TEMPORAL VARIATION IN BAT WING DAMAGE IN THE ABSENCE OF WHITE-NOSE SYNDROME

White-nose syndrome (WNS) is an emerging infectious wildlife disease that has killed an estimated 5.7–6.7 million bats in the eastern United States and Canada since its discovery in winter 2006 (US Fish and Wildlife Service, 2012). The disease is associated with a cold-adapted fungus, Geomyces destructans, which infects the dermis of cave-hibernating bats during winter (Blehert et al., 2009). Geomyces destructans grows optimally between 1 C and 15 C (Gargas et al., 2009), which coincides with the body temperature of cave-hibernating bats that have entered torpor (Cryan et al., 2010). Torpid mammals may have a decreased immune response, and this is hypothesized to allow the fungus to grow unchecked (Cryan et al., 2010). Wing damage associated with WNS has been documented, and researchers have proposed that it could contribute directly to WNS-related mortality by impairing flight (Reichard and Kunz, 2009), disrupting osmoregulation, or other physiologic disruption (Cryan et al., 2010).

Current methods of G. destructans detection include histologic examination of wing tissue, wing punches, or swabs for PCR testing, tape lifts for morphologic identification of fungi and swabs for culturing fungi (Dobony et al., 2011; USGS-NWHC, 2012). These methods can be expensive and time-consuming, and, in the case of histologic examination, they also require wing biopsy of live bats or euthanasia for more complete assessment. To address these challenges, the US Geological Survey-National Wildlife Health Center (NWHC) limits the number of samples that may be submitted from a single site (USGS-NWHC, 2012). The field researcher must, therefore, choose wisely when deciding which bats to include in the sample.

Wing damage has been documented (Meteyer et al., 2011) in bat maternal colonies in WNS-affected areas during early summer (Reichard and Kunz, 2009) and could be used as a screening tool for selecting bats that are more likely to be positive for G. destructans. Reichard and Kunz developed the Wing Damage Index (WDI) for basic field assessment of the severity of wing membrane damage (Reichard and Kunz, 2009). The NWHC currently includes WDI as one of the criteria that researchers may use to determine whether they should submit tissue samples or euthanized animals for WNS testing (USGS-NWHC, 2012). Bat wing damage has become an important topic since WNS, both because of its potential as a screening tool and because wing damage itself may contribute to WNS mortality (Cryan et al., 2010).

Two recent studies have elaborated on the work of Reichard and Kunz (2009), with noticeable differences in results (Francl et al., 2011; Fuller et al., 2011). Fuller and colleagues conducted continued studies at the same New England maternal colonies as Reichard and Kunz (2009) and found that WDI was highest in early summer, with moderate to severe damage occurring in >50% of bats in May, then, significantly lower in early August (Reichard and Kunz, 2009; Fuller et al., 2011). Conversely, in a large-scale, mist-netting survey from mid-May through mid-August at locations spanning five WNS-positive states, Francl and colleagues found that moderate to severe damage was rare, and the decrease in WDI over summer was weak and not statistically significant (Francl et al., 2011). The reason for this disparity is unclear, but Francl and colleagues proposed that the colonies evaluated in the other studies were in a localized area heavily affected by WNS, resulting in a higher occurrence of wing damage. This statement is supported by their observation that WDI was higher in bats surveyed in areas closer to WNS-positive caves. However, wing damage has not been investigated in WNS-negative bat populations, so the relationship between WNS and wing damage remains uncertain. The ability to link WDI with the likelihood of G. destructans infection is limited because we do not know how much wing damage occurs in uninfected populations. The data presented here provide a baseline reference for “typical” wing damage observed among presumed WNS-negative bat populations using the big brown bat, Eptesicus fuscus, as the model species.

We inspected 7,847 bat carcasses submitted for rabies testing in Illinois from 2005 and 2008–2010, then compared the frequencies of non-WNS wing damage categories by month, year, sex, and age group. Carcasses collected in 2006 and 2007 were not included because they had been incinerated. Illinois is presumed to have been WNS-negative during our sample period because the Illinois Department of Natural Resources reported no detection of G. destructans in PCR tests and fungal cultures of wing punches and swabs collected at hibernacula and maternal colonies from 2010 to 2012 (Kath, pers. comm.). We had previously identified species, sex, and age group for bats that had been tested for rabies by the Illinois Department of Public Health (IDPH) and compiled that information into a database including collection date, location, and exposure to humans or domestic animals, then stored the bats at −20 C. We sorted bats into two age groups (adults and juveniles) by transilluminating the wings and observing the gross morphology of the phalangeal-metacarpal joints. Bats with rounded epiphyses at the phalangeal-metacarpal joint were considered adults (Brunet-Rossinni and Wilkinson, 2009). We clustered collection dates into five time frames, which we refer to as “seasons”: spring (15 April–8 June), early summer (9 June–18 July), late summer (19 July–22 September), fall (23 September-5 November), and hibernation (6 November–14 April). These dates approximate the different activity periods of E. fuscus in Illinois, with spring representing the period when females are pregnant, early summer representing the period when females are nursing pups, late summer representing the period when juveniles are independent, and fall representing the “swarming” period when bats leave their summer habitats and return to hibernacula (Hofmann, 2008).

We later thawed the bat carcasses, placed them on a light box to transilluminate the wing membranes, and characterized wing damage as described by Reichard and Kunz (2009). We determined wing damage scores ranging from no damage (score = 0) to severe damage (score = 3) for three categories (discoloration, holes, and membrane loss) then compiled those into a composite WDI (Fig. 1; Reichard and Kunz, 2009). The Reichard classification for flaking and necrosis was not applied to the bats in this study to avoid possible confusion with similar changes we attributed to postmortem freeze artifact. Wing discoloration refers to areas of pallor where the color was lighter than surrounding membrane. Discoloration scores (DS) were determined as follows: five or fewer areas of discoloration, DS = 0; from six areas of discoloration to 50% of membranes discolored, DS = 1; 50–90% of membranes discolored, DS = 2; and >90% of membrane discolored, DS = 3 (Fig. 1A, B). For hole scores (HS), we used the following: no holes or small pinholes, HS = 0; for holes <0.5 cm in diameter, HS = 2; and for holes >0.5 cm in diameter, HS = 3 (Figs. 1A&B). For membrane loss scores (MLS), we used this measure: no loss, MLS = 0; tears at edge <1 cm, MLS = 2; and membrane loss >1 cm from edge, MLS = 3 (Figs. 1 B, C). The greatest of the three scores was taken as the composite index, WDI (Figs. 1 A–C; Reichard and Kunz 2009).

Of the 7,847 bats examined for inclusion in this study, 1,327 (16.9%) met the criteria for selection. Freezing and thawing may cause skin to peel, so we did not assess two other types of documented wing damage: flaking forearm and necrotic tissue. These bats were still assessed for areas of discoloration, holes, and membrane loss if these three categories of damage occurred in isolation from areas of flaking and necrotic tissue. We excluded bats if they had areas of discoloration, holes, or membrane loss that occurred within an area of flaking or necrotic tissue. Bats that were reported as caught by pets, were dehydrated or decomposing, or were apparently damaged when captured (i.e., unhealed tears in wing membrane or unhealed broken limbs) were also excluded. There is inherent uncertainty in determining whether wing injuries occurred before or after capture. To maintain consistency in evaluating individuals, we classified areas of discoloration, holes, or tears that had rounded borders as premortem damage. We classified areas of discoloration, holes, or tears that had jagged edges to be postmortem and did not include that damage in our analyses. Additionally, we excluded any areas of discoloration associated with peeling skin, with the assumption that the discoloration was the result of a loss of layers of skin that may have resulted from freezing and thawing.

We chose bats previously collected for rabies testing in Illinois between 2005 and 2010 because they represent a large sample size of individuals that were very unlikely to have been previously exposed to G. destructans. However, there are inherent biases associated with our sample because of differences in collection and preservation methods and from artifacts of freezing tissues. The sample may overestimate wing damage in natural bat populations because of possible damage during collection, because of rabies testing and freezing, and because bats with wing damage may have had impaired flight that made them more likely to be caught and submitted for testing. We made every effort to minimize the biases of postmortem damage using the criteria described in the previous paragraph. These efforts may have resulted in underestimation of wing damage in natural populations because some wing damage that occurred before collection may have been excluded as postmortem.

Following the selection process, we compiled wing damage scores for 1,327 bat carcasses, including four cave-hibernating species (Eptesicus fuscus, n = 1108; Myotis lucifugus, n = 61; M. septentrionalis, n = 15; and Perimyotis subflavus, n = 1) and four species that do not hibernate in caves (Lasionycteris noctivagans, n = 72; Lasiurus borealis, n = 50; and L. cinereus, n = 13; and Nycticeius humeralis, n = 7). The sample sizes were small for all species except E. fuscus. We did not have individuals representing all five seasons for any species except E. fuscus, so our models include E. fuscus only. We modeled age, sex, season, and year as predictors of wing damage using generalized linear models (GLMs), then used Akaike information criterion (AIC) to select the best models (R Development Core Team, 2010; Pinheiro et al., 2011). We modeled the three damage categories (discoloration, holes, and membrane loss) and composite WDI separately to clarify which categories of damage contribute to differences in wing damage between sexes, age groups, seasons, and years. We evaluated the models by comparing ΔAIC scores, where models with ΔAIC<2 were considered competitive, and compared the probability that any model could predict damage using Akaike weights (Burnham and Anderson, 2002). To distinguish between models of the different categories of damage, we use DS to refer to discoloration scores, HS to refer to hole scores, MLS to refer to membrane loss scores, and WDI to refer to the composite index we created by combining DS, HS, and MLS.

All species examined had individuals with composite WDI scores of 1 or 2, with the exception of P. subflavus in which only one specimen was evaluated. Individuals rating a composite WDI of 3 were observed in all species with a sample size greater than 50 (Fig. 2 A, B). We did not include species as a predictor in our models because the sample size was small for all species except E. fuscus. The small samples of species other than E. fuscus do not represent all five seasons, so data depicted in Figure 2 may overestimate or underestimate typical wing damage in these species due to possible seasonal effects. We did not observe any bats with severe discoloration (DS  =  3).

In our GLMs of E. fuscus, AIC showed strong support for a single model of discoloration. The best model for the discoloration score (DS) included season (clustered into three groups: early summer; late summer; and a combined fall-winter-spring category) and year (clustered into two groups: 2005 and 2008–2010; Table 1). The proportion of moderate discoloration (DS = 2) was highest in early summer, lowest in late summer, and intermediate in fall through spring (Fig. 3A). By year, the proportion of moderate discoloration (DS = 2) was lower in 2005 than it was in other years (Fig. 3A).

All models for wing holes and membrane loss in E. fuscus were weakly supported by AIC. The four competing models for HS included combinations of season, year, and age (Table 1), but each of these models represented only a modest improvement over the intercept-only (null) model (ΔAIC scores 0–1.7 for the top models vs. 2.3 for intercept-only; Table 1). Also, Akaike weights were low for all models of HS (Table 1), indicating a low probability that any one model could be used to predict wing holes. There is a similar lack of resolution between models for MLS, in which the top four models performed only slightly better than the intercept-only model (ΔAIC scores 0–1.4 for the top models vs. 2.2 for intercept-only; Table 1). Akaike weights were low for all models of MLS and failed to resolve a single model with high probability. The data show that membrane loss is rare, with very little difference between groups (Fig. 3C). There were three competitive models for composite WDI scores, all of which included two seasonal groups (late summer vs. others), and which sometimes included year (2005 vs. 2008–2010, or all four years as separate parameters; Table 1). The ΔAIC scores show that all three composite WDI models are much better than the intercept only, but only marginally better than two other composite WDI models that cluster season into three groups (early summer vs. late summer vs. others; Table 1). Again, Akaike weights assign relatively small probabilities to several models. Unlike models for score DS, AIC analyses do not find strong support for any specific predictive models of HS, MLS, or composite WDI in our E. fuscus sample using sex, age, season, or year as predictors.

In our study of presumed WNS-negative bats, we found all three of the wing damage categories (discoloration, holes, and membrane loss) previously documented by researchers in WNS-positive areas. In particular, wing discoloration and holes were common. We observed bats with moderate to severe composite WDI that had been collected in Illinois in 2005 (Fig. 1B, C), 2 yr before WNS was first documented in New York (Blehert et al., 2009). We also observed moderate to severe composite WDI in species that do not hibernate in caves and, therefore, are not known to be susceptible to WNS (Fig. 2A). Moderate and severe wing damage occurs in the absence of WNS.

Twenty-five percent of the E. fuscus that we evaluated had a composite WDI of 2 or 3 (Fig. 2B). Our observed frequencies of moderate and severe composite WDI (2 or 3) were similar to those reported by Reichard and Kunz and by Fuller and colleagues, and much higher than that reported by Francl, despite the fact that their studies were conducted in areas where WNS is widespread (Reichard and Kunz, 2009; Francl et al., 2011; Fuller et al., 2011). The modest differences between our sample and the two New England maternal colony studies could be explained by interspecies variation in wing damage between E. fuscus and M. lucifugus. Although we could not include M. lucifugus in our models because of small sample size, we did observe a lower frequency of composite WDI = 2 and 3 in our sample of 61 individuals (less than 15%; Fig. 2B). However, that observation remains tentative because we did not have M. lucifugus from all seasons studied. Interspecies differences cannot account for the large disparity between our results and those of Francl et al. (2011). They observed composite WDI of 2 or 3 in <1% of E. fuscus and <2% of M. lucifugus. Our sample differed from the live-caught bat studies of Reichard, Fuller, and Francl because our bats were submitted from numerous sources to the IDPH, subjected to rabies testing, and stored in freezers before being examined for wing damage. Therefore, our sample may overestimate wing damage because of our collection and storage methods.

We excluded bats that appeared to have been damaged during capture. We expected that to result in underestimates of damage because we would sometimes be excluding damage that occurred for reasons not related to capture. It is also possible that bats with wing damage were more likely to be captured and submitted for rabies testing, and that additional damage may occur in postmortem handling and freezing, which would positively bias our data. Our analyses of damage by year indicated that damage did not substantially increase with time spent frozen: DS were lower in 2005 than in subsequent years (Fig. 3A), whereas HS, MLS, and composite WDI showed only a slightly higher frequencies in 2005 (Fig. 3B–D). These increases in HS, MLS, and composite WDI were not sufficient to place 2005 as a predictor in any of the competitive models (Table 1).

Of the three categories of wing damage we surveyed, DS was the only category that had a single best model using our parameters (Table 1). The top model for DS included three seasons (early summer vs. late summer vs. the remaining three seasons combined) and year (2005 vs. 2008–2010; Table 1). Our sample size in fall (22 September–5 November) was small (n = 45). Therefore, it is uncertain if our reported frequency of discoloration represents what actually occurs during that season. We found smaller differences by season and year for HS, MLS, and composite WDI, which were not disparate enough to determine a best predictive model (Fig. 3B–D). The AIC ranks the best set of predictive models but does not reject or accept models (Burnham and Anderson, 2002). The numerous ΔAIC scores<2 and low Akaike weights of the models for HS, MLS, and composite WDI indicate that our parameters were not reliable predictors of these categories of damage.

We found that E. fuscus DS increased from spring (15–8 June) to early summer (9 June–18 July), then decreased from early summer to late summer (19 July–22 September; Fig. 3A). The general trend is consistent with composite WDI data for M. lucifugus from WNS-positive regions. Both field studies of wing damage conducted at the New England maternal colonies saw a similar increase and decrease in composite WDI over the maternal season (Reichard and Kunz, 2009; Fuller et al., 2011). Meteyer and colleagues noted that wing damage peaked on day 27 of their captive study of WNS-positive bats (Meteyer et al., 2011). They reported collecting bats from hibernacula in May, which would make day 27 sometime in late May through June (Meteyer et al., 2011). However, those investigators found that composite WDI peaked in late May through June, whereas we saw DS peak in late June to early July. In a more recent study, Meteyer et al. (2012) presented data suggesting that damage increases in severity for weeks after G. destructans–infected bats arouse from torpor because of a prolific inflammatory response that begins when the bat becomes euthermic. It is unclear if our observed seasonal increase in DS is also related to an inflammatory response, or what seasonal factor would trigger this response. It should also be noted that we very rarely observed damage as severe as the damage shown by Meteyer et al. (2012). Some wings with composite WDI = 3 were more severely damaged than others. The addition of a fourth WDI category (WDI = 4) might help to distinguish between the severe damage that is occasionally seen in WNS-negative populations and the extremely severe damage that is frequently seen in WNS-positive populations.

Additional wing damage analyses for other susceptible species in WNS-negative regions would help determine whether wing damage prevalence is similar to our observations of E. fuscus. Also, the Reichard scoring for flaking and necrosis was not applied to the bats in this study to avoid possible confusion with postmortem freeze artifact (as described in “Methods”). Field data from live bats in WNS-negative regions would help determine whether these other types of damage are specific to WNS and could be used more reliably as a prescreening tool. The WDI scoring system is a potentially useful tool for monitoring bat population health generally, and possibly for choosing which bats should be tested for G. destructans infection. Further studies documenting wing damage in WNS-affected and unaffected areas will improve our ability to make that distinction. Until unambiguous indicators or WNS-specific damage are found, these data show that field observation of wing damage alone do not correlate with WNS. Guidelines are available to help minimize euthanasia of bats for WNS testing (USGS-NWHC, 2012), and these will likely change as new data and techniques become available.

We thank Carol Meteyer and two anonymous reviewers for their comments on an earlier version of this manuscript. Funding was provided by Bat Conservation International Student Scholarships, the Illinois Department of Natural Resources Wildlife Preservation Fund, and the University of Illinois Program in Ecology, Evolution and Conservation Biology Graduate Student Summer Research Grants.