rockfish estimated from trawl and submersible surveys. Fish. Bull. 94:282-288. Laidig, T. E., D. L. Watters, and M. M. Yoklavich. 2009. Demersal fish and habitat associations from visual surveys on the central California shelf. Estuar. Coast. Shelf Sci. 83:629-637. Lorance, P., and V. M. Trenkel, 2006. Variability in natural behaviour, and observed re- actions to an ROV, by mid-slope fish species. J. Exp. Mar. Biol. Ecol. 332:106-119. Love, M. S., M. Yoklavich, and L. Thorsteinson. 2002. The rockfishes of the Northeast Pacific, 414 p. Univ. California Press, Berkeley, CA. Love, M. S., M. Yoklavich, and D. M. Schroeder. 2009. Demersal fish assemblages in the Southern California Bight based on visual surveys in deep wa- ter. Environ. Biol. Fishes 84:55-68. Luck, D. G., and T. W. Pietsch. 2008. In-situ observations of a deep-sea ceratioid angler- fish of the genus Oneirodes (Lophiiformes: Oneirodidae). Copeia 2008:446-451. Miller, D. J., and R. N. Lea. 1972. Guide to the coastal marine fishes of California. Calif. Fish Game, Fish Bull. 157, 249 p. Moore, J. A., P. J. Auster, D. Calini, K. Heinonen, K. Barber, and B. Hecker. 2002. False boarfish Neocyttus helgae in the western north Atlantic. Bull. Peabody Mus. Nat. Hist. 49:31-41. Murie, D. J., D. C Parkyn, B. G. Clapp, and G. G. Krause. 1994. Observations on the distribution and activities of rockfish, Sebastes spp., in Saanich Inlet, British Co- lumbia, from the Pisces IV submersible. Fish. Bull. 92:313-323. Pearcy, W. G., D. L. Stein, M. A. Hixon, E. K. Pikitch, W. H. Barss, and R. M. Starr. 1989. Submersible observations of deep-reef fishes of Heceta Bank, Oregon. Fish. Bull. 87:955-965. PFMC (Pacific Fishery Management Council) and NMFS (Na- tional Marine Fisheries Service). 2011. Proposed harvest specifications and management measures for the 2011-2012 Pacific Coast ground- fish fishery and Amendment 16-5 to the Pacific Coast Groundfish Fishery Management Plan to update exist- ing rebuilding plans and adopt a rebuilding plan for Petrale Sole. Final Environmental Impact Statement, February 2011, 501 p. PFMC, Portland, OR. Ryer, C. H., A. W. Stoner, P. J. Iseri, and M. L. Spencer. 2009. Effects of simulated underwater vehicle lighting on fish behavior. Mar. Ecol. Prog. Ser. 391:97-106. Starr, R. M., D. S. Fox, M. A. Hixon, B. N. Tissot, G. E. John- son, and W. H. Barss. 1996. Comparison of submersible-survey and hy- droacoustic-survey estimates of fish density on a rocky bank. Fish. Bull. 94:113-123. Stoner, A. W., C. H. Ryer, S. J. Parker, P. J. Auster, and W. W. Wakefield. 2008. Evaluating the role of fish behavior in surveys conducted with underwater vehicles. Can. J. Fish. Aquat. Sci. 65:1230-1243. Trenkel, V. M., R. I. C. Francis, P. Lorance, S. Mahevas, M. Rochet, and D. M. Tracey. 2004a. Availability of deep-water fish to trawling and visual observation from a remotely operated vehicle (ROV). Mar. Ecol. Progr. Ser. 284:293-303. Trenkel, V. M., and P. Lorance. 2011. Estimating Synaphobranchus kaupii densities: contribution of fish behaviour to differences between bait experiments and visual strip transects. Deep Sea Res. 58:63-71. Trenkel, V. M., P. Lorance, and S. Mahevas. 2004b. Do visual transects provide true population density estimates for deepwater fish? J. Mar. Sci. 61:1050-1056. Uiblein, F., P. Lorance, and D. Latrouite. 2003. Behaviour and habitat utilisation of seven demer- sal fish species on the Bay of Biscay continental slope, NE Atlantic. Mar. Ecol. Prog. Ser. 257:223-232. Yoklavich, M. M., M. S. Love, and K. A. Forney. 2007. A fishery-independent assessment of an overfished rockfish stock, cowcod ( Sebastes levis), using direct ob- servations from an occupied submersible. Can. J. Fish. Aquat. Sci. 64:1795-1804. Yoklavich, M. M., and V. M. O’Connell. 2008. Twenty years of research on demersal communi- ties using the Delta submersible in the northeast Pacific. In Marine habitat mapping technology for Alaska (J. R. Reynolds, and H. G. Greene, eds.), p. 143-155. Alaska Sea Grant College Program Report AK-SG-08-03. Univ. Alaska, Fairbanks, AK. doi: 10.4027/mhmta.2008.10 68 Abstract — Rockfishes ( Sebastes spp.) tend to aggregate near rocky, cobble, or generally rugged areas that are difficult to survey with bottom trawls, and evidence indi- cates that assemblages of rockfish species may differ between areas accessible to trawling and those ar- eas that are not. Consequently, it is important to determine grounds that are trawlable or untrawlable so that the areas where trawl sur- vey results should be applied are ac- curately identified. To this end, we used multibeam echosounder data to generate metrics that describe the seafloor: backscatter strength at normal and oblique incidence angles, the variation of the angle-dependent backscatter strength within 10° of normal incidence, the scintillation of the acoustic intensity scattered from the seafloor, and the seafloor rugos- ity. We used these metrics to develop a binary classification scheme to estimate where the seafloor is ex- pected to be trawlable. The multi- beam echosounder data were verified through analyses of video and still images collected with a stereo drop camera and a remotely operated ve- hicle in a study at Snakehead Bank, -100 km south of Kodiak Island in the Gulf of Alaska. Comparisons of different combinations of metrics derived from the multibeam data indicated that the oblique-incidence backscatter strength was the most accurate estimator of trawlability at Snakehead Bank and that the addi- tion of other metrics provided only marginal improvements. If success- ful on a wider scale in the Gulf of Alaska, this acoustic remote-sensing technique, or a similar one, could help improve the accuracy of rock- fish stock assessments. Manuscript accepted 21 November 2012. Fish. Bull. 111:68-77 (2013). doi:10.7755/FB. 11 1.1.6 The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessar- ily reflect the position of the National Marine Fisheries Service, NOAA. Seabed classification for trawlability determined with a multibeam echo sounder on Snakehead Bank in the Gulf of Alaska Thomas C. Weber (contact author)1 Christopher Rooper2 John Butler3 Darin Jones2 Chris Wilson2 Email address for contact author weber@ccom.unh.edu 1 Center for Coastal and Ocean Mapping University of New Hampshire 24 Colovos Road Durham, New Hampshire 03824 2 Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way NE Seattle, Washington 98115 3 Southwest Fisheries Science Center National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Rockfish ( Sebastes spp.) stocks are difficult to assess because of their propensity to aggregate near the seafloor in areas that are difficult to trawl, such as rocky, cobble, or gener- ally rugged areas. Consequently, data from bottom-trawl surveys conducted in trawlable areas typically are ex- trapolated to all areas within the boundaries of a survey, regardless of whether the seafloor is trawlable or not (Wakabayashi et ah, 1985). Such extrapolation may result in biased biomass indices if, for example, there is a shift in biomass between strata with variable but unknown amounts of untrawlable seafloor (Cordue, 2006). Evidence also indicates that species assemblages differ between trawlable and untrawlable areas (Matthews and Richards, 1991; Ja- gielo et al., 2003; Rooper et al., 2010), and remote-sensing techniques with acoustic or optical sensors may be able to help identify these differ- ences. Equally important is the need to have a quantitative assessment of those grounds that are trawlable or untrawlable to more accurately esti- mate the areas where the results of different stock assessment methods are valid. In many bottom-trawl surveys, trawlability has been assessed through the subjective interpreta- tion of normal-incidence backscatter (echoes) from downward-looking sin- gle-beam echo sounders. These back- scatter echoes are examined by vessel captains with different levels of ex- perience, with different echo sound- ers, and with different echosounder settings. Multibeam echo sounders (MBES), which have been successful previously for characterizion of the seafloor for the purposes of mapping habitat and surficial geology (e.g., Kostylev et al., 2001; Goff et ah, 2004; Brown and Blondel, 2009), may offer an alternative solution for as- sessment of trawlability. In addition to the wider, high-precision coverage of the seafloor that results from the use of multiple beams, MBES offer the potential for more accurate dis- crimination between different types of seafloor substrate (e.g., silt, sand, cobble, and rock) than does the use of downward-looking single beams because of the angle-dependent na- Weber et al: Seabed classification for trawlability determined with a multibeam echo sounder 69 0 rough rock Figure 1 A prediction of the angle-dependent seafloor backscatter strength, Sb (dB), ac- cording to APL [1994], for the beam configuration used for the Simrad ME70 multibeam echo sounder at Snakehead Bank in the Gulf of Alaska during a cruise of the NOAA Ship Oscar Dyson in October 2009. The areas over which the oblique-incidence Sb and the slope of the angle-dependent backscatter within 10° of normal incidence (S^-slope) were calculated are shown. Normal-incidence Sb was calculated at 0° incidence angle. ture of the seafloor backscatter strength, Sb. For example, the normal-incidence (i.e., 0° inci- dence angle) Sb that would typi- cally be expected for both cobble and fine sand are predicted to be very similar but are appreciably different at increased incidence angles (Fig. 1). Angle-dependent metrics that describe the back- scatter from the seafloor have been extracted from MBES data in previous studies to determine the nature of seafloor sediments (e.g., Fonseca and Mayer, 2007). Seafloor backscatter collected with an MBES, as are the pre- dictions shown in Figure 1, are often treated as the ensemble average of a large number of random realizations of scattered acoustic intensity. Higher order statistics that describe the scat- tered intensity may also provide information that can be used to characterize the seafloor. Often, the amplitude of the backscat- ter echoes is expected to follow a Rayleigh distribution, with the underlying assumption that there are a large number of contributors to the backscatter from the seafloor at any instant in time (Jackson and Richardson, 2007). Abraham and Lyons (2002) have linked heavy-tailed, non-Rayleigh distributions of backscatter to a model with a relatively small number of objects on the sea- floor that have high levels of backscatter strength. In other words, the details of the probability density func- tion that describe the amplitude of the acoustic echoes are likely to be related to the size and density of the scattering objects and their relative role in the overall scattering response. Measures that indicate non-Ray- leigh backscatter may give an indication of distributed cobble or rock that would render a seafloor untrawlable. In this study, we examined the angle-dependent na- ture of Sb, as well as measures of non-Rayleigh dis- tribution of the backscatter and the seafloor rugos- ity (roughness) derived from bathymetric soundings, in an attempt to discriminate between trawlable and untrawlable seafloors. The data were collected with a Simrad1 ME70 MBES (Kongsberg AS, Horten, Nor- way) at a study area on Snakehead Bank in the Gulf of Alaska, -100 km south of Kodiak Island (Fig. 2). To test the efficacy of the acoustic measures as classifiers of the seafloor as either trawlable or untrawlable, we compared metrics derived from a MBES with observa- 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. tions collected with a stereo drop camera (SDC) system (Williams et ah, 2010) along with cameras mounted on a remotely operated vehicle (ROV) (Rooper et ah, 2012). The results of this comparison were then extracted to the entire multibeam data set that was collected with the Simrad ME70 during our Snakehead Bank surveys. Methods MBES data were collected with a Simrad ME70 MBES mounted on the hull of the NOAA ship Oscar Dyson. The Simrad ME70 was developed specifically for fish- eries applications (Trenkel et ah, 2008), although it also has been used for bathymetric mapping (e.g., Cut- ter et ah, 2010). The Simrad ME70 is configurable in terms of 1) the number of beams generated, 2) acoustic frequency for each beam, and 3) direction and open- ing angle of the beams. For our surveys at Snakehead Bank, the Simrad ME70 was configured to generate 31 beams at frequencies ranging from 73 to 117 kHz and at beam opening angles that ranged from 2.8° to 11.0°. The 31 beams were steered to 0° in the alongship di- rection and from -66° to +66° in the athwartship direc- tion, with the lowest frequencies steered to the high- est beam steering angles to mimimize the possibility of ambiguities associated with grating lobes (angular regions within a beam pattern of a transducer array that have equal sensitivity to the main angular region, or lobe, and cause ambiguities in the determination of 70 Fishery Bulletin 111(1) 56 10°N 56 05°N 56.00°N 55 95“N 55 90 N 1 54. 2° W 154 1“W 154 0°W 153 9°W 1538°W 153 7°W 1536°W 153 5°W Figure 2 The study area at Snakehead Bank in the Gulf of Alaska, south of Kodiak Island. Bathymetric contours are drawn at 50-m intervals. The locations where data were collected in 2009 with a Sim- rad ME70 multibeam echo sounder from the large-scale trackline and during focused surveys are shown in red (classified as untrawlable) and blue (classified as trawlable). Camera data collected in 2009 and 2010 with a stereo drop camera and a remotely operated vehicle are shown as green squares (untrawlable) and cyan circles (trawlable). target angle direction; the occurrence of grating lobes is specific to the design of the transducer array that generates beams). A pulse duration of 1.5 ms was used for each beam. During transmission and reception, the beam-pointing directions were compensated for pitch and roll of the ship with a GPS-aided inertial motion unit (IMU). The IMU was also used to georeference the data collected with the MBES. The standard target method was used to calibrate the combined transmit- receive sensitivity of each beam (Foote et al., 1987). In comparison with the Simrad ME70, most hydro- graphic MBES are capable of generating an order of magnitude more beams with beam opening angles of a fraction of a degree and, therefore, produce a relatively high density of bathymetric soundings and measure- ments of seafloor backscatter. To achieve a similarly high density of data with fewer beams, we processed the Simrad ME70 data with a hybrid multibeam and phase-differencing technique (Lurton, 2010) that pro- vided hundreds of independent seafloor soundings (each of which was associated with a measure of Sb) over a swath that nominally covered ±60°. At beam angles away from normal incidence, the insonified por- tion of the seafloor (the area on the seafloor defined by the intersection of the sonar pulse within the beam pattern of the transducer array) acts as a discrete tar- get; therefore, each beam was processed as if it were a phase-measuring bathymetric sonar (Lurton, 2010, section 8.2.3). Because this approach is more accu- rate at higher incidence angles (Jin and Tang, 1996), a weighted mean amplitude detection (Lurton, 2010, sec- tion 8.3.3) was used for beams with incidence angles of only a few degrees. For our data, the transition be- tween these 2 bottom detection approaches correspond- ed to an incidence angle of approximately 15°. The raw soundings were then merged with vessel position and attitude data and corrected for refraction through the water column. The georeferenced soundings were used to extract the rugosity in a grid of 25-m squares, or cells, by computing the ratio of the observed surface area within each grid cell to the area of a plane fitted to the same data. A measure of the acoustic power was associated with each bottom detection and was converted to Sb by accounting for system gains and calibration offsets, spherical spreading and absorption in the water col- umn, and area insonified. Area insonified was estimat- ed with the assumption that the seafloor was flat and with the method described by Lurton (2010, section 3.4.3). Applications of these radiometric corrections provided a realization of the angle-dependent seafloor backscatter, which was used to help characterize the seafloor, on each ping. Figure 1 shows predictions of the angle-dependent Sb for different substrate types that range from very fine silt to rough rock, on the basis of a scattering model that includes estimates for acoustic impedance, seafloor roughness, and sediment volume scattering strength (APL, 1994). In general, it can be difficult to disambiguate between the different factors that underlie these scattering curves (Fonseca and Mayer, 2007), but they do offer some separation between different substrate types. On the basis of an Weber et al: Seabed classification for trawlabiiity determined with a multibeam echo sounder 71 examination of the predictions of Sb shown in Figure 1, 3 different metrics that describe Sb were used, similar to those of Fonseca and Mayer (2007): the normal-inci- dence Sb, the slope of the angle-dependent backscatter within 10°of normal incidence (S6-slope), and the aver- age oblique-incidence Sb (30° <0< 60°). The acoustic power associated with each bottom de- tection also was converted to acoustic backscatter in- tensity and used to derive an estimate of the scintilla- tion index, SI, which is defined here as 2 S/ = -H, (1) P-i 2 2 where O/ and Ui = the variance and mean of the backscatter intensity, respectively. The SI is a measure of how the backscatter inten- sity fluctuates: for Rayleigh-distributed backscatter, the SI is equal to 1; for heavier tailed distributions that are a potential indicator of a relatively few strong scatterers contributing to the backscattered echo, the SI would be >1. The SI was calculated independently for each beam with a minimum of 50 samples (pings) and then averaged across beams. One important caveat to such SI estimation is that it is dependent on the sonar footprint on the seafloor (Abraham and Lyons, 2004), which changes as a function of incident angle and seafloor depth for MBES. To reduce changes in SI that were associated with the sonar footprint rather than the substrate type, we used only the beam angles between 34° and 50° to generate this parameter. This restriction of angles essentially reduced the resolution to that of a single multibeam swath. The MBES data were compared with image data (both video and still images) from an SDC and a ROV. The SDC contained identical Sony TRD-900 camcorder units (Sony Corp., Tokyo, Japan) capable of collecting progressive scan video images at a pixel resolution of 1280x720. Both SDC camcorder units were mounted on a sled in an aluminum frame and lowered to the seafloor with a dedicated winch, and illumination was provided by 2 lights mounted above the camera hous- ings inside the aluminum frame (Williams et ah, 2010). MBES data also were compared with data collected with a Phantom DS4 ROV (Deep Ocean Engineering, Inc., San Jose, CA). Video footage was recorded from the ROV with a forward-looking color camera (Sony FCB-IX47C module with 470 lines of horizontal resolu- tion and 18x optical zoom). Two pairs of parallel lasers on the ROV were used to estimate substrate size and horizontal field of view. Data were collected during 3 cruises conducted at Snakehead Bank, south of Kodiak Island in the Gulf of Alaska (Fig. 2). During the first cruise, the Oscar Dys- on and the FV Epic Explorer, a commercial fishing ves- sel, visited the study site on 4-12 October 2009. Data were collected aboard the Oscar Dyson with the Simrad ME70 and ROV, and data were collected with the stereo drop camera aboard the Epic Explorer. Several repeat large-scale surveys were conducted with The Oscar Dy- son along a series of parallel transect lines spaced 2.2 km (1.2 nmi) apart and 9.3-14.8 km (5-8 nmi) long. Three of these surveys were used for this analysis. In addition to the large-scale surveys, 4 small-scale, fo- cused surveys were conducted in the same area dur- ing the first of the 3 cruises. The focused surveys were designed to achieve “full coverage” (i.e., no unsampled regions of the seafloor) of the seafloor with the Simrad ME70 in areas where a relatively strong indication of fish had been observed in the acoustic data. For the small-scale surveys, transects were 1.9-3. 7 km (1-2 nmi) long and spaced 0.2-0. 4 km (0. 1-0.2 nmi) apart (depending on the water depth). The drop camera was deployed 9 times during the October 2009 cruise, and locations were chosen where the acoustic data indicated that rockfishes were most abundant. During each of the drop-camera deploy- ments, the camera sled moved over the bottom at speeds of <1.5 kn as the Epic Explorer drifted along transects that lasted up to 1 h and, as a result, col- lected relatively dense data in 9 small regions. The horizontal field of view of the drop camera averaged 2.43 m (standard error of the mean [SE] =0 . 14). The ROV was deployed in 5 different areas where the acoustic data indicated that rockfishes were most abundant. Each deployment lasted for a few hours. The horizontal field of view for the ROV averaged 2.61 m (SE=0.20). During the other 2 cruises in March and June of 2010, the study site was revisited and the SDC de- ployed 51 times aboard the Oscar Dyson. During these additional deployments, the seafloor was recorded in only 1 of the 2 available stereo cameras, preventing collection of stereographic images. Each of these de- ployments was short: the drop camera was deployed to the bottom for a couple of minutes before it was re- trieved to the surface. The resulting images were all from single, small patches ( <25 m radius) of seafloor, rather than from the drift transects described for the first cruise. The seafloor substrate observed during the under- water video transects was classified with a commonly used scheme (Stein et ah, 1992; Yoklavich et ah, 2000). The classification consisted of 2-letter codes for sub- strate types that denoted a primary substrate with >50% coverage of the seafloor bottom and a second- ary substrate with 20-49% coverage of the seafloor. There were 7 identified substrate types: mud (M), sand (S), pebble (P, diameter <6.5 cm), cobble (C, diameter 6.5-25.5 cm), boulder (B, diameter >25.5 cm), exposed low-relief bedrock (R), and exposed high-relief bedrock and rock ridges (K). The size of substrate particles was measured or estimated from a known horizontal field of view (~2.4 m) for the SDC and estimated with a paired laser system for the ROV. With this classifica- tion scheme, a section of seafloor covered primarily in cobble but with boulders over more than 20% of the surface would receive the substrate-type code cobble- 72 Fishery Bulletin 1 1 1 (1) boulder (Cb), with the secondary substrate indicated by the lower-case letter. Because the video collected with the SDC and ROV provided a continuous display of substrata, the substrate-type code was changed only if a substrate type encompassed more than 10 consecu- tive seconds of video. For this study, the substrate observed in the under- water video transects was further classified as either untrawlable or trawlable with reference to the stan- dard Poly-Nor’eastern 4-seam bottom trawl used in biennial bottom-trawl surveys of the Gulf of Alaska and Aleutian Islands by the Alaska Fisheries Science Center (Stauffer, 2004). The Poly-Nor’eastern bottom- trawl footrope comprised 10-cm disks interspersed with bobbins 36 cm in diameter. The untrawlable ar- eas were defined as any substrate containing boulders that reached >20 cm off the bottom of the seafloor or any substrate with exposed bedrock that was so rough that the standard bottom-trawl footrope would not eas- ily pass over it. Therefore, the trawlable grounds were those areas mostly composed of small cobble, gravel, sand, and mud without interspersed boulders or jagged rocks. The untrawlable grounds were those areas that contained any boulder or high-relief rock substrates. The same experienced observer classified the substrate for both the ROV and SDC video transects. The video data thus classified were partitioned in a grid of 25-m squares, or cells — a length scale that is a rough estimate for the accuracy of the position- ing systems associated with both video systems. The primary and secondary substrate types were given a numeric value based on a nominal substrate size, and each grid cell was assigned substrate types associated with the median values for all data within the cell boundaries. Grid cells also were assigned as trawlable or untrawlable if all data within a cell supported such a classification; otherwise, the grid cell was assigned a “mixed” classification. The gridded video classifications were then compared with the seafloor parameters (e.g., rugosity or normal-incidence Sb) derived from data col- lected with the Simrad ME70, where both types of data existed at the same position, to provide an indication of how each acoustically derived seafloor parameter was able to discriminate between trawlable and untraw- lable areas. This comparison was done for each param- eter separately and then done for various combinations of parameters to find a combination of parameters that best discriminated between trawlable and untrawlable substrate. For each parameter, a f-test was used to de- termine whether it was able to distinguish between trawlable and untrawlable seafloor at the significance level of a=0.05 (i.e. , where erroneous rejection of the null hypothesis is expected 5% of the time), and val- ues of standard difference (the difference between the sample means divided by the pooled standard devia- tion) were computed. When combinations of parameters were tested, a best-fit separation (for the goal of mini- mizing the classification error rate) within the multidi- mensional parameter space was found through exami- nation of the entire parameter space. To maintain a clear link back to the underlying data distribution, the separation between trawlable and untrawlable was as- sumed to be a line, plane, or hyperplane (a generaliza- tion of a plane into more than 2 dimensions), depend- ing on the dimension of the parameter space. Results The data showed a wide range of values and, presum- ably, associated substrate types. The shallowest (<100- m) portion of Snakehead Bank contained the highest oblique-incidence Sb (approximately -12 dB). This re- gion contained similar values for the normal-incidence Sb, and small S6-slope (<0.75 dB/°). Taken together, these data indicate a cobble seafloor on the top of the bank. On the northeastern side of the bank at depths -200 m, the oblique-incidence Sh reached its lowest value of approximately -30 dB with a normal-incidence Sh of -15 dB and S6-slope of -1.1 dB/° — values consis- tent with a substrate composed of very fine silt. The region with the highest normal-incidence Sb (-10 to -7 dB) occurred between 154°W and 153. 9°W and near 56.07°N in the northwest region of the bank. The S6-slope was also high in this region, reaching up to 1.5 dB/°, and the oblique-incidence Sb was between -18 dB and -15 dB. These results for the seafloor pa- rameters are confounding, given that the S6-slope was large enough to indicate a fine sand or silt, but the normal-incidence and oblique-incidence Sb both indi- cated a coarser sediment or a higher-than-anticipated volume scatter contribution due to heterogeneities or gas (Jones et ah, 2012) within the sediment. The SI shows a complicated pattern that did not appear to be well correlated with any certain sub- strate type, although there were large (hundreds of meters) contiguous regions that exhibited high SI val- ues (i.e., the data did not appear to be simply random noise). The rugosity levels show the bank to be rela- tively smooth along the top, except at a sharp transi- tion along its northeastern edge between the 100- and 150-m contours. The rugosity analysis also indicates the appearance of what may be large (wavelength -150 m) sand waves in the extreme southeastern por- tion of the study area and smaller pockmarks in the southwestern portion of the study area. The results of a comparison of the seafloor param- eters derived from the backscatter data that was col- lected with the Simrad ME70 and the substrate types derived from the data collected with the SDC and ROV are shown in Figure 3. These data show that, although substrate types Bb, Cb, and Gb are difficult to distin- guish with backscatter parameters, these 3 types are clearly separate from substrate type Ss. The oblique- incidence Sb values for substrate type Ss appeared to be bimodal, with the majority of the values residing be- tween -17 and -15 dB and a substantial number of val- ues between -29 and -26 dB. According to the notional Weber et al: Seabed classification for trawlability determined with a multibeam echo sounder 73 A B 20: Figure 3 The frequencies of occurrence for major and minor substrate combinations, classified from the data collected in 2009 and 2010 with a stereo drop camera and a remotely operated vehicle as a function of different seafloor characteristics derived from the data collected with a Simrad ME70 multibeam echo sounder. Major (capital letter) and minor (lowercase letter) substrate types included Bb=boulder; C=cobble; Gg=gravel; and Ss=sand. values shown in Figure 1, these 2 regions would cor- respond to sandy gravel and very fine silt, respectively. The lower set of oblique-incidence Sh values were found in the deepwater off the northern side of the bank at depths of 200-250 m and also on the south side of the bank at depths of 120-150 m. On average, the larg- est Sfc-slope and the widest range of normal-incidence Sb were observed on sandy substrate. The normal-in- cidence Sb for areas classified as sandy substrate ex- tended to ranges higher than would be expected, a find- ing that could be a result of unusually high volume- backscatter caused by gas or heterogeneities within the sediment volume. The harder substrates (Bb and Cb) all had small S6-slope, as expected, and on average had higher SI than the sandy sediments. To determine how each parameter discriminated between trawlable or untrawlable seafloor, using clas- sified SDC and ROV video data as verification, the frequencies of occurrence for each parameter were ex- tracted for each substrate type (Fig. 4). T-tests indicat- ed that the distributions of trawlable and untrawlable areas of seafloor were distinguishable at the oc=0.05 significance level (Table 1), although each parameter did not perform equally when discriminating between the 2 classifications. The 3 best individual discrimina- tors were the normal-incidence Sb, Sb- slope, and the oblique-incidence Sb with standard differences of 0.74, 1.12, and 1.89, respectively. Of these 3 parameters, the oblique-incidence Sb demonstrated the clearest separa- tion between trawlable and untrawlable seafloor, with a boundary at -13.4 dB. According to modeled data (Fig. 1), this Sb level discriminates cobble and rock from gravel, sand, and silt. The SI and rugosity were separated less well with standard differences of 0.25 for each. With the oblique-incidence Sb considered alone, the combined error rate (erroneous classifications of both trawlable and untrawlable seafloor) reached a mini- mum of 5.6% (n=303) with a boundary set at S6=-13.4 dB. To determine whether this error rate could be lowered, additional parameters derived from the data collected with the Simrad ME70 were linearly com- bined with the oblique-incidence Sb. Figure 5 shows the combination of the oblique- incidence Sb with each of these other parameters, along with a line that best discriminated between the trawlable and untrawlable classifications. The largest reduction in classification error rate was achieved when the oblique-incidence Sb was combined with either the normal-incidence Sb or the SI, both of which had a marginally improved er- ror rate of 5.0%. When 3 parameters were combined to discriminate between trawlable and untrawlable sea- 74 Fishery Bulletin 111 (1) Table 1 Results of a 2-sample t-test and the standard difference in a comparison of trawlable and untrawlable populations for differ- ent parameters derived from the data collected with the Simrad ME70 multibeam echo sounder during a cruise in 2009 aboard the NOAA Ship Oscar Dyson. These parameters are normal-incidence seafloor backscatter strength (Sb), oblique-incidence Sb, the slope of the angle-dependent backscatter within 10° of normal incidence (S^-slope), scintillation index (SI), and rugosity (roughness). Degrees of Standard t-statistic freedom P -value difference Normal-incidence Sb 6.6 260 2xlO-10 0.74 Oblique-incidence Sb 17.2 170 4xl0-39 1.89 S6-slope (0-10°) 9.9 287 5xlO-20 1.12 SI 2.1 216 0.04 0.25 Rugosity 3.6 418 0.0004 0.25 floor, the error rate did not change apprecia- bly except in the case of a combination of the oblique-incidence Sb, the normal-incidence Sb, and the SI, in which case the class error rate was reduced to 3.8%; similar error rates were found with 4 classes separated by a best-fit hyperplane. Because only marginal improvements in class error rate were achieved when multiple parameters were combined and maintenance of simplicity in the interpretation of the re- sults was desired, the oblique-incidence Sb was chosen as the sole discriminator between the trawlable and untrawlable seafloor at the study site. The classifications of trawlable and untrawlable seafloor classifications area shown in Figure 2 for both the from the Sim- rad ME70 and the data from the SDC and ROV. The classification based on the data from the Simrad ME70 is accurate throughout most Weber et al: Seabed classification for trawlability determined with a multibeam echo sounder 75 of the study site, and the most obvious error occurred on the north-south transect intersected 153. 9°W in an area with high oblique-incidence Sb. Discussion The oblique-incidence Sb and the S6-slope followed the expected trends when separated into trawlable and un- trawlable classes and these trends were verified from video data collected with the SDC and ROV. Untraw- lable areas were expected to have a larger oblique in- cidence Sb and Sfe-slope than trawlable areas on the basis of backscatter models (e.g., Fig. 1). The normal- incidence Sb did not appear to discriminate very well between trawlable and untrawlable seafloor and tended to have a wider distribution of backscatter values than would have been expected on the basis of consideration of the oblique-incidence Sb and the modeled values shown in Figure 1. There are several possible reasons for the lack of discrimination with normal-incidence Sb, including higher-than-expected normal-incidence Sb in the sands and silts caused by gas or heterogeneities within the sediment volume in some trawlable areas and higher-than-expected roughness in the areas of cobble and rock that caused a larger-than-anticipated reduction in the normal-incidence Sb for some untraw- lable areas. Although quite variable throughout the study area, the mode of the SI was slightly higher for the untraw- lable seafloor than it was for the trawlable seafloor. This difference seems plausible when we consider the SI to be a metric for how many scatterers are contrib- uting to the sonar return within a beam footprint. A SI value near 1 suggests that there are a large number of scatterers (i.e., the central limit theorem applies, and the backscatter amplitude is Rayleigh distributed), as might be expected from a sand or silt seafloor. On the other hand, a larger SI indicates that there are only a few dominant scatterers within the beam footprint, as might be expected from a seafloor of cobbles or boul- ders. Although the data indicate a trend in the correct direction, SI alone has not provided a clear separation between trawlable and untrawlable seafloor (e.g., a 76 Fishery Bulletin 111(1) threshold of 1.2 would result in a high classification error rate). Rugosity derived from the data collected with the Simrad ME70 was a poor discriminator of trawlable versus untrawlable seafloor, generally with lower val- ues (e.g., smoother seafloor) in areas where the valida- tion data from the SDC and ROV surveys indicate that the seafloor is untrawlable. The areas that contained high values of rugosity generally were dominated by larger scale features: the ridgeline on the northern edge of the bank, the sand waves in the southeast, or the pockmarks in the southwest. It is likely that the spatial resolution of the MBES was insufficient to pro- vide a useful estimate of the rugosity level and that an MBES with higher frequencies and higher resolution might provide more useful results. The oblique-incidence Sb alone provided a low er- ror rate as a discriminator between trawlable and un- trawlable seafloor. When combined with the other met- rics, it was possible to slightly lower the error rate, but an examination of the scatter plots in Figure 5 in- dicates that the error rates were not been lowered in any meaningful way. For example, the best-fit line that discriminates between the combination of oblique-inci- dence Sb and normal-incidence Sb shows that a com- bination of high oblique-incidence Sb and low normal- incidence Sh gives a better indication of untrawlable seafloor than high oblique-incidence Sb on its own. This finding is contrary to what the modeled seafloor return (Fig. 2) would predict: high oblique-incidence Sb and high normal-incidence Sb are a better predictor of an untrawlable seafloor. Therefore, it is likely that the marginal improvement in classification error rate with these extra parameters combined is simply a result of variations in the tails of the underlying data distribu- tions. With only marginal improvements (5. 6-3. 8%) in classification error rate when up to 4 parameters are combined, with a hyperplane separating the 2 classes, it is reasonable to choose the simpler approach of using only the oblique-incidence Sb as a predictor of traw- lable or untrawlable seafloor. Conclusions The results described here indicate that acoustic re- mote sensing of substrate type with an MBES, and oblique-incidence acoustic Sb in particular, offer useful insight into whether the seafloor is untrawlable. This conclusion is in qualitative agreement with the work of Jagielo et al. (2003), who used seafloor backscatter collected with a sidescan sonar as part of an a priori assessment of trawlability (note that much of the sid- escan record was collected at oblique incidence angles). Whether these types of acoustic metrics can provide a similar level of confidence regarding the distribution of untrawlable seafloor in areas throughout the entire Gulf of Alaska needs to be determined. If successful on a wider scale, this type of acoustic remote sensing can help refine the interpretation of bottom-trawl surveys. In particular, techniques such as those described here could increase the accuracy in identification of areas with seafloor characteristics similar to areas where bottom-trawl surveys of rockfish were conducted (i.e., areas where results frojm the trawl surveys can be ap- plied). As a result, the precision and accuracy of bio- mass estimates from bottom-trawl surveys and their resultant stock assessments would be improved. Acknowledgments Support for this work was provided by the North Pa- cific Research Board (contribution no. 373). Additional support for T. Weber was provided by NOAA (grant NA05N0S4001153). We would like to acknowledge the crews of the NOAA Ship Oscar Dyson and FV Epic Ex- plorer for their help during data collection. We would also like to thank M. Martin, D. Somerton, and W. Pal- sson for their thoughtful reviews of this manuscript. Literature cited Abraham, D., and A. Lyons. 2002. Novel physical interpretations of K-distributed re- verberation. IEEE J. Oce. Eng. 27(4):800-813. Abraham, D., and A. Lyons. 2004. Reverberation envelope statistics and their depen- dence on sonar bandwidth and scattering patch size. IEEE J. Ocean Eng. 29(1):126-137. APL (Applied Physics Laboratory). 1994. APL-UW High-frequency ocean environment acoustic models handbook, TR9407, IV1-IV36. APL, Univ. Washington, Seattle, WA. Brown, C., and P. Blondel. 2009. Developments in the application of multibeam so- nar backscatter for seafloor habitat mapping. Applied Acoustics 70:1242-1247. Cordue, P. 2006. A note on non-random error structure in trawl sur- vey abundance indices. ICES J. Mar. Sci. 64:1333-1337. Cutter, G., L. Berger, and D. Demer. 2010. A comparison of bathymetry mapped with the Simrad ME70 multibeam echosounder operated in bathymetric and fisheries modes. ICES J. Mar. Sci. 67(6):1301— 1309. Fonseca, L., and L. Mayer. 2007. Remote estimation of surficial seafloor properties through the application Angular Range Analysis to mul- tibeam sonar data. Mar. Geophys. Res. 28:119-126. Foote, K. G., H. P. Knudsen, G. Vestnes, D. N. MacLennan, and E. J. Simmonds. 1987. Calibration of acoustic instruments for fish densi- ty estimation: a practical guide. ICES Coop. Res. Rep. 144, 69 p. Goff, J., B. Kraft, L. Mayer, S. Schock, C. Sommerfield, H. Olsen, S. Gulick, and S. Nordfjord. 2004. Seabed characterization on the New Jersey middle and outer shelf: correlatability and spatial variability of seafloor sediment properties. Mar. Geol. 209:147-172. Weber et al: Seabed classification for trawlability determined with a multibeam echo sounder 77 Jackson, D., and M. Richardson. 2007. Chapter 16 in High-frequency seafloor acoustics. 616 p. Springer, New York. Jagielo, T., A. Hoffmann, J. Tagart, and M. Zimmermann. 2003. Demersal groundfish densities in trawlable and untrawlable habitats off Washington: implications for estimation of the trawl survey habitat bias. Fish. Bull. 101:545-565. Jin, G., and D. Tang. 1996. Uncertainties of differential phase estimation as- sociated with interferometric sonars. IEEE J. Ocean Eng. 21(1):53— 63. Jones, D. T., C. D. Wilson , A. De Robertis, C. N. Rooper, T. C. Weber, and J. L. Butler. 2012. Rockfish abundance assessment in untrawlable habitats: combining acoustics with complementary sam- pling tools. Fish. Bull. 110: 332-343. Kostylev V., B. Todd, G. Fader, R. Courtney, G. Cameron, and R. Pickrill. 2001. Benthic habitat mapping on the Scotian Shelf based on multibeam bathymetry, surficial geology and sea floor photographs. Mar. Ecol. Prog. Ser. 219:121-137. Lurton, X. 2010. An introduction to underwater acoustics: princi- ples and applications, 2nd ed., 760 p. Springer- Verlag, Berlin. Matthews, K. R., and L. J. Richards. 1991. Rockfish (Scorpaenidae) assemblages of trawlable and untrawlable habitats off Vancouver Island, British Columbia. N. Am. J. Fish. Manage. 11:312-318. Rooper, C.N., G. R. Hoff, and A. DeRobertis. 2010. Assessing habitat utilization and rockfish (Se- bastes spp.) biomass on an isolated rocky ridge using acoustics and stereo image analysis. Can. J. Fish. Aquat. Sci. 67:1658-1670. Rooper, C. N., M. H. Martin, J. L. Butler, D. T. Jones, and M. Zimmermann 2012. Estimating species and size composition of rock- fishes in acoustic surveys of untrawlable areas. Fish. Bull. 110: 317-331. Stauffer, G. 2004. NOAA protocols for groundfish bottom trawl sur- veys of the nation’s fishery resources. NOAA Tech. Mem., NMFS-F/SPO-65, 205 p. Available online at http://spo.nmfs.noaa.gov/tm/tm65.pdf Stein, D. L., B. N. Tissot, M. A. Hixon, and W. Barss. 1992. Fish-habitat associations on a deep reef at the edge of the Oregon continental shelf. Fish. Bull. 90:540-551. Trenkel, V., V. Mazauric, and L. Berger. 2008. The new fisheries multibeam echosounder ME70: description and expected contribution to fisheries re- search. ICES J. Mar. Sci. 65:645-655. Wakabayashi, K., R. G. Bakkala, and M. S. Alton. 1985. Methods of the U.S. -Japan demersal trawl surveys. In Results of cooperative U.S. -Japan groundfish investi- gations in the Bering Sea during May-August 1979 (R. G. Bakkala, and K. Wakabayashi, eds.), p. 7-29. Int. North Pac. Fish. Comm. Bull. 44. Williams, K., C. N. Rooper, and R. Towler. 2010. Use of stereo camera systems for assessment of rockfish abundance in untrawlable areas and for record- ing pollock behavior during midwater trawls. Fish. Bull. 108:352-362. Yoklavich, M. M., H. G. Greene, G. M. Cailliet, D. E. Sullivan, R. N. Lea, and M. S. Love. 2000. Habitat associations of deep-water rockfishes in a submarine canyon: an example of a natural ref- uge. Fish. Bull. 98:625-641. 78 Abstract — Jumbo squid (Dosidicus gigas) and purpleback squid ( Sthe - noteuthis oualaniensis) (Teuthida: Ommastrephidae) are thought to spawn in the eastern tropical Pa- cific. We used 10 years of plankton tow and oceanographic data collect- ed in this region to examine the re- productive habits of these 2 ecologi- cally important squid. Paralarvae of jumbo squid and purpleback squid were found in 781 of 1438 plankton samples from surface and oblique tows conducted by the Southwest Fisheries Science Center (NOAA) in the eastern tropical Pacific over the 8-year period of 1998-2006. Paralar- vae were far more abundant in sur- face tows (maximum: 1588 individu- als) than in oblique tows (maximum: 64 individuals). A generalized linear model analysis revealed sea-surface temperature as the strongest envi- ronmental predictor of paralarval presence in both surface and oblique tows; the likelihood of paralarval presence increases with increasing temperature. We used molecular techniques to identify paralarvae from 37 oblique tows to species level and found that, the purpleback squid was more abundant than the jumbo squid (81 versus 16 individuals). Manuscript submitted 18 April 2012. Manuscript accepted 27 November 2012. Fish. Bull. 111:78-89 (2013). doi:10.7755/FB. 11 1.1.7 The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessar- ily reflect the position of the National Marine Fisheries Service, NOAA. Distribution of ommastrephid paralarvae in the eastern tropical Pacific Danna J. Staaf (contact author)1 Jessica V. Redfern2 William F. Gilly 1 William Watson2 Lisa T. Ballance2 Email address for contact author: danna|oy@gmail 1 Hopkins Marine Station of Stanford University 120 Oceanview Blvd Pacific Grove, California 93950 2 Southwest Fisheries Science Center National Marine Fisheries Service, NOAA 8901 La Jolla Shores Dr. La Jolla. California 92037 Adult squid of the oceanic family Ommastrephidae are active general- ist predators and key prey for a wide variety of marine fishes, birds, and mammals. They are also the primary targets of the world’s larger squid fisheries (Nigmatullin et al., 2001; Markaida et ah, 2005; FAO, 2011). Many questions remain unanswered about the reproduction and early life history of these oceanic squid (Young et ah, 1985; Boletzky, 2003). Logisti- cal challenges impede direct obser- vation of reproduction and develop- ment in the wild, but the collection of paralarvae in net tows often can be used to elucidate ommastrephid spawning grounds and the habitat needs of early life stages (e.g., Oku- tani and McGowan, 1969; Zeidberg and Hamner, 2002). Two ommastrephid species that reproduce in the eastern Pacific are Dosidicus gigas, the jumbo or Hum- boldt squid, and Sthenoteuthis oual- aniensis, the purpleback squid (Vec- chione, 1999). The jumbo squid is currently the target of the world’s largest squid fishery (628,579 t in 2009 [FAO, 2011]), and commercial interest in purpleback squid is grow- ing (Zuyev et ah, 2002; Xinjun et ah, 2007). The adult ranges of these 2 species overlap in the eastern tropi- cal and subtropical Pacific (Roper et com ah, 1984), but the location and ex- tent of spawning grounds of either species over this large region are not well established. Paralarvae of these species cannot be reliably dis- tinguished morphologically; molecu- lar techniques must be used (Gilly et ah, 2006; Ramos-Castillejos et ah, 2010). When molecular identification is not possible because of formalin preservation or other limitations, paralarvae in this broad geographic region are generally assigned to the “SD complex” IS. oualaniensis and D. gigas [Vecchione, 1999]). Ommastrephid paralarvae are relatively rare off California (e.g., Okutani and McGowan, 1969; Wat- son and Manion, 2011), and none have been attributed to jumbo squid or purpleback squid. Both species, however, have been identified off the Pacific coast of the Baja California Peninsula (Hernandez-Rivas et ah1; Ramos-Castillejos et ah, 2010). With- 1 Hernandez-Rivas, M. E., R. De Silva- Davila, S. Camarillo-Coop, J. Grana- dores-Amores, and R. Durazo. 2007. Ommastrephid paralarvae during 1997- 1999 IMECOCAL cruises. Abstract in California Cooperative Oceanic Fisheries Investigations Annual Conference 2007, Program and Abstracts; San Diego, CA, 2-28 November, p. 41. Calif. Coop. Oceanic Fish. Invest., La Jolla, CA. Staaf et at: Distribution of ommastrephid paralarvae in the eastern tropical Pacific 79 in the Gulf of California only the jumbo squid has been reported to spawn (Gilly et ah, 2006; Staaf et ah, 2008; Camarillo-Coop et ah, 2011), and, to our knowledge, no other adult ommastrephid has been described from this region, although adults of purpleback squid have been reported from the area near the mouth of this gulf (Olson and Galvan-Magana, 2002). In the south- ern hemisphere, the Peru Current System has yielded only jumbo squid paralarvae (Sakai et ah, 2008). In the large intervening equatorial region, paralarvae of both purpleback squid and jumbo squid are present (Oku- tani, 1974; Ueynagi and Nonaka, 1993). The data that form the basis of this knowledge were collected through a variety of methods. Samples from both the Pacific and Gulf coasts of the Baja California Peninsula and from the Peru Current were collected primarily during subsurface oblique tows with bongo nets (Ramos-Castillejos et ah, 2010; Camarillo-Coop et ah, 2011; Sakai et ah, 2008). By contrast, the central region of the eastern tropical Pacific (ETP) has been sampled extensively during surface tows with neuston nets, yielding higher densities of paralarvae (Ueynagi and Nonaka, 1993; Vecchione, 1999). In the ETP, densi- ties can be extremely high, as in the case of more than 10,000 very small paralarvae of the SD complex from a single surface tow conducted during the 1986-87 El Nino (Vecchione, 1999). By contrast, the greatest num- ber of SD-complex paralarvae reported from the Baja California Peninsula is 20, collected with a bongo net (Camarillo-Coop et ah, 2011). Surface tows effectively sample only the top 10-20 cm of the water column, but subsurface oblique tows typically sample from the surface to depths of about 200 m. Because oblique tows sample a broader, deeper range of habitats than surface tows, discrepancies in paralarval abundance and size between the 2 types of tows may reflect different vertical habitat prefer- ences at different stages of development. For example, if recently hatched paralarvae exhibit a preference for surface waters, surface tows would be far more effec- tive at capturing these animals because oblique tows spend very little time at the surface (10-20 cm). And if paralarvae begin to occupy greater depths as they grow, while their numbers decrease because of natu- ral mortality, oblique tows would be likely to capture fewer, larger individuals than would surface tows, as has been seen for the ommastrephid Todarodes pacifi- cus (Yamamoto et ah, 2002; 2007). Although high surface abundances can be represen- tatively sampled by surface tows, any narrow subsur- face band of high abundance, as might occur at a pyc- nocline, would be undersampled by oblique tows. How- ever, a strong association of paralarvae with a subsur- face feature in preference to the surface could still be detected by a greater likelihood of capture in oblique rather than in surface tows, as has been found for the northern shortfin squid ( Illex illecebrosus ), which shows a relationship with the subsurface interface be- tween slope water and the Gulf Stream in the Atlantic (Vecchione, 1979; Vecchione et ah, 2001). Diel vertical migrations, typical of adult ommas- trephids, also could drive different abundances in sur- face and oblique tows. This result was found in loliginid paralarvae (Zeidberg and Hamner, 2002), but the situa- tion is less clear for ommastrephids (Piatkowski et ah, 1993; Young and Hirota, 1990). The few surface tows during which paralarvae of northern shortfin squid were collected in the Middle Atlantic Bight were con- ducted at night (Vecchione, 1979) — a finding that could indicate a nighttime migration to the surface, but the numbers are too small to strongly support this idea. No significant differences in paralarval abundance of pur- pleback squid have been found between daytime and nighttime tows in Hawaii (oblique and horizontal tows from the surface to a depth of 200 m; [Harman and Young, 1985]) or Japan (horizontal tows from the sur- face to a depth of 200 m; [Saito and Kubodera, 1993]). On cruises conducted by NOAA in the ETP, ecosys- tem data (including plankton samples) from a large geographic area have been collected regularly and ar- chived for many years. In this study, we present the first analysis of planktonic squid from this data set, focusing on the ommastrephids jumbo squid and pur- pleback squid. Our aims are 1) to compare surface and oblique tows conducted at the same location and time to determine differences in paralarval distribution and abundance due to sampling method, 2) to address ques- tions of species-specific depth preference and vertical migration, 3) to uncover relationships between paralar- val abundance and oceanographic features, and 4) to use molecular techniques on a subset of samples to de- termine whether the 2 species have distinct spawning areas or habitat preferences within their range overlap. Paralarval distribution is also contrasted with adult distribution data, collected during the 2006 cruise, to confirm that the study region is within the adult range of both species and to enhance our understanding of the ETP as a feeding and spawning area. Materials and methods Study area and data collection The ETP, where the ranges of jumbo squid and purple- back squid overlap, is defined by 3 large surface cur- rents and 2 water masses (Fiedler and Talley, 2006; Fig. 1A). The westward-flowing North and South Equa- torial Currents derive from the temperate California and Peru Currents, respectively. The Equatorial Coun- tercurrent flows eastward from the western Pacific to the coast of Central America. These currents define 2 water masses: Tropical Surface Water and Equato- rial Surface Water, the latter cooler and fresher than the former. Two smaller-scale oceanographic features are prominent: 1) a distinct thermocline ridge at the interface between the North Equatorial Current and 80 Fishery Bulletin 111(1) A B Figure 1 Map of study area in our examination of the distribution of ommastrephid paralarvae in the eastern tropical Pacific with (A) oceanography (after Fiedler and Talley, 2006) and (B) sampling stations from cetacean and eco- system assessment surveys conducted by the Southwest Fisheries Sci- ence Center (NOAA) from 1998 to 2006. Two plankton tows, one each with a manta and a bongo net, were conducted each evening approximately 2 h after sunset. STSW=Subtropical Surface Water. TSW=Tropical Surface Water. ESW=Equatorial Surface Water. the Equatorial Countercurrent, nominally along 10°N latitude (although the exact location varies season- ally) and 2) the Costa Rica Dome, an area of thermo- cline doming, nominally at 9°N latitude, 90°W longi- tude, although this feature too varies in location and degree of development through time, seasonally and interannually. The study area for this research forms a polygon that circumscribes the oceanic waters from the U.S. -Mexico border west to Hawaii, and south to central Peru. Cetacean and ecosystem assessment cruises were conducted in this region by the Southwest Fisheries Science Center (NOAA Fisheries) from late July to early December of 1998, 1999, 2000, 2003, and 2006 (Fig. IB), with the University-National Oceano- graphic Laboratory System (UNOLS) research vessel Endeavor (1998), and the NOAA Ships David Starr Jordan (all years), McArthur (1998, 1999, 2000), and McArthur II (2003, 2006). Plankton were sampled with 2 types of net tows, conducted ~2 h after sun- set each day, for a total of 979 manta (surface) tows and 762 bongo (oblique) tows over the 8-year period. On the McArthur II in 2006, during one leg of the cruise, medium-size jigs and rods were used to fish for adult squid from 1 to 2 h after sunset. Manta nets (Brown and Cheng, 1981) with 0.505-mm mesh were towed for 15 min at a ship speed of 1. 0-2.0 kn, with all deck lights off. Bongo nets (McGowan and Brown2; Smith and Richardson, 1977), consisting of a pair of circular net frames with 0.505-mm or 0.333-mm mesh, were towed for a 15-min double oblique haul to a depth of -200 m at a ship speed of 1. 5-2.0 kn. The net was lowered continuously at about 35 m/min, held at -200 m for 30 s, and then was retrieved at about 14 m/min, with the angle of stray al- ways maintained at -45°. Volume of water filtered during manta and bongo tows was estimated with a flowmeter suspended across the center of the net. Contents of the co- dends were preserved in 5% formalin buffered with sodium borate. In 2003 and 2006, the contents of one codend of each bongo tow were frozen in sea- water at -20° C, and the contents of the other were preserved in 5% forma- lin. Also in 2006, the contents of one codend of every fourth bongo tow (38 samples total) were preserved in 70% ethanol instead of formalin. 2 McGowan, J. A., and D. M. Brown. 1966. A new opening- closed paired zooplankton net. Univ. Calif. Scripps Inst. Oceanogr. Ref. 66-23, 56 p. Scripps. Inst. Oceanogr., Univ. Calif, San Diego, CA. Staaf et at: Distribution of ommastrephid paralarvae in the eastern tropical Pacific 81 Water column data were collected with conductivity- temperature-depth (CTD) profilers 1 h before sunrise and 1 h after sunset on each survey day and with ex- pendable bathythermographs (XBTs) during daylight hours at intervals of ~55 km. Samples of surface water were collected in bottles during the CTD casts and in buckets concurrent with XBT casts at depths from 1 to 3 m. Precruise calibration factors (fluorometer cali- bration factor, F, and acid ratio of pure chlorophyll, x) were used to calculate chlorophyll-a and phaeophytin values from digital fluorometer readings of these sur- face water samples. Sea-surface temperature (SST) and salinity (SSS) were measured continuously (around the clock) with a thermosalinograph while the ship was un- derway. Details of the complete data set are available in NOAA data reports (Philbrick et al., 2001, a-c; Am- brose et al., 2002, a and b; Watson et al., 2002; Jackson et al., 2004; 2008). Sample processing Cephalopods were removed manually from 654 bongo (1998, 2000, 2003, 2006) and 784 manta (1998, 1999, 2003, 2006) samples. Bongo samples with >25 mL of plankton were fractioned to -50% of the original sam- ple volume before they were sorted. The absolute count from each tow was divided by the volume of water filtered during that tow, as computed from flowmeter readings, to give paralarvae densities per cubic meter (following techniques described in Kramer et al., 1972). Adult and paralarva! specimens were identified by morphological characteristics (Wormuth et al., 1992). Adults were identified to species by the presence of a fused funnel-locking cartilage in purpleback squid and the absence of the fused structure in jumbo squid. Ommastrephid paralarvae are known as rhynchoteu- thions; their distinctive form is recognized easily by the presence of a proboscis. For individuals missing the proboscis or in which the proboscis already had separated into tentacles, identification was based on the characteristic inverted-T funnel-locking cartilage of this family. When proboscis suckers were visible, they were checked to separate individuals of the gen- era Hyaloteuthis, Eucleoteuthis, and Ommastrephes (enlarged, lateral suckers on proboscis) from individu- als of the genera Dosidicus and Sthenoteuthis (equal- size suckers on proboscis). Hyaloteuthis, Eucleoteuthis and Ommastrephes are relatively rare in the ETP (all the molecularly identified ommastrephids in this study were Sthenoteuthis or Dosidicus', see also Yatsu3), and 3 Yatsu, A. 1999. Morphological and distribution of rhyncho- teuthion paralarvae of two ommastrephid squids, Dosidicus gigas and Sthenoteuthis oualaniensis, collected from eastern tropical Pacific Ocean during 1997-preliminary report. In Report of the Kaiyo Maru cruise for study on the resources of two ommastrephid squids, Dosidicus gigas and Ommastrephes bartramii, in the Pacific Ocean, during September 11- December 24, 1997 (A. Yatsu, and C. Yamashiro, eds.), p. 193-206. Fisheries Agency of Japan, Tokyo. only 9 specimens were tentatively identified as Eucleo- teuthis and 2 specimens were tentatively identified as Ommastrephes by proboscis suckers and photophores (6 others were excluded from Dosidicus or Sthenoteuthis but were too small to be assignable to the other 3 gen- era). Therefore, any specimens damaged such that the terminal suckers were not preserved were assigned to the SD complex. The presence of paralarvae from other cephalopod families was recorded, but these specimens were not identified to genus or species, or counted. Morphological techniques for reliable differentiation between paralarvae of jumbo squid and purpleback squid are not available. Wormuth et al. (1992) and Yatsu3 used proboscis length and photophores as dis- tinguishing characters, but the muscular proboscis can extend and retract (Staaf et al., 2008), and reactions to fixatives have not been quantified. Additionally, there may be variability in ontogenetic timing of photophore formation (Gilly et al., 2006). Ramos-Castillejos et al. (2010) suggested several distinguishing indices that used morphometric ratios; however, samples in this study were prepared in different fixatives (ethanol for jumbo squid and formalin for purpleback squid) that can distort or shrink specimen proportions. The efficacy of indices for diagnoses of individual speci- mens of unknown species also were not tested. There- fore, we attempted no species-level identification of SD- complex specimens that were preserved in formalin. Molecular identification of SD-complex ommas- trephids from ethanol-preserved samples followed pro- tocols described in Gilly et al. (2006). Two frozen bongo samples were also sent to Hopkins Marine Station for sorting and molecular identification. The frozen sam- ples were selected on the basis of a high abundance of ommastrephid paralarvae in the matching codend, and they were sorted primarily to test whether it is possible to reliably identify paralarvae from a frozen plankton sample. Mantle lengths (ML) of ommastrephid paralarvae from 1998 manta and bongo tows were measured with an ocular micrometer. For tows with 10 or fewer om- mastrephids, all individuals were measured. For tows with more than 10 ommastrephids, 10 individuals were selected for measurement. Selection was arbitrary and aimed to be representative; e.g., the largest (or small- est) specimens were not always included. Data analysis and modeling We constructed a data set of ommastrephid paralarval abundance and 5 in situ oceanographic variables: SST, SSS, mixed-layer depth (MLD), temperature at thermo- cline (TT), and surface concentration of chlorophyll-a (CHL). MLD is defined as the depth at which tempera- ture is 0.5°C less than SST (Fiedler, 2010). TT is tem- perature at the depth of the thermocline as determined by the “maximum slope by difference” method (Fiedler, 2010). MLD, TT, and CHL values were collected from the station nearest the net tow; these data were used 82 Fishery Bulletin 11 1 (1) only if the station was located within 18.5 km (10 nau- tical miles) and was sampled within 12 h of the net tow. SST and SSS were averaged over a 2-h window centered on the time of the net tow. In total, 137 bongo and 164 manta samples were discarded according to these criteria, leaving 517 bongo and 620 manta sam- ples. Many of the discards (56 bongo and 57 manta) were collected aboard the McArthur in 2003, when the thermosalinograph malfunctioned. Three outlier points were also removed: an abnormally low value for each of CHL and SST, and an abnormally high value for MLD. Relationships between ommastrephid abundance and oceanographic variables were explored with gener- alized linear models in the R statistics package, vers. 2.1.1 (R Development Core Team, 2005). We used gen- eralized linear models because of their utility in model- ing relationships between cetaceans and oceanographic habitat (Redfern et ah, 2006) and between cephalopod paralarvae and oceanographic habitat off western Ibe- ria (Moreno et al., 2009). Typical of marine survey counts, our paralarval abundance data were overdis- persed, with a high proportion of zeros and a few very large samples. Therefore, we followed Aitchison (1955) and Pennington (1983) in performance of a 2-step anal- ysis, in which we separated the data into a binomial presence and absence data set (hereafter referred to as paralarval presence ) and an abundance data set that included only stations at which paralarvae were pres- ent (hereafter referred to as paralarval abundance). To analyze paralarval presence, we used a binomial distri- bution with a logit link; for paralarval abundance we used a lognormal distribution. We used an automated forward/backward stepwise approach based on Akaike’s information criterion (AIC) to select the variables for inclusion in the model. Results Abundance of paralarvae Paralarvae of the SD complex were found in 781 of the 1438 formalin-preserved plankton samples. By type of tow, 355 of 656 oblique bongo tows (54.28%) and 426 of 784 surface manta tows (54.34%) contained SD-com- plex paralarvae. The greatest abundance in a single manta tow was 1588 paralarvae versus 64 paralarvae in a single bongo tow. SD-complex paralarvae taken in bongo tows were distributed over a somewhat broader geographical area than were those paralarvae captured in manta tows (Fig. 2), but density of captured paralar- vae was typically at least an order of magnitude great- er in manta tows. Size of paralarvae Average mantle length in manta tows was 1.94 ±1.29 mm fn=779; range 0.7-15 mm ML) versus 1.86 ±1.0 mm (ra = 148; range 0.6—7 mm ML) in bongo tows. No significant difference was found between these distri- butions (1-way analysis of variance [ANOVA], P= 0.44). Relationship of presence and abundance of paralarvae to environmental variables and modeling The stepwise approach for the presence models select- ed SST, SSS, and TT as predictor variables for manta data, and SST and MLD for bongo data (Table 1). The decrease in the AIC values for these models and the increase in the percentage of explained deviance came primarily from SST for both bongo and manta tows, with minimal contribution from MLD, SSS, and TT. Therefore, SST emerged as the strongest predictor for presence of SD-complex paralarvae, and the probability of capture increased monotonically as SST increased from 15°C to 32°C (Fig. 3). Analysis of paralarval abundance, rather than pres- ence, revealed no strong predictors (Table 2). For bongo tows, the stepwise approach selected CHL, TT, and SST in the final model (7.5% explained deviance). For manta tows, CHL, SST, MLD, and TT were all selected (12.1% explained deviance). There appears to be little relation- ship between these variables and nonzero paralarval abundance, which varied over a wide range of each en- vironmental variable for both manta and bongo tows. Species identification In total, 97 SD-complex paralarvae were found in 12 of the 38 ethanol-preserved samples. Of these paralar- vae, 81 were identified genetically as Sthenoteuthis oualaniensis and 16 as Dosidicus gigas. Paralarvae of purpleback squid were found over a much greater area than paralarvae of jumbo squid (Fig. 4A). Eight om- mastrephid paralarvae were removed from the 2 frozen samples and identified genetically as purpleback squid. Non-ommastrephid cephalopods were identified in many of the tows, most commonly as taxa in the teu- thid families Enoploteuthidae, Onychoteuthidae, Gona- tidae, Chtenopterygidae, Cranchiidae, and Brachioteu- thidae and in the octopod genera Argonauta and Tre- moctopus; all have previously been reported from the ETP (Ueyanagi and Nonaka, 1993; Vecchione, 1999). Of the 129 adult squid captured in jigging sessions, 118 were jumbo squid and 11 were purpleback squid. Jumbo squid adults were found primarily in the south- ernmost sampling sites off Peru, but the few purple- back squid adults were more evenly distributed (Fig. 4B). Discussion This study represents the most extensive sampling to date in the ETP of paralarvae of jumbo squid and pur- pleback squid, covering most of their broad equatorial and subtropical region of range overlap in the Pacific during a period of 8 years. Staaf et a!. : Distribution of ommastrephid paralarvae in the eastern tropical Pacific 83 A B Paralarvae/m3 I 1 0—0. 1 6 I 1 0.16-0.25 | 1 0.25—0.3 [ I 0.3-0. 4 Dr: 0.4-0.55 r "I 0.55 -0.82 I 0 82-1 3 SI 1.3-2. 1 20 N Figure 2 Abundance of paralarval purpleback squid ( Sthenoteuthis oualaniensis) and jumbo squid ( Dosidicus gigas) from ail study years (1998-2006) for (A) manta (surface) and (B) bongo (oblique) tows conducted in the eastern tropical Pa- cific. Paralarval abundance was interpolated by using inverse distance weighting with a cell size of 1° and a fixed search radius of 5°. Table 1 Generalized linear models used to relate the presence and absence of ommastrephid paralarvae in manta (surface) and bongo (oblique) tows conducted in the eastern tropical Pacific in 1998-2006 to 5 in situ oceano- graphic variables: sea-surface temperature (SST), sea-surface salinity (SSS), mixed-layer depth (MLD), tem- perature at thermocline (TT), and surface-concentration of chlorophyll-a (CHL). A stepwise approach selected SST, SSS, and TT for the final manta model; SST and MLD were selected for the final bongo model. Better- fitting models have a higher percentage of explained deviance and a lower Akaike’s information criterion (AIC) value. Manta Model Deviance (%) AIC Null 913 SST x SSS x TT 18.8 748.1 SST x TT 18.5 748.5 SST x SSS 18.4 749.6 SSS x TT 12.7 801.6 SST 18.1 750.2 SSS 4.6 873.1 TT 11 814.9 Bongo Model Deviance (%) AIC Null 710.3 SST x MLD 12.2 627.7 SST 10.6 637.1 MLD 0.2 711.1 84 Fishery Bulletin 1 1 1 (1) B 0) o c 10,000 in the 4 Palomares-Garci'a, R., R. De Silva-Davila, and R. Avendano- Ibarra. 2007. Predation of the copepod Oncaea mediter- ranea upon ommastrephid paralarvae in the mouth of the Gulf of California. Abstract in Proceedings of the 1st inter- national CLIOTOP symposium; La Paz, Mexico, 3-7 December. Table 2 Generalized linear models used to relate nonzero abundance of ommastrephid paralarvae in manta (surface) and bongo (oblique) tows conducted in the eastern tropical Pacific in 1998-2006 to 5 in situ oceanographic variables: sea-surface temperature (SST), sea-surface salinity (SSS), mixed-layer depth (MLD), temperature at thermocline (TT), and surface-concentration of chlorophyll-a (CHL). A stepwise approach selected SST, MLD, TT, and CHL for the final manta model and SST, MLD, and CHL for the final bongo model. The re- sulting percentage of explained deviance and the Akaike’s information criteria (AIC) value for these models indicate that none of the oceanographic variables is a strong predictor of nonzero abundance. Manta Bongo Model Deviance (%) AIC Model Deviance (%) AIC Null 1341 Null 782 SST x MLD xTTx CHL 12.1 1303.8 SST x MLD x CHL 7.5 764.6 MLD x TT x CHL 11.3 1305 SST x CHL 6.9 764.8 SST x MLD x TT 11.1 1306 MLD x CHL 6.6 765.6 SST x TT x CHL 9.8 1310.9 SST x MLD 4.5 772.2 SST x MLD x CHL 9.5 1312.2 Vertical distribution of paralarvae We found no difference in the size of paralarvae be- tween surface (manta) and oblique (bongo) tows, in agreement with Yatsu.3 These observations are not consistent with an ontogenetic vertical migration to in- creasing depths within the paralarval stage of develop- ment, as proposed for Todarodes pacificus (Yamamoto et al., 2002; 2007). This feature, therefore, may not be common to all ommastrephids. Although incidence of capture in surface and oblique tows was nearly identical (54% positive samples in Staaf et al Distribution of ommastrephid paralarvae in the eastern tropical Pacific 85 Figure 4 Geographic distribution and abundance of (A) genetically identified paralarvae and (B) morphologically identified adult ommastrephids caught in the eastern tropical Pacific during surveys conducted in 2006. The numbers at each station (small dot) represent the total number of individuals of purpleback squid ( Sthenoteuthis oualaniensis) (outlined in black) and jumbo squid ( Dosidicus gigas) (solid black) captured at that station. At stations where numbers do not appear, no squid were caught. ETP (Vecchione, 1999). The consistency of this result seems surprising, because ommastrephid egg masses are thought to occur near the pycnocline, typically tens of meters deep, and not at the surface (O’Dor and Balch, 1985). The only reported observation of an in situ egg mass of jumbo squid was in the Gulf of Cali- fornia at a depth of 16 m near the pycnocline (Staaf et al., 2008). Presumably, this characteristic is common to purpleback squid, but we are unaware of descriptions of natural egg masses for this species. Not only are egg masses of jumbo squid found at depth, but paralarvae are negatively buoyant. Paralar- vae in the laboratory can swim to the surface but sink as soon as they stop swimming (Staaf et al., 2008); this negative buoyancy indicates that surface tension is in- sufficient for passive retention. We can only assume that purpleback squid paralarvae share this trait, and that tissue density of wild paralaravae is similar to laboratory-reared animals. A preferred surface habitat, in which maintenance of position requires significant energy expenditure, strongly indicates that some benefit is derived from this behavior; the benefit may be access to increased food quantity or to food of higher nutritional value (Yamamoto et al., 2007). Nothing is known of the diet of jumbo squid paralarvae, but amphipods, copepods, and crab zoeae have been found in the digestive tracts of purpleback squid paralarvae (Vecchione, 1991); these and other zooplankton, as well as phytoplank- ton, also have been found in paralarvae of another ommastrephid, lllex argentinus (Vidal and Haimovici, 1998). Furthermore, a case has been made for the use of dissolved and particulate organic material by om- mastrephid paralarvae (O’Dor et al., 1985). At certain times and in certain regions, oceanic surface waters may have high concentrations of these foods. The depth of the chlorophyll-# maximum in the ETP ranges from 60 to 90 m in open-ocean regions to near the surface in coastal boundary regions (Pennington et al., 2006). It would be valuable to examine the vertical distri- bution of paralarvae with systematic oblique or hori- zontal tows at a series of discrete depths through the upper 100-200 m of the water column at a variety of times in a given area. This approach would give a more accurate picture of habitat use and of any association with the subsurface chlorophyll-a maximum or acoustic 86 Fishery Bulletin 1 1 1 (1) scattering layers. To our knowledge, such a dedicated effort to address this problem has not been reported. Oceanography The number of both bongo- and manta-net tows that contained paralarvae increased as SST increased from 15°C to 32°C (Table 1, Fig. 3). This increased paralarval occurrence is consistent with the literature. Paralar- vae of purpleback squid exhibit a preference for warm temperatures (28-31°C) in waters off Japan (Saito and Kubodera, 1993), and extremely large numbers of SD- complex paralarvae in the ETP were captured in in- dividual tows coincident with the 29°C SST isotherm (Vecchione, 1999). In the Gulf of California, paralarvae of jumbo squid are more abundant during the warm months of June and September (SST of 27.7-29.4°C) than during the cooler season of February and April (SST of 15.3-18. 1°C) (Camarillo-Coop et ah, 2011). Be- cause our surveys were conducted only between late July and early December, we were unable to assess seasonal variability in paralarval distribution. We found no evidence for a decrease in paralarval occurrence at the highest SST values, despite the fact that embryonic development in vitro is optimal in the range of 17-25°C and fails to proceed at 30°C (Staaf et ah, 2011). The idea that paralarvae may be better able than developing embryos to withstand warmer tem- peratures would be consistent with a upward vertical migration after hatching. If hatchlings promptly swim from near the pycnocline up to warmer near-surface water, where food may be more readily available, an ontogenetic increase in temperature optima would be advantageous. It also is possible that the upper ther- mal limit for successful development of wild embryos could be higher than the limit observed in laboratory studies. Embryos studied in the laboratory, particularly those embryos obtained through in vitro fertilization, may perish at high temperatures because of microbial infection, which could be inhibited in the wild by the presence of natural egg jelly (Staaf et ah, 2011). Peak abundances of SD-complex paralarvae ob- served in our study were an order of magnitude lower than the abundance levels reported during the 1986-87 El Nino (Vecchione, 1999). This discrepancy could be due to chance in sampling or a real difference in abun- dance. Among our study years, only in 2006 was an El Nino observed, and it was weaker than the one in 1986-87. The other years of our sampling were either in La Nina (1998, 1999, 2000) or neutral (2003) con- ditions (http://ggweather.com/enso/oni.htm). Year was included in our models as a potential explanatory dis- crete variable, but it was determined not to be an infor- mative predictor of paralarval abundance or presence, indicating no difference between El Nino, La Nina, and neutral years. However, the strong positive relation- ship between paralarval occurrence and temperature found in our study is consistent with Vecchione’s (1999) hypothesis that the extraordinarily high paralarval abundances in 1987 were related to the 3.5°C increase in SST during El Nino. Reduced upwelling during the 1986-87 El Nino led to a 50% decline in chlorophyll-a in the region of high- est paralarval abundance (Vecchione, 1999). Similarly, in our study, ommastrephid paralarvae were not as- sociated with upwelling zones or their resultant high primary productivity. In general, zooplankton biomass in the ETP tends to be greatest in the 4 major up- welling regions — the Gulf of Tehuantepec, Costa Rica Dome, Equatorial Cold Tongue, and coast of Peru (Fer- nandez-Alamo and Farber-Lorda, 2006) — but ommas- trephid paralarvae were not especially abundant in any of these regions (Fig. 2). Indeed, we found no rela- tionship between SD-complex paralarvae and primary productivity, as measured by CHL or MLD (where the thermocline is shallow, primary productivity tends to be higher [Pennington et ah, 2006]). Species-specific spawning area Molecularly identified jumbo squid paralarvae have been reported from the Gulf of California (Gilly et ah, 2006), off the Baja California Peninsula (Ramos- Castillejos et ah, 2010), off Peru (Wakayabashi et ah, 2008), and now, in this study, from the ETP. We found that most molecularly identified paralarvae from the ETP were purpleback squid (Fig. 4A), but most adult squid captured by jigging were jumbo squid (Fig. 4B). Although jigging capture rates may have been biased, adult jumbo squid have also been found to outnum- ber purpleback squid as prey items of the Dolphinfish ( Coryphaena hippurus) in the ETP (Olson and Galvan- Magana, 2002). Despite this abundance of adult jum- bo squid, we found jumbo squid paralarvae in only 2 samples, and these samples also contained paralarval purpleback squid in appreciable numbers (Fig. 4A). Neither the geographic locations nor oceanographic features of these 2 sampling sites were distinct from sites where only purpleback squid was found. There- fore, we can say only that purpleback squid paralarvae appear to be far more abundant than paralarvae of jumbo squid because we have no way of assessing bias in the capture rates of the 2 species during plankton tows. Species-level molecular identification of paralarvae was possible in this study only with material from oblique tows. If future work on material from surface tows were to find a similar predominance of purple- back squid, it would support the hypothesis that the purpleback squid is the primary ommastrephid that spawns in the ETR Although jumbo squid can spawn in the ETP or subtropical fringes, its primary spawning grounds may actually lie farther to the north, off the Baja California Peninsula in both the Pacific (Ramos- Catellejos et ah, 2010) and Gulf of California (Staaf et ah, 2008; Camarillo-Coop et ah, 2011), and farther to the south off Peru (Tafur et ah, 2001; Sakai et ah, 2008; Anderson and Rodhouse, 2001). Staaf et al Distribution of ommastrephid paralarvae in the eastern tropical Pacific 87 This view clearly contrasts with the one originally proposed by Nesis (1983) in which the jumbo squid spawns in the ETP and then migrates to feed at higher latitudes in both hemispheres. Available genetic analy- sis instead indicates 2 separate breeding populations, 1 in the northern hemisphere and 1 in the southern hemisphere (Staaf et al., 2010). If the preferred spawn- ing habitat of jumbo squid is indeed subtropical to tem- perate, rather than tropical, it could explain the divi- sion into 2 populations, 1 breeding off Mexico and 1 breeding off Peru. For future collections, we recommend preservation of material from both oblique and surface tows in ethanol. Although we were able to extract and identify paralar- vae from frozen plankton samples, the technique has 2 drawbacks: 1) the difficulty of visual identification of individual specimens in the thawed slurry and 2), if the samples are to be sorted in more than one ses- sion, the damage done to the entire sample by repeated freeze-thaw cycles. Conclusions We found paralarvae in surface and oblique tows to be of equal size, indicating that paralarvae of the 2 om- mastrephid species jumbo squid and purpleback squid do not engage in ontogenetic vertical migration at the paralarval stage. Ommastrephid paralarvae were much more abundant in surface tows than in oblique tows; this finding may indicate an ecological advantage of surface waters — perhaps, related to feeding. Models selected SST as the strongest predictor of paralarval presence in both surface and oblique tows; presence was more likely at higher temperatures. Therefore, warm surface waters appear to be the preferred habi- tat of ommastrephid paralarvae in the ETP. Molecu- lar identification of specimens from a small subset of oblique tows showed that paralarvae of purpleback squid far outnumbered those of jumbo squid in this region. Adults of purpleback squid are broadly dis- tributed in the tropics, whereas adult jumbo squid are abundant in tropical, subtropical, and temperate wa- ters and occasionally present in boreal waters. Results from this study are consistent with the possibility that the purpleback squid spawns primarily in the tropics, and the jumbo squid spawns preferentially in subtropi- cal or, perhaps, even temperate regions. Acknowledgments To the cruise coordinators, the net-towing oceanogra- phers, the plankton-sorting students and contractors, and the commanding officers and crew of the research vessels, we offer our boundless gratitude. We also thank P. Fiedler and staff at the Southwest Fisheries Science Center for processing oceanographic data, M. Ohman for providing ethanol-preserved samples and advice, A. Townsend for oversight of sample processing, L. Loren- zo for sample sorting, C. Elliger and Z. Lebaric for DNA sequencing, G. Watters for project guidance. We are also grateful for support from the Nancy Foster Schol- arship Program of NOAA (to DJS) and the National Science Foundation (OCE0526640 and OCE0850839 to WFG). Literature cited Aitchison, J. 1955. On the distribution of a positive random variable having a discrete probability mass at the origin. J. Am. Stat. Assoc. 50:901-908. Ambrose, D. A., R. L. Charter, H. G. Moser, S. R. Charter, and W. Watson. 2002a. Ichthyoplankton and station data for surface (manta) and oblique (bongo) plankton tows taken dur- ing a survey in the eastern tropical Pacific Ocean July 30-December 9, 1998. NOAA Tech. Memo. NOAA-TM- NMFS-SWFSC-337, 126 p. Ambrose, D. A., R. L. Charter, H. G. Moser, B. S. MacCall, and W. Watson. 2002b. Ichthyoplankton and station data for surface (manta) and oblique (bongo) plankton tows taken dur- ing a survey in the eastern tropical Pacific Ocean July 28-December 9, 2000. NOAA Tech. Memo. NOAA-TM- NMFS-SWFSC-342, 130 p. Anderson, C., and P. G. Rodhouse. 2001. Life cycles, oceanography and variability: ommas- trephid squid in variable oceanographic environments. Fish. Res. 54:133-143. Boletzky, S. v. 2003. Biology of early life stages in cephalopod mol- luscs. Adv. Mar. Biol. 44:143-293. Brown, D. M., and L. Cheng. 1981. New net for sampling the ocean surface. Mar. Ecol. Prog. Ser. 5:225-227. Camarillo-Coop, S., C. Salinas-Zavala, M. Manzano-Sarabia, and E. A. Aragon-Noriego. 2011. Presence of Dosidicus gigas paralarvae (Cepha- lopoda: Ommastrephidae) in the central Gulf of Cali- fornia, Mexico related to oceanographic conditions. J. Mar. Biol. Assoc. U.K. 91:807-814. FAO (Food and Agriculture Organization of the United Nations). 2011. FAO yearbook. Fishery and aquaculture statistics: 2009, 78 p. FAO, Rome. Fernandez-Alamo, M. A., and J. Farber-Lorda. 2006. Zooplankton and the oceanography of the eastern tropical Pacific: a review. Prog. Oceanogr. 69:318-359. Fiedler, P. C. 2010. Comparison of objective descriptions of the ther- mocline. Limnol. Oceanogr. Methods 8:313-325. Fiedler, P. C., and L. D. Talley. 2006. Hydrography of the eastern tropical Pacific: a re- view. Prog. Oceanogr. 69:143-180. Gilly, W. F., C. A. Elliger, C. A. Salinas, S. Camarillo-Coop, G. Bazzino, and M. Beman. 2006. Spawning by jumbo squid ( Dosidicus gigas ) in the Pedro Martir Basin, Gulf of California, Mexico. Mar. Ecol. Prog. Ser. 313:125-133. 88 Fishery Bulletin 111(1) Harman, R. F., and R. E. Young. 1985. The larvae of ommastrephid squids (Cephalop- oda, Teuthoidea) from Hawaiian waters. Vie Milieu 35:211-222. Jackson, A., T. Gerrodette, S. Chivers, M. Lynn, P. Olson, and S. Rankin. 2004. Marine mammal data collected during a survey in the eastern tropical Pacific aboard the NOAA Ships McArthur II and David Starr Jordan , July 29-Decem- ber 10, 2003. NOAA Tech. Memo. NOAA-TM-NMFS- SWFSC-366, 98 p. Jackson, A., T. Gerrodette, S. Chivers, M. Lynn, S. Rankin, and S. Mesnick. 2008. Marine mammal data collected during a survey in the eastern tropical Pacific aboard NOAA Ships Da- vid Starr Jordan and McArthur 11, July 28-December 7, 2006. NOAA Tech. Memo. NOAA-TM-NMFS-SWF- SC-421, 45 p. Kramer, D., M. Kalin, E. G. Stevens, J. R. Thrailkill, and J. R. Zweifel. 1972. Collecting and processing data on fish eggs and larvae in the California Current Region. NOAA Tech. Rep. NMFS Circ. 370, 38 p. Markaida, U., J. J. Rosenthal, and W. F. Gilly. 2005. Tagging studies on the jumbo squid ( Dosidicus gigas ) in the Gulf of California, Mexico. Fish. Bull. 103:219-226. Moreno, A., A. Dos Santos, U. Piatkowski, A. M. P Pantos, and H. Cabral. 2009. Distribution of cephalopod paralarvae in relation to the regional oceanography of the western Iberia. J. Plankton Res. 31:73-91. Nesis, K. N. 1983. Dosidicus gigas. In Cephalopod life cycles, vol. 1: species accounts (P. R. Boyle, ed.), p. 216-231. Academ- ic Press, London. Nigmatullin, C. M., K. N. Nesis, and A. I. Arkhipkin. 2001. Biology of the jumbo squid Dosidicus gigas (Ceph- alopoda: Ommastrephidae). Fish. Res. 54:9-19. O'Dor, R. K., and N. Balch. 1985. Properties of lllex illecebrosus egg masses poten- tially influencing larval oceanographic distribution. Sci. Counc. Stud. NAFO 9:69-76. O’Dor, R. K., P. Helm, and N. Balch. 1985. Can rhynchoteuthions suspension feed? (Mollusca: Cephalopoda). Vie Milieu 35:267-271. Okutani, T. 1974. Epipelagic decapod cephalopods collected by mi- cronekton tows during the EASTROPAC expeditions, 1967-1968 (systematic part). Bull. Tokai Reg. Fish. Res. Lab. 80:29-118. Okutani, T., and J. A. McGowan. 1969. Systematics, distribution, and abundance of the epiplanktonic squid (Cephalopoda, Decapoda) larvae of the California Current, April 1954-March 1957. Bull. Scripps Inst. Oceanogr., vol. 14, 90 p. Univ. California Press, Berkeley, CA. Olson, R. J., and F. Galvan-Magana. 2002. Food habits and consumption rates of common dol- phinfish (Coryphaena hippurus) in the eastern Pacific Ocean. Fish. Bull. 100:279-298. Pennington, M. 1983. Efficient estimators of abundance, for fish and plankton surveys. Biometrics 39:281-286. Pennington, J. T., K. L. Mahoney, V. S. Kuwahara, D. D. Kolber, D. Calienes, and F. P. Chaves. 2006. Primary production in the eastern tropical Pacific: a review. Prog. Oceanogr. 69:285-317. Philbrick, V. A., P. C. Fiedler, J. T. Fluty, and S. B. Reilly. 2001a. Report of oceanographic studies conducted dur- ing the 1998 eastern tropical Pacific Ocean survey on the research vessels David Starr Jordan, McArthur, and Endeavor. ^JOAA Tech. Memo. NOAA-TM-NMFS- SWFSC-307, 36 p. 2001b. Report of oceanographic studies conducted dur- ing the 1999 eastern tropical Pacific Ocean survey on the research vessels David Starr Jordan and McArthur. NOAA Tech. Memo. NOAA-TM-NMFS-SWFSC-308, 29 P- 2001c. Report of oceanographic studies conducted dur- ing the 2000 eastern tropical Pacific Ocean survey on the research vessels David Starr Jordan and McArthur. NOAA Tech. Memo. NOAA-TM-NMFS-SWFSC-309, 29 P- Piatkowski, U., W. Welsch, and A. Ropke. 1993. Distribution patterns of the early life stages of pelagic cephalopods in three geographically different regions of the Arabian Sea. In Recent advances in cephalopod fisheries biology (T. Okutani, R. K. O’Dor, and T. Kubodera, eds.), p. 417-431. Tokai Univ. Press, Tokyo. R Development Core Team. 2005. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vi- enna, Austria. [Available from http://www.r-project.org/, accessed July, 2009.1 Ramos-Castillejos, J.E., C. A. Salinas-Zavala, S. Camarillo- Coop, and L. M. Enriquez-Paredes. 2010. Paralarvae of the jumbo squid, Dosidicus gigas. Invertebr. Biol. 129:172-183. Redfern, J. V., M. C. Ferguson, E. A. Becker, K. D. Hyrenbach, C. Good, J. Barlow, K. Kaschner, M. F. Baumgartner, K. A. Forney, L. T. Ballance, P. Fauchald, P. Halpin, T. Hamazaki, A. J. Pershing, S. S. Qian, A. Read, S. B. Reilly, L. Torres, and F. Werner. 2006. Techniques for cetacean-habitat modeling. Mar. Ecol. Prog. Ser. 310:271-295. Roper, C. F. E., Sweeney, M. J., and C. E. Nauen. 1984. Cephalopods of the world: an annotated and illus- trated catalogue of species of interest to fisheries. FAO Species Catalogue, vol. 3. FAO Fish. Synop. 125, 277 p. FAO, Rome. Saito, H., and T. Kubodera. 1993. Distribution of ommastrephid rhynchoteuthion paralarvae (Mollusca, Cephalopoda) in the Kuroshio Region. In Recent advances in cephalopod fisheries bi- ology (T. Okutani, R.K. O’Dor, and T. Kubodera, eds.), p. 457-466. Tokai Univ. Press, Tokyo. Sakai, M., L. Mariategui, T. Wakabayashi, C. Yamashiro, and K. Tuchiya. 2008. Distribution and abundance of jumbo flying squid paralarvae (Dosidicus gigas ) off Peru and in waters west of the Costa Rica Dome during the 2007 La Nina, p 95-97. 4th international symposium on Pacific squids (E. Acuna, L. Cubillos, and C. Ibanez, (eds.), Coquimbo, Chile. Staaf et al. : Distribution of ommastrephid paralarvae in the eastern tropical Pacific 89 Smith, P. E., and S. L. Richardson. 1977. Standard techniques for pelagic fish egg and larva surveys. FAO Fish. Tech. Pap. No. 175, 100 p. FAO, Rome. Staaf, D. J., S. Camarillo-Coop, S. H. D. Haddock, A. C. Nyack, J. Payne, C. A. Salinas-Zavala, B. A. Seibel, L. Trueblood, C. Widmer, and W. F. Gilly. 2008. Natural egg mass deposition by the Humboldt squid ( Dosidicus gigas) in the Gulf of California and characteristics of hatchlings and paralarvae. J. Mar. Biol. Assoc. U.K. 88:759-770. Staaf, D. J., R. I. Ruiz-Cooley, C. Elliger, Z. Lebaric, B. Campos, U. Markaida, and W. F. Gilly. 2010. Ommastrephid squids Sthenoteuthis oualaniensis and Dosidicus gigas in the eastern Pacific show conver- gent biogeographic breaks but contrasting population structures. Mar. Ecol. Prog. Ser. 418:165-178. Staaf, D. J, L. D. Zeidberg, and W. F. Gilly. 2011. Effects of temperature on embryonic development of the Humboldt squid Dosidicus gigas. Mar. Ecol. Prog. Ser. 441:165-175. Tafur, R., P. Villegas, M. Rabf, and C. Yamashiro. 2001. Dynamics of maturation, seasonality of reproduc- tion and spawning grounds of the jumbo squid Dosidi- cus gigas (Cephalopoda: Ommastrephidae) in Peruvian waters. Fish. Res. 54:33-50. Ueynagi, S., and H. Nonaka. 1993. Distribution of ommastrephid paralarvae in the central eastern Pacific Ocean. In Recent advances in cephalopod fisheries biology (T. Okutani, R.K. O’Dor, and T. Kubodera, eds.), p. 587-589. Tokai Univ, Press, Tokyo. Vecchione, M. 1979. Larval development of Illex (Steenstrup) in the northwestern Atlantic with comments on Illex larval distribution. Proc. Biol. Soc. Wash. 91:1060-1075. 1991. A method for examining the structure and con- tents of the digestive tract in paralarval squids. Bull. Mar. Sci. 49:300-308. 1999. Extraordinary abundance of squid paralarvae in the tropical eastern Pacific Ocean during El Nino of 1987. Fish. Bull. 97:1025-1030. Vecchione, M., C. F. E. Roper, M. J. Sweeney, and C. C. Lu. 2001. Distribution, relative abundance, and developmen- tal morphology of paralarval cephalopods in the West- ern North Atlantic Ocean. NOAA Tech. Rep. NMFS 152. Vidal, E. A. G., and M. Haimovici. 1998. Feeding and the possible role of the proboscis and mucus cover in the ingestion of microorganisms by rhynchoteuthion paralarvae (Cephalopoda: Ommas- trephidae). Bull. Mar. Sci. 63:305-316. Wakabayashi, T., T. Yanagimoto, M. Sakai, T. Ichii, and T. Kobayashi. 2008. Identification of Dosidicus gigas and Sthenoteuthis oualaniensis paralarvae using SSPPCR analysis on- board a ship., p. 111-112. 4th international symposium on Pacific squids (E. Acuna, L. Cubillos, and C. Ibanez, (eds), Coquimbo, Chile. Watson, W., and S. Manion. 2011. Ichthyoplankton, paralarval cephalopod, and sta- tion data for surface (manta) and oblique (bongo) plank- ton tows for California Cooperative Oceanic Fisheries Investigations Survey and California Current Ecosys- tem Survey cruises in 2008. NOAA Tech. Memo. NO- AA-TM-NMFS-SWFSC-481, 173 p. Watson, W., E. M. Sandknop, S. R. Charter, D. A. Ambrose, R. L. Charter, and H. G. Moser. 2002. Ichthyoplankton and station data for surface (manta) and oblique (bongo) plankton tows taken dur- ing a survey in the eastern tropical Pacific Ocean July 28-December 9, 1999. NOAA Tech. Memo. NOAA-TM- NMFS-SWFSC-338, 96 p. Wormuth, J. H., R. K. O’Dor, N. Balch, M. C. Dunning, E. C. Forch, R. F. Harman, and T. W. Rowell. 1992. Family Ommastrephidae Steenstrup, 1857. In “Larval” and juvenile cephalopods: a manual for their identification (M. J. Sweeney, C. F. E. Roper, K. M. Man- gold, M. R. Clarke, and S. V. Boletzky, eds.), p. 105-119. Smithson. Contrib. Zool. 513. Xinjun, C., L. Bilin, T. Siquan, Q. Weiguo, and Z. Xiaohu. 2007. Fishery biology of purpleback squid, Sthenoteuthis oualaniensis, in the northwest Indian Ocean. Fish. Res. 83:98-104. Yamamoto, J., S. Masuda, K. Miyashita, R. Uji, and Y. Sakurai. 2002. Investigation on the early stages of the ommas- trephid squid Todarodes pacificus near the Oki Islands (Sea of Japan). Bull. Mar. Sci. 71:987-992. Yamamoto, J., T. Shimura, R. Uji, S. Masuda, W. Watanabe, and Y. Sakurai. 2007. Vertical distribution of Todarodes pacificus (Ceph- alopoda: Ommastrephidae) paralarvae near the Oki Is- lands, southwestern Sea of Japan. Mar. Biol. 153:7-13. Young, R. E., and J. Hirota. 1990. Description of Ommastrephes bartramii (Cepha- lopoda: Ommastrephidae) paralarvae with evidence for spawning in Hawaiian waters. Pac. Sci. 44:71-80. Young, R. E., R. F. Harman, and K. M. Mangold. 1985. The common occurrence of oegopsid squid eggs in near-surface oceanic waters. Pac. Sci. 39:359-36. Zeidberg, L. D., and W. M. Hamner. 2002. Distribution of squid paralarvae, Loligo opalescens (Cephalopoda: Myopsida), in the southern California Bight in the three years following the 1997-1998 El Nino. Mar. Biol. 141:111-122. Zuyev, G., C. Nigmatullin, M. Chesalin, and K. Nesis. 2002. Main results of long-term worldwide studies on tropical nektonic oceanic squid genus Sthenoteuthis : an overview of the Soviet investigations. Bull. Mar. Sci. 71:1019-1060. 90 Staging ovaries of Haddock (Melanogrammus aeglefinus ): implications for maturity indices and field sampling practices Email address for contact author: katie burchard@noaa.gov Department of Natural Resources Conservation University of Massachusetts-Amherst Amherst, Massachusetts 01003 Present address: Narragansett Laboratory Northeast Fisheries Science Center National Marine Fisheries Service, NOAA 28 Tarzwell Drive Narragansett, Rhode Island 02882 Department of Biology University of Victoria Victoria, BC, Canada V8W 3N5 Marine Ecology and Technology Applications, Inc 23 Joshua Lane Waquoit, Massachusetts 02536 4 Marine Resources Research Institute South Carolina Department of Natural Resources 217 Ft Johnson Rd Charleston, South Carolina 29412 Abstract — We build on recent efforts to standardize maturation staging methods through the development of a field-proof macroscopic ovarian maturity index for Haddock (Me- lanogrammus aeglefinus) for stud- ies on diel spawning periodicity. A comparison of field and histological observations helped us to improve the field index and methods, and provided useful insight into the re- productive biology of Haddock and other boreal determinate fecundity species. We found reasonable agree- ment between field and histological methods, except for the regressing and regenerating stages (however, differentiation of these 2 stages is the least important distinction for determination of maturity or repro- ductive dynamics). The staging of developing ovaries was problematic for both methods partly because of asynchronous oocyte hydration dur- ing the early stage of oocyte matura- tion. Although staging on the basis of histology in a laboratory is gen- erally more accurate than macro- scopic staging methods in the field, we found that field observations can uncover errors in laboratory staging that result from bias in sampling unrepresentative portions of ovaries. For 2 specimens, immature ovaries observed during histological exami- nation were incorrectly assigned as regenerating during macroscopic staging. This type of error can lead to miscalculation of length at matu- rity and of spawning stock biomass, metrics that are used to characterize the state of a fish population. The revised field index includes 3 new macroscopic stages that represent final oocyte maturation in a batch of oocytes and were found to be reli- able for staging spawning readiness in the field. The index was found to be suitable for studies of diel spawn- ing periodicity and conforms to re- cent standardization guidelines. Manuscript submitted 6 February 2012. Manuscript accepted 30 November 2012. Fish. Bull. 111:90-106 (2013). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessar- ily reflect the position of the National Marine Fisheries Service, NOAA. Katie A. Burchard (contact author)1 Francis Juanes2 Rodney A, Rountree3 William A. Roumillat4 An important component of the as- sessment and management of any fish stock is quantification of the stock’s productivity, which is a func- tion of survival, individual growth, and reproductive success of a fish population (Wootton, 1998; Morgan, 2008). There are several factors that can be used to estimate the annual reproductive potential of a fish stock, including but not limited to sex ratio, age and size at maturity, spawning stock biomass, fecundity, and stock recruitment estimates where egg and larval viability are taken into consideration (Jennings et ah, 2001; Morgan, 2008). Regular monitoring and data collection on reproduc- tive potential, including estimation of spawning stock biomass, age and size at maturity, and fecundity, are dependent upon the use of reproduc- tive maturity indices from a sample of the population (Tomkiewicz et ah, 2003). Because the ability to accurately determine reproductive maturity by macroscopic examination of the go- nads alone is fallible, the validity of field reproductive indices has been questioned (Hilge, 1977; Templeman et al., 1978; Saborido-Rey and Jun- quera, 1998; Vitale et ah, 2006). De- termination of maturation stages in the field has been criticized as not be- ing dependable because different re- productive phases may appear simi- lar during gross staging of the gonad. For example, estimates of spawning stock biomass or mean length at ma- turity will depend upon an accurate distinction between adult fishes with regenerating gonads and immature fishes (Forberg, 1982; West, 1990). Similarly, estimates of fecundity in determinate-spawning species, such as Atlantic Cod ( Gadus morhua ) and Haddock, require accurate identifica- tion of ovaries in prespawning stages (Murua et al., 2003). Therefore, it is important that the system used for determination of maturity stage is accurate and unambiguous (Brown- Peterson et al., 2011; Lowerre-Barb- ieri et al., 2011). There have been considerable in- consistencies in the definitions of maturity stages of fishes among the existing indices in the literature. For example, O’Brien et al. (1993) defined Burchard et ai.: Maturity indices and field sampling practices for staging Melanogrammus aeglefmus 91 a female developing ovary as “a mixture of less than 50% yolked eggs and hydrated eggs”; however, accord- ing to Murua et al. (2003), the presence of hydrated oocytes indicate that the spawning process has begun and the gonad is in a “spawning” stage, where “oocytes are either in migratory nucleus stage or hydration stage.” This discrepancy between indices in the defini- tion of a developing ovary could result in different esti- mates of fecundity in determinate-spawning species for which prespawnmg, when the most advanced oocytes in an ovary are in the late vitellogenesis stage, is the optimal phase in reproductive maturity for the collec- tion of samples for accurate estimation of fecundity. If sampling is conducted before this stage, all oocytes des- tined to be spawned may not be developed and would be left out, and, as a result fecundity would be under- estimated. If samples are taken from females that have already spawned, the number of eggs that have already been released cannot be detected, an outcome that also would result in an underestimation of fecundity. Another important difference between the matura- tion indices of Murua et al. (2003) and O’Brien (1993) is the description of a resting ovary. The definition of O’Brien (1993) was based on a description by the NMFS (1989) and Kesteven (1960) and was similarly defined by Waiwood and Buzeta (1989), Tomkiewicz et al. (2003), and Vitale et al. (2006). All these authors described the resting maturity stage as occurring af- ter the spent maturity stage. Conversely, Murua et al. (2003) described the resting stage as an in-between batch state occurring before the spent stage, when some hydrated oocytes from the previous batch may remain and further batches of hydrated oocytes are still to be produced. Therefore, there was a need for greater consistency in definitions and standardization in terminology of reproductive maturity stages of fish- es. In a recent work by Brown-Peterson et al. (2011), a great deal of effort was invested in providing such standardization. Although certain reproductive traits, such as ma- turity phases, are universal among teleost fishes, the temporal patterns of these traits vary among species (Lowerre-Barbieri et al., 2011). Incorporation of tem- poral components into standardized indices potentially could produce more accurate staging results for each species studied, as well as provide additional informa- tion on the reproductive success of a species. A recent study by Tobin et al. (2010), published after our sam- pling was completed in 2006-07, identified the tim- ing and microscopic changes in maturation events of female Haddock as they transition from immaturity to maturity between summer and winter. That study provided evidence that Haddock commit to maturation by October or November with the existence of corti- cal-alveolar-stage oocytes in the ovaries. Knowledge of this maturation commitment can allow research- ers to confidently identify females as either immature, skipped-spawner, or mature after November, improving estimations of spawning stock biomass. Haddock is a batch-spawning species with group- synchronous ovary organization and determinate fecun- dity (Clay 1989; Murua and Saborido-Rey, 2003). This collection of reproductive traits is common in demersal Northwest Atlantic fishes, including but not limited to Atlantic Cod, Yellowtail Flounder ( Limanda ferru- ginea), and Atlantic Halibut ( Hippoglossus hippoglos- sus; see Murua and Saborido-Rey, 2003). The standard number of yolked oocytes immediately before the onset of spawning in a determinate-fecundity spawner can be considered equivalent to the potential annual fecun- dity of that fish (Murua et al., 2003). After the onset of spawning, the individual will hydrate several batches of yolked oocytes throughout the spawning season. The purpose of our study was to develop a standard field-proof, macroscopic ovarian maturity index for Had- dock that is suitable for use in studies of diel spawn- ing periodicity (Anderson, 2011) and conforms to the recent standardization guidelines of Brown-Peterson et al. (2011). Diel spawning periodicity has been widely studied in marine fishes (e.g., Ferraro 1980; Walsh and Johnstone, 1992; Wakefield, 2010) and provides details on the chronology of reproductive processes in species. It has been suggested that diel spawning periodic- ity maximizes fish survival and reproductive success (Ferraro, 1980; Lowerre-Barbieri, 2011). In addition to support for the collection of field data on reproductive stages, we also wanted the index to provide guidance on sampling techniques for the collection of samples for laboratory analysis. First, a staging method devel- oped from unpublished observations and a review of data published before our sampling in 2006-07 was used to stage female Haddock ovaries in the field. The resulting maturity index was then revised compared with a laboratory histological staging method similar to that of Tomkiewicz et al. (2003) for Atlantic Cod in the Baltic Sea. New stages were assessed to determine whether they could be used in future studies to exam- ine diel patterns in spawning (Anderson, 2011). Finally, the relative strengths and weaknesses of both the field and laboratory approaches were assessed. Materials and methods Initial field and laboratory indices A new field macroscopic ovarian maturity index for fe- male Haddock was developed by building on previous published indices (Homans and Vladykoy, 1954; Robb, 1982; Murua et al., 2003; Brown-Peterson et al., 2011) and unpublished observations made in the field (Table 1). The index consists of 8 stages, progressing from im- mature to regressing. To move toward use of standard phraseology, the terminology follows Brown-Peterson et al. (2011). It differs from previously published indi- ces with the addition of 3 stages that represent early to late progression of oocyte maturation (OM; Brown- 92 Fishery Bulletin 1 1 1 (1) Table 1 Field index developed and used to stage the reproductive maturity of female Haddock (Melanogrammus aeglefinus ) caught in the Gulf of Maine in 2006-07 during this study in which macroscopic methods in the field were compared with histological methods in the laboratory. OM=oocyte maturation. Stage Abbreviation Description Immature I Ovaries small and firm, about 1/8 the volume of the body cavity. Membrane thin and trans- parent, gray to pink in color. Contents microscopic: Individual oocytes not visible to the naked eye. Developing D Ovaries larger and plump, about 1/3 to 1/2 the volume of the body cavity. Membrane red- dish-yellow with numerous blood vessels. Contents visible to the naked eye and consist of opaque eggs that give the ovaries a granular appearance. Hydration stage 1 HI Ovaries well developed, reddish-yellow in color, at least 2/3 volume of body cavity. Mem- brane opaque with blood vessels conspicuous. Contents consist mostly of yellow-looking oocytes with <25% of the ovary containing larger translucent oocytes. A batch of oocytes in the early stages of OM where oocytes start to hydrate. Hydration stage 2 H2 Ovaries well developed, reddish-yellow in color, at least 2/3 volume of body cavity. Mem- brane opaque with blood vessels conspicuous. Visible surface of the ovary consists of 25- 50% larger translucent oocytes. Further progression of a batch of eggs in OM. Hydration stage 3 H3 Ovaries well developed, reddish yellow in color, at least 2/3 the volume of body cavity. Membrane opaque with blood vessels conspicuous. Visible surface of the ovary consists of 50-75% larger translucent oocytes. Ovaries may appear a little flabby, indicating the previous release of batch! es) of eggs. Final stages of the maturation of a batch of oocytes before a spawning event. Ripe and running RR Ovaries very large, over 2/3 the volume of the body cavity. Contents consist of mostly large, translucent eggs. Eggs running freely with little to no pressure on the abdomen. Regressing S Ovaries soft, and flabby, about 1/4 the volume of the body cavity. Membrane thick and tough, purplish in color, and bloodshot. Contents empty, few eggs remain, giving the gonad a patchy appearance. Regenerating RE Ovaries small and firm. 1/6 the volume of the body cavity. Membrane thin but less trans- parent than an immature ovary, yellowish-gray in color. Contents microscopic, opaque. Peterson et al., 2011) on the basis of the percentage of hydrated oocytes present (HI, H2, H3; Table 1, Fig. 1). During observations of mature female Haddock ovaries, we noticed that many of them had varying numbers of hydrated oocytes. We did not find an ovar- ian maturity index in the literature that categorized the progression in percentage of hydrated oocytes in a gonad. We were interested in whether the increase in percentage of hydrated oocytes was detectable over time and whether these stages may aid in examination of diel reproductive periodicity (Anderson, 2011). Hydration stage 1 (HI) is an ovary where a batch of oocytes is in the early phase of OM and when <25% of that ovary’s visible surface contains translucent, hydrated oocytes (Table 1). Hydration stage 2 (H2) is an ovary where a batch of oocytes is in the middle phase of OM and when 25- 50% of that ovary’s visible surface contains translu- cent, hydrated oocytes (Table 1). Hydration stage 3 (H3) is an ovary with a batch of oocytes in a late phase of OM and when 50-75% of the visible surface of that ovary contains translu- cent, hydrated oocytes (Table 1). We hypothesized that HI, H2, and H3 occur with each batch of oocytes before it is spawned (Fig. 1). The index also includes for each stage: 1) a macroscopically derived ratio of ovary volume to body cavity volume, similar to the ratio of gonad cavity length to body cav- ity length that Robb (1982) included for some stages; 2) a physical description of the ovary membrane, as Homans and Vladykoy (1954) included for some of the stages; and 3) a grossly assessed oocyte development description, included by Homans and Vladykoy (1954), Robb (1982), and Murua et al. (2003) (Table 1). The histological staging method was derived inde- pendently of the macroscopic ovarian maturity index (i.e., during analysis, field-based stages were not used by laboratory personnel in development of histological stages and vice versa), and it was based on previous work of Tomkiewicz et al. (2003), Roumillat and Brou- wer (2004), and Brown-Peterson et al. (2011) (Table 2). To differentiate the processes of early versus later vitel- logenic activity, 2 histological index stages (2.1 or 2.2) were used to define developing ovaries (Table 2). Be- cause Haddock are classified as possessing determinate fecundity (Murua et al., 2003), all oocytes that will be spawned during the upcoming season develop during these 2 stages, leaving a group of primary oocytes as a reserve for the successive spawning season. However, the developing stages in the histological index (2.1 and Burchard et al. : Maturity indices and field sampling practices for staging Melanogrammus aeglefmus 93 Immature Regenerating Regressing H3 * A \ H2 J > Oocyte maturation (OM) HI Batch 2 J * H3-_ H2 Interbatch period OM re-occurs with every batch before a spawning event = Spawning event Figure 1 The maturation cycle of the female Haddock ( Melanogrammus aeglefinus), including 3 hydration stages and an interbatch period, introduced and used during this study of meth- ods for staging the reproductive maturity of Haddock sampled in the southwestern region of the Gulf of Maine in the spring of 2006 and 2007. Hydration stage 1 (HI), hydration stage 2 (H2), and hydration stage 3 (H3) represent early-to-late progression of final oocyte maturation (OM) of a batch of oocytes, based on the percentage of hydrated oocytes pres- ent. *=spawning event. 2.2) were grouped together as one developing stage (2.0) when the histology results were compared with the field results because those stages could not be dif- ferentiated by macroscopic examination. Three phases of spawning-capable (SC) ovaries were assigned in the histological index as 3.1, 3.2, and 3.3 to differentiate the process of early, middle, and late phases of OM: early germinal vesicle migration (GVM) and germinal vesicle breakdown (GVBD) (Table 2). The gross assess- ments of HI, H2, and H3 are based on morphologically distinct criteria that are corroborated by the histologi- cal sections that effectively separate these stages from each other (Table 2). Two histological index stages (4.1 and 4.2) were defined to categorize SC ovaries that showed evidence of recent ovulation with the presence of recent (4.1) or old (4.2) postovulatory follicles (POFs; Alekseyeva and Tormosova, 1979; Saborido-Rey and Junquera, 1998). POFs are ruptured empty oocyte cas- ings left in the ovary after a spawning event (Table 2; Alday et al., 2010; Saborido-Rey and Junquera, 1998). If a sample contained POFs but also exhibited char- acteristics of another stage, the alternative stage was assigned with a note that the sample contained POFs (e.g., if a sample primarily contained oocytes in stage 3.1 but also contained POFs, it was assigned to the 3.1 stage). Field sampling Commercial fishing vessels were chartered for 25 dedi- cated survey trips in the spring of 2006 (15) and 2007 (10) to collect biological samples of Haddock in the southwestern Gulf of Maine (National Marine Fisher- ies Service Statistical area 514; Fig. 2). Surveys were based on a fixed station design with sampling where Haddock aggregations were known to previously exist. Sampling was conducted during the known spawning season of Haddock in the Gulf of Maine, between Janu- ary and June (Brown, 1998). Haddock were identified in the manner used bj' Collette and Klein-MacPhee (2002). Longlining was the preferred collection method for samples because few discards would result. Ap- proximately 19 m of longline was set and retrieved 3 times at each sampling location over a 12-h period with the objective of having 2 consecutive trips repre- sent sampling over a 24-h period (0100-0000 h; Table 3). Sets were conducted within specific 4-h time bins 94 Fishery Bulletin 1 1 1 (1) Table 2 The reproductive maturity index developed and used in this study of staging methods for female Haddock (Melanogrammus aeglefinus ) during histological analysis with analogous stages from the macroscopic field index. Histological definitions were based on criteria of Brown-Peterson et al. (Table 2 in 2011) CA=cortical alveolar; GVM=germinal vesicle migration; GVBD= germinal vesicle breakdown; NA=not applicable; OM=oocyte maturation; POF=postovulatory follicle; SC*=spawning capable, actively spawning subphase; Vtgl=primary vitellogenic; Vtg2=secondary vitellogenic; Vtg3=tertiary vitellogenic. Histology Stage Macroscopic Histological description Immature Developing (early 1.0 1 Small ovaries, only oogonia and primary growth oocytes present. Ovary wall thin, no muscle bundles evident. developing subphase) 2.1 D Only primary and cortical alveolar oocytes present. Developing 2.2 D Primary growth, CA, Vtgl and Vtg 2 oocytes present. SC* early GVM 3.1 HI Predominance of Vtg3 and early OM and beginning of GVM, yolk coalescence beginning. Few germinal GVBD oocytes observed, although some hydrated oocytes present. SC* GVM 3.2 H2 Both early and late stages of GVM oocytes, obvious yolk coalescence occurring. Greater abundance of GVBD oocytes seen. Increased number of hydrated oo- cytes present. SC* GVBD 3.3 H3 Predominance of GVBD oocytes, many with complete yolk coalescence. Many hydrated oocytes present — immediately before ovulation. SC recent POF 4.1 NA Many recent POFs present, showing few signs of degeneration. Otherwise ad- vanced oocytes consist most noticeably of Vtgl-Vgt3 oocytes. SC older POF 4.2 NA Only older POFs present with advanced structural degeneration. Advanced oocytes consist of Vtgl-Vgt3 oocytes. Regressing 5.0 S Only spawning residue (old POFs) and primary growth oocytes remain in the ovary. Spawning effort for season ceased. Regenerating 6.0 RE Only primary oocytes remain in small ovary. Ovarian wall thickened, muscle bundles present. (0100-0500 h, 0500-0900 h, 0900-1300 h, 1300-1700 h, 1700-2100 h, 2100-0000 h EST) to examine diel periodicity in reproductive maturity (Anderson, 2011). Each longline was fished with 150 to 400 circle hooks set 2 m apart for an average soak time of 2 h. The number of hooks fished per line on each trip was de- pendent on the success of catching Haddock that day. With the intent of sampling at least 50 Haddock from each longline set, the number of hooks was increased if the sample size was not reached or decreased if more fish than were needed were caught. All Haddock were measured by fork length (FL, ±1 mm) and examined externally for signs that indi- cated if they were in the ripe and running maturity stage (classified RR; Table 1). Ovaries were classified as RR when eggs were observed to be running freely from females with little pressure applied to the abdo- men. The first 50 Haddock in each set were sacrificed to determine the stage of development of the gonads. If a fish ovary was observed to be ripe and running, its sex and maturation stage could be determined with- out excisions, and it was automatically classified as RR in the field. A subsample of the 50 sacrificed female Haddock that represented all reproductive stages from each longline set was labeled and reserved on ice. Fish from each of the following length bins were collected from each set if possible to have representation from as many cohorts as possible: 30-40 cm, 40-50 cm, 50-60 cm, and >60 cm FL. Laboratory methods Samples were processed in the laboratory within 24 h of the end of each trip. Total weight (±0.1 kg) and ovary weight (±0.01 kg) of each individual were re- corded. Macroscopic maturity stage of all samples was re-examined by the same field examiner. Digital pho- tographs of whole ovaries were taken from a random subsample of each stage in the field index. To deter- mine the accuracy of macroscopic maturity staging per- formed with our maturation index, histological analysis was conducted on tissue samples of a subsample of 169 ovaries from 1706 macroscopically classified fish repre- sentative of all 8 stages. All histological tissue samples were taken from the forward right lobe of each ovary. It was assumed that this approach was appropriate because, according to Robb (1982), Haddock ovaries are homogeneous in structure throughout both lobes with oocytes present in various stages from the walls to the center of the ovary. Samples of 10-g tissue sections were fixed for at least 14 days in 10% neutral buffered formalin before they were transferred to 50% isopropyl alcohol. Samples were processed with standard histological procedures Burchard et al Maturity indices and field sampling practices for staging Melanogrammus aeglefinus 95 70°12'W ! ! Western Gulf of Maine Closure □ Stellwagen Bank National Marine Sanctuary Figure 2 Map of the locations where mature female Haddock (Melanogrammus aeglefinus) were sampled in the southwestern region of the Gulf of Maine in the spring of 2006 and 2007 for for staging reproductive maturity. (Humason, 1972) through a graded ethanol series, em- bedded in paraffin, and sectioned at 6 p. Tissues were stained with Gill’s hematoxylin and counterstained with eosin-Y. Ovary samples were classified by the oc- currence of specific histological features that represent progressive oocyte maturation stages (Brown-Peterson et ah, 2011) (Table 2). The most progressive feature ob- served in each sample was used to assign the appropri- ate stage. Photomicrographs were taken of a random subsample of stained tissue for each field index stage. Statistical analysis A contingency table was used to compare the results between the macroscopic staging methods used in the field and the histological staging methods used in the laboratory (Table 4). The table cell where the 2 equiv- alent stages cross shows the number of samples for which the data from the 2 methods agreed. Because the 2 indices were developed independently, 2 differ- ent types of percent agreement were calculated. One type was derived by dividing the number of samples for which the 2 methods agreed by the field stage sample size (last row in Table 4). The second type of percent agreement was calculated by dividing the number of samples for which the 2 methods agreed by the histological stage sample size (last column in Table 4). We did not have enough observed frequen- cies in each cell to perform a chi-square statistical analysis. 96 Fishery Bulletin 111(1) Table 3 Dates of trips during which longlines were set and re- trieved in the southwestern region of the Gulf of Maine in the spring of 2006 and 2007 to collect samples of fe- male Haddock (Melanogrammus aeglefinus) over a 12-h period with the objective of having 2 consecutive trips represent sampling over a 24-h period. 24-h period Year Sampling dates 1 2006 3/12, 3/28, 3/31 2 2006 4/7, 4/10, 4/28 3 2006 4/30, 5/4, 5/8 4 2006 5/8, 5/16 5 2007 3/26,3/31,4/10 6 2007 4/10, 4/21, 4/24 7 2007 5/1, 5/22 8 2007 5/24, 5/30 Results The results of each stage are formatted to explain both types of percent agreement as a function of each of the two staging methods. For each stage, the results of the macroscopic field staging method are presented first, followed by the results of the histological laboratory staging method. All 6 ovaries classified as immature (I) with the field index were also classified as the equivalent histological stage (1.0) in the laboratory. In contrast, all but 2 of the 8 samples classified as I (1.0) with the laboratory staging method were also classified as I with the field index (Table 4). Two samples classified as 1.0 in the laboratory were classified as regenerating (RE) with the field index. Only 4 of the 9 ovaries classified as developing (D) with the field index were also classified as developing (2.0) with the laboratory staging method (Table 4). Two of the remaining ovaries classified as D with the field index were classified as the adjacent histological stage 3.1, and 2 samples contained early POFs (stage 4.1) and 1 sample contained late POFs (stage 4.2). In con- trast, 7 of the 12 ovaries classified as 2.0 in the labora- tory were classified as the adjacent HI with the field index, and 1 sample was classified as RE. Twelve of the 32 ovaries classified as HI with the field index were also classified as the equivalent his- tological stage 3.1 (Table 4) in the laboratory. Seven of the ovaries classified as HI with the field index were Table 4 Contingency table showing the results from the cross classification between the histological maturity stag- es (columns) and the field maturity stages (rows) in the indices used in this study of methods for staging the reproductive maturity of female Haddock ( Melanogrammus aeglefinus). The gray squares represent where the cross classification is expected to have the highest frequencies of agreement. n=sample size; PA=percent agreement; NA=not applicable. If NA was used in place of PA, then that stage was not expected to agree with any of the opposing index stages. Maturity-index stages based on field examination 03 C 6 03 X OJ 03 o '&) -C G I D HI H2 H3 RR S RE n PA 1.0 6 0 0 0 0 0 0 2 8 75% 2.0 0 4 7 0 0 0 0 1 12 31% 3.1 0 2 12 0 1 0 1 0 16 75% 3.2 0 0 2 21 2 0 4 0 29 72% 3.3 0 0 5 9 22 17 2 2 57 39% 4.1 0 2 1 1 0 0 0 0 4 NA 4.2 0 1 5 2 0 1 0 0 9 NA 5.0 0 0 0 0 0 1 4 16 21 19% 6.0 0 0 0 0 0 0 1 12 13 92% n 6 9 32 33 25 19 12 33 PA 100% 44% 38% 64% 88% NA 33% 36% -a c Burchard et at: Maturity indices and field sampling practices for staging Melanogrammus aeglefmus 97 classified as the adjacent histological stage 2.0, 2 ova- ries were classified as 3.2, and 5 ovaries were assigned as 3.3. One Hl-classified ovary contained early POFs, and 5 HI ovaries contained late POFs. In contrast, 2 of the 16 samples classified as 3.1 in the laboratory were classified as the adjacent D stage with the field index, 1 sample was classified as H3, and 1 sample was as- signed as regressing (S). Twenty-one of the 33 ovaries classified as H2 with the field index were also classified as the equivalent histological stage 3.2 in the laboratory (Table 4). Nine H2-classified ovaries were classified as the adjacent histological stage 3.3. One ovary contained early POFs, and 2 ovaries contained late POFs. In contrast, 4 of the 29 ovaries classified as the 3.2 stage in the labo- ratory were classified as the adjacent field stages (HI and H3), and 4 of those ovaries were classified as S. The H3-classified samples were most frequently classified as the equivalent histological stage 3.3 (n- 22; Table 4). Two H3-classified ovaries were classified as the adjacent histological stage 3.2, and 1 ovary was classified as 3.1. In contrast, 35 of the 57 ovaries classi- fied as the histological stage 3.3 were classified differ- ently with the field index, with most ovaries classified as H2 (n=9) or RR (n= 17). All but 2 of the ovaries classified as RR («=17) in the field were classified as the histological stage 3.3 (Table 4). The 2 remaining ovaries were classified as the histological stages 4.2 and 5.0. Four of the 12 ovaries classified as S with the field index were assigned the equivalent histological stage 5.0 (Table 4). Four additional ovaries classified as S with the field index were classified as the histological stage 3.2, and 2 ovaries were assigned as 3.3, 2 ova- ries as 3.1, and 1 ovary as 6.0. In contrast, most of the 21 ovaries assigned to the histological stage 5.0 in the laboratory were classified as RE with the field index (/? = 16, 76%); however, 1 ovary was assigned as H3 (Table 4). Twelve of the ovary samples classified as RE with the field index were classified as the equivalent histo- logical stage 6.0 (Table 4). Sixteen samples classified as RE with the field index were classified as the adjacent histological stage 5.0 in the laboratory. Two additional samples classified as RE in the field were classified as histological stage 3.3, and 2 samples were classified as 1.0, and 1 sample was assigned as 2.0. In contrast, all but 1 of the 13 ovaries classified as histological stage 6.0 in the laboratory were also classified as RE with the field index. A final composite ovarian maturity index was cre- ated on the basis of the findings from this study (Table 5). Visual characteristics for both the whole ovary and tissue sample were emphasized as was similarly done by Tomkiewicz et al. (2003) for Altantic Cod in the Bal- tic Sea. The final index consists of 7 stages of ovary reproductive maturity distinguishable at sea. Table 5 includes for each maturity stage an image of the whole ovary, a photomicrograph of equivalent histological tis- sue, and both a macroscopic and microscopic physical description of the ovary. Notes are included to aid the user in correct macroscopic identification of each stage. Sampling techniques for collection of tissue samples are also included for problematic stages. On the basis of comparison with the histological data, we concluded that H3 and RR field stages are identical and grouped them together as a single stage (H3). When we used this revised H3 field stage, 39 of the 44 ovaries as- signed as H3 were assigned the equivalent 3.3 histo- logical stage. Discussion The utility of the field-based staging method for the classification of fish reproductive maturity for fisher- ies management is dependent on its biological accuracy. The findings from this study highlight the problems of development of an accurate error-proof field ovarian maturity index on the basis of macroscopic observation. However, a comparison of field-based and histology- based staging methods of Haddock ovaries presented in this study revealed the need to revise the field staging methods to increase the accuracy of both staging meth- ods. Although laboratory staging done on the basis of histology is inherently more accurate than any macro- scopic field staging method, there was indication that field observations can reveal weaknesses in the labora- tory approach because samples of the ovary taken for histology are not always going to be representative of the whole ovary. The strengths and weaknesses of both approaches for each maturation stage are discussed in the next sections, followed by recommendations for correct identification of each stage and a description of helpful sampling techniques for collection of tissue samples of problematic stages. Immature stage The I stage in the field index was equivalent to the 1.0 histological stage (Tables 1 and 2). The only stage mistaken for immature in the field was RE (Table 1). In both stages, the ovary was small and firm. The RE ovary appeared to be a little larger, less transparent, and grayer in color in comparison with the pink color of an immature ovary. However, in a young mature fish or late immature fish, these differences were less detectable. The imprecision in separation of immature and regenerating mature females also has been en- countered in staging Atlantic Cod ovaries (Tomkiewicz et al., 2003). Comparison of the current mean length at maturity for Haddock with the size of the specimen may help support either maturity stage in the field, but this criterion should not be relied on because length at maturity can change over time (Saborido-Rey and Junquera, 1998; Tobin et al., 2010). In this study, the smallest Haddock caught was 35.5 cm FL, larger than the mean length at maturity re- 98 Fishery Bulletin 111(1) corded for this species in the Gulf of Maine (34.5 cm; Collette and Klein-MacPhee, 2002). The gear type used in this study selected for larger fish, and we suspect that smaller fish avoided the longline hooks. Although to our knowledge skipped spawning (when a mature individual skips a year of spawning) has not been ob- served in Haddock, it is not uncommon in long-lived iteroparous fishes, including Atlantic Cod (Jorgensen et ah, 2006; Rideout et ah, 2006; Fig. 1). Therefore, we could not have assumed that a female was immature if it lacked signs of sexual maturity during the spawn- ing season, as was assumed by Waiwood and Buzeta (1989) because there is the possibility that the fish had skipped spawning that year. The use of microscopic analysis or histological ex- amination of a tissue sample of the ovary was a reli- able way to determine whether the ovary was imma- ture or regenerating. Immature ovaries could be dis- tinguished histologically from regenerating ovaries by the diameter of the primary oocytes (W. Roumillat, per- sonal commun.). Immature ovaries contained primary oocytes that were equal in diameter, but regenerating ovaries had primary oocytes that varied in diameter. Additionally, the RE phase can be differentiated from the I phase by the following features: RE ovaries 1) have a thicker ovarian wall, 2) have more space, inter- stitial tissue, and capillaries around primary oocytes, and 3) have the presence of late-phase atresia and muscle bundles (blood vessels surrounded by connec- tive and muscle tissue) (Brown-Peterson et ah, 2011). Because of the selectivity of the fishing gear for larger- size fish and our limited sampling period, our study did not provide adequate data to fully resolve macroscopic differences between the RE and I stages. Further work should focus on differentiation of a regenerating ova- ry from an immature ovary with sampling conducted further into the summer with less size-selective gear. Proper identification of immature ovaries would great- ly reduce the error in calculation of spawning biomass estimates and improve accuracy of estimates of length at maturity. Developing stage There was disagreement between D and early OM phase, HI (Table 1). We observed that when a Had- dock ovary began OM, some oocytes in the initial batch completed the process before others within the same ovulating batch. Although Haddock ovaries have been reported to be homogeneous in structure throughout all phases of maturity (Templeman et ah, 1978; Robb, 1982), our observations indicate that it is not homoge- neous in structure during this very early phase of OM (HI). This result is supported by Alekseyeva and Tor- mosova (1979), who reported that formation of batches occurs through asynchronous maturation of individu- al groups of oocytes. The histological staging method sometimes resulted in HI ovaries being misclassified as D, likely because they were sampled during initial OM of the first batch of oocytes for the season, when there were no histological characteristics present to indicate that prior batches had been spawned. Initial spawning HI ovaries had so few fully hydrated oocytes (because of the asynchronous maturation of the batch) that collection of a small tissue sample from a central location was sometimes unsuccessful in representing all phases of oocytes present. As a single batch pro- gresses through OM, evidence that spawning has been initiated becomes more pbvious with GVM and yolk co- alescence beginning in oocytes (Table 2; Lowerre-Bar- bieri et ah, 2011). As the season progresses and the ovary initiates OM in later batches of oocytes, a HI tissue sample could be distinguished from a D tissue sample by the presence of POFs. The agreement between macroscopic and histologi- cal staging for D and HI ovaries could be improved if the method used to take tissue samples from the ovary were modified. When ovaries are classified as HI in the field, a larger tissue sample or samples should be taken from multiple places in the ovary to improve the accuracy of the histological results. Our observations demonstrate that determination of the maturation of an ovary based on histological examination alone may not always be accurate. To reduce staging errors based on histological analysis in future studies, it is recom- mended that each tissue sample be documented with a photograph of the whole ovary from which it was extracted and with an estimate of the percentage of hydrated oocytes observed on the visible surface of the ovary. Three ovaries classified as D in the field contained POFs when analyzed histologically, and, by our defini- tion, a D ovary could not have previously spawned that season (Table 1; Fig. 1). Therefore, those specimens had spawned at least one batch of eggs but had not yet hydrated oocytes for the next batch, and the decrease in volume of the ovary after spawning a prior batch of eggs was not evident in field observations. A closely re- lated species, Atlantic Cod, begins to hydrate a batch of oocytes 1-2 days before spawning (Kjesbu, 1991). Final oocyte maturation in cold-water marine fishes with pe- lagic eggs generally lasts 1-2 days (Thorsen and Fyhn, 1996). Trippel and Neil (2004) reported that Haddock had a mean interval of 5.4 days between batches of re- leased eggs, and Hawkins et al. (1967) and Alekseyeva and Tormosova (1979) reported an interval of 26-40 h. These findings combined indicate that there is an in- terbatch period between the spawning of a batch and the next batch that is beginning to hydrate, a period described by Murua et al. (2003) as the resting stage (Fig. 1). Consequently, there was the possibility that a ma- ture ovary could be incorrectly classified as D in the field if it was between ovulation events during this in- terbatch period. Therefore, we concluded that it is not always possible to be certain that an individual has begun spawning for the season on the basis of macro- scopic observation alone and this uncertainty can pose Burchard et at: Maturity indices and field sampling practices for staging Melanogrammus aeglefinus 99 a problem for fecundity studies where ovary weight is used as a factor in determining fecundity. For the same reason, we also concluded that it is not possible to ac- curately stage an ovary as D by macroscopic observa- tion alone. This issue poses a problem for studies that use gravimetric counting of vitellogenic oocytes and oocyte density to determine fecundity. The D stage, when the most advanced oocytes in the ovary are in the late vitellogenesis phase, is the optimal stage from which samples should be taken to determine fecundity. Therefore, we recommend that ovary samples be col- lected from fishes classified as D on the basis of mac- roscopic observations to confirm through microscopic or histological analysis that the ovary is in a prespawning state. Hydration stages A challenge in the use of the field index was the subjec- tive evaluation of the percentage of hydrated oocytes in an ovary that was used to assign the consecutive HI, H2, and H3 stages. Therefore, histological samples were often assigned to a stage adjacent to the stage that was reported in the field. There were 5 instances where an ovary was macroscopically classified as HI with the field index but microscopically classified as the histological stage 3.3. This difference in staging was likely due to some variation in individual and tem- poral batch fecundity (Trippel et al., 1998). However, this error was rare and the hydration stages were cor- rectly staged consistently enough that we do not con- sider this misclassification problematic in identification of the correct hydration stage for the purpose of assess- ing diel reproductive patterns. The histology-based laboratory staging method un- derestimated the HI stage because the ovary typi- cally appears to be heterogeneous during this stage and, therefore, was not adequately represented in the tissue samples. An Hl-classified ovary could be incor- rectly identified as D based on histological examination under these conditions. However, as an ovary matured further, the oocytes appeared to hydrate in unison and evenly throughout the ovary and nuclear migration and globule yolk coalescence became more evident. These criteria reduced the bias in the sampling method in later phases of HI and eliminated it in later stages H2 and H3. Histological analysis verified that H3-stage ova- ries were in a state where the next batch of oocytes to be spawned were in final OM phase (GVBD), with most oocytes fully hydrated. This consistent result is important because both the field H3 and histological 3.3 stages can be confidently used to identify spawning readiness, and, therefore, we concluded that they will be well suited for use in studies of diel spawning peri- odicity in Haddock (Anderson, 2011) and other fishes. Ripe and running stage When the ovaries of RR females were examined mac- roscopically, they exhibited characteristics of the H3 stage. Furthermore, the tissue samples from these ova- ries were classified as 3.3 (SC GVBD; Table 2) with histology-based methods. On the basis of results from the histological analysis conducted on ovaries classified as RR in the field and from the portion of the RR ovary full of hydrated oocytes during macroscopic observa- tion, we decided to combine the RR and H3 field stages into a single stage in the final index (H3; Table 5). Use of the RR field stage proved problematic be- cause of the sampling method, and we recommend cau- tion in its use in future studies. Homans and Vladykoy (1954) reported that female Haddock stop feeding dur- ing spawning — behavior that would make it difficult to catch actively spawning fish with baited gear and possibly result in an underestimation of RR females in the population. In addition, RR may be overestimated because of premature ovulation induced by stress or barotrauma. It is hypothesized that the barotrauma caused by forcing specimens to ascend to the surface from an average depth of 90 m during sampling can cause premature ovulation of hydrated oocytes. An increased level of cortisol in fishes is an indication of severe stress, but it is also involved in the natural pro- cess of ovulation (Billard et al., 1981; Wendelaar Bon- ga, 1997). The 2-h average soak time of the hooks in this study could have been enough time for the stress response to induce ovulation in an H3-stage fish before it landed on board the fishing vessel. For the same reason, histological stage 4.1 may be overestimated, because the premature ovulation caused by barotrauma results in POFs appearing before they normally would. We concluded that it is difficult to catch a Haddock in the act of spawning, especially with baited hooks; therefore, use of H3-stage fish to estimate spawning readiness would be more accurate. However, the practice of macroscopically staging a RR Haddock through application of pressure to the abdomen and observation of the excretion of hydrated oocytes is a method that can be used to classify a female as spawn- ing ready without need to sacrifice the fish. Regressing stage The S ovary stage was the most problematic for macro- scopic identification. The regressing condition is partic- ularly difficult to detect in a species such as Haddock with asynchronous development, where batches of eggs are spawned multiple times over a prolonged season (Hickling and Rutenberg, 1936; West, 1990). Species with determinate fecundity complete a spawning sea- son by the maturation and spawning of the entire co- hort of oocytes developed that year. When only a single batch of oocytes was left in the ovary to be spawned, it was termed “last spawn.” This stage was evident only during histological analysis. Of the ovaries classified 100 Fishery Bulletin 1 1 1 (1) Table 5 The final female reproductive maturity index developed from findings with the macroscopic and microscopic method for staging the maturity of female Haddock ( Melanogrammus aegleftnus). ,t jv.ii ^ Sij Immature (I) Macroscopic: The ovary is small and firm, and approximately 1/8 the volume of the body cavity. The membrane is thin, transparent, and gray to pink in color. Individual oocytes are not visible to the naked eye. *Note: This stage can look similar to a resting-stage ovary. Use of microscopic analysis or histology on a tissue sample of the ovary may be the only way to determine that the ovary is immature and not resting. Microscopic: The ovary contains germ cells, oogonia, and primary oocytes. The ovary wall is thin and the primary oocytes vary little in diameter. No muscle bundles can be seen. The nucleus is relatively large with the most advanced oocytes having peripheral nucleoli (magnification lOOx). Developing (D) Macroscopic: The ovary is plump and approximately 1/3 to 1/2 the volume of the body cavity. The membrane is reddish-yellow and has numerous blood vessels. The contents are visible to the naked eye and consist of opaque eggs, giving the ovaries a granular appearance. *Note: If hydrated oocytes are visible, the ovary should be classified as HI (see the next stage below). Hydrated oocytes will be large in diameter and translucent in color. A large tissue sample should be taken from all ovaries macroscopically classified as developing and analyzed microscopically to confirm that postovulatory follicles are not present and that the ovaries are in a prespawning state. It may be helpful to document the tissue sample with a photograph of the whole ovary. Microscopic: Primary and cortical alveoli oocytes, and primary and secondary vitellogenic oocytes are present. There is no evidence of postovulatory follicles (magnification 40x). in the field as S, 58% (N= 7) were classified as being in 1 of the 3 OM histological phases. The most plausible explanation for this result, other than observational er- ror, is that these particular specimens were maturing the last batch of eggs to be spawned that season (last spawn) and the ovary at this point had lost its rigid- ness and, therefore, looked as though it was in the S stage. Last spawn was observed in 8 (5%) of the his- tological samples, 5 of which were classified as S in the field. Last spawn also was observed in Haddock in the North Sea (Alekseyeva and Tormosova, 1979). Near the end of the spawning season, the ovary can lose its rigidness, although it still has 1-2 batches of oocytes to spawn and appears as S. The outside membrane thick- ens, which increases the difficulty of staging the ovary through examination of just the outside (Templeman et ah, 1978). Staging on the basis of the flabbiness of the ovary alone is not recommended, and the inside of the ovary should be examined for hydrated oocytes. If any oocytes during final oocyte maturation (OM) remain, the ovary is most likely not in the S stage and could be in last spawn. Histological examination of a sample of an ovary can be an effective way to determine if an ovary is regressing. Burchard et al Maturity indices and field sampling practices for staging Melanogrammus oeg/efinus 101 Table 5 continued Hydration stage 1 (HI) Macroscopic: The ovary is well developed, reddish-yellow in color, and approximately 2/3 the volume of the body cavity. The membrane is opaque and has prominent blood vessels. The contents consist mostly of yellow-looking oocytes and <25% of the ovary contains large translucent (hydrated) oocytes. *Note: In the early phase of the HI stage, the ovary is not visually homogeneous and hydrated oocytes can be unevenly scattered throughout. If microscopic analysis will be conducted on a subsample, take care to obtain a representative tissue sample that in- cludes translucent, hydrated oocytes. Document with a photograph of the whole ovary if possible. Microscopic: There is a predominance of tertiary vitellogenic oocytes, with many oocytes showing oocyte maturation, germinal vesicle migration and germinal vesicle breakdown. A small percentage of oocytes (<25%) will have completed oocyte maturation and are hy- drated. Postovulatory follicles may be present (magnification 100x). Hydration stage 2 (H2) Macroscopic: The ovary is well developed, reddish-yellow in color, and approximately 2/3 the volume of the body cavity. The membrane is opaque with blood vessels conspicuous. The visible surface of the ovary consists of 25-50% of large translucent oocytes. *Note: There are gradients between the consecutive HI and H2 stages as well as the H2 and H3 stages, where it is difficult to assign one or the other stage. In these cases, the ovary is at a state where it is either close to entering the H2 stage or close to advanc- ing to H3. In both cases the ovary is near if not in an intermediate phase of final oocyte maturation and may be accurately classified as H2. Microscopic: There is a predominance of oocytes showing germinal vesicle migration and germinal vesicle breakdown. Approximately 50% of the advanced oocytes are hydrated. Postovulatory follicles may be present (magnification 40x). Regenerating stage The histological results for RE stage ovaries reflected the difficulty in distinguishing between a regenerating and regressing ovary in the field, with 46% of the ova- ries classified as RE in the field assigned as S during histological analysis. The plausible explanation for this result is observational error. As the ovary progressed into the RE stage, it became easier to differentiate from the S stage, but, because of the short sampling period, it was difficult to differentiate between the 2 stages during the time when regenerating fish were captured. For future studies, we recommend that sam- pling be conducted from well before to well after the known spawning season and that a photograph of each ovary be taken for comparison with histology-based staging results. Such documentation of the changes observed in different phases, from spent to regressing, could improve the ability to distinguish between these 2 stages. However, extension of the sampling period too far into the fall and winter may make it more difficult to distinguish the D and RE stages from spawning stag- es (Tomkiewicz et al., 2003). Histological examination of a sample of an ovary was an effective way to determine if an ovary was in the RE stage. If a regenerating ovary was observed from a fish near or larger in size than the mean length at maturity during the peak spawning period, it is possible that 102 Fishery Bulletin 111(1) Tabie 5 continued Hydration stage 3 (H3) Macroscopic: The ovary is well developed, reddish-yellow in color, and approximately 2/3 the volume of the body cavity. The membrane is opaque with blood vessels conspicu- ous. Greater than 50% of the visible surface of the ovary consists of large translucent oocytes. Microscopic: There is a predominance of oocytes showing germinal vesicle migration and germinal vesicle breakdown. Greater than 50% of the advanced oocytes are hydrated. Postovulatory follicles may be present (magnification 40x). Regressing (S) Macroscopic: The ovary is soft and flabby and approximately 1/4 the volume of the body cavity. The membrane is thick and tough, purplish in color, and bloodshot. The inside of the ovary is almost empty and few oocytes remain, giving the gonad a patchy appearance. *Note: Toward the end of the spawning season, the ovary loses its rigidness although it still has 1-2 batch(es) of oocytes to spawn. Staging should not be based only on the flab- biness of the ovary, and the ovary should be inspected internally. The ovary is likely not yet spent if any hydrated oocytes remain. Microscopic: An abundance of postovulatory follicles are present. Oogonia and primary oocytes are evident. The ovary wall is thick, and muscle bundles are visible (magnifica- tion 40x). vM o A&i&L §gr Iff! ♦, ' w it spawned much earlier that season or skipped that year’s spawning season (Fig. 1). One mature regenerat- ing female was observed during the peak of the spawn- ing season. Skipped spawning is a response to various physiological and ecological conditions (Jorgensen et al., 2006) and often a trade-off between present re- production and survival for future reproduction (Bull and Shine, 1979; Rideout et ah, 2005). Because it is not possible to determine the existence and frequency of skipped spawning and its effect on recruitment, it is difficult to determine spawning stock biomass and, hence, difficult to conduct stock assessments and man- age such species (i.e., stock-recruitment models may overestimate recruitment and underestimate survival; Rideout et ah, 2005). Postovulatory follicles POFs were commonly found in ovary samples classified as HI, H2, H3, and S in the field, but these POFs of- ten were in various phases of atrophy. The observation of early and late phases of POFs in the same ovary indicated that POFs from the 2 previous batches still existed during the OM of the next batch to be spawned (Table 2). Evidence indicates that the complete atro- phy of a POF in Haddock could take up to 10 days, considering that Haddock have an average interval of 5.4 days between spawned batches (Trippel and Neil, 2004), and that final oocyte maturation in marine fish- es with pelagic eggs generally lasts 1-2 days and ends with ovulation (Thorsen and Fyhn, 1996). The atrophy Burchard et al.: Maturity indices and field sampling practices for staging Melanogrammus aeglefinus 103 Table 5 continued Macroscopic: The ovary is small and firm, and approximately 1/6 the volume of the body cavity. The membrane is thin but less transparent, yellowish-gray. Contents are micro- scopic, opaque. Microscopic: The ovary wall is thick. There is often indication of past spawning with rem- nants of unabsorbed material. The ovary contains primary oocytes that vary largely in diameter (magnification 100x). Regenerating (RE) *Note: If a resting ovary is observed from a fish greater in size than the mean length at maturity during the peak spawning period, then it is probable that the fish skipped that year’s spawning season. of POFs occurs for the Spotted Seatrout (Cynoscion nebulosus) in 24-36 h in water temperatures >2°C (Roumillat and Brouwer, 2004) and for the Northern Anchovy ( Engraulis mordax) in 48 h at 19°C (Hunter and Macewicz, 1985). The atrophy of Haddock POFs may take much longer because this species prefers to spawn in cold temperatures (4-7°C; Overholtz, 1987) — an actuality that may be widespread in boreal fishes. The slow degeneration of POFs in cold-water species is supported by Brown-Peterson et al. (2011) and noted by Saborido-Rey and Junquera (1998). Aging of POFs has been used in other species to de- termine spawning frequency or duration of time since the female last spawned a batch of eggs (Hunter and Macewicz, 1985; Roumillat and Brouwer, 2004). No de- finitive information on diurnal timing of spawning was clear from our inspection of Haddock POFs because none of them appeared to have been very recently cre- ated. Fish collections were concentrated in an area where active spawning took place, and those Had- dock that had finished spawning may not have been available for capture. Observation of many ovaries in spawning condition that also showed many phases of POF atrophy indicated that these residual tissues had a very slow rate of atrophy and were of little use in making accurate assessments of diel timing of ovula- tion. A more advanced study of aging POFs in cold-wa- ter species similar to the studies done for clupeiforms by Alday et al. (2010) and Haslob et al. (2012) is need- ed and would increase our knowledge on the timing of spawning in cold waters. There were no equivalent field index stages for the histological stages 4.1 and 4.2. Samples classified as 4.1 or 4.2 were typically assigned to an ovary in a state between the last batch of oocytes spawned and the next batch to be spawned, a state that we did not attempt to identify in the field. In ovaries of this state, no oo- cytes for the next batch had yet progressed to OM and the only oocytes present were in a vitellogenic devel- oped phase equivalent to the resting stage described by Murua et al. (2003). We found that this stage was not easily or accurately ascertainable through macroscopic observation of the ovary. A trained eye may be able to recognize a degree of flaccidity of an ovary that has spawned already. Many of the ovaries assigned as 4.1 or 4.2 exhibited characteristics of an ovary that was classified as the D stage in the field. The overestima- tion of the D stage in this study indicates the need to conduct histology on a subsample of ovaries classified as D stage in the field to assure there is no indication, on the basis of the presence of POFs, that females thus classified have started spawning that season. Conclusions Working independently, we came to the same conclu- sion as Brown-Peterson et al. (2011): standardization of maturation staging methods and terminology are need- ed. Our study confirms the importance of these efforts but extends them with the development of a new ovar- ian maturity index specifically for examination of diel spawning periodicity while using the maturation ter- minology established by Brown-Peterson et al. (2011). Comparison of macroscopic and microscopic observa- tions of ovaries helped us to improve the initial field index and sampling methods, as well as to provide use- ful insight into the reproductive biology of Haddock. 104 Fishery Bulletin 111(1) Noting the apparent longevity of POFs helped us un- derstand the duration and cyclical process of OM in this species and potentially other boreal or cold-water fishes. Because reproductive maturation occurred over a prolonged period of time, OM occurred throughout 3 distinct field stages (HI, H2, and H3) and histol- ogy stages (3.1, 3.2, and 3.3). This finding supports the conclusion of Alekseyeva and Tormosova (1979) that Haddock exhibits asynchronous maturation of in- dividual groups of oocytes. We believe that the asyn- chronous maturation of oocytes in a batch results in heterogeneous ovaries during early phases of OM and can lead to misclassification of HI ovaries as D stage in the field. However, Robb (1982) and Templeman et al. (1978) previously reported that Haddock ovaries are homogeneous in structure throughout all phases of maturity. Studies of follicle size-frequency distributions throughout OM are needed to confirm our observation of apparent heterogeneity of ovaries during early matu- ration to clarify how future studies should be modified to ensure accurate staging in the field and laboratory. Additional work should be focused on differentiation of a regenerating ovary from an immature ovary. This differentiation is the most important distinction in de- termination of maturity or reproductive dynamics of a stock because of the use of these numbers in estima- tion of spawning stock biomass. The timing of the sampling in this study, although restricted, was focused around the known spawning season of Haddock in the Gulf of Maine. This focus likely increased the reliability of staging SC fish be- cause the closer in time to the spawning season the more developed the ovary becomes, as was observed by Tomkiewicz et al. (2003). Alternatively, reliability in staging SC fish in the fall and winter is tenuous because ovary development is just beginning (Tomkie- wicz et ah, 2003). Therefore, the optimal time to collect data to be used to estimate spawning stock biomass should span across the spawning season, and we cau- tion against the use of SC data collected off season in estimation of spawning stock biomass. It is anticipated that the revised ovarian maturity index (Table 5) presented in our study will be useful to Haddock resource managers. The H2 and H3 stages appear to be useful indicators of spawning readiness for Haddock ovaries in the field. We suspect that the progression of OM is detectable in other boreal spe- cies with the same reproductive traits as Haddock and that the later stages could also be used to examine diel periodicity in these species. Although this index was developed for studies on diel reproductive periodicity, we feel it would also be useful for study of other short- term temporal reproductive patterns related to tidal, lunar, or solar zenith cycles. The revised field index in- cludes pointers to help users stage ovaries and take ap- propriate samples (Table 5). Although this revised field index will improve accuracy in the determination of the maturity stage of Haddock in the field, evidence has shown that field indices alone may not be enough to correctly classify a fish in problematic stages. However, the observations in our study also demonstrate that de- termining the maturation of an ovary by histological examination alone may not always be accurate, high- lighting the importance of field staging. In addition to field staging with the index presented here, appropri- ate tissue samples should be collected and analyzed microscopically or histologically to verify problematic stages, especially when field data are used in assess- ment and management of ^a fish stock. Acknowledgments This publication is the result of research spon- sored by The Massachusetts Institute of Technol- ogy Sea Grant College Program, under National Oce- anic and Atmospheric Administration grant number NA060AR4170019 and project number 2005-R/RD-29. The authors thank the cooperative work and generos- ity of fishermen T. Hill, P. Powell, and J. Montgomery. We also thank C. Goudey, S. Cadrin, and R. McBride for project advice and support. The assistance of vari- ous volunteers in the field and laboratory work is appreciated. Literature cited Alday, A., M. Santos, A.Uriarte, I. Martin, U.Martinez, and L.Motos. 2010. Revision of criteria for the classification of post- ovulatory follicles degeneration, for the Bay of Biscay anchovy (Engraulis encrasicolus L.). Rev. Invest. Mar. 17:165-171. Alekseyeva, Y. L, and I. D. Tormosova. 1979. Maturation, spawning and fecundity of the North Sea haddock, Melanogrammus aeglefinus. J. Ichthyol. 19:56-64. Anderson, K. A. 2011. Reproductive maturation and diel reproductive pe- riodicity in western Gulf of Maine haddock. M.S. the- sis, 77 p. Univ. Massachusetts, Amherst, MA. Billard, R., C. Bry, and C. Gillet. 1981. Stress, environment and reproduction in teleost fish. In Stress and fish (A. D. Pickering, ed.), p. 185- 208. London Academic Press, London. Brown, R. W. 1998. Haddock. In Status of fishery resources off the Northeastern United States for 1998 (S.H. Clark, ed.), p. 53-56. NOAA Tech. Memo. NMFS-NE-115. Brown-Peterson, N. J., D. M. Wyanski, F. Saborido-Rey, B. J. Macewicz, and S. K. Lowerre-Barbieri. 2011. A standardized terminology for describing repro- ductive development in fishes. Mar. Coast. Fish. 3:52-70. Bull, J. J., and R. Shine. 1979. Iteroparous animals that skip opportunities for reproduction. Am. Nat. 114:296—303. Burchard et at: Maturity indices and field sampling practices for staging Melanogrammus aeglefmus 105 Clay, D. 1989. Oogenesis and fecundity of haddock (Melanogram- mus aeglefinus L.) from the Nova Scotia shelf. ICES J. Mar. Sci. 46:24-34. Collette, B. B., and G. Klein-MacPhee. 2002. Bigelow and Schroeder’s fishes of the Gulf of Maine, 748 p. Smithsonian Inst. Press, Washington, D.C. Ferraro, S. P. 1980. Daily time of spawning of 12 fishes in the Peconic Bays, New York. Fish. Bull. 78:455-464. Forberg, K. G. 1982. A histological study of development of oocytes in capelin, Mallotus villosus villosus (Muller). J. Fish Biol. 20:143-154. Haslob, H., G. Kraus, and F. Saborido-Rey. 2012. The dynamics of postovulatory follicle degenera- tion and oocyte growth in Baltic sprat. J. Sea Res. 67:27-33. Hawkins, A. D., K. J. Chapman, and D. J. Symonds. 1967. Spawning of haddock in captivity. Nature 215:923-925. Hickling, C. F., and E. Rutenberg. 1936. The ovary as an indicator of the spawning period of fishes. J. Mar. Biol. Assoc. U.K. 21:311-317. Hilge, V. 1977. On the determination of the stress of gonad ripe- ness in female bony fishes. Meeresforschung 25:49-55. Homans, R. E. S., and V. D. Vladykoy. 1954. Relation between feeding and the sexual cycle of the haddock. J. Fish. Res. Board Can. 11:535-542. Humason, G. L. 1972. Animal tissue techniques, 661 p. W. H. Freeman & Co., San Francisco. Hunter, J. R., and B. J. Macewicz. 1985. Measurement of spawning frequency in multiple spawning fishes. In An egg production method for es- timating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed.), p. 79-94. NOAA Tech. Rep. NMFS 36. Jennings, S., M. J. Kaiser, and J. D. Reynolds. 2001. Marine fisheries ecology, 432 p. Blackwell Publ., Malden, MA. Jorgensen, C., B. Ernande, 0. Fiksen, and U. Dieckmann. 2006. The logic of skipped spawning in fish. Can. J. Fish. Aquat. Sci. 63:200-211. Kesteven, G. L. 1960. Manual of field methods in fisheries biology, 160 p. FAO, Rome. Kjesbu, O. S. 1991. A simple mthod for determining the maturity stages of northeast Arctic Cod ( Gad us morhua L) by in- vitro examination of oocytes. Sarsia 75:335-338. Lowerre-Barbieri, S. L., K. Ganias, F. Saborido-Rey, H. Murua, and J. R. Hunter. 2011. Reproductive timing in marine fishes: variabil- ity, temporal scales, and methods. Mar. Coast. Fish. 3:71-91. Morgan, M.J. 2008. Integrating reproductive biology into scientific ad- vice for fisheries management. J. Northwest Atl. Fish. Sci. 41:37-51. Murua, H., and F. Saborido-Rey. 2003. Female reproductive strategies of marine fish spe- cies of the North Atlantic. J. Northwest Atl. Fish. Sci. 33:23-31. Murua, H., G. Kraus, F. Saborido-Rey, P. R. Witthames, and S. Junquera. 2003. Procedures to estimate fecundity of marine fish species in relation to their reproductive strategy. J. Northwest Atl. Fish. Sci. 33:33-54. NMFS (National Marine Fisheries Service). 1989. Finfish maturity sampling and classification schemes used during Northeast Fisheries Center bottom trawl surveys, 1963-89. NOAA Tech. Memo. NMFS-F/ NEC-76, 14 p. O'Brien, L., J. Burnett, and R. K. Mayo. 1993. Maturation of nineteen species of finfish off the northeast coast of the United States, 1985-1990. NOAA Tech. Rep. NMFS 113, 66 p. Overholtz, W. J. 1987. Factors relating to the reproductive biology of Georges Bank Haddock ( Melanogrammus aeglefinus ) in 1977-83. J. Northwest Atl. Fish. Sci. 7:145-154. Rideout, R. M., M. J. Morgan, and G. R. Lilly. 2006. Variation in the frequency of skipped spawning in Atlantic cod ( Gadus morhua ) off Newfoundland and Labrador. ICES J. Mar. Sci. 63:1101-1110. Rideout, R. M., G. A. Rose, and M. P. M. Burton. 2005. Skipped spawning in female iteroparous fishes. Fish Fish. 6:50-72. Robb, A. P 1982. Histological observations on the reproductive bi- ology of the haddock, Melanogrammus aeglefinus ( L. ). J. Fish Biol. 20:397-408. Roumillat, W. A., and M. C. Brouwer. 2004. Reproductive dynamics of female spotted seatrout (Cynoscion nebulosus) in South Carolina. Fish. Bull. 102:473-487. Saborido-Rey, F., and S. Junquera. 1998. Histological assessment of variations in sexual maturity of cod ( Gadus morhua L.) at the Flemish Cap (north-west Atlantic). ICES J. Mar. Sci. 55:515-521 Templeman, W., V. M. Hodder, and R. Wells. 1978. Sexual maturity and spawning in haddock, Me- lanogrammus aeglefinus , of the Southern Grand Bank. ICNAF Res. Bull. 13:53-65. Thorsen, A., and H. J. Fyhn. 1996. Final oocyte maturation in vivo and in vitro in marine fishes with pelagic eggs; Yolk protein hy- drolysis and free amino acid content. J. Fish Biol. 48( 6 ): 1 1 95—1209. Tobin, D., P. J. Wright, and M. O’Sullivan. 2010. Timing of the maturation transition in had- dock Melanogrammus aeglefinus . J. Fish Biol. 77:1252-1267. Tomkiewicz, J., L. Tybjerg, and A. Jespersen. 2003. Micro- and macroscopic characteristics to stage gonadal maturation of female Baltic cod. J. Fish Biol. 62:253-275. Trippel, E. A., C. M. Doherty, J. Wade, and P. R. Harmon. 1998. Controlled breeding technology for haddock (Mela- nogrammus aeglefinus ) in mated pairs. Bull. Aquacult. Assoc. Can. 98(3):30-35 Trippel, E. A., and S. R. E. Neil. 2004. Maternal and seasonal differences in egg sizes and spawning activity of northwest Atlantic haddock (Me- 106 Fishery Bulletin 111(1) lanogrammus aeglefinus ) in relation to body size and condition. Can. J. Fish. Aquat. Sci. 61:2097-2110. Vitale, F., H. Svedang, and M. Cardinale. 2006. Histological analysis invalidates macroscopically determined maturity ogives of the Kattegat cod ( Ga - dus morhua ) and suggests new proxies for estimating maturity status of individual fish. ICES J. Mar. Sci. 63:485-492. Waiwood, K. G., and M. I. Buzeta. 1989. Reproductive-biology of southwest Scotian Shelf haddock ( Melanogrammus aeglefinus). Can. J. Fish. Aquat. Sci. 46(Sl):sl53-sl70. Wakefield, C. B. 2010. Annual, lunar and diel reproductive period-icity of a spawning aggregation of snapper Pagrus auratus (Sparidae) in a marine embayment on the lower west coast of Australia. J. Fish Biol. 77:1359-1378. Walsh, M., and A. D. F Johnstone. 1992. Spawning behavior and diel periodicity of egg pro- duction in captive Atlantic mackerel, Scomber scombrus L. J. Fish Biol. 40:939-950. Wendelaar Bonga, S. E. 1997. The stress response in fish. Physiol. Rev. 77: 591-625. West, G. 1990. Methods of assessing ovarian development in fish- es: a review. Aust. J. Ma^Freshw. Res. 41:199-222. Wootton, R. J. 1998. Ecology of teleost fishes, 2nd ed., 392 p. Kluwer Academic Pubis., Dordrecht, The Netherlands. Errata Page 355: Figure 4 should read as follows: Fishery Bulletin tt0:344-360 (2012). Barlow, Paige F., and Jim Berkson Evaluating methods for estimating rare events with zero-heavy data: a simulation model estimating sea turtle bycatch in the pelagic longlme fishery Corrections: Page 354. The last paragraph in the right column should read as follows: The GLMs only outperformed the delta-lognormal methods in the fully uniform scenario ( Turtles , J uniform7 Sets , ). In this spatial scenario, the GLMs were the most accurate esti- mation method, but they produced more positive outliers. The co-occur- rence clumping scenario (Turtles clump, Sets . , „ ) was the only spa- tial scenario in which the GLMs did not produce more outliers than the delta-lognormal methods. The GLMs were biased lower than the delta-lognormal methods in the co-occurrence clumping scenario ( Turtle , , Sets . , ) and sets- only clumping scenario (Turtles , , u r ° unilorm7 Sets . . ). No substantial differ- clump-sets ence was seen between GLM-P and GLM-NB performance in any spatial scenario. Page 357. The third paragraph in the right column should read as follows: The GLMs were more accurate than the delta-lognormal methods in the fully uniform scenario (Turtles , Sets uniform^ because this spatial sce- nario was the only one that did not violate the GLM-P assumption that counts are independent and randomly distributed in space (McCracken 2004, Sileshi 2006). A O CD 0) > n, ±, etc.), use the symbols provided by the MS Word program and italicize all variables. Do not use photo mode when cre- ating these symbols in the general text. Literature cited section comprises published works and those accepted for publication in peer-re- viewed journals (in press). Follow the name and year system for citation format in the “Literature cited” section (that is to say, citations should be listed al- phabetically by the authors’ last names, and then by year if there is more than one citation with the same authorship. Abbreviations of serials should conform to abbreviations given in Cambridge Scientific Ab- stracts (http://www.csa.com/ids70/serials_source_list. php?db=aquclust-set-c). Authors are responsible for the accuracy and com- pleteness of all citations. Literature citation format: Author (last name, followed by first-name initials). Year. Title of article. Abbreviated title of the journal in which it was published. Always include number of pages. If there is a sequence of citations in the text, list chrono- logically: (Smith, 1932: Green. 1947; Smith and Jones, 1985). If a reference contains URL or DOl code, one or the other (preferably DOI code) is added at the end of the citation. Cite all software and special equipment or chemical solutions used in the study within parenthe- ses in the text (e.g., SAS, vers. 6.03, SAS Inst., Inc., Cary, NC). Guidelines for authors 109 Footnotes are used for all documents that have not been formally peer reviewed and for observations and communications. These types of references should he cited sparingly in manuscripts submitted to the journal. All reference documents, administrative reports, inter- nal reports, progress reports, project reports, contract reports, personal observations, personal communica- tions, unpublished data, manuscripts in review, and council meeting notes are footnoted in 9 pt font and placed at the bottom of the page on which they are first cited. Footnote format is the same as that for formal literature citations. A link to the online source (e.g., [http://www/ , accessed July 2007.]), or the mail- ing address of the agency or department holding the document, should be provided so that readers may ob- tain a copy of the document. Tables are often overused in scientific papers; it is seldom necessary or even desirable to present all the data associated with a study. Tables should not be ex- cessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All un- usual symbols must be explained in the table legend. Other incidental comments may be footnoted with italic numeral footnote markers. Use asterisks only to indi- cate significance in statistical data. Do not type table legends on a separate page; place them above the table data. Do not submit tables in photo mode. Figures must be cited in numerical order in the text. Graphics should aid in the comprehension of the text, but they should be limited to presenting patterns rather than raw data. Figures should not exceed one figure for every four pages of text. Figures must be la- beled with the number of the figure. Avoid placing la- bels vertically (except for the y axis). Figure legends should explain all symbols and abbreviations seen in the figure and should be double-spaced on a separate page at the end of the manuscript. Color is allowed in figures to show morphological differences among spe- cies (for species identification), to show stain reactions, and to show gradations in temperature contours within maps. Color is discouraged in graphs, and for the few instances where color may be allowed, the use of color will be determined by the Managing Editor. • Notate probability with a capital, italic P. • Provide a zero before all decimal points for values less than one (e.g., 0.07). • Capitalize the first letter of the first word in all la- bels within figures. • Do not use overly large font sizes in maps and for units of measurements along axes in figures. • Do not use bold fonts or bold lines in figures. • Do not place outline rules around graphs. • Use a comma in numbers of five digits or more (e.g., 13,000 but 3000). • Place a North arrow and label degrees latitude and longitude (e.g., 170°E) in maps. • Use symbols, shadings, or patterns (not clip art) in maps and graphs. Failure to follow these guidelines and failure to correspond with editors in a timely manner will delay publication of a manuscript. Copyright law does not apply to Fishery Bulletin, which falls within the public domain. However, if an author reproduces any part of an article from Fishery Bulletin in his or her work, reference to source is con- sidered correct form (e.g., Source: Fish. Bull 97:105). Submission Submit manuscript online at http://mc.manuscriptcentral. com/fisherybulletin. Commerce Department authors should submit papers under a completed NOAA Form 25-700. For further details on electronic submission, please contact the Associate Editor, Kathryn Dennis, at kathryn.dennis@noaa.gov When requested, the text and tables should be submit- ted in Word format. Figures should be sent as PDF files (preferred), Windows metafiles, TIFF files, or EPS files. Send a copy of figures in the original software if con- version to any of these formats yields a degraded ver- sion of the figure Questions? If you have questions regarding these guidelines, please contact the Managing Editor, Sharyn Matriotti, at sharyn . matriotti@noaa .gov Questions regarding manuscripts under review should be addressed to Kathryn Dennis, Associate Editor. Fishery Bulletin Subscription form Superintendent of Documents Publications Order Form *5178 1 I YES, please send me the following publications: Subscriptions to Fishery Bulletin for $32.00 per year ($44.80 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. ITS EASY! Please Choose Method of Payment: ] Check Payable to the Superintendent of Documents | | GPO Deposit Account -□ ] VISA or MasterCard Account To fax your orders (202) 512-2104 (Credit card expiration date) (Authorizing Signature) Mail To: U.S. Government Printing Office P.O. Box 979050, St. Louis, MO 63197-9000 Thank you for your order! Also available online at http://bookstore.gpo. gov/actions/GetPublication.do?stocknumber=703-023-00000-2