Skin Deep: An Assessment of New Zealand Blue Whale Skin Condition

Skin condition assessment of wildlife can provide insight into individual and population health. Yet, logistics can limit skin condition assessment of large whales. We developed a standardized, quantitative protocol using photographs to assess skin condition of blue whales in New Zealand, and demonstrate the value gained by testing hypotheses, documenting new morphologies, and establishing baselines that can be monitored for change. We reviewed a photo-identification catalog to compile common markings, categorized markings according to existing definitions, and described markings not previously documented. Photographs of blue whale skin (n=1,466) were assessed to quantify marking prevalence, severity, and co-occurrence patterns. Of the whales assessed (n=148), 96.6% had cookie cutter shark bites, 80.4% had blister lesions, 56.0% had pigmentation blazes on the dorsal fin, and 33.7% had holes in the dorsal fin. Additionally, 35.8% had “starburst” lesions, a previously undocumented marking. Blister and cookie cutter shark bite severity did not accumulate linearly, indicating that the two marking types are unrelated. There was a positive relationship between blister severity and number of starbursts, indicating that the two could be related; based on morphological similarities, starburst lesions may derive from ruptured blisters. Whales with holes in their dorsal fin had significantly higher blister severity than those without, indicating that these markings could be related; this is supported by observed blisters on dorsal fins of blue whales. There was a significantly higher probability of fresher cookie cutter shark bites on whales observed at more northerly latitudes, but no relationship between blister severity or number of starbursts and latitude. These latitudinal patterns indicate that blue whales in New Zealand accumulate cookie cutter shark bites at more northerly latitudes; this finding is supported by the known range of cookie cutter sharks in New Zealand waters. Of the eight individual whales re-sighted across multiple years, there was no uniform pattern in lesion change over time, however individual cases revealed lesion healing over a multi-year timeframe. Our protocol for quantifying skin condition can be applied to any cetacean photo-identification catalog, and can be used to compare across individuals and populations, and explore causal links between skin condition and cetacean health.

cookie cutter shark bites", comparable in appearance to blue whales observed in Australia and 63 better in appearance than blue whales observed in Antarctica. However, no quantitative 64 assessment of skin condition has been conducted to-date for New Zealand blue whales. In this the leading edge of the dorsal fin. Holes were classified as any circular, hollow punctures through the dorsal fin, excluding notches or tears (Table 1). 121 Lesions were further categorized into sub-classes: cookie cutter shark bites, blister-like vesicles 122 (hereafter referred to as "blisters"), and starbursts. Cookie cutter shark bites were defined as 123 small, crater-like wounds of variable depth that are round, oval, or crescent in shape (Best and 124 Photopoulou, 2016;Dwyer and Visser, 2011;Jones, 1971). Cookie cutter shark bites were 125 subsequently further classified by phase of healing based on bite morphology and pigmentation 126 ( Table 1). Blisters were defined as air or fluid-filled elevations of the epidermis, observed either 127 as a single lesion or a cluster, distinguishable from mottled skin pigmentation by their three-128 dimensionality (Brownell et al., 2007;Hamilton and Marx, 2005). Blisters were of comparable 129 size to cookie cutter shark bites or slightly larger, but rounder in shape than the typically oval 130 bite marks (Brownell et al., 2007). Starbursts were not previously described in the literature, yet 131 we noted their presence on the epidermis of many blue whales; hence, we introduce this new 132 skin lesion type, defined as light-colored markings with a clear central origin and spindly tendrils 133 that extend outward, away from the origin (Table 1).

134
When lesion presence was determined, the photographs were evaluated for lesion severity.

135
Reference images for each severity category were obtained and agreed upon by all analysts prior 136 to the assessment process. Each severity category for cookie cutter shark bite and blister lesions 137 was assigned a numerical score (Table 2).

138
Data processing: To minimize bias arising from how much of the whale's body was visible in 139 the photographs, we designated an "assessment area" of the body (Hamilton and Marx, 2005).

140
The assessment area was defined by creating a gridded box and overlaying it on the photo(s) in 141 which the greatest amount of the whale's body was visible for each identified individual, using 142 the dorsal fin as a visual control. The box was aligned so that the upper bound crossed the 143 intersection of the trailing edge of the dorsal fin and the dorsal ridge, and resized so that 25% of 144 the box was caudal to the dorsal fin and 75% was cranial to the dorsal fin ( Fig. 1). For each 145 whale assessed, the percentage of the assessment area visible above the water was recorded for 146 each side of the whale. Images of each assessed whale with the assessment area illustrated (e.g.,

147
Fig. 1) were compiled for skin condition assessment by analysts.

148
Photo brightness, sharpness, contrast, and saturation were manipulated in the Windows Photos 149 application to increase visibility of markings. Each assessment was also given an overall photo 150 quality score of poor, fair, good, or excellent based on distance, clarity, brightness, and focus of 151 the images (Supplementary material, Table S1). The skin condition assessment was conducted 152 for the side with poorer quality photos first to minimize observer bias from assumptions based on 153 prior knowledge of markings. All individuals photographed in each calendar year were evaluated 154 for the presence or absence of lesions, blazes, and holes within the assessment area. For cookie 155 cutter shark bites and blisters, the overall severity score was also recorded ( Table 2). For cookie cutter shark bites, the presence or absence of each phase (Table 1) was also determined. For of the whale. Where presence or severity could not be determined due to photo quality it was 159 scored as "could not be determined" (CBD).

160
Two analysts (DRB and ALP) conducted the skin condition assessment for all blue whales 161 independently. The reference images demarking the assessment area for each individual ensured 162 that the two analysts evaluated exactly the same area on the whale. Subsequently, the two 163 independent assessments were reconciled and in instances where there was a difference (n=19, 164 12.8%) a third, independent analyst (LGT) assessed the photographs for which there was a 165 discrepancy to assign the final classifications and scoring. This process was meant to minimize 166 bias from any one observer and ensure that our assessment scoring was robust (Bradford et al., 167 2009;Maldini et al., 2010;Yang et al., 2013).

168
Statistical Analysis: Once the data were processed, statistical analyses were conducted to test a 169 series of six hypotheses (Table 3). A portion of the photo-identification dataset only included 170 images of one side of the whale. Therefore, we tested the hypothesis that (H1) skin condition on 171 one side of a whale was representative of the other side. We compared markings on both sides of 172 whales where data were available and photo quality was rated as fair or better. Paired t-tests were 173 used to compare cookie cutter shark bite and blister severity scores, the number of starbursts, and 174 blaze presence between the left and right sides.

175
The occurrence rate of skin lesions was assessed by calculating the proportions of whales with 176 no, mild, medium, and severe cookie cutter shark bites and blisters. This assessment was 177 conducted separately for adults and calves to address the hypothesis that (H2) these lesions 178 accumulate with age, and therefore could potentially be used as a proxy for age class. To test the 179 hypothesis that (H3) whales accumulate cookie cutter shark bite and blister severity at a similar 180 rate, a confusion matrix was generated to examine their co-occurrence pattern. A heatmap was 181 produced from the confusion matrix to visualize this co-occurrence pattern. 182 We hypothesized that (H4) starburst lesions derive from blisters, and evaluated this theory by 183 assessing the relationship between starbursts and blister severity, and starbursts and cookie cutter 184 shark bite severity. The total number of starbursts counted per whale were first standardized by 185 the proportion of the assessment area evaluated (left-side count + right-side count * proportion of 186 total assessment area visible on both sides). The standardized number of starbursts metric was 187 log-transformed for further analysis to account for skew toward zero. Linear regression was used 188 to examine the relationship between the standardized number of starbursts and both blister 189 severity score and cookie cutter shark bite severity score, using the relationship with cookie 190 cutter shark bites to test the null hypothesis that starburst are not related to cookie cutter shark 191 bites.

192
Furthermore, we also hypothesized that (H5) holes in the dorsal fin of blue whales derive from 193 ruptured blisters on the dorsal fin. T-tests were used to test for any difference in blister severity  Zealand.

218
Using individuals with fair or better quality photographs for both sides (n=23), we determined no 219 significant difference in cookie cutter shark bite severity score, blister severity score, presence or 220 absence of blazes, or the number of starbursts between the left-and right-hand sides (paired t-221 tests, t<1.2, p>0.2 for all). With this support for H1 we justifiably assumed the same skin 222 markings were applicable for both sides of whales where only one side was photographed or 223 photo quality was poor for one side.

224
Marking rates and severity: Of the whales assessed (n=148) present, 18.1% had mild blister severity, and no calves had medium or severe blistering (Fig. 2), 231 lending support for H2. More whales were given a score of CBD for blister severity than for 232 cookie cutter shark bite severity, indicating that once presence of the lesion was determined, 233 severity was more difficult to ascertain for blisters than for cookie cutter shark bites (Fig. 2).

234
Co-occurrence of marking types: According to the confusion matrix and heat map (Fig. 3) blister 235 severity and cookie cutter shark bite severity do not accumulate linearly, thus rejecting H3. The 236 most common co-occurrence pattern was medium cookie cutter shark bites with medium blistering, followed by severe cookie cutter shark bites with no or mild blistering, and mild 238 cookie cutter shark bites with severe blistering (Fig. 3).

264
Cookie cutter shark bite severity increased for two individuals (0 to 1 over a one-year period and 265 1 to 2 over a three-year period, respectively), declined for one individual (3 to 2 over a seven-

266
year period), and did not change for five individuals (Supplementary material, Figure S2).

267
However, examination of healing of individual cookie cutter shark bites over time for one 268 individual whale revealed progression from phase 2 to 3 over a 38-month period, from phase 2 to 269 4 over a 38-to 69-month period, and from phase 3 to 4 over a 31-to 69-month period (Fig. 5). It 270 should be noted that these healing times represent a maximum amount of time between healing 271 phases considering the gaps between photographic records.

293
The distribution of cookie cutter sharks is thought to be circumpolar in tropical and subtropical 294 waters, with a geographic range that is limited by water temperature (Jahn and Haedrich, 1987;295 Jones, 1971;Nakano and Tabuchi, 1990). The southernmost record of a cookie cutter shark is -  Table S2). 309 We determined that New Zealand blue whales also have a high rate of blister presence (80.4%). reported to be between 60-80%, which is substantially higher than the prevalence of blisters for 315 sperm whales (Physeter macrocephalus, 40-60%) and fin whales (Balaenoptera physalus, 0-316 10%) in the same study region (Martinez-Levasseur et al., 2011). Blisters have also been Marx, 2005). It therefore appears that blue whales may exhibit higher blister prevalence than 320 other large whale species, and that among blue whale populations, the New Zealand population 321 may exhibit comparable or somewhat elevated rates of blister presence.

322
The hypothesis that cookie cutter shark bite and blister severity accumulate at a similar rate (H3) 323 was rejected (Figure 3, Table 3), indicating that the causes for these two lesion types are distinct 324 and unrelated. There was a positive relationship between blister severity and the number of 325 starburst lesions, indicating that these two lesion types could be related (H4). Based on 326 morphological characteristics, it is possible that ruptured blisters could lead to starburst lesions, 327 however this causal link is purely speculative. Blister severity was also significantly higher for which may also result in more UV exposure during summer foraging. Although not quantified in 337 this study, analysts noted that blistering appeared common on the caudal peduncle of blue 338 whales, which may be a region of the whale subjected to increased UV radiation, lending support 339 to this hypothesis.

340
The hypothesis that skin lesions accumulate over time and can be used as a proxy for age class 341 (H2) could not be fully explored in this study due to a low sample size of individual re-sightings 342 and no known-age individuals (other than calves). It would therefore be valuable for this skin 343 condition assessment method to be applied to cetacean photo-identification catalogs for 344 populations with more re-sightings and known-age individuals. In addition to accumulating 345 markings over time, some lesions heal over time, as evidenced by the example in Figure 5. A 346 skin condition assessment with more re-sighted individuals than available for this study would 347 therefore also be useful in elucidating rates of lesion accumulation and healing over time.

348
In this study, we did not observe the tattoo-like skin disease documented on a blue whale in 349 Chile (Brownell et al., 2007) or rake mark scars from killer whale predation that were described not prey primarily on marine mammals in New Zealand waters; rather they specialize on feeding 356 on elasmobranchs such as stingrays (Visser, 1999(Visser, , 2005. The absence of killer whale rake mark

H1:
Skin condition on one side of whale is representative of the other side. Supported

H2:
Lesions accumulate with age and can be used a proxy for age class, therefore adults will have higher cookie cutter shark bite and blister severity scores.
Inconclusive (rejected for cookie cutter shark bites, supported for blisters)

H3:
Whales accumulate cookie cutter shark bite and blister severity at a similar rate. Rejected

H4:
Starburst lesions are derived from blisters, and will therefore be positively related to blister severity score and not cookie cutter shark bite severity score.

H5:
Holes in the dorsal fin are derived from blistering, therefore whales with holes in their dorsal will have higher blister severity scores.

H6:
Cookie cutter shark bite phase is related to latitude, and fresher bites are more likely at more northerly latitudes. Supported is aligned such that 25% of the total area is caudal to the dorsal fin, and 75% is cranial to the 522 dorsal fin. In this image, 100% of the assessment area is visible.