Estimating humpback whale (Megaptera novaeangliae) entanglement rates on the basis of scar evidence

Transcription

1 Estimating humpback whale (Megaptera novaeangliae) entanglement rates on the basis of scar evidence Report to the Northeast Fisheries Science Center National Marine Fisheries Service 166 Water Street Woods Hole, MA Order Number 43EANF May 13, 2004 Submitted by: Jooke Robbins and David Mattila 1 Center for Coastal Studies P.O. Box 1036 Provincetown, Massachusetts

2 ABSTRACT Entanglement in fishing gear is a serious source of humpback whale, Megaptera novaeangliae, injury and mortality in the Gulf of Maine. The flukes and caudal peduncle are commonly implicated in entanglements and consistently presented during the terminal dive. Between 2000 and 2002, photographs were systematically obtained of these features while parallel to the whale and slightly ahead of its flukes during the terminal dive. Images were coded to reflect the presence or absence of wrapping scars, notches and other injuries that were believed to be entanglement-related. Assumptions were successfully tested against documented caudal peduncle entanglements during the study period. Approximately half (48%-57%) of each annual sample exhibited scars that were likely to have resulted from a prior entanglement. However, as individuals acquire and lose scars over their lifetime, the inter-annual acquisition rate ( %) was considered more likely representative of events during the study period. Composite (lifetime) rates of high probability scarring, inter-annual scar acquisition and the standardized minimum number of events were all comparable to the previous three-year period ( ). However, neither recent documented entanglements, nor scar acquisition rates supported prior evidence of a higher male risk of entanglement. Based on the weight of the current evidence, it does not appear that breeding ground injuries alone can explain the higher frequency of healed caudal peduncle injuries among Gulf of Maine males. Continued efforts to refine scar analysis techniques and examination of historic caudal peduncle data may help to clarify this issue. With further development, the study of raw entanglement-related injuries may allow for annual estimates to be developed without a baseline sample. Nevertheless, caudal peduncle monitoring can only provide a relative index of entanglement, as it does not account for events that do not involve the caudal peduncle, nor those that result in the death of the animal prior to sampling. The present work confirmed that even animals that die with evidence of fisheries interactions do not necessarily exhibit extensive external injuries. Thus, tracking even apparently minor entanglements may be important in detecting and preventing increases in severe injuries or death. Given the low rate at which these events continue to be reported (10%), management initiatives should focus on prevention, in addition to human intervention (disentanglement). 1 Present address: Hawaiian Islands Humpback Whale National Marine Sanctuary, Maui, HI USA 1

3 INTRODUCTION Entanglement in fishing gear is a known source of injury and mortality for Gulf of Maine humpback whales, Megaptera novaeangliae (National Marine Fisheries Service 1991). Although animals are sometimes anchored in place by entangling gear in this region, they are more commonly reported after having broken free from a portion of the gear. The rate at which events are reported therefore depends on several factors, including the actual number of events, the likelihood of event detection (overlap between observers and whales), observer awareness and willingness to report. Until it is possible to quantify these factors, reporting rates will remain an uncertain basis for studying the magnitude of this impact and to monitor it over time. Photographic monitoring programs provide an alternate source of information on the incidence of entanglement. In a study of scarification on the North Atlantic right whale, Kraus (1990) found that 57% of the population carried entanglement-related scars at the anterior caudal peduncle. The flukes and caudal peduncle are also commonly implicated in humpback whale entanglements and consistently presented during the terminal dive. Between 1997 and 1999, the Center for Coastal Studies performed systematic photographic sampling to evaluate relative rates and severity of entanglement, as well as the proportion of the population affected (Mattila and Robbins 1998, Robbins and Mattila 1999, 2000, 2001a). That work has confirmed that even short-term, mitigated entanglements produced noticeable scars on the peduncle that remained diagnostic from one year to the next. An annual, baseline sample of peduncle photographs was initiated and a scar-based coding system was developed. The latter was tested using documented entanglements and shown to accurately indicate an animal s entanglement status. Based upon the proportion of the animals carrying high probability scars, at least half of the individuals sampled were likely to have been entangled at least once during their lifetime. Males were significantly more likely than females to exhibit high probability scars, while juveniles and mature whales exhibited comparable levels of high probability scarring. Nevertheless, the entanglement status of an average of 28% remained unclear. Individuals were found to accumulate scars over their lifetime, and some that were known to be entanglement-related healed to the point that they were no longer diagnostic. Recent entanglements and more serious entanglements, regardless of their timing, were more likely to be detected and accurately interpreted. Finally, scar evidence suggested that no fewer than 31 Gulf of Maine humpback whales were entangled between 1997 and 1999 in addition to reported incidents. Less than 3% of these corresponded to events that were detected and reported in progress. This report describes the continuation of this study through 2002, including information presented in two annual contract reports (Robbins and Mattila 2001b, 2003). The primary goals of this work were to maintain baseline photographic coverage of the population, to continue to refine and validate techniques, and to evaluate the incidence and trend in entanglement-related injuries over time. 2

4 METHODS Between 2000 and 2002, caudal peduncle photographs were obtained to serve as a basis for monitoring entanglement related scarring. In this study, the terms caudal peduncle and tail stock actually refer to several features at the insertion point of the flukes, including the posterior caudal peduncle and leading edges of the flukes. These body parts are rarely captured during standard ventral fluke photography and so images were obtained from a forward position, alongside the animal and ahead of its flukes when it began its terminal dive. Photographers were instructed to document this part of the body whenever it was presented, without regard for scar patterns observed in the field. In some cases, this included non-standard approaches, such as when the base of the flukes and caudal peduncle were exposed by rolling or lob tailing behaviors. The latter was particularly important for calves, which are less likely to perform a standard fluking dive than older animals. Images were obtained using standard or digital SLR cameras equipped with 300-mm telephoto or mm zoom lenses. When film was used, it was 400 ASA black & white and the best images were digitally enlarged to a minimum resolution of 150 dpi. Digital images were shot in 24 bit color at a minimum resolution of 2160 x 1440 pixels. Animals were photographed both from directed research platforms and opportunistically from whale watching vessels. Directed sampling was performed during photo-identification and biopsy-oriented cruises throughout the U.S. and Canadian Gulf of Maine. Opportunistic coverage was obtained primarily by naturalists aboard commercial whale watch boats operating in the coastal waters of Massachusetts, particularly in the vicinity of the Gerry E. Studds-Stellwagen National Marine Sanctuary (Stellwagen Bank). The likelihood of a previous tail entanglement was estimated using previously developed techniques (Mattila and Robbins 1998, Robbins and Mattila 1999, 2000, 2001a). The caudal peduncle and insertion point of the flukes were divided into six general areas for coding purposes: the right and left flanks, the leading edge of the flukes (right and left), the dorsal peduncle and the ventral peduncle. A single individual (JR) examined each area for evidence entanglement, such as wrapping scars, notches or more extensive tissue damage. Low quality photographs (based upon distance, angle and focus) were excluded from analysis to prevent a bias toward more obvious injuries. While all six coding areas were not necessarily available for each individual, those with at least two areas of coverage were included in subsequent analyses. This minimum coverage requirement was intended to increase the likelihood that a scar pattern across regions could be detected, if present. Animals with wrapping scars or injuries in at least two coding areas were assigned a high probability of a prior entanglement. Those with no diagnostic injuries or scars were considered to have a low probability of prior entanglement. When wrapping injuries were detected in only one coded area such that entanglement was neither strongly supported nor ruled out, the animal was assigned an uncertain probability of previous entanglement. Examples of scar pattern interpretation are shown in Figure 1. Although the coding scheme was intended to standardize entanglement status determinations, they remained somewhat subjective, based upon the pattern and severity of scarring observed across all of the documented regions. However, auxiliary 3

5 information, such as documented entanglement histories, was not factored into the coding process. Coding composite scars in a single set of images provided insight into whether or not an individual had an entanglement history, but rarely provided detail on the number or timing of specific events. Additional analyses were therefore performed to investigate rates of entanglement-related scar acquisition and loss. Each image obtained contributed to a baseline data set for that individual. After entanglement status coding, a direct photographic comparison was made to the baseline sample to determine whether specific diagnostic scars had increased, decreased or remained stable over the study period. Of particular interest was the inter-annual rate of scar acquisition, which was the proportion of cases in which new entanglement-related injuries were obtained since the previous year. When color images were routinely available, the persistence of raw injuries was also specifically studied because of their potential value in quantifying events without a baseline sample from a previous year. Unique injuries that could be definitively linked to the period were summed to create a minimum number of interactions with gear. These consisted of injuries acquired between samples and those on calves and juveniles born after As this estimate was based on scar patterns, it did not include animals that were reported as entangled, if they were not also sampled in this study. However, the overlap between documented entanglements and entanglementrelated scar acquisition (described below) was used to estimate the reporting success rate for the period. Given the subjective nature of scar coding, efforts continued to validate underlying assumptions and the consistency of application. The ability of the coding techniques to detect entanglements was tested based on accumulated data from confirmed entanglement events. The Center for Coastal Studies operates the only federally authorized large whale disentanglement program along the eastern seaboard of the United States, including a region-wide reporting network and awareness training. When humpback whale entanglements are reported in this region, an effort is made to obtain documentation of both the entanglement and the individual so that it can be reidentified with or without the entangling gear. These data made it possible to study the injuries created by entanglements and to correlate the individuals in this study to documented events after the coding process was completed. Particular attention was paid to the subset of entanglement cases in which there was clear documentation of the caudal peduncle at the time the animal was entangled. While this excludes some cases that were known to involve the caudal peduncle, it was intended to ensure that the configuration of gear and/or raw injuries could be directly compared to later photographic samples. It was also considered that other sources of injury might confound a scar-based study of entanglement. Initial work identified a male bias among animals with a likely history of entanglement and there was concern that results might be confounded by scars produced by agonistic displays between males on low latitude breeding grounds. These behaviors are thought to produce male-biased scarring at the dorsal fin (Chu and Nieukirk 1988) and the caudal peduncle and flukes are also regularly involved in displays. Breeding ground injuries were considered more likely to mask physical evidence of entanglement, than to resemble those injuries. However, a preliminary test of this assumption was performed using a suite of 4

6 comparable samples obtained from the Hawaiian Island breeding ground in Although from a separate oceanic population, it was assumed that the agonistic behaviors of concern would be sufficiently similar between oceanic populations. Despite the fact that a single individual performed all coding since 1997, there was nevertheless the potential for coding practices to drift over time, or to be affected by slight differences in photographic quality or coverage between samples. Coding consistency was investigated by examining the frequency with which the changes in the annual coded status of an animal were not supported by side by side image comparisons of scar patterns. This study focuses on only one of several potential entanglement sites on the body of humpback whales. While documented entanglements provide information on attachment sites, those observations may be biased by highly visible nature of the peduncle relative to the mouth or flipper. Whenever possible, comprehensive photographic coverage was therefore obtained from Gulf of Maine humpback whales that stranded during the study period, regardless of the cause of death. Images were examined for evidence of entanglement at all potential attachment sites, including the insertion of the flukes and flippers and the mouth. This had the added advantage of providing data on the apparent severity of external injuries associated with mortality. Identifying shots of each individual were matched to a photo-identification catalog of Gulf of Maine humpback whales maintained by the Center for Coastal Studies. Annual population monitoring performed since the late 1970s has produced an extensive database of individual information that is potentially useful to evaluating entanglement risk. Sexes of Gulf of Maine humpback whales were determined by genetic analysis of a tissue sample (Palsbøll et al. 1992, Clapham et al. 1995, Bérubé and Palsbøll 1996b, Bérubé and Palsbøll 1996a), a photograph of the genital slit (Glockner 1983) or, in the case of females, documentation of at least one calf. An exact age was calculated for each individual cataloged in the Gulf of Maine during its first year of life. For animals without a known year of birth, a minimum age was assigned by assuming that the whale was at least 1 year old the first year it was sighted. Female humpback whales in the Gulf of Maine have been shown capable of reaching sexual maturity as early as age four, and averaging age 5 (Clapham 1992). Consequently, juvenile whales were assumed to be those that were first cataloged as calves and were less than five years old in the year that they were sampled. Whales were considered to be sexually mature if they were known to be at least five years old or were first sampled as an independent whale at least four years prior to being sampled. A maturational class was not definitively assigned to whales without a known year of birth that were first cataloged less than four years prior to this study. When reported, the 95% confidence interval (CI) of percentages were calculated based on the standard error, as follows: CI = 1.96 p *(100 p) n Where: p = the percentage of interest and n = total number of animals examined. Statistical differences between samples were evaluated using a G-test with a William's correction. Results 5

7 were compared to a chi-square distribution and Freeman-Tukey deviates were used to identify particularly high or low cell values (Sokal and Rohlf 1981). RESULTS Evaluation of methodology Caudal peduncle images are available from 12 documented entanglements since 1997 (Table 1). Half of these events involved the peduncle at the time of the report and the rest exhibited raw injuries that suggested involvement at an earlier stage. Injuries took the form of wrapping scars, notches and other tissue damage in one or more areas of the caudal peduncle/flukes (Table 1, Figure 2). Six individuals were re-sighted at a later date, after the gear had been lost through shedding or disentanglement. In each of these cases, the scars remained diagnostic at the time of the most recent sighting, after as many as 1,920 days (Table 1). Re-sightings of previously entangled animals have provided an opportunity to study the fate of specific entanglement-related scars over time. Scar patterns can be remarkably stable, particularly when they involve major injuries to the leading edges of the flukes. While healing was observed to reduce scar patterns, it was unusual for individuals to loose diagnostic injuries, particularly between consecutive years. However, dorsal and ventral peduncle injuries that began as a combination of raw wrapping injuries and notches (such as those shown on Figure 2) often reduced to notches alone, without the white scar tissue commonly produced at the leading edges of the flukes. Injuries to the flanks of the peduncle also tended to heal to the native skin color, even when large sections of skin were removed. As shown in Figure 3, even very prominent flank injuries have been lost or greatly reduced between consecutive years. The nature of healing at the flanks of the caudal peduncle may explain the low rate at which high probability scars have been detected there in this study. In 2002, 68% of the animals sampled exhibited diagnostic injuries at one or both leading edges, 58% at the dorsal peduncle, 44% at the ventral peduncle and only 11% at the flanks of the peduncle. Of these areas, the ventral peduncle was the least likely to be documented in a standard lateral approach. Therefore, in this study, the leading edges of the flukes and the dorsal peduncle tended to contribute the most information on the likely entanglement status of an individual. Between , 17-26% of each annual sample could not be positively attributed to a likely composite entanglement category, despite the presence of at least one entanglement-like element. Since the uncertain category likely contained animals with entanglement-related scars that have healed beyond recognition, high probability values reported herein were the minimum number of caudal peduncle entanglements in this sample. Similarly, five (4%) of the animals sampled in two consecutive years alternated between a high probability of entanglement code and a low probability or uncertain status, despite the fact that a direct photographic comparison indicated that no noteworthy change had actually occurred. Although the minimum documentation requirements had been met in these cases, slight differences in lighting, angle and coverage nevertheless affected the way that marks were interpreted. Comparison of the images involved confirmed that the lower code had likely under-estimated the entanglement history of the individuals involved. 6

8 Full body examinations were performed for five animals that stranded between 2001 and 2003 (Table 2). Four exhibited diagnostic scarring at the caudal peduncle/flukes, and two also exhibited evidence of entanglement at other body parts. None exhibited entanglement evidence without corresponding injuries at the caudal peduncle. Of particular interest was the severity of caudal peduncle scarring on the two stranded animals with recent entanglement evidence. The injuries of animals that died (Figure 4) were not more visibly severe than those of animals sampled alive during this study. However, the immediate cause of these deaths was not known with certainty. In the case of MAL02170MN, a whale that was disentangled from gear four days earlier, no necropsy was ultimately performed. Based on the necropsy report, the working hypothesis for case CCSN was anoxia and hyperthermia due to anchoring in gear (Cape Cod Stranding Network 2002). When images were obtained from a humpback whale breeding ground, 24% (n=36) exhibited raw injuries, presumably resulting from agonistic interactions. The vast majority of the photographs (73%, n=112) were obtained from agnostic (competitive) groups and only 26 images were obtained from animals known or strongly suspected to be female. Raw marks occurred in all areas of the peduncle, but rarely took the form of a consistent pattern of wrapping marks. Furthermore, preliminary analysis indicates that only 14% (n=22) of the images obtained there exhibited healed scarring that would have been interpreted as entanglement-related. Thus, while males obtain injuries from breeding ground interactions, it appears unlikely that the scar patterns observed in the Gulf of Maine can be explained by an accumulation of such injuries. The transition to high quality, digital-source images with 24-bit color data has encouraged efforts to study entanglement based on the frequency of raw injuries. Thirteen (6.5%) of the animals sampled in 2002 exhibited raw, high probability injuries at the caudal peduncle. Examples of these are shown in Figure 5. None were known to correspond to a documented entanglement event and it was generally not possible to determine when the injuries were obtained, as all were present at the time of the first sighting. However, four events had already occurred by the end of May, and one as early as April 12. Three animals were seen on at least one subsequent day and their injuries remained raw between sightings, which spanned 11, 60 and 83 days. The latter animal was re-sighted in 2003 and 2004, at which point the injuries remained diagnostic but were no longer raw. It is, however, possible for injuries to appear to be raw long after an entanglement event. Tiara has been without gear since at least 1994, but his injuries remained raw and aggravated in 2002 (Figure 1d). Without additional knowledge about this case, it might not have been possible to differentiate this old event from a more recent case. Inferred entanglement rates Between 2000 and 2002, images were obtained from an average of 183 individuals per year (Table 3). While not all images were considered to be of equal quality for determining entanglement status in any particular year, all were deemed potentially valuable for monitoring portions of the caudal peduncle over time. Modern baseline coverage is now available for 569 unique individuals sampled at least once since On average, 58% of the images were obtained from whale watching platforms, resulting in particularly strong representation of the Massachusetts Bay and Stellwagen Bank National Marine Sanctuary areas. 7

9 Approximately half (48-57%) of the sample was coded as having a high likelihood of prior entanglement. However, a more time-sensitive measurement was the inter-annual acquisition rate, based on comparable peduncle coverage from two consecutive years. The latter suggests that entanglements have occurred at a minimum rate of 8.0% to 10.4% per year (Table 4). With six years of sampling now complete, there is little evidence that entanglement rates have changed over that period (Tables 5 and 6). Although annual variations in composite high probability scarring approached significance (G=12.825, p=0.0557, df=5), this was primarily due to one high value in When this point was excluded, there was no evidence of significant differences between the remaining years. A similar pattern was noted in inter-annual scar acquisition rates, with exhibiting more than twice the frequency of events than any other period. However, this difference also failed to reach significance (G=7.569, p=0.1193, df=4). The demographic characteristics of animals that acquired high probability scars, and those involved in recent documented entanglements are shown in Table 5. The two samples are not entirely comparable, as scar acquisition was based on animals sampled in two consecutive years or observed with raw entanglement-related injuries. Our ability to detect the latter was limited prior to 2002 and so calves, in particular, may be under-represented in the scar-based estimate, while mature whales may be over-represented. However, if it is assumed that animals in the unknown maturational class tend to be juveniles (Robbins 2003), then the characteristics of inferred and documented entanglements were roughly similar. When the sex of the animal was known, both documented and inferred events involved more females than males. Unlike previous years, overall (composite) high probability scarring in 2002 was close to parity, with only 50.6% (n=41) of cases corresponding to males. Inferred entanglement events and reporting rates Scars acquired by Gulf of Maine humpback whales between 2000 and 2002 suggest that a minimum of 49 interactions with gear took place during that three-year period. Most of these events were inferred from scars acquired after a baseline sample (n=23) or were simply raw at the time of sampling (n=10). The remaining cases involved calves (n=7) or juveniles (n=9) that were not alive prior to the period of interest. Injuries ranged from outwardly minor to severe and only five (10%) corresponded to entanglement events reported during the period. In order to compare the past three years to the period, it was necessary to exclude the first inter-annual sample ( ) and any raw injuries that could not be detected with black & white media. That adjustment produced an equal number of events as in the previous period (n=31). DISCUSSION The primary goal of this study was to continue to accumulate a data set suitable for evaluating trends in entanglement rates over time. Photographic monitoring has successfully produced large annual samples since 1997 and the coding system has reliably detected documented caudal peduncle entanglements during the study period. Based on these data, there is presently little 8

10 evidence that entanglement rates have changed over the study period. Both the overall (composite) frequency of entanglement-related scars and the inter-annual rates of scar acquisition appear to have remained relatively stable since scar monitoring began. Both suggest that there may have been an increase in activity during late 1998 or 1999; however, if that was the case, it does not appear to be a continuing trend. Data accumulated to date may now be sufficient for use in mark-recapture based models of scar acquisition, as suggested by Pace (2003). At least 49 events were required to generate the scarring observed on Gulf of Maine whales in the past three years, in addition to documented entanglements that were not also represented in this study. However, that is still likely an underestimate of the number of events that have occurred. A less conservative value could be calculated by applying the inter-annual scar acquisition rate to a current abundance estimate of the Gulf of Maine population. While no precise number exists at this time, Clapham et al. (2000) reported estimates ranging from 497 to 902 animals. Applying the lowest inter-annual estimate (8.0%) to the smallest available abundance estimate would still result in at least 39 animals encountering gear each year. Even this likely underestimates the actual number of entanglements, as our methods do not yet account for events that do not involve the caudal peduncle, nor those that result in the death of the animal prior to sampling. Based on composite scar patterns, it was originally believed that Gulf of Maine males were more likely to become entangled than females (Robbins and Mattila 2000). However, this hypothesis has not been borne out by the demographic characteristics of documented entanglements, nor scar acquisition since One potential explanation of this discrepancy is that males are exposed to other sources of injury that confound the interpretation of scar patterns. The most obvious source of such injury would be agonistic interactions on the breeding ground. However, several factors suggest against this as an important source of bias in this study. First, in a preliminary study off Hawaii, raw breeding ground injuries did not appear to strongly resemble entanglement scarring. Secondly, if agonistic behavior were a major source of error in this study, then one might expect a high apparent entanglement rate in the Hawaii sample, which was strongly biased toward males. Instead, high probability scars were detected at a lower rate than in the Gulf of Maine. It has, of course, been assumed that both oceanic populations make similar use of their caudal peduncles during breeding ground activities. However, even if this is not the case, Gulf of Maine males should continue to acquire their own form of injuries each winter, and these would then be detected by inter-annual photographic comparison. Based on the weight of the current evidence, it does not appear that breeding ground injuries alone can explain the higher frequency of healed caudal peduncle injuries among Gulf of Maine males. It is also possible that there has been a true change in the portion of the population affected by entanglement in the Gulf of Maine. The greater frequency of high probability composite scars among females in 2002 suggests they may now, in fact, be approaching the overall exposure of males. This could have resulted from a long-term change in the distribution of high-risk gear. Habitat use within the Gulf of Maine is known to vary in relation to age and maturational class (Robbins et al. 2001), so a geographic change in gear has the potential to affect some classes disproportionately. Alternatively, the foraging patterns of females may have changed since the 1980s. Indeed, habitat use patterns in the southwestern Gulf of Maine has been highly variable 9

11 in the past decade, presumably in relation to fluctuations in the availability of sand lance (Weinrich et al.1997). If geography is a factor in these results, then the large volume of regionally representative caudal peduncle data obtained during the MoNAH project in 2003 may help to identify it. By contrast, a temporal change may be clarified by efforts to estimate entanglement rates in the early 1980s based on archival opportunistic data. Despite its highly regional nature, whale watching vessels have been invaluable in maintaining large annual sample sizes. If established in this region, a 100-meter distance regulation will greatly limit the value of opportunistic platforms in this study. In addition to addressing the question of male-biased injuries, the parallel effort in Hawaii has permitted a preliminary comparison of entanglement rates between two humpback whale populations. Although a relatively large sample size was obtained in Hawaii, fewer animals exhibited high probability scars and no clear instances of severe entanglement-related injury were noted. Of course, the two samples are not directly comparable, as Hawaiian scarring likely reflects entanglement probabilities in multiple feeding areas, characterized by different types and densities of gear. In a scar-based study of live whales, a lower perceived rate of entanglement could also result if there were reduced survivorship of entanglement events. It was theorized that if humpbacks in the North Pacific were more likely to become entangled in heavy, offshore gear then they might also be less likely to survive those encounters. However, in that event we would have also expected to encounter a higher number of severe-entanglement related injuries. In fact, this was not the case. These results instead suggest that entanglement may be a larger issue in the Gulf of Maine than elsewhere. Future efforts in the North Pacific would benefit from focusing on specific feeding grounds so that results can be interpreted in light of regional information on gear and documented entanglements. A comparative study using these techniques is now part of a Master s project at the University of Alaska at Fairbanks. Aside from crude inter-population comparisons, composite scar patterns are no longer preferred for assessing the incidence of entanglement. As more data are accumulated, it is clear that individuals accumulate scars over their lifetime and some scars ultimately heal to the point that their source can not be identified. Recent entanglements and more serious entanglements, regardless of their timing, are more likely to be detected and accurately interpreted in a composite sample. In addition, although low quality images are rejected from analysis, some variation in image quality is unavoidable due to slight differences in angle, distance and lighting. Such differences can influence the coding process in a small percentage of cases. Thus far, animals have not been coded as entangled when other evidence suggested that they were not. Indeed, it is more likely for entanglement risk to be underestimated. However, it remains clear that some variability in coding exists in subtle cases, even when there is no notable change in scarring patterns. By contrast, annual caudal peduncle monitoring provides sensitive and less ambiguous data from which to infer entanglement events. While time-sensitive changes in entanglement rates can also potentially be determined from the proportion of high probability scars on calves and juveniles, it remains difficult to sample those age classes in sufficient numbers. Unfortunately, entanglement rates based on direct photographic comparison are necessarily biased toward the best-represented group, mature whales. If it is true that juvenile whales are at higher risk, then entanglement rates 10

12 affecting them will be underestimated by that method. With further development, the study of raw entanglement-related injuries may allow for greater precision and power in an annual sample. Raw injuries are the easiest to interpret correctly and do not necessarily require a previous sample from the same individual. However, for raw injuries to be useful, their persistence must be determined. There is presently evidence that injuries remain raw for at least 80 days and this may be sufficient for within-season studies. However, even if raw injuries are stable over those time frames, events that occur after an individual is sampled will be missed. Thanks to the MoNAH project, we anticipate that sample sizes in 2003 will be sufficiently large to allow for additional development of this method. In the first few years of this study, we attempted to estimate rates of severe entanglement related injuries separately from those that appeared to involve less tissue damage. Severe injuries were considered likely to be of management interest on their own and it was also assumed that their frequency might parallel entanglement-related mortality. However, this practice was discontinued in 2002 for several reasons. First, it is not uncommon for injuries appear more severe when they are raw than once they have healed. Secondly, a separate estimate diminishes the importance of other entanglement events. We presently have no clear understanding of the factors that lead to a minor entanglement-related injury versus a major one. Even animals that die with evidence of fisheries interactions do not necessarily exhibit extensive external injuries. Thus, the frequency of severe injuries on live whales may bear little relationship to the actual mortality rate. Tracking even apparently minor entanglements may be important in detecting and preventing increases in severe injuries or death. Lateral caudal peduncle photographs were used for this study because they provide the most consistent opportunities to view injuries at the base of the flukes. However, wrapping injuries at the leading edges of the flukes and peduncle are also sometimes visible in photo-identification images. Furthermore, humpbacks have been observed to sustain entanglement-related injuries to other parts of the flukes, such as along the trailing edge. A reliable technique for estimating entanglement rates from the ventral flukes alone would be useful, both because it could be applied retrospectively to pre-existing photo-identification catalogs and because it would reduce the amount of photographic data required in the future. Because of its strong predictive powers and large sample sizes, the present study has been used to develop and evaluate those techniques. Woodhead-Buotte (2002) examined the reliability of ventral fluke scarring as an indicator of entanglement status, based on documented entanglements and those inferred here from caudal peduncle scarring. In her study, half of the 20 animals with documented entanglements received injuries to the ventral flukes, and 20.0% (n=4) provided clear evidence that an entanglement had occurred. Similarly, 26.3% of the 57 animals assigned a high likelihood of entanglement based on caudal peduncle scarring were coded as such based on the flukes alone. Entanglements were most consistently detected in the case of extreme entanglement-related tissue damage, with five of six animals correctly assigned. With further development, it is possible that the techniques employed by Woodhead-Buotte (2002) may assist in determining the cause of severe fluke injuries. However, as previously noted, a technique that only reliably detects the small fraction of severe injuries is of questionable conservation value, particularly in the case of endangered species. 11

13 ACKNOWLEDGEMENTS Aaron Avellar, the Beneficia Foundation, Martine Bérubé, Moriah Bessinger, Carole Carlson, the CCS disentanglement program, the Dolphin Fleet whale watching company, Mark Gilmore, Clarisse Hart, Dana Hartley, the Hawaiian Islands Humpback Whale National Marine Sanctuary, Joanne Jarzobski, Pauline Kamath, Philip Kibler, Saeko Kumagai, Scott Landry, Ed Lyman, the Marine Animal Lifeline, the Marine Mammal Commission, the National Fish and Wildlife Foundation, Owen Nichols, Per Palsbøll, Nancy Scaglione-Peck, Lisa Sette, Rosie Seton, and Sal Testaverde. Directed approaches were performed under U.S. federal research permit numbers and REFERENCES CITED Bérubé, M. and P.J. Palsbøll. 1996a. Erratum of identification of sex in cetaceans by multiplexing with three ZFX and ZFY specific primers. Molecular Ecology 5: 602. Bérubé, M. and P.J. Palsbøll. 1996b. Identification of sex in cetaceans by multiplexing with three ZFX and ZFY specific primers. Molecular Ecology 5: Cape Cod Stranding Network CCSN necropsy report. October 3, Available from the Cape Cod Stranding Network, P.O. Box 287, Buzzards Bay, MA Chu, K. and S. Nieukirk Dorsal fin scars as indicators of age, sex, and social status in humpback whales Megaptera novaeangliae. Can. J. Zool. 66(2): Clapham, P.J., and C. A. Mayo Reproduction of humpback whales (Megaptera novaeangliae) observed in the Gulf of Maine. Rept. Intl. Whal. Commn. Special Issue 12: Clapham, P. J Age at attainment of sexual maturity in humpback whales, Megaptera novaeangliae. Can. J. Zool. 70: Clapham, P. J., M. Bérubé and D.K. Mattila Sex ratio of the Gulf of Maine humpback whale population. Mar. Mammal Sci. 11(2): Clapham, P. J., J. Barlow, M. Bessinger, T. Cole, D. Mattila, R. Pace, D. Palka, J. Robbins and R. Seton Abundance and demographic parameters of humpback whales from the Gulf of Maine, and stock definition relative to the Scotian Shelf. Journal of Cetacean Research and Management 5: Glockner, D.A Determining the sex of humpback whales in their natural environment. Communication and Behavior of Whales. R. Payne. Colorado, AAAS Selected Symposium 76. Westview Press: Kraus, S.D Rates and potential causes of mortality in North Atlantic right whales, Eubalaena glacialis. Marine Mammal Science. 64(4): National Marine Fisheries Service Final recovery plan for the humpback whale (Megaptera novaeangliae). Prepared by the Humpback Whale Recovery Team for the National Marine Fisheries Service, Silver Spring, Maryland. 105 pp. Pace, R Two new methodological approaches useful for estimating scarring rates of cetaceans from mark-recapture data. Proceedings of the 15 th Biennial Conference on the Biology of Marine Mammals. Greensboro, NC. December 14-19,

14 Palsbøll, P. J., A. Vader, I. Bakke. and M.R. El-Gewely Determination of gender in cetaceans by the polymerase chain reaction. Can. J. Zool. 70: Robbins, J. and D. K. Mattila Monitoring entanglement scars on Gulf of Maine humpback whales. Center for Coastal Studies. Order number 40ENNF Robbins, J. and D. K. Mattila Monitoring entanglement scars on the caudal peduncle of Gulf of Maine humpback whales: Center for Coastal Studies. Order number 40ENNF Robbins, J. and D. K. Mattila. 2001a. Monitoring entanglements of humpback whales (Megaptera novaeangliae) in the Gulf of Maine on the basis of caudal peduncle scarring. Paper submitted to the Scientific Committee of the International Whaling Commission. SC/53/NAH25. Robbins, J. and D. K. Mattila. 2001b. Gulf of Maine humpback whale entanglement scar monitoring results, Center for Coastal Studies. Order number 43EANF Robbins, J. and D. K. Mattila Gulf of Maine humpback whale entanglement scar monitoring results, Report to the National Marine Fisheries Service. Order number 43EANF Robbins, J Age structure of the Gulf of Maine humpback whale population. Report to the Northeast Fisheries Science Center, National Marine Fisheries Service. Order number 40EQFN pp. Sokal, R.R. and F.J. Rohlf Biometry: the principles and practice of statistics in biological research. W.H. Freeman and Co. New York. 859pp. Weinrich, M. T., M. Martin, R. Griffiths, J. Bove and M. Schilling A shift in distribution of humpback whales, Megaptera novaeangliae, in response to prey in the southern Gulf of Maine. Fishery Bulletin 95(4): Woodhead Buote, Carolyn S Detecting humpback whale/fisheries interactions in the Gulf of Maine: the effectiveness of fluke scar analysis. Master s Thesis. University of Massachusetts, Boston Massachusetts. 13

15 Table 1: Entanglement events used to evaluate methods. These are a subset of all entanglement cases, in which there was clear documentation of the caudal peduncle at the time the animal was entangled. While this excludes some cases that were known to involve the caudal peduncle, it was intended to ensure that the configuration of gear and/or raw injuries could be directly compared with later samples. Event First Clean Evidence of Injuries sustained?* Most recent Elapsed Coded Report (max) Involvement Leading Edges Dorsal Peduncle Ventral Peduncle Flanks Sighting Days likelihood of entanglement Taper 04/08/98 08/13/02 E Y Y Y N 07/18/ High Entropy 05/15/98 05/15/98 E Y N Y N 08/17/03 1,920 High Adirondack 05/27/98 05/27/98 R Y Y Y Y Putter 06/22/98 07/02/98 E Y Y Y N 08/26/03 1,881 High CCS /25/00 08/25/00 R Y Y X X Tribble 10/20/ E Y Y X X Overcast 09/18/01 09/23/01 R Y Y Y Y CCS /17/02 06/17/02 E Y Y Y Y Perseus 09/09/02 09/10/02 R Y Y X X Hat Trick 05/31/03 07/10/03 R Y Y Y N 07/30/03 20 High Tanith 06/23/03 06/24/03 R Y N X N 08/28/03 65 High Ravine 07/14/03 07/14/03 E Y Y Y N 10/17/03 95 High E= gear observed at feature, R= raw injuries indicative of entanglement, Y = wrapping injuries, notches or other tissue damage, N = no damage observed, X = could not determine 14

16 Table 2: Full body examination of carcasses for patterns of entanglement-related scarring. Note that when high probability scarring was present anywhere on the body, it was also present on the caudal peduncle/flukes Strandiug Date Catalog ID Stranding Network ID Recent Entanglement Evidence 1 Disentangled Prior to Death? Locations of high probability scarring Caudal Flippers Mouth Body Peduncle/Flukes 05/27/01 CCS-0100 MH01777MN No No Y N N N 10/01/01 Pitfall MH01975MN No No Y N N N 06/07/02 CCS-0207 MAL02170MN Yes 06/03/02 Y Y Y N 10/02/02 CCS-0235 CCSN Yes No Y Y Y Y 10/18/02 Squiggle 02 calf CCSN No No N N N N 1 Based on the presence of raw injuries or sighting with gear 15

17 Table 3: Sampling and analysis summary, order number 43EANF Year Number of Images Examined Percent from whale watching platforms Unique Individuals Cumulative Individuals (1997+) % % % Table 4: Inter-annual increases in diagnostic scarring, Period N* Scar acquisition rate Start End % ± % ± % ± % ± % ± 7.64 *Number of animals with comparable coverage in two consecutive years. Table 5: Demographic characteristics of animals that acquired high probability scars versus those involved in documented entanglements, The results shown are the percentage of events, not of the class category. Sample sizes are shown in parentheses. Class High probability scars 1 Documented entanglements Calf 5.4% (3) 12.5% (3) Juvenile 28.6% (16) 12.5% (3) Unknown % (11) 44.4% (12) Mature 46.4% (26) 33.3% (9) Female 71.7% (35) 63.6% (14) 1 Injuries acquired after a baseline sample, or obviously raw during the sampling period; however, the ability to detect raw injuries (such as on calves) was limited prior to The vast majority of animals in the unknown maturational class are thought to be juveniles (Robbins 2003) 16

18 Figure 1: Examples and interpretations of caudal peduncle scarring a). Scarring present, but not clearly characteristic of a previous entanglement. b). Apparently minor wrapping on the leading edge of the flukes and notches in the dorsal peduncle. c). Wrapping injuries and tissue loss on the leading edges of the flukes and notches on the dorsal peduncle. d). Deformation of the flukes due to extensive damage at the leading edges. A confirmed entanglement was shed in the early 1990s, but the tissue remained aggravated in 2002.

19 Figure 2: Scars acquired by Overcast in All images were obtained during a disentanglement event on September 23, No caudal peduncle images are available prior to the entanglement. a). Left side of the caudal peduncle. b). Right side of the caudal peduncle. c). Right side of the caudal peduncle. d). Left side of the fluke. Note the wrapping lines at the leading edge.

20 Figure 3: Examples of rapid healing at the dorsal caudal peduncle and flanks. a). Caudal peduncle of Nile on October 4, 2001, after having shed a documented entanglement. Note the prominent white scuffing from a caudal peduncle wrap. b). Nile on September 29, The prominent caudal peduncle wrap has been reduced to a minor notch on the dorsal caudal peduncle. c) Left caudal peduncle of Freckles on July 17, Note the raw injuries on the leading edge of the flukes, the flanks and the dorsal caudal peduncle. d) Freckles on May 6, Note that the injuries to the leading edges have healed to white, wrapping scars, while injuries to the flanks and dorsal peduncle have been reduced to minor notches. The majority of the healing had already occurred in 2003.

21 Figure 4: Humpback whale carcasses with entanglement evidence. Note that the injuries sustained were no more outwardly significant than observed on animals known to have survived entanglements a) CCS-0207 (MAL02170MN), an unnamed humpback whale that stranded four days after a successful disentanglement. Photo credit Marine Animal Lifeline. b) CCS-0235 (CCSN02-255), an unnamed humpback whale calf that stranded with entanglement evidence, including the imprint of line (right). Thought to be the same animal reported entangled (dead) one day prior.

22 Figure 5: Examples of raw injuries interpreted to be entanglement related in None are known to correspond to reported entanglement events.