Field trial of a probiotic bacteria and a chemical , chitosan , to protect bats from white-nose 1 syndrome 2 3

syndrome 2 3 Joseph R. Hoyt, Kate E. Langwig, J. Paul White, Heather M. Kaarakka, Jennifer A. 4 Redell, Katy L. Parise, Winifred F. Frick, Jeffrey T. Foster, A. Marm Kilpatrick 5 Dept. Ecology and Evolutionary Biology, University of California, Santa Cruz, 95064, USA 6 Dept. Biological Sciences, Virginia Tech, Blacksburg, VA, 24061, USA 7 Wisconsin Department of Natural Resources, Bureau of Natural Heritage Conservation, 8 Madison, WI 53707, USA 9 Department of Molecular, Cellular & Biomedical Science, University of New Hampshire, 10 Durham, NH 03824, USA 11 5 Current Address: Center for Microbial Genetics and Genomics, Northern Arizona University, 12 Flagstaff, AZ 86011, USA 13 6 Bat Conservation International, Austin, Texas 78716 USA 14 *To whom correspondence should be addressed: akilpatr@ucsc.edu (A.M.K.) or 15 hoytjosephr@gmail.com (J.R.H) 16 17 18 19


Introduction
White-nose syndrome (WNS), caused by the fungal pathogen, Pseudogymnoascus destructans, has caused widespread declines in bat populations throughout eastern and midwestern North America and threatens several species with extinction 1-3 .Three species (Myotis lucifugus, Myotis sodalis, and Perimyotis subflavus) have declined by 70-90% across multiple states, and a fourth species, Myotis septentrionalis, has been extirpated from most sites within three years of WNS detection [2][3][4] , in part, due to highly connected bat communities 5 .
Although a few populations of M. lucifugus appear to be persisting at 10-25% of pre-WNS colony sizes, most colonies of this species have declined by >90% 4,6 .Several previously common species of hibernating bats are now relatively rare across large regions of the northeast USA 2,4,7 .Management interventions to reduce the impact of WNS on bat populations are needed to prevent further declines and restore bat populations.
Over the past seven years, several treatments for WNS have been explored and are in various stages of development, but none have been successfully tested in the field.Potential treatments to enable bats to survive hibernation have included volatile compounds released by bacteria, vaccination, chemical anti-fungals, and probiotic microbes (Table S1).The outcome of most lab and field trial studies is unclear, and there are currently no published reports of effective treatments from field trials (Table S1).Thus, at present, there are few tools for managers to reduce the impacts of WNS, and developing control options to reduce the severity of this disease among bats is a high priority 8,9 .
Our goal was to determine the efficacy of two treatments, Pseudomonas fluorescens and chitosan, in reducing WNS mortality in a field setting.Pseudomonas fluorescens is a ubiquitous bacterial species complex that is used as a fungal biocontrol agent in agriculture, and has been tested as a treatment for chytridiomycosis in amphibians [10][11][12] .A previous study on multiple isolates of P. fluorescens isolated from different species of bats showed a range of anti-P.destructans properties in vitro 13 .One strain, isolated from a hibernating Eptesicus fuscus in Virginia, reduced the number of lesions, and increased survival of little brown bats when applied at the time of infection in a laboratory in vivo trial.Chitosan is a biopolymer polysaccharide extract from crustacean shells, has powerful antimicrobial and wound-healing properties, is biodegradable and non-toxic, and is a widely used anti-fungal in agriculture 14 .Chitosan has shown promise in inhibiting growth of P. destructans in vitro and in reducing mortality in in vivo lab experiments 15 .
We performed a field trial with two simultaneous experiments to balance the strengths and weaknesses of each approach.In the free-flying experiment, we treated bats, and attached an integrated passive transponder (PIT) tag to determine the date they emerged from the site.This experiment allowed bats to behave normally and roost freely throughout the site.However, there was additional uncertainty in determining the survival of free flying bats (e.g., bats could move to another site midwinter, die at the site and be eaten by predators or escape from the site by an unknown exit and not be detected by the PIT tag receiver).To balance these unknowns, we also performed an experiment with bats in cages.Placing bats in cages, as has been done in most in vivo experiments to date 1, [16][17][18] , prevents bats from leaving the site and provides certainty about the survival of each bat.However, caging bats may alter their behavior (e.g.bats in the same cage may be disturbed when other bats arouse from hibernation).
. CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which

Results
On the day of treatment, P. destructans infection prevalence, fungal loads and weights of bats were very similar among treatment groups in both experiments (Figure 1).This suggests that the randomization of bats to treatment groups and experiments did not result in any initial differences.Infection prevalence and fungal loads in November were very similar to loads observed on M. lucifugus at other sites where the fungus has been present for at least one previous winter 19,20 .

Free-flying experiment
Of the 44 bats we treated, only one bat (from the control group) appeared to have left the site due to the disturbance of being handled/treated (it was detected by the PIT tag reader on the day of treatment November 20, 2015 and never again).We detected 17 of the remaining 43 bats on the PIT tag reader between December 9, 2015 and April 17, 2016, with 6 of 7 (P.fluorescens), one of six (control), and two of four (chitosan) bats having left the site on or after the assumed overwinter survival date of March 7, 2016 (Figure 2).We found three additional bat carcasses inside the site (two control and one chitosan-treated bats).The fraction of bats known to be alive and detected by the PIT tag reader after March 7 th (apparent overwinter survival) was 46.2% (6/13) for P. fluorescens-treated bats, which was significantly higher than 8.5% (1/12) for control bats (Figure 3; the remaining 7 bats had lost their PIT tag; see Methods).Apparent overwinter survival was 18.0% (2/11) for chitosan-treated bats which was not significantly different from control bats (Figure 3).
The last date a bat was detected on the PIT tag reader was significantly later for P. fluorescens-treated bats than control bats, and overall, was earlier for bats with higher fungal 6 loads in November (Figure 4).The last detection dates for chitosan-treated bats were not significantly different than untreated controls (Figure 4).We did not compare differences in fungal loads or UV-fluorescence among treatment groups in March for bats in the free-flying experiment because only three bats were found and recaptured when we visited the site on March 8.The remaining bats were likely in difficult-to-access portions of the mine.
Unlike in the free-flying experiment, P. destructans fungal loads in November were not a significant predictor of survival (likelihood ratio test: P = 0.60).Most of the bats still alive in the cages were in very poor condition, and only three of the sixteen bats that survived to be released were subsequently detected by the PIT tag reader (two chitosan and one control bat) (Fig 5).
Secondary measures of disease severity from the cage experiment showed non-significant differences or patterns that contradicted the survival results.Fungal loads on bats in March were not significantly different among treatment groups (Figure 6a) and disease severity, as measured by UV-fluorescence, was significantly higher for chitosan-treated bats than control bats (Figure 6b).

Discussion
White-nose syndrome has caused widespread declines in multiple species of bats throughout eastern and midwestern North America, with declines in M. lucifugus colonies in the first year of WNS detection averaging 79% 2,3 .Survival of untreated bats in the free-flying experiment was similarly severe, with 91% (95% CI: 62-99%) of control bats likely dying over the winter.Treatment with the probiotic, P. fluorescens, increased apparent overwinter survival more than five-fold by extending the last date of detection by a month into early spring.
Although this effect is substantial, over half of P. fluorescens-treated bats still likely died from WNS over the winter.In contrast, support for a protective effect of treatment with chitosan in reducing WNS mortality was mixed.Chitosan treatment increased survival in the cage experiment until March 8th, but few of these bats were subsequently detected by the PIT tag reader emerging onto the landscape, and disease severity, as measured by UV fluorescence, was significantly higher in chitosan-treated bats.
If treatment efficacy with P. fluorescens could be improved, P. fluorescens could provide a useful tool for conserving populations of M. lucifugus declining from WNS.One potentially important factor for future efforts with P. fluorescens, based on work in other systems, is whether the bacterial treatment persists or proliferates on the host species 21 .Increasing persistence and growth of P. fluorescens on bats by altering the dosage, or treatment frequency, or adding components to P. fluorescens solutions to encourage the formation of biofilms could help increase treatment efficacy 21 , assuming these alterations wouldn't have more deleterious .CC-BY-NC-ND 4.0 International license It is made available under a was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which side effects.Previous in vitro studies suggested that many different strains of P. fluorescens have anti-P.destructans effects 13 .Future studies could examine alternate strains of P. fluorescens isolated from different populations (e.g., persisting populations of M. lucifugus; 6 ) or other species of bats (the strain in this study was isolated from E. fuscus; 13 ).
The effect of P. fluorescens or chitosan in reducing WNS impacts on other species also has yet to be tested.The most important species to protect from WNS is M. septentrionalis, which suffers nearly 100% mortality, and is on a pathway to extinction 2 .To date, no treatments have been developed for, or tested on, this species in either the lab or field.This is despite M. septentrionalis being the most heavily affected by WNS, with few hibernacula still containing this species in the US 2,4,22 .
In addition, our results offer potential insights for the experimental design of future field treatment trials aimed at reducing WNS mortality in hibernating bats.Researchers often have to choose between cage-artifacts and concerns about bats leaving the site following treatment and uncertainty in the survival outcome for some free-flying bats.Our data suggest that the freeflying experiment was a better experimental design, despite some challenges.Bats in this experiment were able to roost and behave normally, and only one of 44 bats left the site on the day of the experiment, suggesting that disturbance of handling and treatment are unlikely to compromise experiments if treatment can be done quickly (treatment, weighing and banding required ~1 hr.underground in this study).In addition, lower November fungal loads prolonged apparent survival, as would be expected if bats were dying from WNS 22 .The main challenge of the free-flying experiment was uncertainty associated with the fate of animals that were never detected by the PIT tag reader but not found dead within the site.However, as noted above, the extent of mortality in control bats inferred in the free-flying experiment was very similar to the 9 WNS declines observed in populations of M. lucifugus sites, supporting our assumption that most bats that were not detected by the PIT tag reader after March 7 th did not survive the winter.
One final challenge with mixed treatment free-flying experiments is that mixing of treatment groups (e.g., probiotic bacteria being transferred from treated to control bats) might occur through direct social interactions or indirect contact via the environment.The significant differences we observed between survival of bats in the P. fluorescens and control treatment groups suggest that direct or indirect contact was insufficient to transfer significant amounts of probiotic bacteria among bats.
The cage experiment suffered from several shortcomings that, in hindsight, indicate this was a problematic design.In our experiment each cage contained all the bats in each treatment, due to a limited availability of space for mounting cages to natural substrate in a predatorprotected room, and to allow social bats to roost in groups.This resulted in pseudo-replication in this experiment, as in most previous laboratory studies on WNS 1, [16][17][18] .This is particularly problematic for studies of WNS, because in small cages bats appear to disturb other bats when they arouse from hibernation 23 , and increased arousal frequency is thought to be a key mechanism of WNS mortality 1, 24,25 .The fact that survival in the cage experiment was correlated with initial body mass, but not fungal loads, suggests that disturbance from other bats, or an inability to move to other locations within hibernacula, was more important than WNS in determining survival in this experiment.Together, these results indicate that the ideal design for a field trial (and for WNS challenge experiments more generally) is a free-flying experiment with mixed treatment groups in each site where bats have to pass through a PIT tag antenna to leave the site or are prevented from leaving the site (e.g. by sealing entrances, which may be very difficult).Sites where dead bats are relatively easy to find and are not eaten by predators (e.g.mice, rats, and raccoons) would reduce uncertainty in survival outcomes.If an experiment requires constraining bats within a site, one could use replicated cages (constructed of metal, as we used, to prevent mice from chewing into the cages) with a single bat in each cage to prevent cascading disturbances from infected bats, or groups of bats that are analyzed as individual data points.In addition, barriers to prevent larger predators (e.g.raccoons) from accessing the cages and eating the bats are an absolute necessity.Cages are not ideal in that they limit bats' movement within sites, but they offer higher certainty in terms of knowing the survival of each individual.
In conclusion, preventing population declines due to WNS in M. lucifugus and other species will likely require a combination of multiple approaches 8 .Potential strategies that could be combined with treatment include reducing the environmental reservoir of P. destructans 3,26 , protecting and facilitating growth of populations of M. lucifugus that are now persisting with WNS (possibly due to resistance that limits fungal growth to moderate loads 6 or increased fat stores that allow bats to tolerate infection 27 ), and improving summer and fall habitat for bats to increase reproduction and fat storage for hibernation.The latter two strategies would facilitate the evolution of resistance or tolerance which reduces the need for perpetual management action 28,29 .Finally, any strategy which slows or stops the very rapid local extirpations of M. septentrionalis colonies is urgently needed to prevent this species from extinction.

Methods
We performed the field trial on M. lucifugus bats in the winter of 2015-16 at an inactive mine in southwest Wisconsin where P. destructans was detected the previous winter 2014- 15.
The mine has one large (~3m tall by 5 m wide) entrance that was gated several years earlier, and a single smaller entrance that was sealed with a fine mesh metal screen the year of the gating.
We selected a site where P. destructans had been detected the previous year because lab trials with P. fluorescens had indicated that treating bats at the time of infection was more beneficial than treatment prior to infection 16 , and previous work suggests that most bats become infected early in the second year following P. destructans invasion, likely due to build-up of an environmental reservoir 3 .This site had 226 M. lucifugus in November 2014, before P. destructans was detected, but the colony had declined to 82 bats by March 2015.The average temperature where M. lucifugus roosted was 7.0C  0.4C.We screened 55 samples for P. fluorescens by PCR from bats collected in the winter of 2014-15 to confirm that P. fluorescens naturally occurred on bats found in the site to address concerns regarding using a live bacterium as a treatment.We found DNA from P. fluorescens present in 20% of samples and from all four species sampled (Little brown myotis (Myotis lucifugus), Northern long-eared myotis (Myotis septentrionalis), Tri-colored bat (Perimyotis subflavus), and Big brown bat (Eptesicus fuscus)).
In September 2015 we installed a PIT-tag reader (IS1001 and HPR reader, Biomark Inc., Boise, ID) at the site entrance (Figure S1), and 3 metal screen cages (46x30x51 cm Fresh Air Screen Habitat, Zilla Products, Franklin, WI, USA) in a small chamber in the back of the mine where bats roosted in previous winters.We removed the top of each cage and mounted cages directly to the ceiling to allow bats direct access to mine substrate for roosting, and to allow for natural infection and reinfection.We also installed chicken wire with a hinged gate at the entrance of the cage room to prevent large predators (e.g.raccoons) from entering.
We briefly visited the site on Nov 16 th (total time underground 14 minutes) to count the number of bats present and to assess the P. destructans infection status of the bats at the site.We counted approximately 95 M. lucifugus and sampled six of them by dipping a sterile polyester swab in sterile water to moisten it and then rubbing the swab five times across both the forearm and muzzle of a bat 20 .We tested these samples for P. destructans DNA using qPCR 30 , and all six samples tested positive.
We returned to the site for the experimental treatment on Nov 20, 2015.We sealed off the entrance to the site (Figure S1) using fine mesh cloth to prevent bats from leaving the site during the treatment.We collected all M. lucifugus we could find at the site (89 bats; 23 females and 66 males) and placed them individually in paper bags and brought them to a processing station near the entrance of the site.We weighed bats to the nearest 0.1 g with an electronic scale, but we did not take a length measurement (e.g.forearm) to minimize handling time and disturbance.Recent work has shown that body mass is equally accurate in predicting fat stores (as measured by quantitative magnetic resonance) as body condition indices 31 .We sampled bats for P. destructans as described above and banded each bat with an aluminum band (2.9mm; Porzana Ltd., Icklesham, E. Sussex, U.K.), that had a PIT tag attached (see Supplemental Methods text).
We randomly assigned bats to each of three treatment groups: control (29), chitosan (29), and P. fluorescens (31).We treated each bat by spraying ~2ml of a solution containing P. fluorescens or chitosan solution (see Supplemental Methods text) on their wings and tail with a spray bottle (FantaSea, Blaine, WA).For control bats we replicated the handling disturbance but did not spray any liquid onto bats because both of our treatments could only be applied in liquid form and the goal of our study was to determine the effect of treatment compared to untreated bats.We split bats in each treatment group into the two experiments based on a power analysis (Figure S2)cage (15 for each treatment group in a single cage for each treatment; 5 females and 10 males per cage) and free-flying (16, 14, and 14 bats in the P. fluorescens, chitosan and control groups, respectively).After treatment bats were released into cages or into the site onto a chitosan would have higher survival, have a later last detection (by the PIT tag reader) date, and lower UV fluorescence than the control group and that apparent overwinter survival would increase with initial body mass and lower initial fungal loads in both experiments.
All research was performed as described under protocol Kilp1509 approved by the University of California, Santa Cruz's IACUC committee.

fluorescens (PF).
There were 14-16 bats in each of the six treatment-experiment groups.There were no significant differences among treatment groups in fungal loads, prevalence or body mass in November (likelihood ratio-tests for treatment effect in linear models of mass, generalized linear models with a binomial distribution for prevalence, and linear models for log-transformed fungal loads: all P-values > 0.57).Columns show the fraction of bats in each treatment group that were known alive and detected by the PIT tag reader on or after March 7 (with binomial 95% CI; sample sizes for each group, left to right, were 13, 12, 11; the remaining 7 bats had lost their PIT tag; see Methods).Differences among treatments were significantly different, (Fisher's exact test for all three treatments: two-tailed P = 0.012).P. fluorescenstreated bats had higher apparent survival than the control group (Fisher's exact test: one-tailed P = 0.046; logistic regression coef.= 2.24  1.18, z = 1.896, one-tailed Pvalue = 0.029), but there were no significant differences between chitosan-treated and control bats (Fisher's exact test: one-tailed P = 0.47; logistic regression coef.= 0.89  1.3, z = 0.69, onetailed P-value = 0.49).

Figure Legends Figure 1 .
Figure LegendsFigure1.Bat body mass (A), P. destructans prevalence (B), and fungal loads (C) of M. lucifugus in November in three treatment groups (chitosan (CH), control (CO), and P. fluorescens (PF).There were 14-16 bats in each of the six treatment-experiment groups.There were no significant differences among treatment groups in fungal loads, prevalence or body mass in November (likelihood ratio-tests for treatment effect in linear models of mass, generalized linear models with a binomial distribution for prevalence, and linear models for log-transformed fungal loads: all P-values > 0.57).Figure2.External air temperature, emergence date and fungal loads of bats in the freeflying bat experiment.The top panel shows air temperature measured at the site entrance.In the lower panel, line color indicates the P. destructans fungal load (in nanograms) measured on bats in November.Lines end on the last date that each bat was detected by the PIT tag reader, and dotted vertical lines show the mean emergence date for each treatment group, indicated on the Y-

Figure 2 .
Figure LegendsFigure1.Bat body mass (A), P. destructans prevalence (B), and fungal loads (C) of M. lucifugus in November in three treatment groups (chitosan (CH), control (CO), and P. fluorescens (PF).There were 14-16 bats in each of the six treatment-experiment groups.There were no significant differences among treatment groups in fungal loads, prevalence or body mass in November (likelihood ratio-tests for treatment effect in linear models of mass, generalized linear models with a binomial distribution for prevalence, and linear models for log-transformed fungal loads: all P-values > 0.57).Figure2.External air temperature, emergence date and fungal loads of bats in the freeflying bat experiment.The top panel shows air temperature measured at the site entrance.In the lower panel, line color indicates the P. destructans fungal load (in nanograms) measured on bats in November.Lines end on the last date that each bat was detected by the PIT tag reader, and dotted vertical lines show the mean emergence date for each treatment group, indicated on the Y-

Figure 4 :
November P. destructans fungal loads and the late date of detection for individual bats in the three treatment groups (Pf -P.fluorescens, Co -Control; Ch -Chitosan).P. fluorescens-treated bats were last detected later than Control and Chitosan treated bats, and the last date of detection decreased (was earlier) with higher November P. destructans loads (P.fluorescens vs. control coef = 1.05  0.57, t = -1.91,one-tailed P-value = 0.039, early winter fungal loads: coef = 0.61  0.32, t = 1.84, one-tailed P-value = 0.040).

Figure 5 .
Survival, subsequent detection, and body mass of bats in the caged bat experiment.Each horizontal line represents a single bat, with line color indicating the body mass of that bat measured in November.Bats surviving until release March 8 (gray vertical bar) are shown by darker lines, and bats that were subsequently detected by the PIT tag reader are extended to the last date they were detected.Treatment is indicated by the first two letters of the bat identification number on the Y-axis (PF -P.fluorescens, CO -Control; CH -Chitosan).

Figure 6 .
P. destructans (a) fungal loads, and (b) disease severity (UV) for bats in the cage experiment that survived to March 8, and (c) differences in November mass between bats in the cage experiment that survived to March 8, or died.March fungal loads did not differ among treatment groups (likelihood ratio-tests for treatment effect in models of log-transformed fungal loads: P-value > 0.77).Disease severity (ultraviolet fluorescence) was significantly higher for chitosan-treated bats than control bats (regression on arc-sin sqrt transformed data: control vs. chitosan: coef = 0.15  0.068, z = 2.273, P-value = 0.037; control vs. P. fluorescens: coef = 0.017  0.073, z = 0.234, P-value = 0.82).

Figure S2 .
Figure S2.Power analysis used to design both experiments of the field trial.The x-axis Cited 1 Warnecke, L. et al.Inoculation of bats with European Geomyces destructans supports the novel pathogen hypothesis for the origin of white-nose syndrome.Proc.Natl.Acad.Sci.U. S. A. 109, 6999-7003 (2012). 2 Langwig, K. E. et al.Sociality, density-dependence and microclimates determine the persistence of populations suffering from a novel fungal disease, white-nose syndrome.Ecol Lett 15, 1050-1057 (2012).