Investigation of Foraging Habits and Prey Selection by Humpback Whales (Megaptera Novaeangliae) Using Acoustic Tags and Concurrent ﬁsh Surveys

MARINE MAMMAL SCIENCE, 24(3): 516–534 (July 2008) C 2008 by the Society for Marine Mammalogy DOI: 10.1111/j.1748-7692.2008.00193.x

Investigation of foraging habits and prey selection by humpback whales (Megaptera novaeangliae) using acoustic tags and concurrent ﬁsh surveys

BRIANA H. WITTEVEEN University of Alaska Fairbanks, School of Fisheries and Ocean Sciences, 118 Trident Way, Kodiak, Alaska 99615, U.S.A. E-mail: [email protected]

ROBERT J. FOY1 University of Alaska Fairbanks, School of Fisheries and Ocean Sciences, 118 Trident Way, Kodiak, Alaska 99615, U.S.A.

KATE M. WYNNE University of Alaska Fairbanks, School of Fisheries and Ocean Sciences, 118 Trident Way, Kodiak, Alaska 99615, U.S.A.

YANN TREMBLAY University of California Santa Cruz, Long Marine Laboratory, 100 Shaffer Road, Santa Cruz, California 95060-5730, U.S.A.

ABSTRACT Tags containing acoustic time-depth transmitters (ATDT) were attached to four humpback whales near Kodiak, Alaska. Tags allowed for whale dive depths to be recorded in real time. Acoustic and mid-water trawl surveys were conducted concurrent with tagging efforts within the study area to quantify available ﬁsh resources and describe potential prey selection by humpback whales. Recorded dives were grouped through visual assessment and t-tests. Dives that indicated likely foraging occurred at a mean maximum depth of 106.2 m with 62% of dives occurring between 92 m and 120 m. Acoustic backscatter from ﬁsh surveys was attributed to potential humpback prey based on known target strength values and 10 net tows. Capelin comprised 84% of the total potential prey abundance in the region followed by age 0 (12%) and juvenile pollock (2%), and eulachon (<1%). Although hori- zontally segregated in the region, both capelin and age 0 pollock were distributed at depths exceeding 92 m with maximum abundance between 107 m and 120 m. The four-tagged humpbacks were found to forage in areas with greatest capelin

1Current address: NOAA Fisheries, Kodiak Laboratory, Kodiak Fisheries Research Center, 301 Research Court, Kodiak, Alaska 99615, U.S.A.

516 WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 517

densities but bypassed areas of high age 0 pollock abundance. The location and diving behavior of tagged whales suggested that whales were favoring capelin over pollock as a prey source. Key words: acoustic tag, cetacean, dive behavior, foraging, prey, ﬁsh acoustic sampling, humpback whale, Megaptera novaeangliae.

Humpback whales (Megaptera novaeangliae) are considered top-level predators and may consume significant amounts of prey on feeding grounds. Studies have reported that regional prey removal due to cetacean consumption approaches or exceeds re- movals due to commercial fishing (Laws 1977, Laevastu and Larkins 1981, Bax 1991, Markussen et al. 1992, Nordøy et al. 1995, Kenney et al. 1997, Witteveen et al. 2006). Such high levels of consumption can have significant effects on the distribution and abundance of prey species and the structure of marine communities (Perez and McAlister 1993, Kenney et al. 1997, Croll et al. 1998). In the North Pacific, humpback whales are considered generalists in their prey selection and seasonally feed on euphausiids (Thysanoessa spp.) and pelagic fish species up to 30 cm in length, including capelin (Mallotus villosus), Pacific herring (Clupea pallasii), and juvenile walleye pollock (Theragra chalcogramma), (Nemoto 1957, 1959, Krieger and Wing 1984, 1986). Many of these species are targeted by other marine predators and commercial fisheries or are linked to fisheries through complex food webs. Therefore, examining consumption by humpback whales contributes information about complex ecosystem linkages and the long-term sustainability of marine resources (Perez and McAlister 1993, Kenney et al. 1997, Tamura and Ohsumi 2000). To determine the ecological role and the potential for competition among humpback whales and other upper level predators, reliable species-specific estimates of their seasonal abundance, diet composition, and prey consumption are needed. Identifying prey in cetacean diets with certainty requires analysis of the stomach contents of harvested or beached whales (Thompson 1940, Klumov 1963), but both demand the availability of dead animals. Direct observation of prey in the mouths of surface feeding animals can also elucidate diet composition, but because such direct observations are infrequent (for humpback whales in particular) diet assessments may be biased toward prey types in which surface foraging is necessary. Fecal samples from free-swimming whales may also provide dietary insights, but their collection is rare as most fecal types disintegrate and sink. Other methods of determining whale prey have included associations based on whale movements, distribution, and abundance (Payne et al. 1990, Moore et al. 2000) and, more recently, regional sampling of prey populations (Croll et al. 1998, Baumgartner and Mate 2003, Baumgartner et al. 2003, Tynan et al. 2005). While these methods are effective at describing potential prey populations, none can definitively identify, which potential species are being targeted. To more accurately describe targeted prey species, prey sampling needs to occur both in the presence of foraging whales and at depths matching those of their foraging dives. The most efficient means of determining whale dive depth is to tag animals with time-depth recorders (TDR). There are currently three categories of TDR tags used to obtain whale diving behavior data: archival (Croll et al. 1998, Hooker and Baird 1999), satellite radio-linked (Kingsley et al. 2001, Mate and Urban-Ramirez 2003), 518 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008 and acoustic transmitter (Goodyear 1993, Baumgartner and Mate 2003). Of the three types of tags, only the acoustic transmitter allows for continuous dive depth data to be received in real time. Previous efforts have shown the utility of combining tagging and prey sampling techniques to describe prey selectivity and habits of a number of baleen whale species, including blue (Balaenoptera musculus), fin(Balaenoptera physalus), and North Atlantic right whales (Eubalaena glacialis) (Croll et al. 1998, Acevedo-Guitierrez et al. 2002, Baumgartner and Mate 2003). However, studies employing tagging and prey sampling techniques that are temporally and spatially linked have not yet been applied to humpback whales. The nearshore waters northeast of Kodiak Island, Alaska support seasonal and year-round foraging by a large number of piscivorous whale species, including an estimated 157 (95% CI: 114–241) humpback whales (Witteveen et al., in press). To date, cetacean studies conducted near Kodiak Island have examined the distribution, population size, habitat use of, and consumption by local whale populations (Wynne et al. 2005, Witteveen et al. 2006). Consumption by humpback whales northeast of Kodiak Island was estimated to assess the current and historic impact of their predation on local prey populations (Witteveen et al. 2006). Results showed that the potential removal of prey biomass by humpback whales is significant. However, the study assumed humpback whales were feeding in proportion to pelagic prey availability as estimated by acoustic fish surveys and did not, therefore, account for any degree of prey selectivity. Methods that allow for the concurrent observation of whale foraging and available prey populations will improve consumption models by testing hypotheses made about prey selectivity. Our study reports on first efforts to investigate humpback whale foraging habits and prey selectivity by recording real- time whale dive profiles with acoustic tags while concurrently assessing fish prey available to tagged whales through acoustic and mid-water trawl surveys.

MATERIALS AND METHODS Study Area and Period Between 3 and 8 August 2004, humpback whales were tagged and ﬁsh were sampled northeast of Kodiak Island, approximately 10 km east of Spruce Island (Fig. 1). The majority of whale tagging and ﬁsh sampling effort was focused in an area of approximately 144 km2 where a large aggregation of humpback whales (30–50 animals) was consistently found during the study.

Tag Construction, Deployment, and Tracking Each tag consisted of a syntactic foam housing containing both an acoustic time depth transmitter (ATDT) (V22P continuous transmitter, VEMCO, Halifax, NS) and VHF transmitter (MOD125, Telonics, Mesa, AZ) (Fig. 2). An 8-cm diameter suction cup was attached to the tag body with clear tubing. Tags weighed ca. 360 g. The ATDTs had a depth range of 0–204 m with a resolution of 1.2 m and a frequency range of 34.0–48.0 kHz. Tags were deployed with a crossbow and attached to free swimming whales via suction cups from the 8.2-m R/V Soundwave. After a tag was successfully attached, WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 519

Figure 1. Map of northeastern Kodiak Island showing site of humpback whale tagging and pelagic ﬁsh sampling in 144 km2 diagonally striped area. The contour is the 100-m isobath. the whale was closely followed for either the duration of the attachment period or until conditions (i.e., darkness, sea state) prevented further observation. Acoustic signals from ATDTs were received in real time with a vessel-mounted directional hydrophone (VEMCO VR-60) and recorded into a text ﬁle. Signals could be received at distances up to 1 km. In instances when this distance was exceeded or the acoustic signal was lost, the tagged whale was located by listening for the VHF signal, which could be received when the whale was at the surface and within several kilometers. VHF signals were also used to locate tags that had been shed. Dive data were time- linked to the position (latitude and longitude) of R/V Soundwave, which due to its constant close proximity to the whale, was assumed to represent the track of the 520 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008

Figure 2. Image of tags deployed by crossbow. tagged whale. The location and range of dive depths observed for tagged whales were reported in real time to the accompanying prey survey vessel to coordinate acoustic and net sampling in the vicinity of tagged whales.

Analysis of Dive Data and Dive Categorization Raw dive data were analyzed using algorithms contained within the IKNOS toolbox (Y. Tremblay, unpublished data). Each dive was designated with a code (Type 1 through 6) that was visually assigned based on the general shape of the dive (Fig. 3). Dives in which data were missed (i.e., signal interruption prevented portions of the dive from being recorded) were designated as Type 9 and were not considered for additional analyses. The IKNOS toolbox calculates a total of 14 dive parameters for each dive, including maximum dive depth, dive duration, and intra-depth zone (IDZ) dives (Table 1). The IDZ dives were classiﬁed as consecutive dives that re- turned to the same depth zone (Tremblay and Cherel 2000). The percentage of IDZ dives (%IDZ) was deﬁned as the number of IDZ dives divided by the total number of dives for each of the six dive categories and was the only parameter calculated independently of the IKNOS toolbox. Each tagged whale was considered an inde- pendent observation; therefore, means for all dive parameters were calculated for WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 521

Figure 3. General shape patterns and associated codes used for initial categorization of dive type. each tagged whale and then averaged across whales for reporting and interpreting results. Following dive categorization, mean values for dive parameters were tested for differences using paired-sample t-tests (Zar 1984) in which signiﬁcance was determined at = 0.05.

Correlations Between Dive Parameters Some dive parameters were also used to test correlations between these parameters and maximum dive depth for comparison with previous studies of humpback whale foraging (Dolphin 1987a, b). Only parameters examined in both Dolphin’s study and

Table 1. Description and units of dive parameters from the INKOS toolkit (Trembly, unpublished data).

Parameter Unit Description No. of dives Total number of dives No. of dives Total number of dives within the intra-depth zone in IDZ % IDZ Percentage of total dives in the intra-depth zone Maxdepth m Mean maximum depth reached Dduration s Total duration of the dive Botttime s Duration spent at the bottom during the dive DescTime s Time spent traveling from the surface to the bottom of the dive DescRate ms−1 Rate of descent AscTime s Time spent traveling from the bottom to the surface of the dive AscRate ms−1 Rate of ascent PDI Post Dive Interval. Times spent at the surface between successive dives VOD Number of vertical oscillations greater than tag depth resolution counted during the descent phase VOB Number of vertical oscillations greater than tag depth resolution counted during the bottom phase VOA Number of vertical oscillations greater than tag depth resolution counted during the ascent phase BottDist m Sum of vertical distance traveled during the bottom phase BottRange m Range of depth transited during the bottom time Efﬁciency Botttime/(Dduration + PDI) 522 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008 this study were selected for correlation analysis and were thus limited to ascent and descent rates, time to ascent and descent, and dive duration.

Prey Surveys and Sampling A commercial fishing vessel, the F/V Alaskan, was used to conduct acoustic and mid-water trawl fish surveys. Two types of fish acoustic surveys were initiated once whales were sighted and tagged. The first was a traditional zigzag survey of the entire region occupied by the feeding aggregation of humpback whales to estimate fish species composition and abundance. The second type of survey was a focal-follow acoustic survey conducted in two areas within the study region used by specific tagged whales. The focal-follow survey immediately followed the tracks of tagged whales to provide a finer scale assessment of fish density targeted by whales within each area. Fish abundance and species composition were estimated acoustically and assessed with mid-water trawl nets. Acoustic surveys were conducted during daylight hours when fish were vertically aggregated and concurrent to collection of whale dive data. Pelagic volume backscatter was estimated with a Simrad EK60 echo sounder using a hull-mounted split-beam 38 kHz transducer with a 12◦ beam angle, 1.024 ms pulse length, and 0.5 s−1 ping rate. The echo sounder system was calibrated with a copper sphere of known acoustic backscatter (Foote et al. 1987). The vessel speed during acoustic surveys was approximately 13.2 km/h to maintain a constant engine rpm throughout the surveys. Raw acoustic backscatter data were echo-integrated in 185-m intervals along track lines between 5 m of the surface and 0.5 m of the bottom using Echoview 3.5 software (SonarData Pty Ltd, Tasmania, Australia). Trawl nets (headrope/footrope length: 39 m; vertical opening: 20 m; 2.22 cm codend liner mesh) designed to target pelagic schooling fish were deployed for 10– 20 min to determine the species composition and size distribution of the fish species responsible for patterns of acoustic backscatter in a localized area. The catchability of the main species encountered in the study was assumed to be equal in all trawls. All fish were separated by species and counted; fish of each species were randomly subsampled for length and weight measurements. Fish density was estimated by apportioning the acoustic backscatter based on species proportions from the trawls and on expected species- and size-specific target strengths (TS) (Simmonds and MacLennan 2005). Species composition and species lengths were applied to acoustic backscatter within a localized region near the tow that had similar visual characteristics (school shape, size, and density) and TS distribution. The TS-to-length equation used for this study was

TS = 20 log L T + b where LT was the total fish length and the constant b was −69.3 for capelin, −67.2 for walleye pollock, and −84.5 for eulachon (Thaleichthys pacificus) (Gauthier and Horne 2004). The analysis area, totaling 144 km2, was defined as the area bounding both acoustic surveys (region and area) and within 2.0 km of all whale tracklines during the study. The depth of whale feeding was defined as the mean humpback whale feeding depth ±1 SD (see Analysis of Dive Data and Dive Categorization). Over the entire region, densities (fish/km2) and biomasses (kg/km2) for each fish species were averaged over all 185-m trackline intervals within the mean depth range of whale feeding. In areas 1 and 2, mean densities (fish/km2) of each species WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 523 were estimated in 14 m depth bins over the entire water column where focal follow surveys occurred. Only species less than 30 cm in length, our assumed maximum size of humpback whale prey (Nemoto 1959), were included in these analyses. In areas 1 and 2, prey density and the number of humpback dives in each depth bin were then examined for correlation.

RESULTS Tagging and Tracking Tags were attached to free-swimming humpback whales on four occasions and tracked for a total of nearly 17 h. Tags 2 and 3 were shed after 5.78 h and 1.91 h, respectively. Tracking of whales with tags 1 and 4 was abandoned after 6.22 h and 2.82 h, respectively, due to worsening sea conditions. These tags were subsequently shed and recovered. Following initial disturbance from vessel approach and tag deployment, attachment of the suction cup tag did not appear to affect the behavior of whales.

Dive Categorization A total of 122 dives made by four individual humpback whales were recorded during the study period. Of these dives, 18 were discarded as Type 9 dives and the remaining 104 were classiﬁed as shown in Table 2.

Dive Parameters Estimates and Comparisons Based on the theory that foraging behavior of diving animals seeks to maximize time spent at depth while minimizing travel and recovery time, visual assessment of dive data suggested Type 3 and Type 5 dives represented foraging dives (Houston and Carbone 1992; Thompson et al. 1993; Carbone and Houston 1994, 1996; Carbone et al. 1996; Kramer 1998; Thompson and Fedak 2001). Paired t-tests (t3 = 3.18) between Type 3 and Type 5 dives against all remaining dive types (Types 1, 2, 4, and 6) revealed significant differences in mean values of six parameters: Dduration, Botttime, AscRate, VOB, Bottdist, and Efficiency.In addition to these six parameters, Type 5 dives were found to be significantly different from Types 1, 2, 4, and 6 for

Table 2. Counts of dives, by individual tag, for each of seven dive categories. Categories are based on general dive shape and Type 9 dives were discarded prior to analysis (see Fig. 3).

Dive category Tag no. 1 2 3 4 5 6 9 Total 1 452 615106 48 23011231635 3 101063314 43113131325 Total 11 6 5 10 57 15 18 122 Total analyzed 104 524 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008 mean values of%IDZ (P < 0.001) and BottRange (P = 0.008), while Type 3 dives differed for mean values of Maxdepth (P = 0.014) and DescRate (P = 0.004). Based on results of paired t-tests, Type 3 and Type 5 dives were classiﬁed as foraging while the remaining four dive types were considered to represent non- foraging behavior, including travel, rest, or searching for prey. Type 3 and Type 5 dives were combined to determine mean depth of humpback whale foraging dives and are described collectively as foraging dives from this point forward. Foraging dives accounted for 59.6% of all dives (62 of 104). Foraging dives had a mean maximum dive depth range of 81.9 m to 117.5 m, while non-foraging dives ranged from 23.2 m to 89.63 m. Foraging dives were 1.7 times deeper (106.2, SD 14.17 m) on average than non-foraging dives (70.9, SD 27.40 m) (Table 3). The

Table 3. Means of dive parameters for non-foraging and foraging (Type 3 and Type 5) dives averaged across four tagged humpback whales. Means are shown with SD below in italics. The P values for paired sample t-tests (t3 = 3.18 for all tests) are shown with signiﬁcant results indicated by bold font.

Dive type Non-foraging Type 3 Type 5 Parameter n = 42 n = 5 Pn= 57 P % IDZ 7.9% 5.5% 0.649 68.5% <0.001 0.10% 0.04% 20.14% Maxdepth 64.4 112.1 0.014 100.4 0.060 33.33 5.14 18.75 Dduration 265.4 391.9 0.049 410.1 0.008 90.65 142.26 35.83 Botttime 96.8 231.3 0.007 244.2 <0.001 56.94 122.91 16.32 DescTime 72.6 78.8 0.774 78.2 0.788 40.22 18.87 8.20 DescRate 0.8 1.4 0.004 1.1 0.061 0.32 0.40 0.32 AscTime 96.0 81.9 0.483 87.7 0.685 37.78 15.19 24.03 AscRate 0.6 1.3 <0.001 1.1 0.008 0.30 0.19 0.30 PDI 69.2 94.4 0.258 105.0 0.090 37.79 35.96 16.45 VOD 1.9 1.0 0.250 1.2 0.378 1.54 0.00 0.15 VOB 5.2 15.0 0.011 13.3 0.006 4.89 8.79 2.09 VOA 2.0 1.0 0.168 1.1 0.228 1.39 0.00 0.27 BottDist 27.4 47.7 0.042 107.1 <0.001 15.70 8.18 19.46 BottRange 16.3 17.3 0.820 29.8 0.008 7.91 8.18 6.66 Efﬁciency 0.24 0.45 0.015 0.47 0.004 0.130 0.142 0.016 WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 525

Number of Dives 0 5 10 15 20 25 30

<8 9-22 Non-Foraging Dives 23-36 Foraging Dives 37-50 51-64 65-78 79-92

Depth (m) Depth 93-106 107-120 121-134 135-148 149-162 163-176

Figure 4. Frequency of occurrence of mean maximum dive depths in 14 m (1 SD of mean maximum dive depth) depth bins for dives recorded from tagged humpback whales. distribution of foraging dives was such that a majority (61.3%) fell within mean maximum depth range (107–120 m) (Fig. 4). Mean values for Dduration, Botttime, BottDist, VOB, and Efﬁciency were 1.5, 2.4, 1.7, 2.9, and 1.9 times greater for foraging dives compared to non-foraging dives, while means for both AscTime and DescTime were nearly identical for both foraging and non-foraging dives (Table 3). Type 3 and Type 5 dives were differentiated from one another by the BottRange parameter (paired t-test; t3 = 3.18, P = 0.02). The mean value of this parameter was nearly twice as large for Type 5 dives (29.8, SD 6.66 m) compared to Type 3 dives (17.3, SD 8.18 m).

Correlations Between Dive Parameters and Maximum Depth of Dive There were no signiﬁcant correlations found between maximum dive depth and any of the tested parameters. However, certain comparisons suggested that a positive relationship may exist, such as maximum depth vs. ascent rate (r = 0.84, F = 4.83, P = 0.159) and descent time (r = 0.90, F = 8.36, P = 0.102) (Table 4).

Prey Surveys and Sampling On 7 August 2004, numerous feeding humpback whales, including tag 2, were encountered approximately 25 km northeast of Kodiak Island. Based on the apparent distribution of whales, a 36 km zigzag survey design was employed to assess ﬁsh abundance within the region occupied by this feeding aggregation (approximately 144 km2) (Fig. 4). Throughout the day, ﬁsh acoustic data were collected within 1–3h 526 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008

Table 4. Comparison of correlation values found between maximum depth of dive and other dive parameters for Dolphin (1987a, b) and this study. Note that values of r for this study were not found to be signiﬁcant (P > 0.05)

r Parameter Dolphin This study Dduration 0.91 0.69 DescTime 0.96 0.89 DescRate 0.24 0.38 AscTime 0.97 0.49 AscRate 0.7 0.84 of the whale dive data recorded within the region. On 8 August 2004, humpback whales were again observed in the same region. To identify a fine scale pattern of potential prey abundance, two tagged humpback whales (tags 3 and 4) were closely followed while assessing acoustic backscatter. Within two specific areas (Fig. 5), focal- follow acoustic surveys were initiated immediately after a whale was tagged so that prey data were collected within 10 min of whale dive data collection. Potential humpback prey, including capelin, eulachon, and pollock <30 cm (age 0 and juveniles), were identified from both surveys (region and area) based on 10 net tows and known TS values (Table 5). The remaining backscatter was either >30 cm and assumed too large to be humpback prey or was not identified with our sampling gear and assumed to be mostly invertebrates. Ten tows were assumed to adequately assess all patterns of acoustic backscatter given the limited complexity in species composition encountered over an area of this size. All tows were deployed with the headrope depth between 60 m and 150 m corresponding to whale dive depths and maximum acoustic backscatter. Backscatter was minimal or above the TS threshold typical for fish species in the rest of the water column. The abundance of potential humpback prey (fish <30 cm) found between depths of 92 m and 120 m (range of whale dive depths) was estimated from these surveys. Capelin and age 0 pollock dominated the fish encountered at these depths and were caught in different areas of the region surveyed (Fig. 5). Capelin were caught where total bottom depths were between 100 m and 150 m, age 0 pollock were caught over total bottom depths of 150 m and greater, and eulachon were caught throughout the region. A total of 1,276 fish were measured for length and 224 fish for weight (Table 6). Age 0 pollock were the smallest potential whale prey encountered (mean length = 7.0 cm, mean weight = 2.3 g) followed by capelin (mean length = 10.9 cm, mean weight = 6.2 g), eulachon (mean length = 16.0 cm, mean weight = 21.2 g), and juvenile pollock (mean length = 17.0 cm, mean weight = 36.8 g). The maximum densities (fish/km2)offish per 185 m trackline interval ranged from 1,000 eulachon/km2 to 53,329,000 capelin/km2 while the maximum biomass densities ranged from 13 kg eulachon/km2 to 330,637 kg capelin/km2 over the entire region (Fig. 5). Abundance estimates were also determined in two discrete areas used by tagged whales (Fig. 5). In both areas, capelin dominated the total fish abundance. In area 1 maximum capelin abundance was 17 million fish/km2 (Fig. 6). Mean capelin abundance was greatest at depths exceeding 92 m with a maximum abundance of 918,298 fish/km2 at depths between 107 m and 120 m (Fig. 7a). No pollock <30 cm were caught in the direct vicinity of feeding whales WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 527

Figure 5. Whale tracklines from 3, 7, and 8 August 2004 relative to the distribution of fish species (<30 cm) found between depths 92 m and 120 m. Solid circles are tow locations for groundtruthing acoustic fish data. The largest vertical bar represents 53 million fish/km2. The contour in the inset map is the 100-m isobath. in area 1 and minimal eulachon abundances were found in depths exceeding 107 m. The acoustic backscatter (m2/km2) that was not attributed to fish in area 1 was distributed throughout the water column with 22% found between 9 m and 22 m, 32% found between 107 m and 120 m, and 44% found deeper. These acoustic targets were believed to be smaller plankton in the surface waters and potentially larger plankton such as euphausiids in the deeper water based on their relatively 528 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008

Table 5. Data for each ﬁsh tow including the date, depth range covered by the tow, species composition, mean lengths and weights, and proportion of ﬁsh <30 cm in each tow used to apportion the acoustic backscatter. Tow 7 was not included in the analyses due to net fouling.

Mean Average Depth length weight Proportion Tow Date range Species (cm) (g) (%) 1 7 August 2004 100–120 capelin 10.9 6.2 96 pollock (age 0) 7.0 2.3 3 eulachon 16.8 24.9 1 2 7 August 2004 125–150 capelin 10.6 5.6 98 pollock (age 0) 6.5 1.9 2 3 7 August 2004 100–120 capelin 8.4 2.5 100 4 7 August 2004 60–120 capelin 10.9 6.2 100 5 7 August 2004 100–125 pollock (age 0) 6.5 1.9 100 6 7 August 2004 120–150 pollock (juvenile) 17.0 36.7 90 eulachon 16.0 21.1 10 7 8 August 2004 na na na na na 8 8 August 2004 100–125 capelin 11.1 6.6 100 eulachon 15.1 17.7 9 8 August 2004 110–140 capelin 8.8 3.0 100 eulachon 19.9 41.6 10 8 August 2004 90–120 capelin 9.9 4.5 100 smaller mean volume backscattering strengths (Simmonds and MacLennan 2005). Area 2 was also dominated by capelin where the highest densities of 305,030 fish/km2 were estimated between 107 m and 120 m (Fig. 7b). Smaller abundances of juvenile pollock and eulachon were also caught in area 2 at or below the average whale dive depth. The acoustic backscatter (m2/km2) that was not attributed to fish in area 2 was found between 9 m and 22 m (49%), deeper than 121 m (46%) and had relatively smaller mean volume backscattering strengths compared to the areas where fish were caught. Correlation analyses showed strong relationships between mean prey depths and mean dive depths. Specifically, mean capelin depth was positively correlated with

Table 6. Mean length and weight used to calculate abundance and biomass densities for each pelagic species found at depths between 92 m and 120 m in the entire study region.

Length (cm) Weight (g) Abundance Biomass (count×103/km2) (kg/km2) Mean Species Mean SD n Mean SD n Mean (maximum) (maximum) Capelin 10.9 1.87 500 6.2 3.17 100 779 (53,329) 4,689 (330,637) Pollock 7.0 0.73 500 2.3 0.58 48 106 (14,641) 251 (33,674) age 0 Pollock 17.0 1.50 250 36.8 9.24 50 17 (601) 6,408 (220,638) juvenile Eulachon 16.0 2.64 26 21.2 13.58 26 <1 (1) <1 (13) Total 902 11,347 WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 529

capelin1 Whale tracklines

00.511.520.25 Km Figure 6. Whale tracklines from 8 August 2004 relative to the distribution of fish species (<30 cm) found between depths 92 m and 120 m in area 1. The largest vertical bar represents 17 million fish/km2. mean depth of foraging dives in area 1 (r = 0.91, P < 0.001, n = 13) and area 2 (r = 0.91, P < 0.001, n = 13). Moderate, but not significant, correlation was also found between mean capelin depth and mean depth of non-foraging dives in area 1 (r = 0.46, P = 0.113, n = 13), but not in area 2 (r =−0.21, P = 0.377, n = 13). A positive correlation was also found between mean age 0 pollock depths and mean depths of foraging dives in area 2 (r = 0.91, P < 0.001, n = 13).

DISCUSSION Type 3 and Type 5 dives were significantly different than the other four dive types and were indicative of foraging behavior because of high mean values of Dduration, Botttime, VOB, BottDist, and Efficiency. When considered together, these parameters indicate that foraging dives are longer in total duration and that a greater portion of total time and distance is spent at depth and support prior assumptions that max- imizing time at spent at depth in synonymous with foraging behavior (Thompson and Fedak 2001). In particular, higher Efficiency values are consistent with increased relative bottom time and reduced transit time to and from the surface, an assumed energetic benefit while foraging (Thompson et al. 1993). Type 3 and Type 5 dives are differentiated from one another by the BottRange parameter, with mean values nearly twice as large for Type 5 dives. This is due to 530 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008

(a) Area 1

Capelin count * 104/km2 Fish count * 104/km2 0 20406080100 0 0.01 0.02 0.03 0.04 0.05 <8 <8 9-22 9-22 eulachon 23-36 23-36 37-50 37-50 51-64 51-64 65-78 65-78 79-92 79-92 93-106 93-106 107-120 107-120 Depth bins (m) 121-134 Depth bins (m) 121-134 135-148 135-148 149-162 149-162 163-176 163-176

(b) Area 2

Capelin count * 104/km2 Fish count * 104/km2 0 5 10 15 20 25 30 35 0 0.2 0.4 0.6 0.8 <8 <8 9-22 9-22 eulachon 23-36 23-36 pollock age-0 37-50 37-50 51-64 51-64 65-78 65-78 79-92 79-92 93-106 93-106 107-120 107-120 Depth bins (m) Depth bins (m) 121-134 121-134 135-148 135-148 149-162 149-162 163-176 163-176 Figure 7. Capelin vs. other pelagic fish abundance in 14 m depth bins where humpback whales were actively feeding in (a) Area 1 and (b) Area 2. Note that the x-axis scale varies among plots. vertical lunging occurring at depth in Type 5 dives (Fig. 3). Vertical lunges at depth have been used to imply foraging behavior in other balaenopterid species (Croll et al. 2001, Acevedo-Guitierrez et al. 2002, Goldbogen et al. 2006). Therefore, if both types do represent foraging, Type 3 dives likely indicate straight horizontal foraging, while Type 5 dives indicate vertical lunging behavior. It is possible that horizontal lunges were occurring in both dive types, but the tags used in this study were not capable of recording those movements. Significant correlations between dive parameters were expected as has been found in humpback whale diving studies in Southeast Alaska (Dolphin 1987a, b). The lack of significant correlations in this study is likely an artifact of small sample size. A post hoc power analysis based on variation among tags from this study (P = 0.6) shows that increasing the current sample size from four to eight tags would be required to detect significant correlations. Although not significant, positive correlations were found between depth of dive and dive duration (r = 0.69, P = 0.307), descent time = 0.90, P = 0.102), and ascent rate (r = 0.84, P = 0.159) (Table 4). Similarly, depth of dive was positively correlated to dive duration (r = 0.91) and descent time (r = 0.96) of humpbacks in southeast Alaska (Dolphin 1987a, b). Horizontal humpback whale distributions corresponded with pelagic forage fish abundance in the study region (Figs. 5, 6). Examination of the fish abundance along WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 531 the tracks of the tagged whales (Figs. 5, 6) shows humpbacks foraged in areas with greatest capelin densities but bypassed areas of high age 0 pollock abundance. Because four separate whales foraged in areas of high capelin abundance, it appears humpbacks were selecting capelin over pollock as a prey source in the study region. When capelin and age 0 pollock co-occurred spatially, as in area 2, humpback whales may have been feeding on both species in proportion to their relative densities. This assumes that whales were not targeting single-species aggregations of prey that were smaller than detectable in the 185-m trackline intervals by which the acoustic data were averaged. Further, the conclusion that humpback foraging was focused on capelin is based on the premise that the whales’ tracks were accurately represented by the research vessel’s tracks. Although undoubtedly present, the error associated with this method is likely to be small during the tracking time due to the range limitations of the ATDT. Vertical humpback whale distributions corresponded with pelagic forage fish abundance in the study areas. Type 3 and Type 5 dives occurred at a mean maximum depth of 106.2 m with the majority of dives occurring between 92 m and 120 m, a depth range that corresponded to the highest numbers of available forage fish in areas 1 and 2 (Figs. 4, 7). Specifically, foraging humpback dive depths were highly correlated to depths of maximum capelin density in both areas 1 and 2. While non-foraging mean maximum dive depths were moderately correlated to capelin depths in area 1, they were negatively correlated in area 2. Foraging humpback dive depths were also correlated to age 0 pollock depths in area 2 but age 0 pollock densities in this area were much lower than in the rest of the region (Fig. 5). It should be noted that the capelin distribution was closely associated with higher bathymetric relief, which may influence local oceanographic conditions and may have affected whale dive patterns. Additionally, it is recognized that the sample size of tagged whales was low and may have therefore led to biased conclusions about prey selection. However, the occurrence of humpback whales in the Atlantic Ocean was shown to be strongly correlated to both the presence and year-class strength of capelin (Whitehead and Carscadden 1985, Piatt et al. 1989, Christensen et al. 1990). If humpback whales were favoring capelin within the study region it may be the result of several factors. An increased selection of capelin may simply be due to the fact that these fish were present in higher numbers than juvenile or age 0 pollock. Although the total density by weight of juvenile pollock was greatest in the region, the capelin density by number was greater than the other species in the region. It is generally believed that whales have some threshold level at which foraging becomes productive. For baleen whales, it is has been shown that this level corresponds to the number of fish schools but it is unknown whether density or biomass of the schools determines the threshold (Piatt and Methven 1992). Other factors that may influence selectivity may include the energetic benefit associated with consumption of capelin over juvenile pollock. Capelin in the Kodiak Island region are known to have a higher lipid content than pollock of a similar size class (Payne et al. 1999, Wynne et al. 2005). Regardless of the mechanism, prey selection is probable and should be further assessed to accurately estimate the total prey consumption of humpback whales (Witteveen et al. 2006). The presence or absence of large zooplankton was recorded during the acoustic fish surveys, but their abundance was not quantitatively assessed during this study. Because small targets cannot be accurately detected at 38 kHz, we assumed large zooplankton were a large component of the acoustic backscatter that was not identified as juvenile or adult fish. The majority of this backscatter occurred at depths greater than maximum whale dive depths, suggesting tagged whales were not diving through 532 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008 the deep zooplankton layer. Although it as been suggested that euphausiids comprise 5%–30% of the humpback whale diet (Perez and McAlister 1993, Kenney et al. 1997), our data suggest zooplankton were not targeted by whales in this study. Future prey sampling efforts need to include assessment of zooplankton distribution and abundance so their importance within the humpback whale diet can be discussed and evaluated against other potential prey species. Tracking humpback whales using ATDTs proved to be a useful means of recording dive patterns. The greatest advantage of ATDTs over other tagging methodologies was the ability to know dive depth in real time and thus concurrently assess prey resources available to foraging humpback whales on a scale not previously studied. Future efforts of this nature will continue to improve understanding of whale diets and foraging habits.

ACKNOWLEDGMENTS The authors thank the captain and crew of the F/V Alaskan, as well as Jordy Thomson, Casey Clark, and Petra Reimann for their assistance in the field. We thank Robin Baird and Gregg Schorr for graciously sharing their knowledge of tag deployment and construction, as well as Don Croll and Kelly Newton for their guidance on field methodology. We further thank three anonymous reviewers for their comments and help in improving this article. Financial support for this project was provided by NOAA Grant #NA16FX1270 and # NA04NMF4390158 to the University of Alaska Gulf Apex Predator-Prey project. All whale research was conducted under provisions of NMFS Scientific Research Permit No. 1049-1718- 00 and UAF IACUC#02-38. Fish collections were conducted under UAF IACUC#04-21, ADF&G#CF-04-031, and NMFS LOA#2004-06 permits.

LITERATURE CITED

ACEVEDO-GUITIERREZ, A., D. A. CROLL AND B. R. TERSHY. 2002. High feeding costs limit dive time in the largest whales. The Journal for Experimental Biology 205:1747–1753. BAX, N. J. 1991. A comparison of the fish biomass flow to fish, fisheries, and mammals in six marine ecosystems. ICES Marine Science Symposium 193:217–224. BAUMGARTNER,M.F.,AND B. R. MATE. 2003. Summertime foraging ecology of North Atlantic right whales. Marine Ecology Progress Series 264:124–135. BAUMGARTNER, M. F., T. V. N. COLE,R.G.CAMPBELL,G.J.TEEGARDEN AND E. G. DURBIN. 2003. Associations between North Atlantic right whales and their prey, Calanus finmarchicus, over diel and tidal time scales. Marine Ecology-Progress Series 264:155– 166. CARBONE, C., AND A. I. HOUSTON. 1994. Patterns in the diving behaviour of the pochard, Aythya ferina: A test of an optimality model. Animal Behaviour 48:457–465. CARBONE, C., AND A. I. HOUSTON. 1996. The optimal allocation of time over the dive cycle. An approach based on aerobic and anaerobic respiration. Animal Behaviour 51:1247– 1255. CARBONE, C., J. J. DELEEUE AND A. I. HOUSTON. 1996. Adjustments in the diving time budgets of tufted duck and pochard: Is there evidence for a mix of metabolic pathways? Animal Behaviour 51:1257–1268. CHRISTENSEN, I., T. HAUG AND N. OIEN. 1990. A review of the distribution, migrations, food, reproduction, exploitation and present abundance of humpback whales (Megaptera novaeangliae) in the Northeast Atlantic. International Council for the Exploration of the Sea 10:1–14. CROLL, D. A., B. R. TERSHY,R.P.HEWITT,D.A.DEMER,P.C.FIEDLER,S.E.SMITH,J.M. ARMSTRONG,T.POPP,V.R.KEIKHEFER,J.URBAN LOPEZ AND D. GENDRON. 1998. WITTEVEEN ET AL.: HUMPBACK FORAGING HABITS 533

An integrated approach to the foraging ecology of marine birds and mammals. Deep-Sea Research 45:1353–1371. CROLL, D. A., A. ACEVEDO-GUTIERREZ´ ,B.R.TERSHY AND J. URBAN´ -RAMı´REZ. 2001. The diving behavior of blue and fin whales: Is dive duration shorter than expected based on oxygen stores? Comparative Biochemistry and Physiology Part A 129:797–809. DOLPHIN, W.F. 1987a. Ventilation and dive patterns of humpback whales, Megaptera novaeangliae, on their Alaskan feeding grounds. Canadian Journal of Zoology 65:83–90. DOLPHIN, W.F.1987b. Dive behavior and estimated energy expenditure of foraging humpback whales in southeast Alaskan. Canadian Journal of Zoology 65:354–362. FOOTE, K. G., H. KNUDSEN,G.VESTNES,D.MACLENNAN AND E. SIMMONDS. 1987. Cal- ibration of acoustic instruments for fish density estimation: A practical guide. ICES Cooperative Research Report 144, 69 pp. GAUTHIER, S., AND J. K. HORNE. 2004. Acoustic characteristics of forage fish species in the Gulf of Alaska and Bering Sea based on Kirchhoff-approximation models. Canadian Journal of Fisheries and Aquatic Sciences 61:1839–1850. GOLDBOGEN, J. A., J. CALAMBOKIDIS,R.ESHADWICK,E.M.OLESON,M.A.MCDONALD AND J. A. HILDEBRAND. 2006. Kinematics of foraging dives and lunge-feeding in fin whales. The Journal of Experimental Biology 209:1231–1244. GOODYEAR, J. D. 1993. A sonic/radio tag for monitoring dive depths and underwater movements of whales. Journal of Wildlife Management 57:503–513. HOOKER,S.K.,AND R. W. BAIRD. 1999. Deep-diving behavior of the northern bottlenose whale, Hyperoodon ampulatus (Cetacea: Ziphiidae). Proceedings of the Royal Society of London B Biological Sciences 266:671–676. HOUSTON, A. I., AND C. CARBONE. 1992. The optimal allocation of time during the diving cycle. Behavioral Ecology 3:255–265. KENNEY, R. D., G. P. SCOTT,T.J.THOMPSON AND H. E. WINN. 1997. Estimates of prey consumption and trophic impacts of cetaceans in the USA Northeast continental shelf ecosystem. Journal of Northwest Atlantic Fishery Science 22:55–171. KINGSLEY, M. C. S., S. GOSSELIN AND G. A. SLENO. 2001. Movements and dive behavior of belugas in northern Quebec. Arctic 54:262–275. KLUMOV, S. K. 1963. Food and helminth fauna of whalebone whales in the main whaling regions of the world ocean. Trudy Instituta Okeanologii 71:94–194. KRAMER, D. L. 1998. The behavioral ecology of air breathing by aquatic animals. Canadian Journal of Zoology 66:89–94. KRIEGER,K.J.,AND B. L. WING. 1984. Hydroacoustic surveys and identification of humpback whale forage in Glacier Bay, Stephens Passage, and Frederick Sound, southeastern Alaska, Summer 1983. NOAA Technical Memorandum NMFS F/NWC-66. 60 pp. KRIEGER, K. J., AND B. L. WING. 1986. Hydroacoustic monitoring of prey to determine humpback whale movements. NOAA Technical Memorandum NMFS F/NWC-98. 62 pp. LAEVASTU,T.,AND H. A. LARKINS. 1981. Marine Fisheries Ecosystems, Its Quantitative Evaluation in Management. Fishing News Books, Farnham, Surrey, UK. LAWS, R. M. 1977. Seals and whales in the Southern Ocean. Philosophical Transactions of the Royal Society of London Series B Biological Sciences 279:81–96. MARKUSSEN, N. H., M. RYG AND C. LYDERSEN. 1992. Food consumption of the NE Atlantic minke whale (Balaenoptera acutorostrata) population estimated with a simulation model. ICES Journal of Marine Science 49:317–323. MATE, B., AND J. URBAN-RAMIREZ. 2003. A note on the route and speed of a gray whale on its northern migration from Mexico to central California, tracked by satellite-monitored radio tag. Journal of Cetacean Research and Management 5:155–157. MOORE, S. E., J. M. WAITE,L.L.MAZZUCA AND R. C. HOBBS. 2000. Mysticete whale abundance and observations of prey associations on the central Bering Sea shelf. Journal of Cetacean Resource Management 2:227–234. NEMOTO, T. 1957. Foods of baleen whales in the northern Pacific. Scientific Reports of the Whales Research Institute, Tokyo 12:33–89. 534 MARINE MAMMAL SCIENCE, VOL. 24, NO. 3, 2008

NEMOTO, T. 1959. Foods of baleen whales with reference to whale movements. Scientific Reports of the Whales Research Institute, Tokyo 14:244–290. NORDøY, E. S., L. P. FALKOV, P.-E. MATENSSON˚ AND A. S. BLIX. 1995. Food requirements of northeast Atlantic minke whales. Pages 307–317 in A. S. BLIX,L.WALLøE, and Ø.ULLTANG, eds. Whales, seals, fish, and man. Elsevier Science, Amsterdam, The Netherlands. PAYNE, P. M., D. N. WILEY,S.B.YOUNG,S.PITTMAN,P.J.CLAPHAM AND J. W. JOSSI. 1990. Recent fluctuations in the abundance of baleen whales in the southern Gulf of Maine in relation to changes in selected prey. Fishery Bulletin 88:687–696. PAYNE, S. A., B. A. JOHNSON AND R. S. OTTO. 1999. Proximate composition of some north-eastern Pacific forage species. Fisheries Oceanography 8:159–177. PEREZ, M. A., AND W. B. MCALLISTER. 1993. Estimates of food consumption by marine mammals in the eastern Bering Sea. U.S. Department of Commerce, NOAA Technical Memorandum NMFS-AFSC-14. PIATT,J.F.,AND D. A. METHVEN. 1992. Threshold foraging behavior of baleen whales. Marine Ecology Progress Series 84:205–201. PIATT, J. F.,D. A. METHVEN,A.E.BURGER,R.L.MCLAGAN,V.MERCER AND E. CREELMAN. 1989. Baleen whales and their prey in a coastal environment. Canadian Journal of Zoology 67:1523–1530. SIMMONDS, E. J., AND D. N. MACLENNAN. 2005. Fisheries Acoustics: Theory and practice. Blackwell Publishing, Oxford, UK. TAMURA,T.,AND S. OHSUMI. 2000. Regional assessments of prey consumption by marine cetaceans in the world. International Whaling Commission document SC/52/E6. THOMPSON, R. J. 1940. Analysis of stomach contents taken during the years 1937 and 1938 from the North Pacific. M.Sc. thesis, University of Washington, Seattle, WA. 82 pp. THOMPSON, D., AND M. A. FEDAK. 2001. How long should a dive last? A simple model of foraging decisions by breath-hold divers in a patchy environment. Animal Behaviour 61:287–296. THOMPSON, D., A. R. HILBY AND M.A. FEDAK. 1993. How fast should I swim? Behavioral implications of diving physiology. Symposia of the Zoological Society of London 66:223– 232. TREMBLAY,Y.,AND Y. C HEREL. 2000. Benthic and pelagic dives: A new foraging behavior in rockhopper penguins. Marine Ecology Progress Series 204:257–267. TYNAN, C. T., D. G. AINLEY,J.A.BARTH,T.J.COWLES,S.D.PIERCE AND L. B. SPEAR. 2005. Cetacean distribution relative to ocean processes in the northern California Current System. Deep-Sea Research 52:145–167. WHITEHEAD, H., AND J. E. CARSCADDEN. 1985. Predicting inshore whale abundance – whales and capelin off the Newfoundland coast. Canadian Journal of Fisheries and Aquatic Science 42:976–981. WITTEVEEN, B. H., R. J. FOY AND K. M. WYNNE. 2006. Potential current and historic prey removal due to consumption by humpback whales (Megaptera novaeangliae) near Kodiak Island, Alaska. Fishery Bulletin 104:10–20. WITTEVEEN, B. H., T. J. QUINN II AND K. M. WYNNE. In press. A feeding aggregation of humpback whales Megaptera novaeangliae near Kodiak Island, Alaska: Historical and current abundance estimation. Alaska Fisheries Research Bulletin. (In press) WYNNE, K. M., R. J. FOY AND C. L. BUCK. 2005. Gulf Apex Predator-prey Study (GAP). Final report FY2001-2003 for NOAA grant NA16FX1270. 249 pp. Avail- able from University of Alaska Fairbanks, 118 Trident Way, Kodiak, AK 99615 or http://www.sfos.uaf.edu/gap/publications/annual reports.htm. ZAR, J. H. 1984. Biostatistical analysis. 2nd edition. Prentice-Hall, Inglewood Cliffs, NJ. Received: 28 July 2006 Accepted: 26 November 2007