Environmental Biology of Fishes

, Volume 88, Issue 4, pp 361–368

Foraging ecology of Cookiecutter Sharks (Isistius brasiliensis) on pelagic fishes in Hawaii, inferred from prey bite wounds

Authors

    • Hawaii Institute of Marine BiologyUniversity of Hawaii at Manoa
  • Brad M. Wetherbee
    • Department of Biological SciencesUniversity of Rhode Island
  • John O’Sullivan
    • Monterey Bay Aquarium
  • Gwen D. Goodmanlowe
    • Department of Biological SciencesCalifornia State University Long Beach
  • Christopher G. Lowe
    • Department of Biological SciencesCalifornia State University Long Beach
Article

DOI: 10.1007/s10641-010-9649-2

Cite this article as:
Papastamatiou, Y.P., Wetherbee, B.M., O’Sullivan, J. et al. Environ Biol Fish (2010) 88: 361. doi:10.1007/s10641-010-9649-2

Abstract

The Cookiecutter Shark (Isistius brasiliensis) is an ecto-parasitic predator of numerous large pelagic fish and mammals. However, little is known of its foraging ecology due to its elusive foraging tactics in the pelagic environment. We used bite scar patterns on pelagic fishes landed at the Honolulu Fish Auction to assess some of the Cookiecutter Shark foraging habits. Swordfish (Xiphias gladius) had the greatest percentage of bites (87.9 ± 25.0% of individuals had healed scars) followed by Opah (Lampris guttatus, 33.0 ± 8.3% of individuals). Most fish with scars only had one Cookiecutter Shark bite per individual with the exception of Swordfish, which often had >5 bites per individual. Furthermore, Swordfish had a higher proportion of healed bite scars meaning they had been attacked while free-swimming. Seasonal changes in the probability of hooked fish being bitten by sharks were apparent for Swordfish, Bigeye Tuna and Opah. Based on bite scar diameter, larger Cookiecutter Sharks may preferentially attack Swordfish rather than the other species of pelagic fish. When taken in conjunction with diving behavior of pelagic fish, and fishing depths, the results add further support to the hypothesis that Cookiecutter Sharks perform diel vertical migrations.

Keywords

Cookiecutter SharkLongline fisheryOpahPredationSwordfish

Introduction

The Cookiecutter Shark (Isistius brasiliensis) is a small squaloid shark found in the pelagic waters of tropical and sub-tropical oceans (Jahn and Haedrich 1987; Nakano and Nagasawa 1996; Compagno et al. 2005). Cookiecutter Sharks feed on cephalopods, but also have a unique parasitic foraging strategy which enables them to prey on animals much larger than themselves. Cookiecutter Sharks have a collar of pigmented cells just posterior to the head, surrounded on either side by luminous photophores, which are thought to act as a lure for large upward-looking visual predators (Widder 1998). The shark’s unique mouth, teeth and tongue morphology then allows it to remove a circular plug of tissue from the larger animal (Jones 1971; Shirai and Nakaya 1992). As such, Cookiecutter Sharks are able to prey on or bite a wide variety of pelagic fishes (Jones 1971; Muñoz-Chápuli et al. 1988), whales (e.g. Evans et al. 2002; McSweeney et al. 2007) seals (e.g. Le Boeuf et al. 1987; Hiruki et al. 1993), and even submarines and oceanographic equipment (Johnson 1978). Cookiecutter Sharks have been caught as deep as 3,800 m, although most are caught at shallower depths, particularly in surface trawls at night, leading to the hypothesis that these sharks perform diel vertical migrations (Jahn and Haedrich 1987; Le Boeuf et al. 1987; Nakano and Tabuchi 1990; Compagno et al. 2005). Other than these basic observations, very little is known about Cookiecutter Shark foraging ecology or behavior.

Due to it being rarely encountered by humans, its small size, and pelagic and elusive nature, one of the few available methods used to study Cookiecutter Shark foraging ecology is through the patterns and frequency of bite scars on prey animals. Fresh bites on Swordfish caught in the longline fishery were used to quantify the distribution of Cookiecutter Sharks in the North Eastern Atlantic, as well as look for evidence of shark size segregation (Muñoz-Chápuli et al. 1988). To date, no other studies have used bite scars on pelagic fishes to make inferences on the foraging ecology of Cookiecutter Sharks. In the 1980s a large pelagic longline fishery developed off the Hawaiian Islands for Swordfish and tunas. In addition, there are numerous local fishers that commercially land pelagic fishes caught around the main Hawaiian Islands, at the Honolulu Fish Auction, one of the largest public fish auctions in the world. Understanding the foraging ecology of Cookiecutter Sharks can have economic implications as the bites inflicted by sharks on pelagic fish can lower the market value (B. Takenaka, manager of the Hawaii Fish Auction, pers. comm.).

We conducted weekly surveys of the Hawaii Fish Auction on Oahu, Hawaii to examine the patterns of bite scars on pelagic fishes. Our goals were to 1) quantify seasonal changes in bite scars on pelagic fishes, 2) quantify species specific differences in the frequency and patterns of bite scarring on pelagic fishes, 3) use bite diameter to estimate the size of Cookiecutter Sharks attacking different pelagic fishes, 4) make inferences on the foraging ecology of Cookiecutter Sharks from bite scarring patterns.

Methods

We visited the Honolulu fish auction weekly from February 2007 until February 2008 to examine pelagic fish brought in by the longline fleets. The surveys of the fish auction were conducted at 06:00 h on the same day each week, to standardize the sampling regime. We counted the number of fish of each target species on the auction floor and determined what percentage of fish had Cookiecutter Shark bite scars. We then selected a sub-sample of ten individuals for each species to make a more detailed examination of bite patterns. Fish are laid out on pallets, so we sampled fish in order along the floor to prevent sampling bias. For the sub-sampled fish, we determined the number of bites, diameter of each bite, and scar stage. Scar stage was classified as follows: 1 (fresh bite—deep crater wound with no indication of new tissue formation), 2 (partially healed—crater wound shallow and membrane and tissue formation started), and 3 (healed scar—crater wound filled in with new tissue leaving just a dermal scar) (Fig. 1). Based on their high annual occurrences, the following species were selected for counts: Swordfish (Xiphias gladius), Bigeye Tuna (Thunnus obesus), Yellowfin Tuna (Thunnus albacares), Skipjack Tuna (Katsuwonus pelamis), Sickle Pomfret (Taractichthys steindachneri), Opah (Lampris guttatus), Wahoo (Acanthocybium solandri), Pacific Blue Marlin (Makaira mazara), Striped Marlin (Tetrapturus audax), and Shortbill Sailfish (Tetrapturus angustirostris). Fish are arranged on one of their sides along the auction floor, which means that counts of bites were only of one side. All fish sold at the Honolulu fish auction are required to be presented with bite marks or scars facing upward. As such, all estimates of counts and frequency are perhaps an under-estimate of predation rates by Cookiecutter Shark.
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-010-9649-2/MediaObjects/10641_2010_9649_Fig1_HTML.gif
Fig. 1

Photo of Cookiecutter Shark bites on Swordfish at various scar stages. Includes fresh bites (stage 1), partially healed (stage 2) and fully healed bites (scale 3)

We determined seasonal changes in the number of bites, and scar stage for each species. We used one-way ANOVA’s with Tukey-Kramer aposteriori tests, to examine seasonal and species differences in bite rates. All data were arcsine squareroot transformed before statistical testing. Further testing was done for the four species with the highest frequencies of Cookiecutter Shark bites, and largest sample sizes: Swordfish, Opah, Bigeye and Yellowfin Tuna. For each month, we determined the probability that a particular species would be bitten by a Cookiecutter Shark while on the longline. For each species we determined the frequency of fish with fresh scars (bite scar stage 1), and calculated the probability (P) of a fish on a longline being bitten, based on a binomial distribution, \( P = {1} - {\left( {{1} - {\hbox{f}}} \right)^2} \), where f is the frequency of fish with fresh bites. We also calculated the probability that a free swimming fish would be attacked by a Cookiecutter Shark, based on the frequency of stage 2 and 3 bite scars combined throughout the year (i.e. not differentiated seasonally). We estimated the size of sharks that inflicted bites on pelagic fish from the diameter of bite scars, by using the Cookiecutter Shark mouth width (MW)-total length (TL) regression, \( {\hbox{TL}} = {4}.{51}\left( {\hbox{MW}} \right) + {82}.{8} \), where TL is shark total length (mm) and MW is shark mouth width (mm), as described by Cadenat and Blache (1981, in Muñoz-Chápuli et al. 1988).

This project required certain assumptions that included 1) all crater wounds were inflicted by I. brasiliensis, 2) for fresh bites, the bite was inflicted while the fish was on the long-line, and 3) bite diameter of fresh scars corresponds with shark mouth diameter.

Results

We conducted 51 weekly surveys of the Honolulu fish auction where we surveyed a total of 15 107 fishes. The largest numbers of fish counted were for Bigeye Tuna, followed by Sickle Pomfret, Yellowfin Tuna, and Opah (Fig. 2a). There were clear differences in the percentage of fish with fresh Cookiecutter Shark bites, between the species (ANOVA: F8, 99 = 7.15, p < 0.0001; Fig. 2a). Swordfish had the highest percentage of fresh bites with 16.9 ± 15.6% (±1 SD) of fish having scars, followed by Opah (13.9 ± 10.5%; Fig. 2a). There were no significant differences between the other species (range fresh bites 1.7–3.4%). Pacific Blue Marlin were the only species surveyed for which no bites were observed (429 individuals). There were also clear species specific differences in the percentage of fish with old healed scars (ANOVA: F8, 99 = 45.85, p < 0.0001; Fig. 2b). Again, Swordfish (87.9 ± 25.0%, probability 0.97), and Opah (33.0 ± 8.3%, probability 0.44) had the greatest percentage of bites, while there were no differences between the other species (range old scars 0–6%). The exception was for Sickle Pomfret, Skipjack Tuna, and Blue Marlin, all of which had significantly lower frequencies of healed scars than the other species (Fig. 2b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-010-9649-2/MediaObjects/10641_2010_9649_Fig2_HTML.gif
Fig. 2

Percentage of fish with fresh (a) or healed (b) Cookiecutter Shark bites. Pelagic fish are listed vertically to correspond with the depth at which fish are caught (Swordfish caught shallow at night down to Bigeye Tuna caught deep during the day). Error bars are standard deviations. Bars with the same letter are statistically the same. Numbers in parenthesis in (a) represent sample size, while those in (b) are the probability of a free swimming fish being bitten, based on a binomial distribution

For the three species with ranked percentage with bites, there were no obvious seasonal changes in the percentage of fish with bites. Swordfish (ANOVA: F11,39 = 1.49, p = 0.17), Bigeye Tuna (ANOVA: F11,39 = 0.84, p = 0.60) and Opah (ANOVA: F11,39 = 1.97, p = 0.06), had relatively consistent bite scar frequencies throughout the year (Fig. 3). The decline in Swordfish on the auction floor from August–December was due to longline quotas having been met, hence much smaller sample sizes during those months. However, when looking at the probability of these three species having fresh bites (scar stage 1), some seasonal patterns appeared (Figs. 3, 4). For Opah and Bigeye Tuna, the probability of being bitten was highest in February and March and October–December, while the lowest probability occurred during January (Fig. 4a, c). For Swordfish, the probability of a fish being bitten peaked in May (0.68), after which there was a linear decrease until September (probability 0, r2 = 0.97, F = 143.9, p = 0.0006) and only an increase in December–January (probability 0.36, Fig. 4b). Again, however, probability values calculated in September–December were based on smaller sample sizes than other months.
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-010-9649-2/MediaObjects/10641_2010_9649_Fig3_HTML.gif
Fig. 3

Seasonal changes in the number of fresh, partially healed, and fully healed bites for Swordfish, Bigeye Tuna and Opah. Numbers above bars are sample size

https://static-content.springer.com/image/art%3A10.1007%2Fs10641-010-9649-2/MediaObjects/10641_2010_9649_Fig4_HTML.gif
Fig. 4

Seasonal changes in the probability of fish receiving fresh Cookiecutter Shark bites for Swordfish, Opah and Bigeye Tuna

The majority of fish only had one Cookiecutter Shark bite, particularly Skipjack Tuna where no fish had >1 bite (Fig. 5). Sickle Pomfret (7%), Shortbill Sailfish (15%), Opah (20%), Bigeye Tuna (7%), Yellowfin Tuna (12%), and Swordfish (62%) had appreciable numbers of individuals with 2 bites. The only species with significant numbers of individuals with >2 bites were Swordfish, where 5% of fish had >5 bites per individual (Fig. 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-010-9649-2/MediaObjects/10641_2010_9649_Fig5_HTML.gif
Fig. 5

Species specific differences in the number of cookie cutter bites per individual

We estimated that the Cookiecutter Sharks that inflicted wounds on pelagic fish ranged in size from 13 cm to 53 cm TL (Fig. 6b). However, Swordfish tended to have larger bite wounds than the other species, suggesting that they were attacked by larger sharks (Fig. 6a). Estimated shark size preying on Swordfish (34.1 ± 7.9 cm), were larger than for Bigeye Tuna (29.5 ± 6.7 cm), Opah (27.2 ± 7.0 cm) or Yellowfin Tuna (28.4 ± 5.9 cm, ANOVA: F3, 411 = 12.9, p < 0.0001; Fig. 6a).
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-010-9649-2/MediaObjects/10641_2010_9649_Fig6_HTML.gif
Fig. 6

a Predicted sizes of cookiecutter sharks that inflicted bites on Swordfish, Opah, Bigeye and Yellowfin Tuna. Error bars are standard deviation; bars with the same letter are statistically identical to each other. b Frequency histogram of the size range of Cookiecutter Sharks inflicting bites on pelagic fishes

Discussion

The Cookiecutter Shark is clearly a versatile pelagic predator as indicated by the large variety of pelagic fish with wounds. In addition to the fish sampled, other species on the auction floor all have been seen with Cookiecutter Shark scars, including Dolphin Fish (Coryphaena hippurus), Oil Fish (Ruvettus pretiosus) and various species of shark (Y. Papastamatiou, pers. obs.). In Hawaiian waters, whales and seals are also seen with Cookiecutter Shark bite scars (Hiruki et al. 1993; McSweeney et al. 2007), and Hawaii is also the location of the first documented Cookiecutter Shark attack on a human swimmer (G. Burgess, International Shark Attack File, pers. comm.). However, there were clear differences in the degree of predation on the various species of pelagic fishes.

Differences in bite scar frequencies between pelagic fishes are going to be related to both the behavior of fish prey, the behavior of Cookiecutter Sharks, and the characteristics of the longline fishery which targets the different species. Analysis of the longline fisheries reveals that in general, sets that catch Swordfish are predominantly set at night, in shallow water (median depth of 60 m), and 70% are caught outside the Hawaiian Exclusive Economic Zone (EEZ) (extending up to 45°N; He et al. 2006; Bigelow et al. 2006, PIFSC log book report 2007). Tunas and other fishes including Wahoo, Opah, and Sickle Pomfret, tend to be caught on sets in deeper water (median depth of 268 m), set during the day, and almost 70% are within the Hawaiian EEZ (Bigelow et al. 2006; He et al. 2006; PIFSC log book report 2007). Swordfish and Opah had the highest percentage of fresh bites, most of which were presumably inflicted while the fish were hooked on the longlines. When taken in conjunction with the characteristics of the fishery, our data further supports the hypothesis that Cookiecutter Sharks are diel vertical migrators, occupying deeper depths during the day, and moving close to the surface at night (Le Boeuf et al. 1987; Nakano and Tabuchi 1990; Compagno et al. 2005). Although there were no seasonal shifts in the number of fish with scars, there did appear to be some seasonal changes in the probability of hooked Swordfish, Bigeye Tuna, and Opah, being bitten by Cookiecutter Sharks. These changes in bite probabilities could indicate some degree of north-south seasonal migration of Cookiecutter Sharks during the winter months. However, the small winter sample sizes, and the fact that we do not know exact capture locations of fishes, means that presently the seasonal migration is little more than a hypothesis.

Healed scars were almost certainly bites that were inflicted on free-swimming prey. When examining the percentage of fish with healed scars, and species with >1 bite per individual, Swordfish have the greatest probability of being bitten while free-swimming, followed by Opah. The vertical diving behavior of these two species, consist of deep dives during the day (100–600 m), and shallower dives (0–150 m) at night (Carey and Robison 1981; Polovina et al. 2008). These depths are considerably deeper than the habitat used by Blue Marlin, Striped Marlin and Yellowfin Tuna, all of which rarely dive below the mixed layer, which in Hawaii is typically less than 100 m depth (Holland et al. 1990; Brill et al. 1993, 1999), and may further reflect the diel vertical migratory behavior of Cookiecutter Sharks. However, Bigeye Tuna also use deeper habitats (e.g. Dagorn et al. 2000), but are targeted less frequently by sharks than Swordfish or Opah, so some other factors must also influence species specific foraging rates. Characteristics of dermal tissue may also influence predation rates, as no bite scars were found on any Blue Marlin specimens. Unlike the other species, Blue Marlin have characteristic thick, thorny scales which may make it difficult for Cookiecutter Sharks to successfully remove a plug of tissue. Overall, the probability of free-swimming pelagic fishes being bitten by Cookiecutter Shark was fairly low.

Finally, there may be some evidence of prey size selection by Cookiecutter Sharks, as larger bites were recorded on Swordfish than for any other pelagic fish species. This could be indicative of size or sex selection/segregation. For example, male Cookiecutter Sharks are rarely found in excess of 40 cm Total Length (TL), while females reach up to 51 cm TL (Jahn and Haedrich 1987; Nakano and Tabuchi 1990, Compagno et al. 2005). Over 14% of bites on Swordfish were made by sharks >40 cm TL, while only 3% of Opah, Bigeye and Yellowfin Tuna bites were from sharks >40 cm. Therefore, a higher proportion of bites on Swordfish may have been inflicted by female sharks. Fishing records from trawling, and bite scar patterns have also suggested some degree of horizontal sex and size segregation of Cookiecutter Sharks (e.g. Muñoz-Chápuli et al. 1988; Nakano and Tabuchi 1990). However, the mechanical properties of the tissue of pelagic fishes are unknown; and it is also possible that Swordfish skin stretches more after being bitten by equally sized sharks, thereby giving the appearance of larger shark bites. It is also unknown if there are species specific differences in healing rates, or if scars may eventually disappear altogether.

The low abundance, oceanic habitat use and elusive nature of Cookiecutter Sharks, requires us to use indirect methods to quantify their foraging behavior. Yet despite the numerous assumptions, this study provides insight into the foraging ecology of this elusive species. The data presented provides new hypotheses of Cookiecutter Shark behavior and also provides quantitative information on which pelagic fish are targeted and when. Future studies should investigate the economic impacts of Cookiecutter Sharks on the pelagic fisheries, and obviously, should a source of live animals present itself, attempt to quantify in situ behavior to test these hypotheses.

Acknowledgements

We would like to thank B. Takenaka for providing us with continuous access to the Honolulu fish auction. We would like to thank J. Dale, N. Whitney, and L. Davis for helping with auction sampling. Finally, we would also like to thank J. McCosker and two anonymous reviewers whose comments improved the manuscript. This is a contribution of the MBA/HIMB Collaborative Research Program.

Copyright information

© Springer Science+Business Media B.V. 2010