Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Effects of Hierarchical Roost Removal on Northern Long-Eared Bat (Myotis septentrionalis) Maternity Colonies

  • Alexander Silvis ,

    Affiliation Department of Fish and Wildlife Conservation, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, United States of America

  • W. Mark Ford,

    Affiliations Department of Fish and Wildlife Conservation, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, United States of America, US Geological Survey, Virginia Cooperative Fish and Wildlife Research Unit, Blacksburg, Virginia, United States of America

  • Eric R. Britzke

    Affiliation US Army Engineer Research and Development Center, Environmental Laboratory, Vicksburg, Mississippi, United States of America

Effects of Hierarchical Roost Removal on Northern Long-Eared Bat (Myotis septentrionalis) Maternity Colonies

  • Alexander Silvis, 
  • W. Mark Ford, 
  • Eric R. Britzke


Forest roosting bats use a variety of ephemeral roosts such as snags and declining live trees. Although conservation of summer maternity habitat is considered critical for forest-roosting bats, bat response to roost loss still is poorly understood. To address this, we monitored 3 northern long-eared bat (Myotis septentrionalis) maternity colonies on Fort Knox Military Reservation, Kentucky, USA, before and after targeted roost removal during the dormant season when bats were hibernating in caves. We used 2 treatments: removal of a single highly used (primary) roost and removal of 24% of less used (secondary) roosts, and an un-manipulated control. Neither treatment altered the number of roosts used by individual bats, but secondary roost removal doubled the distances moved between sequentially used roosts. However, overall space use by and location of colonies was similar pre- and post-treatment. Patterns of roost use before and after removal treatments also were similar but bats maintained closer social connections after our treatments. Roost height, diameter at breast height, percent canopy openness, and roost species composition were similar pre- and post-treatment. We detected differences in the distribution of roosts among decay stages and crown classes pre- and post-roost removal, but this may have been a result of temperature differences between treatment years. Our results suggest that loss of a primary roost or ≤ 20% of secondary roosts in the dormant season may not cause northern long-eared bats to abandon roosting areas or substantially alter some roosting behaviors in the following active season when tree-roosts are used. Critically, tolerance limits to roost loss may be dependent upon local forest conditions, and continued research on this topic will be necessary for conservation of the northern long-eared bat across its range.


Roosts provide bats with sites for day-time sheltering as protection from weather and predators, mating, and social interaction. For species in temperate areas that form maternity groups in forested landscapes, roosts also provide thermal benefits for successful juvenile development [14]. Because of their importance in both survival and recruitment, roosts long have been considered a critical habitat feature for bats [5, 6]. Approximately half of all known bat species use plants as roosts [6]; in North America, roosts most commonly are found in snags or live trees with cavities or defects. Roosts such as snags in forests are ephemeral [7, 8]. Ephemerality of the roost resource strongly suggests that bats experience roost loss at some low constant background level, with periodic pulses of increased roost loss after intense disturbances from fire, wind throw, ice damage, insect outbreak, or certain types of forest management actions [912]. It seems likely, therefore, that bats are adaptive to roost loss. This plasticity often is ignored as many managers tasked with bat conservation often view roosts and roosting areas as fixed landscape elements that are decoupled from stochastic environmental processes [13, 14].

Bat conservation in forested landscapes often involves identification of roost sites with subsequent limitations on management activities (e.g., forestry) within these areas. Conservative approaches to roost habitat management may seem warranted, but this strategy may interrupt natural processes or anthropogenic management actions that are vital to create suitable roosts in the present or provide roosts in the future. Impacts of management actions that result in roost loss are unknown as few studies directly have assessed the effect of roost loss on bat roosting behavior in controlled, manipulative studies. Evidence from roost exclusion studies suggests that exclusion from permanent structures can decrease site fidelity, alter home range size, lower reproductive recruitment, and reduce colony size and the strength of association among individuals [1518]. Conversely, several lines of evidence suggest that tree roosting bats may be tolerant of roost loss up to some threshold point. For example, bats have exhibited positive roosting responses to prescribed fire at short-term and long-term temporal scales [1923]. Positive responses to prescribed fire may be due to rapid, increased snag recruitment that offsets the loss of existing snags [2426]. Clearly, natural forest disturbance processes also can remove and create bat roosts. Natural forest disturbance processes contrast with many types of forest harvest that remove potential and available roosts without creating new roosts in the short-term. However, if applied on the landscape properly, it is possible that forest harvesting may mimic natural processes that also create suitable roosting areas or possibly enhance the quality of existing roosts, i.e., reduce canopy shading of remaining boles.

Tolerance limits to roost loss are unclear and probably highly variable among bat species and the forest systems wherein they reside [1518, 27, 28]. For colonial species, insight into the impacts of roost loss will require understanding both of individual and colony level factors [29]. Responses to roost loss may be apparent in demographics, survival, roost use, space use, and sociality. Unfortunately, demographic changes are exceedingly difficult to ascertain for bats that roost-switch frequently and exhibit fission-fusion behavior. Within the context of roost use, resilience to roost loss generally may be visible as either a shift in overall uses of individual roosts without a change in overall space use or social structure, or alternatively, as a shift in roosting area and roosts without a change in social structure. Conversely, if colonies are not robust to disturbance, the colony may either dissolve such that social structure at the site is not maintained, or dissolve to the point where no bats are present on the site [27]. Within the network of roosts used by colonies of bats, individual roosts frequently are used differentially, with some receiving intense use (primary roosts) and others limited use (secondary roosts) [2931]. Roost switching studies have provided insight on why bats may switch roosts, but the underlying causes for differences in the relative level of roost use have not been investigated widely. Regardless, differential roost use suggests that individual roosts may either serve different functions for colonies and individual bats therein or vary in their value. If so, loss of heavily used or primary roosts may impact colonies more strongly than loss of less frequently used roosts [28, 29].

Our objective was to experimentally examine how hierarchical loss of roosts affects roosting social structure along with roost and space use by female northern long-eared bats (Myotis septentrionalis) during the maternity season at both the colony and individual level. Northern long-eared bats occur in forests throughout the eastern United States and southern Canada [3238], but foraging activity consistently is greatest in closed-canopy forests [34, 3944]. During the maternity season (May-July), female northern long-eared bats form non-random assorting colonies in upland forests under the exfoliating bark or within cavities of snags or declining live trees [10, 33, 36, 44]. This species is a proposed for listing as endangered and currently of high conservation concern in North America (Federal Register § 78:61045–61080) due to severe population declines following the onset and spread of White-nose Syndrome in eastern North America. An improved understanding of the effects of roost loss on this species will be important for development of future conservation efforts.

Accordingly, we evaluated the impacts of primary and multiple secondary roost loss specifically to reflect discussion in the literature by Rhodes et al. [29] and Silvis et al. [27] that suggests that loss of either a single primary of >20% of total roosts might result in colony fragmentation, a negative conservation outcome of substantial concern. We assessed changes in colony roost and space use, roost selection, and social structure, as well as changes in individual behaviors related to roost switching. We specified several a priori hypotheses related to the differing levels of roost site disturbance based on previous research on multiple species [15, 16, 18, 27, 29]. For primary roost tree removal, we proposed 2 hypotheses:

  1. H1: At the colony level, loss of the primary roost will result in an alternate tree receiving increased use, subsequently causing a previously less-used roost to become the primary roost [15, 16]; bats will not display evidence of roost seeking behavior. Bats will display an affinity for the same roosting area, but the core use area would re-center around the new primary roost, and roost selection would be consistent. At the individual level, loss of the primary roost will not impact roost switching behavior or distances moved between sequentially used roosts.
  2. H2: At the colony level, loss of the primary roost will result in dissolution of the colony [29]. Space use will either be random across the former roosting area or will be nonexistent. Bats will display characteristics of roost searching, and the characteristics of selected roosts will differ [18]. At the individual level, loss of the primary roost will increase roost switching frequency and the distances moved between sequentially used roosts.

For secondary roost loss, we proposed three hypotheses:

  1. H1: At the colony level, loss of multiple secondary roosts will not impact roosting behavior, social structure, space use, or roost selection by northern long-eared bat maternity colonies [27]. At the individual level, loss of multiple secondary roosts will not impact roost switching behavior or distances moved between sequentially used roosts. Roost characteristics will not differ.
  2. H2: At the colony level, loss of multiple secondary roosts will result in dissolution of the colony [27]. Space use will either be random across the former roosting area or will be nonexistent. Bats will display characteristics of roost searching and roost characteristics will differ [18]. At the individual level, loss of multiple secondary roosts will increase roost switching frequency and the distances moved between sequentially used roosts.
  3. H3: At the colony level, loss of multiple secondary roosts will result in increased social cohesion and increased use of the primary roost, and roosting area will decrease. Roost characteristics will not differ. At the individual level, loss of multiple secondary roosts will decrease the number of roosts used by individual bats and the distances moved between roosts.


We conducted our study at 3 sites on the Fort Knox military reservation in Meade, Bullitt, and Hardin Counties, Kentucky, USA (37.9°N, −85.9°E, WGS84). Our sites lie in the Western Pennyroyal subregion of the Mississippian portion of the Interior Low Plateau physiographic province of the upper South and lower Midwest portion of the USA [45]. Forest cover is predominantly a western mixed-mesophytic association [46], with second- and third-growth forests dominated by white oak (Quercus alba), black oak (Q. velutina), chinkapin oak (Q. muehlenbergii), shagbark hickory (Carya ovata), yellow poplar (Liriodendron tulipifera), white ash (Fraxinus americana), and American beech (Fagus grandifolia) in the overstory, and sassafras (Sassafras albidum), redbud (Cercis canadensis), and sugar maple (Acer saccharum) in the understory [47].

We initially captured northern long-eared bats over small woodland pools from May through July 2011 (pre-roost removal) and 2012 (post-roost removal). We attached a radiotransmitter (LB-2, 0.31 g: Holohil Systems Ltd., Woodlawn, ON, Canada) between the scapulae of each female bat using Perma-Type surgical cement (Perma-Type Company Inc., Plainville, CT, USA). A uniquely numbered lipped band was attached to the forearm of all captured bats. After identifying a small number of roosts, we maximized number of bats captured by erecting mist nets around roosts located while radiotracking bats. Captured bats were released within 30 minutes of capture at the net site. Using TRX-1000S receivers and folding 3-element Yagi antennas (Wildlife Materials Inc., Carbondale, IL, USA), we attempted to locate radio-tagged bats daily for the life of the transmitter or until the unit dropped from the bat. For each located roost, we recorded tree species, diameter at breast height (dbh; cm), height (m), canopy openness (%), decay class ([48]; live [1], declining [2], recent dead [3], loose bark [4], no bark [4], broken top [6], broken bole [7]) and crown class ([49]; i.e., suppressed [S], intermediate [I], codominant [CO], dominant [D]). We estimated size of individual colonies by performing 5 exit counts per colony at day-roosts used by radiotracked bats.

We followed the methods of Silvis et al. [27] in defining a northern long-eared bat maternity colony as all female and juvenile bats connected by coincident roost use. We represented colonies graphically and analytically as two-mode networks that consisted of bats and roosts (hereafter “roost network”) [30, 31]. We used these roost network representations to describe patterns of roost use by colonies and to identify roosts for our removal treatments. To reduce bias resulting from uneven tracking periods and observing only a portion of each colony, we considered relationships to be binary (i.e., presence or absence of a connection) [50]. We assessed roost network structure using mean degree, network degree centralization, network density, and clustering. Within networks, degree is a count of the number of edges incident with a node [51]; high degree values indicate a large number of connections to a node. Network degree centralization, density, and clustering all have values between 0 and 1 (0 = low, 1 = high). Network degree centralization describes the extent that a network is structured around individual nodes, whereas network density and clustering describe the distribution of connections among nodes [5256]. We calculated two-mode degree centralization and density using the methods of Borgatti and Everett [52] and clustering using the method of Opsahl [57] for our roost network. To determine whether our observed network values differed from those of random networks, we performed 999 Monte Carlo simulations and compared observed network metrics to random network metrics using two-tailed permutation tests [58, 59]; random networks [60] were generated with the same number of nodes as our observed networks and with a constant probability of link establishment. We then compared the relative difference from random networks pre-post treatment to assess whether colony social dynamics and roost use patterns were disrupted.

In February 2012 when bats were hibernating and not occupants of trees and snags, we implemented two roost removal treatments and one control following the identification and delineation of 3 colonies in 2011. For our primary roost removal treatment, we felled the single roost with the highest degree centralization value via chainsaw. For the secondary roost removal treatment, we similarly felled 5 randomly selected roosts (24% of colony total) with degree centralization values less than the colony maximum, but greater than the colony minimum in our secondary roost removal treatment group. This number was selected to specifically test the simulation-based predictions of Silvis et al. [27] that colonies may fragment with loss of >20% of roosts.

We used conditional Wilcoxon 2-sample tests and conditional Chi-squared tests to compare continuous (height, dbh, and canopy openness) and categorical roost characteristics (species composition, decay stage, and crown class) pre- and post-treatment and among groups; we corrected for multiple comparisons using the Bonferroni method. Conditional tests were performed using Monte Carlo simulations with 999 permutations. We examined the roost switching behavior of individual bats by creating a Poisson regression model describing the number of roosts used by a bat relative to the total number of relocations, reproductive condition, and interaction of treatment identity and year. We used this Poisson model to conduct general linear hypothesis tests with Tukey’s adjustment for multiple comparisons to determine whether the number of roosts used by bats differed within or among treatment areas. We evaluated the fit of our Poisson model using maximum-adjusted D2 [61]. We assessed the spatial component of roost switching behavior by individual bats by comparing the distances that bats within treatment areas moved between sequentially used roosts with general linear hypothesis tests, also with Tukey’s adjustment for multiple comparisons. We performed our general linear hypothesis tests for distances moved on a linear mixed model containing year, group, their interaction term, and reproductive condition as fixed effects, and bat identity as a random effect; we used a log transformation to normalize distance data. We assessed the fit of our linear mixed model using the conditional (R2c) and marginal (R2m) coefficients of determination [62].

We evaluated roost removal impacts on colony roosting area space use for each treatment group using Bhattacharya’s affinity (BA) [63] and the difference in roosting area centroids between years. The BA uses the joint distribution of 2 utilization distributions to quantify similarity between utilization distributions and is appropriate for comparisons of utilization distributions for the same individual or group [63]. These values range from 0 to 1, with values close to 1 indicating highly similar utilization distributions [63]. We calculated 95% utilization distributions from the pooled locations of all bats within a colony using bivariate normal fixed kernel methodology. To reflect the concentration of roost use, we weighted roost locations by the number of times a roost was used by radio-tagged bats [64]. We used the reference method for smoothing parameter estimation as appropriate for weighted locations [65]; that also allowed us to consider our estimates of colony space use as liberal. In cases where roosting areas of separate colonies overlapped to an appreciable extent, we calculated the utilization distribution overlap index (UDOI) to determine if space use was independent; UDOI values range from 0 to infinity, with values <1 indicating independent space use, and values >1 indicating non-independence [63].

We assessed overall changes in colony roost use patterns by comparing pre- and post-roost removal network degree centralization, density, and clustering for the roost networks. We used this same comparative network approach to assess changes in colony roosting social structure for the single mode projections of our 2-mode roost networks [66]. This projection allowed us to focus on existing direct and indirect connections among bats in a colony. Because comparing values from networks of differing size may yield inappropriate inferences [67], we used indirect comparisons of network characteristics. In these, we compared the relative difference between a roost or social network and its equivalent random network pre- and post-treatment. All analyses were performed in the R statistical program version 3.0.2 [68]. We calculated conditional tests using the coin package [69], linear mixed models using lme4 [70], and utilization distributions, BA, and UDOI values using the adehabitatHR package [71]. We used the igraph [72] and tnet libraries [57] to visualize networks and calculate metrics. Lastly, network Monte Carlo simulations were performed using a custom script with dependencies on the igraph and tnet libraries. We used an α = 0.05 for all tests of statistical significance.

Ethics statement

Our study was carried out in accordance with state requirements for capture and handling of wildlife (Kentucky Department of Fish and Wildlife Resources permit numbers SC1111108 and SC1311170) and did not involve any endangered species at the time of the study. Capture and handling protocol followed the guidelines of the American Society of Mammalogists [73] and was approved by the Virginia Polytechnic Institute and State University Institutional Animal Care and Use Committee (protocol number 11–040-FIW). We received explicit permission to conduct work on the Fort Knox military reservation from the reservation staff biologists and Fort Knox Range Control. Data used in this study are archived in the Virginia Polytechnic Institute and State University VTechWorks institutional repository (DOI: 10.7294/W4H41PBH).


We captured 58 female northern long-eared bats pre-treatment in 2011. Based on patterns of coincident roost use, we assigned 36 of these bats (11 gestating, 20 lactating, 1 post-lactation, and 4 non-reproductive) to 3 colonies. Exit counts for these 3 colonies generated minimum estimated colony sizes of 13, 18, and 14 bats, respectively. We captured 67 bats post-treatment in 2012, 62 of which (4 gestating, 45 lactating, 10 post-lactation, and 3 non-reproductive) we were able to assign to the 3 colonies identified in 2011. We recaptured only 3 individuals banded in 2011 during 2012. Exit counts indicated that the 2012 colonies contained a minimum of 24, 20 and 25 bats, respectively. We located 58 roosts over 204 relocation events for the 3 colonies identified in 2011 and 100 roosts (7 of which were used in 2011) over 324 relocation events in 2012. We recorded a mean (± SD) of 5.7 (± 1.5) locations per bat in 2011 and 5.2 (± 2.9) in 2012.

We identified between 4 and 33 roosts per colony pre-roost removal, and between 23 and 42 roosts per colony post-removal (Table 1). When controlling for the total number of relocations of an individual bat and reproductive condition, the number of roosts used by individual bats was similar between pre- and post-treatment and among colonies, with the exception of the control colony, pre-removal, that differed from all other groups (model D2 = 0.74; Tables 1, 2).

Table 2. Factors influencing the number of roosts used by individual female northern long-eared bats.

Neither roost dbh nor height differed between treatments or among colonies (Table 3). Canopy openness was similar between pre- and post-treatment, but some individual colonies differed from one another (Table 3). Distribution of roosts among decay stages differed pre- and post-treatment within the primary removal colony but not in the control colony or the secondary removal colony (Table 3). Distribution of roosts among crown classes differed pre- and post-treatment for the primary removal colony but not in the control or secondary removal colony (Table 3). Distribution of roosts among decay stage and crown classes did differ among colonies in some cases (Table 3). We found no difference in roost species composition between pre- and post-treatment or among any of our groups (Table 3). Sassafras (Sassafras albidum) trees or snags were the most commonly used roost species, accounting for between 43 and 57% of roosts used in each group.

Table 3. Summary of female northern long-eared bat roost characteristics.

Distances moved between sequentially used roosts were non-normally distributed with right skew; median distances were between 111.1 and 219.4 m (Table 1). Distances between sequentially used roosts differed only pre- and post-roost removal in our secondary roost removal treatment group (model R2c = 0.18, R2m = 0.08; Tables 1, 4). Overall colony roosting areas were between 1.3 and 58.5 ha (Table 1). Patterns of roosting area space use largely were consistent between pre- and post-treatment in our primary and secondary roost removal treatment groups, particularly evident in the distances between weighted colony roosting area centroids (Table 1, Fig. 1). However, space use by and roosting area centroids of our control colony differed substantially between years (Table 1).

Figure 1. Northern long-eared bat maternity colony roosting areas.

Roosting areas (95% utilization distribution) of 3 northern long-eared bat (Myotis septentrionalis) maternity colonies subjected to different levels of roost removal on the Fort Knox military reservation, Kentucky, USA, pre- and post- roost removal (2011 and 2012)

Table 4. Factors influencing distances moved between roosts by female northern long-eared bats.

Roost network degree centralization significantly was greater than random for primary removal and control colonies, but not the secondary roost removal colony pre-treatment (Table 1). Roost network clustering differed from random networks in both the primary and secondary roost removal colonies post-treatment, but, for all other colonies, there was no difference from random networks (Table 1). Roost network density did not significantly differ from random networks for any group (Table 1). As represented in the social networks, bats shared between 3.5 and 15.9 social connections with other bats within colonies (Table 5). Social network degree centralization differed from random networks only for the control colony pre-treatment and the primary roost removal treatment post-treatment; the former was significantly less than and the latter significantly greater than equivalent random networks (Table 5). Social network clustering significantly was greater than that of random networks for colonies except the secondary roost removal treatment colony pre-treatment (Table 5). Social network density did not differ from random networks pre-treatment, but was greater in all other cases (Table 5).

Table 5. Northern long-eared bat maternity colony social network metrics.

Visual inspection of the roost network maps indicated that the secondary roost removal colony was split into 2 groups connected only by a single roost post-treatment (Fig. 2). Because these 2 halves possibly represented 2 separate colonies connected by a single ‘chance’ roost use, we conducted a post-hoc analysis wherein we removed the roost connecting the 2 network sections (subcolony 1 and subcolony 2) and re-calculated spatial metrics. Roosting area was 46.37 ha for subcolony 1 and 27.43 ha for subcolony 2. Roosting areas of these 2 sections overlapped substantially (UDOI = 1.26).

Figure 2. Northern long-eared bat maternity colony roost network map.

Pre- and post- roost removal treatment (2011 and 2012) 2-mode roost network map of a northern long-eared bat (Myotis septentrionalis) maternity colony subjected to removal of 5 secondary roosts on the Fort Knox military reservation, Kentucky, USA. Edge width is scaled by the number of connections between a bat and an individual roost.


In our manipulative roost removal experiment, treatments did not result in abandonment of roosting areas by northern long-eared bats. Persistence after exclusion from a roost also has been observed in big brown bats (Eptesicus fuscus) in northern forest-prairie transitions zones in Canada [15] and disc-winged bats (Thyroptera tricolor) in Costa Rican tropical forests [18], species that both exhibit relatively frequent roost switching. In contrast, syntopic little brown bats (Myotis lucifugus), that form larger colonies and roost-switch less than northern long-eared bats, appear to abandon roosting areas after exclusion [16]. Persistence after roost loss may be related to the greater number of roosts used by colonies and to roost ephemerality. Roost fidelity is less in species with more ephemeral roosts [74], therefore, having a variety of alternate roosts or some degree of flexibility in what roosts may be selected may be an adaptation for tolerating roost loss for the northern long-eared bat.

Northern long-eared bat maternity colony roosting areas did not appear to change as a result of either of our roost removal treatments. In contrast, Chaverri and Kunz [18] found that exclusion resulted in larger individual roosting home ranges in disc-winged bats [18] and Borkin et al. [17] found that roost loss resulted in smaller home ranges in New Zealand long-tailed bats (Chalinolobus tuberculatus) [17]. Increased home range size in disc-winged bats was related to the need to locate a limiting resource—suitable roosts [18]. However, northern long-eared bats are not extreme roost specialists [32, 75, 76] and potential roosts are not limited on our sites [77]. On the other hand, decreased home range size in New Zealand long-tailed bats as a result of roost loss following clear-cutting, reflected the lack of available roosts and alternative roosting areas in the harvested areas [17]. Locally, large numbers of available roosts may explain why so few roosts were used in both years of our study and why colony locations did not change.

It was surprising that so few roosts were used both pre- and post-treatment, but could be the result of tracking different bats in each year. We captured a substantial proportion of the bats within individual colonies (range 0.62–1.0, ). As such, it is unlikely that our low recapture rate was due to sampling effort. Regardless, roost removal treatments did not impact the number of roosts used by individual bats within treatment areas when controlling for the number of total locations and reproductive condition. The lack of difference in the number of roosts used differs from Borkin et al. [17], who found that bats used fewer roosts post-roost loss. The number of roosts used per bat was fewer in 2011 than in 2012 in our control colony, but this is likely due to the fact that the colony was captured and tracked during parturition in 2011 [78]; the number of roosts used per bat in the control colony in 2012 was consistent with that of all other groups. Given the positive relationship between the number of roosts located and the number of days a bat was tracked, differences in the total number of roosts located per colony were not unexpected.

Northern long-eared bats are known to exhibit inter-annual site fidelity of at least 5 years in a mixed pine-deciduous system in Arkansas [79], but our low recapture rates relative to our sampling effort suggest that bats marked during the first year of our study largely were not present in the second. Whether this is due to high annual adult mortality or some other socio-spatial assortment dynamic is unknown, but Perry [79] also recaptured few banded individuals. Consistent patterns of space use between years suggest that, although colony composition changed, colony identity did not. Northern long-eared bat maternity colonies [80] as well as those of some other species [81] contain maternally-related individuals, and it is possible that primarily juveniles from the first year returned in the second. In the context of having tracked different bats within colonies, our data may be interpreted best not as changes in behavior of individual bats resulting from removal treatments, but as differences in patterns of colony behavior at our treatment sites.

In contrast to Chaverri and Kunz [18], we observed no change in roost species selection post-roost removal. This is consistent with the high roost availability at our sites [27]. Roost decay stage and crown class in the primary removal colony were the only roost characteristics to differ between pre- and post-treatment. Selection for more advanced stages of decay in 2011 appears to be correlated with crown class, as trees in advanced stages of decay at our sites are primarily in suppressed crown classes. Although the difference in decay stage and crown class pre- and post-treatment is statistically significant only for the primary removal colony, a similar trend in reduced selection for suppressed roosts in later stages of decay was visible across all colonies in 2012. It is possible that by random chance roost removal caused the difference in roost decay stage and crown class in our findings, but given the lack of difference between roost dbh, height, and canopy openness in the primary removal colony, this seems unlikely. Higher summer temperatures in 2011 than in 2012 on our study site may have caused bats to select trees in more suppressed crown classes, thereby reducing solar heating of roosts. Mean minimum temperature during June–July was 1.78 C° greater in 2011 than in 2012 (National Oceanic and Atmospheric Administration station GHCND: USC00154955); similarly small temperature differences have been found to affect roost selection by Bechstein’s bats (Myotis bechsteinii) [82] and development of juvenile greater mouse-eared bats (Myotis myotis) [83].

Patterns of northern long-eared bat roost use and association, as assessed through roost and social networks, displayed a mix of random and non-random characteristics. The overall character of roost networks relative to random networks was similar within and among treatments. Although there were minor differences in roost and social networks pre- and post-treatment, northern long-eared bat social network structure changes with reproductive condition [84, 85]. After accounting for reproductive condition, the character of the roost networks post-treatment differed only for roost network clustering. The change in roost network clustering from not significantly different from random networks to significantly greater than random networks also was reflected through increased social network density. An increase in roost network clustering and social network density may be an adaptive response to maintain colony stability after roost loss. Such an adaptive response to roost loss could suggest co-evolution between northern long-eared bats and these mixed mesophytic forests and other systems with similar stand dynamics and disturbance patterns, but replication of our study across more regions and forest types is required to document this.

For the secondary roost removal colony, we observed a segmented roost network and the only statistically significant difference in the distance moved between sequentially used roosts. Division of this network into 2 halves as a result of the removal of 24% of roosts would be consistent with previous simulation based outcomes showing that loss of approximately 20% of roosts generates a 50% chance of colony fragmentation [27]. Connection of the 2 halves of this network by a single roost may reflect an incomplete division of the colony. An incomplete division may indicate that colony fragmentation occurs incrementally as roosts are lost, an outcome that theoretically should be most likely to occur if individual roosts are important locations for social interaction. Incomplete colony fragmentation is consistent with our finding that the 2 sections of this colony shared a single roosting area—an observation that was contrary to our a priori prediction that colony fragmentation would result in random use of the roosting area, but that may be related to the difference in distances moved between roosts by bats in this colony. Alternately, apparent division also could be the result of unwarranted joining of two separate neighboring colonies as a result of chance use of single roost. Silvis et al. [27] speculated that roost sharing may be infrequent and inconsequential at the periphery of the roosting area for northern long-eared bats. In this case, the shared roost was not at the periphery of the colony roosting area and the roosting areas of the 2 sections of the colony overlapped extensively in terms of both extent and concentration of use. Research from other bat species in both temperate and tropical regions suggests that roosting areas are exclusive relatively to individual colonies [17, 30, 31]. Whether this apparent fragmentation is a result of roost removal treatments or some other process remains speculative.


In their review of conservation concerns for bats in the United States, Weller et al. [86] identified a need to transition conservation priorities from focal threats to diffuse threats. In the context of the White-nose Syndrome enzootic that is threatening many species, including the northern long-eared bat, with widespread extirpation, it is necessary to link focal and diffuse threats through understanding of the impacts of specific changes to roosting habitats. Although our study contains limited replicates of our individual treatments, it is to our knowledge the only study to perform targeted roost removal treatments for colonial bats in a temperate forest ecosystem. Clearly, caution should be taken in interpreting the results of individual treatments, particularly with regard to changes in roost and social network structure. However, our results are consistent with previous predictions and anecdotal observations that northern long-eared bats would be robust to low levels of roost loss [20, 22] particularly if loss of these naturally ephemeral roost resources are lost at or below rates of tree mortality / snag loss in temperate forests. Clearly, the maximum levels of annual or cumulative multi-year roost loss that northern long-eared bats can tolerate remains to be determined. It is important to consider that roosts were not limiting at our study sites similar to much of the temperate forested environments where northern long-eared bats occur [10, 87]. However, in more roost limited areas, e.g., in agricultural landscapes with greater forest fragmentation or in industrial forest settings skewed towards younger forest age classes, roost loss may have different consequences for northern long-eared bats.

Monitoring of sufficient numbers of colonies for robust inference is largely infeasible within a single study. Therefore, replication across studies is needed to better confirm or modify the patterns we have observed. With the ongoing spread of White-nose Syndrome in North America, and continued rapid declines in northern long-eared bat populations, replication of this study in disease-free areas is urgently needed. Moreover, a better understanding the impacts of roost loss, whether natural or anthropogenic, on survival and recruitment remains a critical gap in our knowledge of bat ecology.


We thank Jimmy Watkins, Mike Brandenberg, and Charlie Logsdon for their assistance in supporting this project. The Kentucky Department of Fish and Wildlife Resources graciously provided field housing for this project. We thank James Fraser, James Parkhurst, Tim Carter, Steven Castleberry and an anonymous reviewer for comments that improved this manuscript. Use of trade, product, or firm names does not imply endorsement by the US government.

Author Contributions

Conceived and designed the experiments: WMF ERB AS. Performed the experiments: AS WMF ERB. Analyzed the data: AS WMF ERB. Contributed reagents/materials/analysis tools: WMF ERB. Wrote the paper: AS WMF ERB.


  1. 1. Grindal SD, Collard TS, Brigham RM, Barclay RMR (1992) The influence of precipitation on reproduction by Myotis bats in British Columbia. Am Midl Nat 128: 339–344.
  2. 2. Willis CK., Lane JE, Liknes ET, Swanson DL, Brigham RM (2005) Thermal energetics of female big brown bats (Eptesicus fuscus). Can J Zool 83: 871–879.
  3. 3. Speakman JR (2008) The physiological costs of reproduction in small mammals. Philos Trans R Soc B Biol Sci 363: 375–398.
  4. 4. Frick WF, Reynolds DS, Kunz TH (2010) Influence of climate and reproductive timing on demography of little brown myotis Myotis lucifugus. J Anim Ecol 79: 128–136. pmid:19747346
  5. 5. Fenton MB (1997) Science and the conservation of bats. J Mammal 78: 1–14.
  6. 6. Kunz TH, Lumsden LF (2003) Ecology of cavity and foliage roosting bats. In: Kunz TH, Fenton MB, editors. Bat Ecology. Chicago, Illinois: University of Chicago Press. pp. 2–90.
  7. 7. Moorman CE, Russell KR, Sabin GR, Guynn Jr DC (1999) Snag dynamics and cavity occurrence in the South Carolina Piedmont. For Ecol Manag 118: 37–48.
  8. 8. Wisdom MJ, Bate LJ (2008) Snag density varies with intensity of timber harvest and human access. For Ecol Manag 255: 2085–2093.
  9. 9. McShea WJ, Healy WM (2002) Oak forest ecosystems : Ecology and management for wildlife. Baltimore, Maryland: Johns Hopkins University Press. 432 p.
  10. 10. Menzel MA, Carter TC, Menzel JM, Ford WM, Chapman BR (2002) Effects of group selection silviculture in bottomland hardwoods on the spatial activity patterns of bats. For Ecol Manag 162: 209–218.
  11. 11. Lorimer CG, White AS (2003) Scale and frequency of natural disturbances in the northeastern US: Implications for early successional forest habitats and regional age distributions. For Ecol Manag 185: 41–64.
  12. 12. Copenheaver CA, Matthews JM, Showalter JM, Auch WE (2006) Forest stand development patterns in the southern Appalachians. Northeast Nat 13: 477–494.
  13. 13. Dixon MD, Heist K, Tinsley K (2013) The state of bats in conservation planning for the National Wildlife Refuge System, with recommendations. J Fish Wildl Manag 4: 406–422.
  14. 14. Stone EL, Jones G, Harris S (2013) Mitigating the effect of development on bats in England with derogation licensing. Conserv Biol 27: 1324–1334. pmid:24112694
  15. 15. Brigham RM, Fenton MB (1986) The influence of roost closure on the roosting and foraging behaviour of Eptesicus fuscus (Chiroptera: Vespertilionidae). Can J Zool 64: 1128–1133.
  16. 16. Neilson AL, Fenton MB (1994) Responses of little brown myotis to exclusion and to bat houses. Wildl Soc Bull 22: 8–14.
  17. 17. Borkin KM, O’Donnell C, Parsons S (2011) Bat colony size reduction coincides with clear-fell harvest operations and high rates of roost loss in plantation forest. Biodivers Conserv 20: 3537–3548.
  18. 18. Chaverri G, Kunz TH (2011) Response of a specialist bat to the loss of a critical resource. PLoS ONE 6: e28821. pmid:22216118
  19. 19. Boyles JG, Aubrey DP (2006) Managing forests with prescribed fire: Implications for a cavity-dwelling bat species. For Ecol Manag 222: 108–115.
  20. 20. Johnson JB, Edwards JW, Ford WM, Gates JE (2009) Roost tree selection by northern myotis (Myotis septentrionalis) maternity colonies following prescribed fire in a central Appalachian mountains hardwood forest. For Ecol Manag 258: 233–242.
  21. 21. Johnson JB, Ford WM, Rodrigue JL, Edwards JW, Johnson CM (2010) Roost selection by male Indiana myotis following forest fires in central Appalachian hardwoods forests. J Fish Wildl Manag 1: 111–121.
  22. 22. Lacki MJ, Cox DR, Dodd LE, Dickinson MB (2009) Response of northern bats (Myotis septentrionalis) to prescribed fires in eastern Kentucky forests. J Mammal 90: 1165–1175.
  23. 23. Womack KM, Amelon SK, Thompson FR (2013) Resource selection by Indiana bats during the maternity season. J Wildl Manag 77: 707–715.
  24. 24. Bagne KE, Purcell KL, Rotenberry JT (2008) Prescribed fire, snag population dynamics, and avian nest site selection. For Ecol Manag 255: 99–105.
  25. 25. Hutchinson TF, Long RP, Ford RD, Sutherland EK (2008) Fire history and the establishment of oaks an maples in second-growth forests. Can J For Res 38: 1184–1198.
  26. 26. Signell SA, Abrams MD, Hovis JC, Henry SW (2005) Impact of multiple fires on stand structure and tree regeneration in central Appalachian oak forests. For Ecol Manag 218: 146–158.
  27. 27. Silvis A, Ford WM, Britzke ER, Johnson JB (2014) Association, roost use and simulated disruption of Myotis septentrionalis maternity colonies. Behav Processes 103: 283–290. pmid:24468215
  28. 28. Silvis A, Kniowski AB, Gehrt SD, Ford WM (2014) Roosting and foraging social structure of the endangered Indiana bat (Myotis sodalis). PLoS ONE 9: e96937.
  29. 29. Rhodes M, Wardell-Johnson GW, Rhodes MP, Raymond B (2006) Applying network analysis to the conservation of habitat trees in urban environments: A case study from Brisbane, Australia. Conserv Biol 20: 861–870. pmid:16909578
  30. 30. Fortuna MA, Popa-Lisseanu AG, Ibáñez C, Bascompte J (2009) The roosting spatial network of a bird-predator bat. Ecology 90: 934–944. pmid:19449689
  31. 31. Johnson JS, Kropczynski JN, Lacki MJ, Langlois GD (2012) Social networks of Rafinesque’s big-eared bats (Corynorhinus rafinesquii) in bottomland hardwood forests. J Mammal 93: 1545–1558.
  32. 32. Foster RW, Kurta A (1999) Roosting ecology of the northern bat (Myotis septentrionalis) and comparisons with the endangered Indiana bat (Myotis sodalis). J Mammal 80: 659–672.
  33. 33. Lacki MJ, Schwierjohann JH (2001) Day-roost characteristics of northern bats in mixed mesophytic forest. J Wildl Manag 65: 482–488.
  34. 34. Menzel MA, Owen SF, Ford WM, Edwards JW, Wood PB, et al. (2002) Roost tree selection by northern long-eared bat (Myotis septentrionalis) maternity colonies in an industrial forest of the central Appalachian mountains. For Ecol Manag 155: 107–114.
  35. 35. Broders HG, Forbes GJ, Woodley S, Thompson ID (2006) Range extent and stand selection for roosting and foraging in forest-dwelling northern long-eared bats and little brown bats in the Greater Fundy ecosystem, New Brunswick. J Wildl Manag 70: 1174–1184.
  36. 36. Perry RW, Thill RE (2007) Roost selection by male and female northern long-eared bats in a pine-dominated landscape. For Ecol Manag 247: 220–226.
  37. 37. Morris AD, Miller DA, Kalcounis‐Rueppell MC (2010) Use of forest edges by bats in a managed pine forest landscape. J Wildl Manag 74: 26–34.
  38. 38. Johnson JB, Ford WM, Edwards JW (2012) Roost networks of northern myotis (Myotis septentrionalis) in a managed landscape. For Ecol Manag 266: 223–231.
  39. 39. Brooks RT, Ford WM (2005) Bat activity in a forest landscape of central Massachusetts. Northeast Nat 12: 447–462.
  40. 40. Ford WM, Menzel MA, Rodrigue JL, Menzel JM, Johnson JB (2005) Relating bat species presence to simple habitat measures in a central Appalachian forest. Biol Conserv 126: 528–539.
  41. 41. Ford WM, Menzel JM, Menzel MA, Edwards JW, Kilgo JC (2006) Presence and absence of bats across habitat scales in the upper coastal plain of South Carolina. J Wildl Manag 70: 1200–1209.
  42. 42. Jung TS, Thompson ID, Titman RD, Applejohn AP (1999) Habitat selection by forest bats in relation to mixed-wood stand types and structure in central Ontario. J Wildl Manag 63: 1306–1319.
  43. 43. Loeb SC, O’Keefe JM (2006) Habitat use by forest bats in South Carolina in relation to local, stand, and landscape characteristics. J Wildl Manag 70: 1210–1218.
  44. 44. Owen SF, Menzel MA, Edwards JW, Ford WM, Menzel JM, et al. (2004) Bat activity in harvested and intact forest stands in the Allegheny mountains. North J Appl For 21: 154–159.
  45. 45. Arms ELP, Mitchell MJ, Watts FC, Wilson BL (1979) Soil survey of Hardin and Larue Counties, Kentucky. USDA Soil Conserv Serv.
  46. 46. Braun EL (1950) Deciduous forests of eastern North America. Philadelphia, Pennsylvania: Blakiston Company. 596 p.
  47. 47. Cranfill R (1991) Flora of Hardin County, Kentucky. Castanea 56: 228–267.
  48. 48. Cline SP, Berg AB, Wight HM (1980) Snag characteristics and dynamics in Douglas-fir forests, western Oregon. J Wildl Manag 44: 773–786.
  49. 49. Nyland RD (2002) Silviculture: Concepts and applications. Second edition. New York, NY: McGraw-Hill. 704 p.
  50. 50. Goodreau SM, Kitts JA, Morris M (2009) Birds of a feather, or friend of a friend? Using exponential random graph models to investigate adolescent social networks. Demography 46: 103–125. pmid:19348111
  51. 51. Boccaletti S, Latora V, Moreno Y, Chavez M, Hwang D-U (2006) Complex networks: Structure and dynamics. Phys Rep 424: 175–308.
  52. 52. Borgatti SP, Everett MG (1997) Network analysis of 2-mode data. Soc Netw 19: 243–269.
  53. 53. Dong J, Horvath S (2007) Understanding network concepts in modules. BMC Syst Biol 1: 24. pmid:17547772
  54. 54. Freeman LC (1978) Centrality in social networks conceptual clarification. Soc Netw 1: 215–239.
  55. 55. Wasserman S, Faust K (1994) Social network analysis: Methods and applications. Cambridge University Press. 852 p.
  56. 56. Watts DJ, Strogatz SH (1998) Collective dynamics of “small-world” networks. Nature 393: 440–442. pmid:9623998
  57. 57. Opsahl T (2009) Structure and evolution of weighted networks [Dissertation]. London, United Kingdom: Queen Mary, University of London. Available: Accessed 1 April 2013.
  58. 58. Hope ACA (1968) A simplified Monte Carlo significance test procedure. J R Stat Soc Ser B Methodol 30: 582–598.
  59. 59. Davison AC (1997) Bootstrap methods and their application. Cambridge, New York: Cambridge University Press. 598 p.
  60. 60. Erdős P, Rényi A (1960) On the evolution of random graphs. Math Inst Hung Acad Sci 5: 17–61.
  61. 61. Guisan A, Zimmermann NE (2000) Predictive habitat distribution models in ecology. Ecol Model 135: 147–186.
  62. 62. Nakagawa S, Schielzeth H (2013) A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol 4: 133–142.
  63. 63. Fieberg J, Kochanny CO (2005) Quantifying home-range overlap: The importance of the utilization distribution. J Wildl Manag 69: 1346–1359.
  64. 64. Popa-Lisseanu AG, Bontadina F, Mora O, Ibáñez C (2008) Highly structured fission–fusion societies in an aerial-hawking, carnivorous bat. Anim Behav 75: 471–482.
  65. 65. Gitzen RA, Millspaugh JJ, Kernohan BJ (2006) Bandwidth selection for fixed-kernel analysis of animal utilization distributions. J Wildl Manag 70: 1334–1344.
  66. 66. Zhou T, Ren J, Medo M, Zhang Y-C (2007) Bipartite network projection and personal recommendation. Phys Rev E 76: 046115.
  67. 67. James R, Croft DP, Krause J (2009) Potential banana skins in animal social network analysis. Behav Ecol Sociobiol 63: 989–997.
  68. 68. R. Development Core Team (2014) R: A language and environment for statistical computing. Vienna, Austria. Available:
  69. 69. Hothorn T, Hornik K, van de Wiel MA, Zeileis A (2006) A lego system for conditional inference. Am Stat 60: 257–263.
  70. 70. Bates D, Maechler M, Bolker B, Walker S (2014) lme4: Linear mixed-effects models using Eigen and S4. R package version 1.1–5.
  71. 71. Calenge C (2006) The package “adehabitat” for the R software: A tool for the analysis of space and habitat use by animals. Ecol Model 197: 516–519.
  72. 72. Csardi G, Nepusz T (2006) The igraph software package for complex network research. InterJournal Complex Systems: 1695.
  73. 73. Sikes RS, Gannon WL, the Animal Care and Use Committee of the American Society of Mammalogists (2011) Guidelines of the American Society of Mammalogists for the use of wild mammals in research.Mammal J 92: 235.
  74. 74. Lewis SE (1995) Roost fidelity of bats: A review. J Mammal 76: 481–496.
  75. 75. Carter TC, Feldhamer GA (2005) Roost tree use by maternity colonies of Indiana bats and northern long-eared bats in southern Illinois. For Ecol Manag 219: 259–268.
  76. 76. Timpone JC, Boyles JG, Murray KL, Aubrey DP, Robbins LW (2009) Overlap in roosting habits of Indiana bats (Myotis sodalis) and northern bats (Myotis septentrionalis). Am Midl Nat 163: 115–123.
  77. 77. Silvis A, Ford WM, Britzke ER, Beane NR, Johnson JB (2012) Forest succession and maternity day roost selection by Myotis septentrionalis in a mesophytic hardwood forest. Int J For Res 2012: 8.
  78. 78. Olson CR, Barclay RMR (2013) Concurrent changes in group size and roost use by reproductive female little brown bats (Myotis lucifugus). Can J Zool 91: 149–155.
  79. 79. Perry RW (2011) Fidelity of bats to forest sites revealed from mist-netting recaptures. J Fish Wildl Manag 2: 112–116.
  80. 80. Patriquin KJ, Palstra F, Leonard ML, Broders HG (2013) Female northern myotis (Myotis septentrionalis) that roost together are related. Behav Ecol 24: 949–954.
  81. 81. Metheny J, Kalcounis-Rueppell M, Willis C, Kolar K, Brigham R (2008) Genetic relationships between roost-mates in a fission–fusion society of tree-roosting big brown bats (Eptesicus fuscus). Behav Ecol Sociobiol 62: 1043–1051.
  82. 82. Kerth G, Weissmann K, König B (2001) Day roost selection in female Bechstein’s bats (Myotis bechsteinii): A field experiment to determine the influence of roost temperature. Oecologia 126: 1–9.
  83. 83. Zahn A (1999) Reproductive success, colony size and roost temperature in attic-dwelling bat Myotis myotis. J Zool 247: 275–280.
  84. 84. Garroway CJ, Broders HG (2008) Day roost characteristics of northern long-eared bats (Myotis septentrionalis) in relation to female reproductive status. Ecoscience 15: 89–93.
  85. 85. Patriquin KJ, Leonard ML, Broders HG, Garroway CJ (2010) Do social networks of female northern long-eared bats vary with reproductive period and age? Behav Ecol Sociobiol 64: 899–913.
  86. 86. Weller T, Cryan P, O’Shea T (2009) Broadening the focus of bat conservation and research in the USA for the 21st century. Endanger Species Res 8: 129–145.
  87. 87. Ford WM, Sheldon FO, Edwards JW, Rodrigue JL (2006) Robinia pseudoacacia (Black Locust) as day-roosts of male Myotis septentrionalis (Northern Bats) on the Fernow Experimental Forest, West Virginia. Northeast Nat 13: 15–24.