Effective conservation of subterranean‐roosting bats

Bats frequently inhabit caves and other subterranean habitats and play a critical role in subterranean food webs. With escalating threats to subterranean ecosystems, identifying the most effective measures to protect subterranean‐roosting bats is critical. We conducted a meta‐analysis to evaluate the effectiveness of conservation and management interventions for subterranean‐roosting bats. We used network analyses to determine to what extent interventions for bats overlap those used for other subterranean taxa. We conducted our analyses with data extracted from 345 papers recommending a total of 910 conservation interventions. Gating of roost entrances was applied to preserve bat populations in 21 studies, but its effectiveness was unclear. Habitat restoration and disturbance reduction positively affected bat populations and bat behavior, respectively, in ≤4 studies. Decontamination was assessed in 2 studies and positively affected bat populations, particularly in studies focused on reducing fungal spores associated with white‐nose syndrome in North America. Monitoring of bat populations as an effective conservation strategy was unclear and infrequently tested. Only 4% of bat studies simultaneously considered other subterranean organisms. However, effective interventions for bat conservation had similarities with all other organisms. If other subterranean organisms are considered when applying interventions to conserve bats, they might also benefit.


INTRODUCTION
It is commonplace for speleologists and cavers to look toward the ground when exploring a cave to not lose footing.Yet, by directing the light source upward, one may be lucky enough to notice clusters of nocturnal animals hanging from the ceiling.
Additionally, human-wildlife conflict is on the rise, particularly in parts of Africa and Asia, where urbanization is rapidly encroaching on natural areas (Seoraj-Pillai & Pillay, 2017).This creates a further problem for bat conservation related to conflicts between humans and bats (Frick et al., 2020).Unfortunately, negative perceptions of bats (e.g., Bhattacharjee et al., 2018;Kingston, 2016) have been exacerbated in the era of COVID-19 (Nanni et al., 2022).At the onset of the pandemic, bats were suggested as the probable origin for SARS-CoV-2, the causative agent of COVID-19 (Zhang et al., 2020).Although bats are reservoirs for several viruses (Letko et al., 2020;Van Brussel & Holmes, 2022;Wu et al., 2016), including harboring related coronaviruses (Li et al., 2005;Murakami et al., 2020), the idea that bats are the origin of the coronavirus pandemic is based on unproven assumptions and oversimplifies the process of disease spillover (Plowright et al., 2017).A result of these assumptions is the strengthening of the negative public opinions of this diverse group (e.g., Lu et al., 2021;Pereira et al., 2020).
For all the aforementioned reasons, the protection of subterranean-dwelling bats is a challenging endeavor, involving the need to account for multiple threats and their combined effects.Yet, current knowledge on effective conservation measures is far from systematic or exhaustive.Recently, we reviewed practical conservation of subterranean environments considering a diversity of threats, organisms, and conservation actions (Mammola et al., 2022).The results of this review suggested that bats are one group for which there is a significant amount of knowledge relative to other cave-dependent organisms.Twenty-seven percent of papers in the review focused on bats, and a substantial fraction of these studies yielded quantitative estimates that could be used in a meta-analysis.
Here, we examined whether we could harness this knowledge to identify effective measures of protection and the extent to which effective conservation actions can be used to protect subterranean-roosting bats.We used a meta-analytical approach to extract effect sizes from experimental and observational studies that tested the effectiveness of conservation interventions.Given that vertebrates are often viewed as more charismatic than invertebrates (Miralles et al., 2019), we examined, using indirect evidence obtained through a network analysis, whether subterranean-dwelling bats can be useful umbrella species for the protection of the entirety of subterranean ecosystems.

Focus of the study
We focused on subterranean-roosting bats, intended in a broad sense.Bats make up around one quarter of all mammal species, occupying all continents except Antarctica.To date, 323 bat species are listed on the International Union for the Conservation of Nature (IUCN) Red List as threatened (critically endangered, endangered, or vulnerable).Of these, 142 use subterranean habitats, and 56% are cave-dependent roosting species (IUCN Red List, version 2020-3).However, the subterranean habitats used by bats are much broader than caves.Bats roost in a wealth of artificial subterranean habitats (e.g., culverts, cellars, mines, tunnels, bunkers, and other human-made structures in urban environments [e.g., Herter, 2007;Meierhofer et al., 2019;Webala et al., 2022;Whitaker et al., 2006]) and in fissural systems, such as screes and fractures in rocks (Blejwas et al., 2021;Lewis et al., 2022).For some species, any shelter that is lightless and climatically buffered is potentially a suitable roost for bats, although species show differences in preferences and requirements (Dorward et al., 2022).Therefore, we considered the literature focused on the full breadth of these subterranean habitat spaces.

Inclusion criteria
We conducted a systematic literature review to compile an extensive list of publications that discussed or tested conservation actions with a focus on subterranean-roosting bats (Appendix S1).Specifically, we used detailed information specific to bats from our initial literature survey (Mammola et al., 2022) and supplemented these data with a refined systematic literature search to ensure thoroughness (see "Systematic literature search").We included studies in which the effectiveness of conservation interventions (e.g., gating to prevent access to caves) was statistically tested based on quantitative variables describing the status of bat species (e.g., population, range trends, behavior); conservation interventions for bats were discussed, suggested, or recommended without testing their effectiveness; research priorities for conservation of bats and performing risk assessments were discussed; and surface management or protection measures that affect subterraneanroosting bats were the focus.We excluded studies focusing on nonsubterranean bats (e.g., forest species).

Systematic literature search
Our systematic search was based on Mammola et al.'s (2022) database, which included the collation of data from 708 publications on conservation interventions on different subterranean organisms based on a standardized literature search in the Clarivate Analytics Web of Science (WoS), of which 203 referenced bats.Although this source provided a reliable literature baseline for our analyses based on a standardized literature survey method, we decided to carry out additional searches to maximize coverage.Thus, on 9 September 2021, we performed an additional standardized literature search in WoS to supplement the original search from Mammola et al. (2022).For this supplemental WoS search, we used the following search string designed to target bats and bat-related topics: ALL = ("bats" OR "bat" OR "Chiropter*") AND ALL = ("cave" OR "subterranean biology" OR "mines" OR "tunnel" OR "bunker" OR "artificial subterranean habitat" OR "hibernacula" OR "scree" OR "rocky habitats" OR "boulder field") AND ALL = ("COVID-19" OR "disease" OR "attitud*" OR "communicat*" OR "conserv*" OR "managem*" OR "restorat*" OR "preserv*" OR "policy" OR "policies" OR "politic*" OR "protect*" OR "reintroduc*" OR "regulat*" OR "legislat*" OR "IUCN" OR "CITES" OR "sustainabil*" OR "priorit*").
The search returned 1,065 papers, of which 348 were previously captured by Mammola et al. (2022).The remaining papers (717) were screened for eligibility to be included in the review.First, we screened these papers based on their titles and abstracts, excluding 629 papers that did not meet the inclusion criteria defined above.We examined the full text of the references retained to determine whether they addressed our research questions.Ultimately, we included 34 additional papers in the final database.
Similarly, we incorporated 83 papers extracted from the DarkCideS database (Tanalgo, Tabora, et al., 2022).The papers in the DarkCideS database include bat cave studies, particularly georeferenced species occurrence from field surveys.We included only full texts from the database that met our inclusion criteria.We counterchecked the extracted literature from the DarkCideS with the main database to ensure there was no duplication in the final database.
Finally, we conducted an unstandardized search of the gray literature (Haddaway et al., 2020), including articles not in English (Nuñez & Amano, 2021), retrieving 25 papers that were previously missed.

Metadata extraction
For all new articles (i.e., those not included in Mammola et al. [2022]), we extracted the type of publication; year of study; geographic and taxonomic scope, including family and genus where applicable; domain; system; and threat type associated with the conservation intervention.
Adopting the methods of Mammola et al. (2022), we identified the following threat categories: non-native species and pathogens, climate change, overexploitation and poaching, pollution, surface habitat change, subterranean habitat change, visitors, and not identified.Likewise, we identified indirect conservation measures (i.e., education, legislation, monitoring, prioritization, risk assessment) and direct conservation measures (i.e., eradication, ex situ conservation, gating, habitat creation, habitat restoration, protected area, reintroduction, access regulation, decontamination).
When available, we collected all associated statistical measures to carry out a meta-analysis.Specifically, we collected all statistical tests used to measure effectiveness, their test statistic, degrees of freedom, number of observations, p value, and direction of effect.When studies presented partial statistics, we contacted the corresponding authors and requested missing information, which yielded further usable data from 3 studies (response rate 75%).Using standard conversion formulas (Lajeunesse, 2013), we converted all test statistics that described the effect of the conservation actions on subterranean-roosting bats to Pearson's r.Pearson's r is a common measure of the effect size (ranging continuously between from −1.0 to 1.0) that expresses the strength of a given linear association between the predictor and the response variable.We calculated Pearson's r for any comparison within a given study, such as when different conservation interventions or bat's responses were tested by the author or authors.Given that in individual studies a wide variety of response variables were analyzed, we aggregated them into 6 macrocategories for analyses: behavior (e.g., aborted exit or entry, flight, emergence, bat calls [activity]), occupancy (e.g., presence of a species, species richness), population (e.g., abundance of hibernating bats, magnitude of change in population, colony or population size, disease severity, proportion of sites with population increase, survival), pathogen (e.g., fungal loads, spore count, number of fungal species), trait (e.g., body size), and suitability (e.g., cave dimensions, habitat quality).

Meta-analysis
We conducted the meta-analysis in R 4.1.0(R Core Team, 2021) with the R package metafor 2.4.0 (Viechtbauer, 2010).We constructed a set of meta-analytic linear mixed-effects models (metafor function rma.mv), 1 for each combination of response variable (as defined in "Metadata extraction") and predictor (the conservation intervention), to assess the extent to which different conservation interventions affected the response of subterranean-roosting bats.In all models, we specified a publication-level nesting factor to account for study-level nonindependence (often, there were multiple measurements per study; mean [SD] = 6.89 estimates/publication [6.34]).We only tested the effect of a combination of response variable and predictor variables when the sample size was above 2 studies.In all models, we converted Pearson's r to Fisher's z to approximate normality (Rosenberg et al., 2013).However, for ease in presentation of the results, we back transformed Fisher's z values and their 95% confidence intervals to Pearson's r.We interpreted model-derived estimates of Pearson's r as the strength of the standardized effect, which we considered significant when the 95% confidence intervals did not overlap zero.We evaluated publication bias with fail-safe number analyses (metafor function fsn).Specifically, we used Rosenthal's method (Rosenberg, 2005;Rosenthal, 1979) to calculate the number of studies.We averaged negative results that would have to be added to the given set of observed outcomes (predictors) to reduce the combined significance level to a target alpha level (0.05).We found no evidence of strong publication bias (Table 1).

Network analysis
To identify the extent to which conservation interventions implemented for bats benefit other subterranean-dwelling organisms (umbrella effect), we examined papers that considered bats and other organisms in their testing or recommendation of conservation interventions.Only 4% (14 papers out of 345) considered other organisms in addition to bats, preventing us from making any meaningful inference.Therefore, we explored this question with indirect evidence obtained through network analyses.
We constructed and manipulated networks with the R packages igraph 1.2.6 (Pedersen, 2021) and tidygraph 1.2.0 (Pedersen, 2020).We started from the full database provided in Mammola et al. (2022) complemented with additional literature on bats.However, we excluded studies focusing on aquatic subterranean systems (including marine systems) because they do not host bats and studies focusing on unspecified organisms.Furthermore, we considered only direct conservation interventions, resulting in a total sample size of 291 papers and 640 conservation interventions.With this data set, we generated a bipartite directed network that connected organisms with their recommended direct conservation interventions.In the network, the first node type represented subterranean organisms grouped in 6 macrocategories: bats, arthropods, microorganisms (an artificial category including archaea, bacteria, and unicellular fungi), other invertebrates (all nonarthropod invertebrate groups), other vertebrates (all vertebrates other than bats), and plants (plant organisms typically occurring in the cave twilight zone).The second node type represented the proposed direct conservation intervention.We weighted edges (links) connecting nodes by the number of times a given conservation intervention was proposed for each organism.On this matrix, we expressed similarity in conservation interventions among organisms with the Bray-Curtis index calculated as 1the complement of the beta function in the R package BAT 2.7.1 (Cardoso et al., 2015(Cardoso et al., , 2022)).Edge values were inputted as abundances.

Summary of the literature
In total, 852 papers were entered in the final database (2,162 conservation interventions across all organisms).Of these, 345 papers were specific to bats and recommended 910 conservation interventions.Sources included 277 research articles, 38 technical reports, 24 review or opinion pieces, and 6 books.

General trends
Two-hundred ninety-four out of 910 (34%) conservation and management interventions were tested using statistical techniques.Testing of these interventions predominantly occurred in the Nearctic and Palearctic biogeographic regions (Figure 1).Limited testing was conducted in each of the Afrotropical, Australasian, and Indomalayan regions, and no testing of any conservation interventions was documented in the Neotropical region (Figure 1).
Perhaps the clearest result was that gating was the most studied and tested direct conservation intervention (73%; 138 out of 190 interventions  1).Across all regions, there were several similarities among threats.Visitors were the most studied threat across regions (except for the Australasian biogeographic region).The Nearctic and Palearctic regions more consistently applied and tested conservation interventions across threats than other regions (Figure 2).Of all threats, non-native species and climate change were the least mentioned.
The most tested conservation intervention was gating, with few studies testing decontamination and restoration efforts, monitoring techniques, and disturbance reduction (Figure 3).Population, behavior, and occupancy were the most studied responses to gates, showing a generally positive yet not significant trend.Although there was a positive effect of restoration on bat populations, there were only 4 studies.Similarly, decontamination efforts significantly reduced pathogens, but were only tested in 2 studies.Disturbance reduction had a near sig-nificant positive effect on bat behavior, but was tested in just 3 studies.Finally, 3 studies tested the effectiveness of different monitoring techniques in detecting changes in bat populations, yielding an overall positive, but not significant, effect.

Umbrella effect of bats
When considering all organisms, bats did not stand out as having strong similarity in direct conservation interventions relative to other organisms (Figure 4).Conservation interventions for bats were most similar to interventions for arthropods (0.13).For all other organisms, interventions had little similarity (<0.05).Arthropod interventions had the greatest similarity with interventions for all other organisms (all values >0.10, except with microorganisms).However, the number of species and the number of studies included in each group were quite different.

Effective protection of subterranean-roosting bats
Although there is growing quantitative knowledge on the effectiveness of conservation interventions for bats on the surface (Sutherland et al., 2021), our results suggest that limited   1s knowledge is available for broad and effective protection of subterranean populations.Our findings summarize information for only 7 species in 3 families (Molossidae, Vespertilionidae, and Rhinolophidae) in areas where most of the research was focused (Neartic and Paleartic biogeographic regions), highlighting major gaps in knowledge.Indeed, the most diverse parts of the planet with some of the highest degrees of threat (Tanalgo, Oliveira, et al., 2022)  interventions.These findings included only a very small fraction of the 1,453 known species of bats (Simmons & Cirranello, 2022), although nearly 40% use the subterranean environment (Furey & Racey, 2016).Yet, it is difficult to conclude whether effective measures are being implemented for the conservation of subterranean-roosting bats given the general lack of testing.However, from our meta-analysis on 5 of 13 interventions measured, the trend appears positive for the responses studied.
Our results are consistent with previous literature describing mixed success in the use of gates to protect bats (Tobin & Chambers, 2017).This may result, in part, from the lack of publications reporting successful cave gating.Indeed, in the years since Tobin and Chambers' (2017) article, only 3 studies in our data set tested the effect of gates on bats despite significant conservation attention.There is still a clear lack of an effect of gates on the response of bats, yet gates are frequently constructed.These mixed outcomes may result due to gates being effective only when bat populations are in decline, as opposed to when populations are growing (Crimmins et al., 2014).Understanding current population trajectories is essential when installing gates and setting expectations for conservation outcomes.The mixed output on the effectiveness of gating could further result from the change in gate styles over time and the different species studied.Certainly, there is a critical need for more studies to discern whether gates considered bat friendly are effective across species because some species may respond better than others to gating given their morphology and behavior (e.g., Rodrigues, 1996).
Although it is likely that successful conservation outcomes through gating are underreported, this intervention is not a silver bullet and cannot be expected to address the diverse threats facing bat populations (e.g., Souza & Bernard, 2016).Gates only address threats posed by human visitors to subterranean environments and the failure of bat gates to achieve a con-servation outcome may relate to a mismatch between a threat and management response (Johnson et al., 2021).Although it is understandable that gating was the most frequently tested conservation measure in our review, given that visitors were the most frequently identified threat to bats in subterranean environments, our results indicate that this pair of threat and management action overshadows others.
We examined several papers in which conservation measures were applied without prior identification of a threat.In these instances, it was difficult to assess the efficacy of gating or any other intervention because their potential for success was limited by the appropriateness of their application.This illustrates a need to be more thoughtful, including more rigorous attempts to pair management interventions with existing threats.Given the limited overlap in conservation actions between bats and other organisms (see "Umbrella effect of bats"), inspiration for alternative approaches may come from more integration between bat researchers and other biologists working with subterranean organisms.For instance, climate change poses a real and present threat to bats (Festa et al., 2022) and entire subterranean ecosystems, yet the effects of warming on underground environments are not well documented (Mammola, Piano, et al., 2019).This threat represents a need for stewards of subterranean bat populations to test the utility of additional interventions because gating will not be an effective action to take against climate change (Mammola et al., 2022), and the impact of changes in climate on species hibernation as well as pathogen risks requires further monitoring to assess.
Nonetheless, there are times when an action proposed to mitigate the effects of one threat might also be considered a conservation action for another threat.For example, a study by Turner et al. (2022) implemented a management intervention with the intention of mitigating WNS, a disease of hibernating bats.The researchers successfully modified a select few subterranean structures to alter microclimates by excavating new entrances to alter airflow.Although the modifications were implemented as a mitigation measure for a bat-specific disease, one could consider this altering of microclimates a potential management strategy to buffer the effects of climate change (Meierhofer et al., 2022).In this way, scientists working in subterranean environments could act synergistically, moving beyond individual specialties with the intentions of more holistic management practices.
Although we encourage consideration of threats other than those posed by visitors to the subterranean environment, bats that congregate in large aggregations in sites such as caves are especially vulnerable to disturbance (Frick et al., 2020;Furey & Racey, 2016).Disturbance reduction had clear positive results on the behavior of bats, but limited testing.It is not shocking that reducing disturbance would have an overall positive effect on bats.What is striking though is that only 1 study was conducted on tour caves (Mann et al., 2002)-surprising given the ease of study design and access to tour caves and the continued pressures on bats in these environments.The other 2 studies measuring impacts of disturbance reduction on the behavior of bats sought to determine the effects of white flash photography as a method for monitoring bats to reduce disturbance (Krivek et al., 2022) and to identify which, if any, lighting source could be used to illuminate caves without causing significant disturbance (Straka et al., 2020).Provided that disturbance of bats is ranked fourth as a major threat type for threatened bat species (Frick et al., 2020), it goes without saying that it is imperative that understanding of unintentional disturbance be measured and mitigated.
Restoration of the subterranean environment, a challenging and frequently costly endeavor (Medellín et al., 2017), has been tested predominantly in North America in relation to WNS; only a single study has been conducted in Central Europe.The interventions varied greatly even when combating the same threat-WNS.For instance, Kwait et al. (2022) focused on whole-room sanitization in treating environmental reservoirs of Pseudogymnoascus destructans to restore the site to pre-WNS condition.They tested different setups in a lab setting to identify the best distance to tackle the problem.A second treatment study conducted in the field at a mine focused on understanding the efficacy of probiotic bacterium, Pseudomonas fluorescens, to reduce impacts of WNS (Hoyt et al., 2019).A third strategy to combat WNS included the modification of select sites to alter the microclimatic environment (Turner et al., 2022).Finally, a single study in Europe, which used historical bat counts as a baseline to measure recovery of bats at a hibernation site, focused on halting the banding of bats and reducing human activity at a hibernation site (Gaisler & Chytil, 2002).Thus, restoration efforts of the subterranean environment are relying on single studies with vastly differing techniques, and yet in all cases these speciesspecific actions appear to work well.However, these findings are still anecdotal in that we were only able to test a handful of cases in our meta-analysis, and recommendations are only broadly generalizable.Wherever restoration efforts are applied, it is important to be cognizant of the feedback effect (Meierhofer et al., 2022) and the need for monitoring the efficacy of these restorative efforts.
Several strategies exist for monitoring bats, from more intrusive techniques that may result in increased investigator disturbance, resulting in a potential for additional energy expenditure for the bat, such as physical counts, to more passive methods, such as acoustic monitoring (e.g., Hayes et al., 2009;Kloepper et al., 2016Kloepper et al., , 2017)), videography (e.g., Betke et al., 2008), and camera traps (e.g., Krivek et al., 2022).Acoustic monitoring has been tested in surface environments (e.g., Adams et al., 2012;Hogue & McGowan, 2018) and there is some consensus on appropriate methodologies.When comparing the effectiveness of different monitoring techniques, there was a positive, although not significant, effect on bat populations.The testing of different monitoring techniques is beneficial given the discussion and lack of recent knowledge on how disturbance affects roosting bats (Boyles, 2017).Yet, with the range in monitoring techniques and technologies available to monitor bats, there is still limited testing done to compare the different approaches to determine the method that best benefits bats.This is of particular importance because bat populations are typically monitored seasonally-in the breeding season and during hibernation, both of which are considered energetically stressful periods.
Although other interventions can address more than 1 type of threat, interventions such as decontamination are very targeted.The act of decontamination, although infrequently tested, had a positive effect on reducing pathogens, particularly in studies focused on reducing P. destructans fungal load, the causative agent of WNS (Lorch et al., 2011).Although controversial, several studies with a focus on WNS mitigation via decontamination have been attempted, but only 2 studies measured the effectiveness of such an intervention: 1 study focused on decontamination of field equipment (Zhelyazkova et al., 2020) and 1 study focused on a field trial of a treatment (Hoyt et al., 2019).This is disconcerting because altering the environment without measuring the effects can have long-lasting ramifications.
It is clear from our results that limited management interventions are broadly applied to tackle a breadth of threats.Perhaps to be successful in our conservation, one must consider applying multiple interventions simultaneously and ensure that one correctly links the intervention to the threat, which may in turn assist with the conservation of multiple organisms.However, it must be stressed that these interventions, taken together, should also be statistically measured for their overall effectiveness and their effect on the subterranean domain (Mammola et al., 2022).From our own participation in regional and international conferences, we are aware that additional information on these topics exists in unpublished reports and databases.Although those sources are often shared among close colleagues at local and regional scales (e.g., in conferences, workshops, informal meetings), they remain hard to discover, if not undiscoverable, to the larger scientific community.Thus, many conservation measures may not appear to be evidence based when, in fact, they are.Easily accessible information on the effectiveness of strategies not only can improve biodiversity conservation, but also has the potential to protect humans (e.g., by reducing the risk of disease spillover) (e.g., Cunningham et al., 2017;Morand et al., 2014).Therefore, we strongly encourage the publication of these records to make conservation information more readily usable by current and future generations.

Considering the broader subterranean domain
A lack of collaboration among bat biologists and overall speleologists is a genuine problem, and many caving societies lack exchanges with bat researchers.When applying conservation and management interventions, bat researchers typically fail to consider other organisms.As illustrated by our findings, only 4% of bat studies simultaneously considered other organisms, suggesting that researchers focused on bats do not often collaborate with other subterranean biologists and vice versa.Despite this stark statistic, it appears that there is some similarity with conservation measures across all organisms.Yet, the similarities in conservation interventions between bats and all other organisms were low in comparison with all other organisms.This suggests that interventions were bat oriented, conceptualized, and executed by bat researchers and that there was a lack of communication with other taxa experts or, alternatively, that proper interventions were not identified.Measures that limit access or reduce disturbance (particularly with issues such as guano extraction) could have significant ecosystem-wide benefits.In addition, the importance of guano to subterranean ecosystems means that diverse bat caves have the capacity to support other taxa.Thus, in such cases, prioritization and protection of key sites for bats may infer benefits to other taxa.The lack of apparent overlap with success of certain types of intervention actions is perhaps not a surprising outcome, given the special niche bats occupy in comparison with other organisms and the different types of threats and how they are measured.Although most other subterranean fauna are limited to the subterranean environment, bats occupy space inside and outside these spaces; individuals move between the surface and subterranean environments daily or use caves only in certain seasons (e.g., for hibernation or swarming).This suggests that bats may require some unique conservation measures not always suitable or necessary for other organisms.This may pertain more to identified threats (e.g., WNS), for which some methods, such as decontamination, have the intended goal of conserving a specific organism.However, most other conservation actions (e.g., restoration) should infer effective conservation to the whole system provided all organisms are considered when the intervention is applied.
Given the lack of overall testing of conservation measures targeting bats and other organisms (Mammola et al., 2022), it is difficult to discern how effective interventions are at protecting other organisms and if the proper interventions are being applied given the identified threat.More testing of intervention strategies should occur to elucidate whether the overlap of bat conservation interventions with interventions for other organisms could be broadened and to further identify whether intervention strategies are being used inappropriately.Indeed, it would behoove researchers to be cognizant of intervention strategies suggested as important for other taxa to try to expand the focus onto those groups through their conservation efforts, thereby taking them under their wing.
To realize and sustain collaborative efforts, then, conservationists must acknowledge the important role of bats in the ecosystem.Perhaps unsurprising, the intersection for collaborative efforts may lie underfoot-in guano.Bats play a key role as sources of energy input into these subterranean environments through the deposition of guano (Pimentel et al., 2022), to the point of potentially bioengineering cave systems (Piló et al., 2023).Many organisms in these environments are functionally phylogenetically unique and have a high degree of endemism (Chagas & Bichuette, 2018;Zhang et al., 2018), relying on the injection of energy from the outside environment (Ferreira, 2019).The guano deposits in bat caves have also been linked to increases in the species richness and diversity of fungi (Cunha et al., 2020).These examples illustrate the importance of protecting and maintaining cave systems to ensure the provision of key ecological services (Medellín et al., 2017).
Our meta-analysis, being strongly focused on the peerreviewed literature and to a lesser extent gray literature, may have partly underestimated the total volume of work that is being done for cave bat conservation.The overall low sample size of our meta-analysis (see Fox [2022] for comparative data) is symptomatic of this, possibly reflecting how the volume of relevant papers is dwarfed by the low numbers of researchers working in conservation.There is a need for researchers and biologists alike to work with unpublished small data sets and compile them into a discoverable source.Conservation actions are sometimes based on information not readily available and obtained through conversations, but there is danger in this approach.At best, this information is not being made discoverable to future conservationists.At worst, interventions are being based on studies with limited sample sizes that might not be reliable.When there are emergent threats, such as WNS, for instance, the lack of discoverable information (i.e., available in readily accessible literature) can prevent the timely implementation of effective conservation and management at scale.Thus, without the bigger picture, conservation efforts may be misguided.

Moving forward in the conservation of subterranean-roosting bats
Our systematic survey of the literature and analyses revealed that few (∼1/3) conservation and management interventions were tested using statistical techniques; testing of conservation interventions was heterogeneous, focused on a handful of species and concentrated mainly in Europe, the United States, and Canada; visitors were the most mentioned and frequently tested impact and bat gating the most adopted intervention; and bats can potentially be used as umbrella species for conservation of the subterranean environment.
Multiple anthropogenic impacts threaten the conservation of subterranean-dwelling bats and need to be considered in management efforts.This is challenging, particularly when knowledge on subterranean biodiversity and effective intervention strategies is limited.To address and correct the problems identified in our review, we highlight that threats to caves and subterranean-dwelling bats must be accurately identified and documented; effectiveness of interventions needs to be tested to enable implementation of effective mitigation strategies for identified threats; testing bat-oriented interventions must be considered a priority in the Neotropics, Afrotropical, Australasian, and Indomalayan regions, where most of the world's bat species richness and diversity are concentrated and where cave and bat protection are lagging; management interventions identified and implemented for specific threats should be tested not only at the population level, but also across species (apparent particularly in gating efforts for which results of effectiveness were unclear); bat-oriented subterranean biologists should collaborate with other taxonomic specialists to ensure more inclusive conservation efforts and broader interventions; and successes and failures must be better documented and diffused to effectively use time and financial resources and ensure the willingness of decision-makers to protect caves and bats.

FIGURE 1
FIGURE 1 Number of studies of bat conservation interventions (345 total studies) by type of intervention and biogeographic region (dotted lines separate direct and indirect conservation interventions).

FIGURE 2
FIGURE 2 Number of studies that did and did not test conservation interventions to address threats to bats (345 total studies recommending 910 conservation interventions) by biogeographic region (multiple, ≥3 threat groups considered; none identified, studies that discussed or tested conservation intervention without explicitly referring to a threat [i.e., proactive interventions]).

TABLE 1
(Rosenthal, 1979)arameters for the metafor models and evaluation of publication bias using the fail-safe number(Rosenthal, 1979)a .
a Significant values indicate that no publication bias was detected.b Number of estimates.
almost entirely lack studies on