Bat activity following repeated prescribed fire in the central Appalachians, USA

Fire occurrence was widespread in the eastern United States pre-European settlement due to Native American ignition and lightning strikes (Nowacki 2008), resulting in a landscape considerably modified and maintained by fire. As a result of the suppression era beginning in the 1920s, the frequency and intensity of fire decreased through the 1960s in the central Appalachians of western Virginia and elsewhere in the mid-Atlantic Highlands, resulting in profound forest composition shifts that now favor fire-intolerant species (Abrams 1992; Yarnell 1998). Through “mesophication” (Nowacki 2008), shade-intolerant and fire-dependent species fail to regenerate and self-replace as competing shade-tolerant and fire-intolerant species begin capturing canopy or light gaps in the forest stand (Kreye et al. 2013). These alterations result in an alternative stable forest condition whereby shading promotes cool, damp microclimates and the production of non-flammable fuels. As this progresses, it becomes increasingly difficult to reverse, and forests may be locked into a steady mesophytic state wherein only shade-tolerant, fire-intolerant plant species occur (Nowacki 2008). As such, land managers are prioritizing prescribed fire as a tool for maintaining current and transitioning fire-dependent communities in the East, particularly in the southern and central Appalachian Mountains (USDA Forest Service [US department of Agriculture Forest Service] 2006; Hessl et al. 2011).

Throughout North America, including the central Appalachians, bats currently are of great conservation concern (Carter and Ford 2002; Perry 2012) due to the impacts of white-nose syndrome (WNS; Francl et al. 2012, Reynolds et al. 2016) and the proliferation of wind energy development (Arnett 2013). Of the suite of bats in the region, the endangered Indiana bat (Myotis sodalis [Miller and Allen, 1928]; MYSO) and the threatened northern long-eared bat (Myotis septentrionalis [Trouessart, 1897]; MYSE) are two species potentially impacted by prescribed fire (Austin et al. 2018b). Prescribed fire use may alter non-hibernating season day-roosts, (i.e., trees and snags) or change foraging habitat conditions (Carter and Ford 2002; Perry 2012; Ford et al. 2016b; Silvis and Perry 2016b). Because there has been limited work examining the relationships of these and other bat species in the central Appalachians relative to the return of fire as a prescriptive tool, land managers are often challenged to show that burning is not additive in negative impacts to these already stressed species (Johnson et al. 2010b; Ford et al. 2016a).

Research on the short-term effects of fire on bats suggests that bats display species-specific responses to fire-modified vegetation (Owen et al. 2004; Cox et al. 2016; Austin et al. 2018a), with larger-bodied bats benefiting from vegetation clutter reduction that simplifies flight (Norberg 1985; Aldridge 1988). Snags in burned areas may be of higher quality for day-roosts because the newly created canopy gaps allow for increased canopy light penetration that aids in thermoregulation and expedites juvenile bat development (Zahn 1999; Boyles 2006; Johnson et al. 2009). Bats may even be robust to some roost tree loss (Silvis and Ford 2015), as may result from fire. Additionally, newly created roosts may offset loss (Ford et al. 2016a) as remaining trees are likely more conducive to roosting due to fire-modified cavity, bark conditions, and canopy characteristics (Perry 2012). However, Reilly et al. (2016) found that repeated prescribed fire in the southeastern Piedmont changed composition and structure of trees in the midstory but not in the overstory, indicating that foliage-roosting species may be less affected by fire. Many insect taxa, including Lepidoptera, that serve as prey for bats, also benefit from prescribed fire and associated increases in nectar-producing plants (Rudolph 2000) as well as new growth that provides a substrate on which to lay eggs and feed larva (Rudolph 2000; Evans et al. 2013).

Although most bat and fire research in central Appalachians has focused on day-roost ecology (Johnson et al. 2009, 2010b; Ford et al. 2016a, 2016b), acoustics have been used to monitor fire effects on bat activity (Johnson 2012). Cox et al. (2016) found that bat activity in the Cumberland Plateau was higher after spring and fall prescribed burns in savannah conditions than after spring and fall prescribed burns in woodlands. In the upper Piedmont and western Allegheny Hills, respectively, Loeb (2008) and Silvis and Gehrt (2016a) found that thinning and burning treatments yielded higher total bat activity than control stands. However, there has been relatively limited research on the effect of repeated prescribed fire relative to bats as used for restoration and maintenance of fire-dependent communities in the central and southern Appalachians. In the Chicago, Illinois, metropolitan area, repeated prescribed fire had a positive effect on bat activity (Smith 2010); and in a longleaf pine (Pinus palustris Miller)−wiregrass (Aristida stricta Michaux) ecosystem in Florida, higher bat activity was associated with sites that experienced short fire-return intervals (Armitage 2012).

With the continued spread of WNS, the severe population declines of many cave-dwelling species of bats that use forests for summer day-roosting and foraging, and the endangered status of MYSO and the threatened status of MYSE as regulatory drivers modifying land management activities, understanding of how practices such as prescribed fire affect bats is critical. Regionally, in the central Appalachians, managers are challenged to show that burn programs for other natural resource purposes will benefit bats or, at minimum, will not exacerbate population reductions from WNS (Ford et al. 2016a). To investigate this, we compared bat activity in burned and unburned forest habitat and examined edge effects associated with burning in the central Appalachians of western Virginia. Previous research in the central Appalachians found strong species-specific responses to forest habitat and structural characteristics (Ford et al. 2005); thus, we predicted that bats would have distinct species-specific responses to repeated prescribed fire, as well as to the resulting fire-modified habitat and vegetation characteristics (Austin et al. 2018b).

We collected data on 707 detector nights at 149 sites total for up to nine nights at each site. Uneven sampling periods among sites occurred due to periodic detector failures or black bear (Ursus americanus [Pallas, 1780]) damage. When visually examining calls, we identified a systematic error whereby insect noise was classified both as LACI and LABO calls. To address this issue, we visually examined all calls identified by Kaleidoscope using AnalookW v. 3.9f (Titley Electronics, Ballina, New South Wales, Australia) and removed erroneously identified insect noise. After removing noise, we repeated call analysis in Kaleidoscope to obtain corrected nightly counts by species. Kaleidoscope identified 24 180 total call files, post visual noise removal, and assigned them to nine unique species: EPFU (10 039), LABO (2665), LACI (795), LANO (2836), MYLE (414), MYLU (3066), MYSE (1045), MYSO (3036), and PESU (284). For LANO, this represented an unusually high number of calls for the summer, as this species typically is caught only in May as it migrates through the state (Cryan 2003). However, post WNS, anecdotal reports of greater summer captures have been occurring (M. St. Germain, Virginia Tech Conservation Management Institute, Blacksburg, Virginia, USA, personal communication). Nonetheless, it was possible that many LANO calls were actually misclassified EPFU calls (Betts 1998; USFWS [US Fish and Wildlife Service] 2017); therefore, we combined all EPFU and LANO calls into one group and refered to them as EPFU/LANO. We also presented results for all high-frequency bats (Myotis spp. and PESU) combined to account for program identification uncertainty among Myotis species, and lastly, all bat species combined to examine fire and habitat effects on overall bat activity (O’Keefe et al. 2013, 2014).

Burn condition and elevation were the most important variables for explaining activity levels of EPFU/LANO, MYLU, MYSE, MYSO, high-frequency bats, and total activity and had a positive effect on all species (Table 1; Table 2); the confidence interval for model-averaged burned habitat overlapped zero, indicating a neutral or marginal effect of burned habitat on EPFU/LANO, LABO, MYLU, MYSO, and total bat activity. However, within burn condition, edge had a positive effect on LABO, MYLU, MYSE MYSO, and all high-frequency bats combined (Fig. 2; Table 2). Burn quantity was included in the set of competing models for EPFU/LANO; the effect was marginally positive with confidence intervals overlapping zero. Basal area was also included in the set of competing models for MYLU; the effect was marginally negative with confidence intervals overlapping zero. Six models describing LABO activity contained combinations of burn condition, canopy cover, burn quantity, elevation, and basal area were competing (Table 2). Model-averaged confidence intervals overlapped zero for burned habitat, burn quantity, elevation, canopy cover, and basal area. The elevation model and the canopy cover model were the top models for LACI with canopy cover having a negative effect on activity and elevation having a slightly positive effect on activity. Confidence intervals for both variables overlapped zero. There were seven competing models that described PESU activity, with the top model being the null, indicating little evidence to suggest fire had an effect on this species (Table 2). Accordingly, we did not model-average PESU results. Lastly, we were unable to analyze burn and habitat effects on MYLE due to model convergence error.

Species	Model(s)	df	logLik	QAICc	ΔQAICc	Weight
Eptesicus fuscus/ Lasionycteris noctivagans EPFU/LACI	burn condition + elevation	7	−487.19	148.00	0.00	0.24
	elevation	5	−507.47	149.00	1.01	0.15
	burn condition + burn quantity	7	−494.17	149.80	1.87	0.10
	null	4	−522.08	150.70	2.77	0.06
	global	19	−483.70	176.60	28.60	0.00
Lasiurus borealis LABO	burn condition + canopy cover	7	−178.37	150.30	0.00	0.18
	burn condition	6	−181.87	150.70	0.38	0.15
	canopy cover	5	−186.05	151.60	1.30	0.10
	burn condition + burn quantity	7	−180.37	151.80	1.49	0.09
	burn condition + elevation	7	−180.72	152.00	1.75	0.08
	burn condition + basal area	7	−180.79	152.10	1.80	0.07
	null	4	−190.90	153.00	2.75	0.05
	global	19	−174.02	176.60	26.28	0.00
Lasiurus cinereus LACI	canopy cover	5	−162.28	155.20	0.00	0.56
	elevation	5	−164.08	156.80	1.58	0.25
	global	19	−147.94	176.60	21.37	0.00
	null	4	−176.28	165.30	10.13	0.00
Myotis lucifugus MYLU	burn condition + elevation	7	−179.90	149.90	0.00	0.24
	burn condition	6	−183.09	150.00	0.12	0.22
	burn condition + basal area	7	−182.51	151.80	1.93	0.09
	null	4	−195.41	154.70	4.85	0.02
	global	19	−176.05	176.60	26.69	0.00
Myotis septentrionalis MYSE	burn condition + elevation	7	−102.24	158.50	0.00	0.93
	global	19	−93.96	176.60	18.07	0.00
	null	4	−118.95	175.00	16.51	0.00
Myotis sodalis MYSO	burn condition + elevation	7	−132.29	152.30	0.00	0.69
	null	4	−146.18	159.90	7.58	0.02
	global	19	−127.17	176.60	24.29	0.00
Perimyotis subflavus PESU*	null	4	−67.73	143.70	0.00	0.16
	burn condition	6	−65.98	144.60	0.81	0.10
	aspect	5	−67.07	144.60	0.82	0.10
	elevation	5	−67.45	145.30	1.57	0.07
	burn condition + aspect	7	−65.29	145.40	1.64	0.07
	burn condition + elevation	7	−65.33	145.50	1.71	0.07
	canopy cover	5	−67.62	145.70	1.91	0.06
	global	19	−62.47	168.80	25.10	0.00
High frequency	burn condition + elevation	7	−272.75	149.50	0.00	0.26
	burn condition	6	−278.25	150.00	0.44	0.21
	null	4	−297.53	154.90	5.43	0.02
	global	19	−267.62	176.60	27.04	0.00
Total activity	burn condition + elevation	7	−524.61	147.80	0.00	0.23
	elevation	5	−546.64	148.90	1.05	0.14
	burn condition	6	−540.12	149.40	1.63	0.10
	burn condition + burn quantity	7	−532.47	149.80	1.96	0.09
	null	4	−560.64	150.20	2.37	0.07
	global	19	−521.48	176.60	28.75	0.00

Species	Variable	Coefficient	SE	LCI	UCI
Eptesicus fuscus/ Lasionycteris noctivagans EPFU/LANO	intercept	0.65	0.41	−0.15	1.49
	burn condition: burn	0.57	0.45	−0.35	1.44
	burn condition: edge	2.10	1.35	−0.68	4.74
	elevation	0.70	0.38	−0.04	1.45
	burn quantity	0.18	0.37	−0.55	0.90
Lasiurus borealis LABO	intercept	−4.32	0.83	−5.95	−2.66
	burn condition: burn	1.40	0.90	−0.49	3.19
	burn condition: edge	3.78	1.57	0.17	7.03
	canopy cover	−0.36	0.48	−1.34	0.61
	burn quantity	0.10	0.31	−0.50	0.70
	elevation	0.06	0.20	−0.33	0.44
	basal area	−0.06	0.20	−0.44	0.33
Lasiurus cinereus LACI	intercept	−1.57	0.31	−2.19	−0.95
	canopy cover	−0.54	0.38	−1.28	0.21
Myotis lucifugus MYLU	intercept	−4.20	0.79	−5.76	−2.64
	burn condition: burn	1.23	0.74	−0.24	2.71
	burn condition: edge	4.83	0.97	2.91	6.74
	elevation	0.35	0.46	−0.54	1.24
	basal area	−0.06	0.19	−0.43	0.32
Myotis septentrionalis MYSE*	intercept	−5.27	1.03	−7.84	−3.59
	burn condition: burn	2.10	0.98	0.18	4.02
	burn condition: edge	4.99	1.05	3.16	7.49
	elevation	1.26	0.35	0.62	2.04
Myotis sodalis MYSO*	intercept	−5.17	1.04	−7.21	−3.13
	burn condition: burn	1.12	0.93	−0.70	2.94
	burn condition: edge	4.86	1.10	2.70	7.01
High-frequency bats	intercept	−2.30	0.48	−3.26	−1.35
	burn condition: burn	1.26	0.54	0.20	2.33
	burn condition: edge	4.18	0.67	2.86	5.51
	elevation	0.43	0.41	−0.39	1.22
Total activity	intercept	0.82	0.41	0.02	1.70
	burn condition: burn	0.70	0.46	−0.29	1.61
	burn condition: edge	2.45	1.28	−0.38	5.00
	elevation	0.58	0.42	−0.27	1.41
	burn quantity	0.13	0.32	−0.49	0.75

We found some evidence for fire effects on MYSE, MYSO, MYLU, EPFU/LANO, high-frequency bats combined, and total activity. For all species and species groups, activity was the highest in edge habitat between burned and unburned portions of the surveyed transects and slightly higher in burned than unburned habitat. EPFU/LANO, LABO, and total bat activity responded to multiple burns; at our study site,there had been up to three fires for some individual stands over a period of 8–12 years. As noted by earlier research, larger-bodied bats such as EPFU and LANO benefit from recent fires due to clutter reduction because it facilitates flight (Norberg 1985; Aldridge 1988; Brooks 2005). However, we observed some evidence that this was true for the smaller-bodied Myotis as well. Hutchinson and Sutherland (2005) found that ≥3 fires were required over the course of eight years to achieve statistically significant differences in basal area of midstory trees and reductions in basal sprouting. Similarly, this suggests that multiple prescribed burns would be required before long-term benefits to larger-bodied bat species are evident (Austin et al. 2018a). Previous research in the central Appalachians (Austin et al. 2018a), as well as in the Coastal Plain of South Carolina (Ford et al. 2006; Hein and Castleberry 2009) has found that all bat activity, regardless of body size, is higher in corridors where flight efficiency is high for bats; it seems likely that this also occurs on WSM in and along the roads serving as fire breaks. Similar to riparian corridors (Rogers et al. 2006), the creation of fire breaks may benefit bats by providing efficient travel pathways connecting roosting and foraging areas. These areas of concentrated activity may offer potential bat monitoring locations to assess fire effects in this area in the future.

Nonetheless, both MYLU and MYSO displayed only slight positive to neutral responses to prescribed fire. These species, while considered more clutter-adapted than LABO and less clutter-tolerant than MYSE (Broders and Findlay 2004; Brooks 2005; Broders et al. 2006), are likely tolerant of a wide range of forest conditions. However, it is widely accepted that higher acoustic activity indicates better habitat conditions (Johnson et al. 2010a; Coleman et al. 2014). For example, Brooks (2005) noted that MYLU was present across all habitat types in New England. Moreover, in bottomland hardwood forests of Illinois (Carter 2005) and in the agriculture landscape of central Ohio (Kniowski 2014), MYSO readily utilized bottomland habitat, an area likely unaffected by fire, for roosting or foraging.

The clutter-adapted MYSE, a gleaning species, displayed a positive response to prescribed fire. Immediate effects of a single fire, post suppression, include mortality of understory vegetation and small trees (although dependent on fire severity and slope position), and prolific basal sprouting shortly post fire (Elliott et al. 1999). In the absence of midstory clutter, basal sprouts may provide additional substrate for gleaning insects (Ratcliffe 2003). Indeed, Silvis and Gehrt (2016a); Silvis and Perry (2016b) found that Myotis were positively related to low-strata vegetative clutter. Several studies also have documented positive fire effects on MYSE roosting habitat (Johnson et al. 2009; Lacki et al. 2009; Ford et al. 2016a). In all burn conditions, activity had a slightly positive relationship with increased elevation. Because burn intensity is greater on upper slopes (Mladenoff 1999), reductions in clutter that improve foraging condition may have negated previously negative relationships with bat activity and increasing elevation in the central Appalachians at elevations where oak-dominated types are replaced by northern hardwood to the west on the Allegheny Plareau (Ford et al. 2005; Ford et al. 2016b).

Similar to our findings, Ford et al. (2005) also documented a positive relationship between LACI activity and minimum canopy gap width in the central Appalachians. Previous research has found that LACI use rapid, straight flight and low-attenuating, high amplitude echolocation to pursue prey in open habitats (Barclay 1985).

Our results show lack of support for connection between most burn and habitat variables and LABO activity with the exception of edge habitat between the burned and unburned stands. In the central Appalachians, LABO tend to be forest habitat generalists (Hutchinson 1999; Ford et al. 2005; Austin et al. 2018b). Ford et al. (2005) documented a positive trend with minimum canopy gap size for LABO. Similarly, LABO in our study area used the woods’ roads present at our edge sites between burned and unburned stands. Whether in burned forests or unburned forests, woods’ roads provide linear canopy openings or, at minimum, a less cluttered corridor when a covering canopy was present where LABO activity often is high (Estrada and Coates‐Estrada 2001; Ford et al. 2006; Hein and Castleberry 2009; Austin et al. 2018a).

In the central Appalachians, fire is used for conservation of fire-dependent, oak-dominant communities and to promote the montane pine systems, as well as to help land managers meet other stewardship goals (i.e., preservation of biodiversity and control of invasive plant species). Our research helps to elucidate effects of repeated prescribed fire in the central Appalachians in the summer. Overall, we found weak positive to neutral effects of fire on bats. Small sample sizes for species such as PESU and MYLE resulting from WNS-related population declines (Frick et al. 2010; Ford et al. 2011; Francl et al. 2012) further limit the inferences we can draw about fire effects on bats in this landscape. Further research is needed to determine whether these negligible impacts extend to critical fall swarm and spring emergence periods when habitat associations of bats are less well understood (Muthersbaugh 2018). Relative to most of the GWNF, the WSM landscape has a higher rate of prescribed fire both spatially and temporally. Indeed, despite prioritization of prescribed burning regionally (Brose 2014), burned land comprises a small percentage of public lands overall in the central Appalachians (Ford et al. 2016a). The bat activity response we observed at WSM suggests that bat presence should not serve as an impediment for burn programs on the central Appalachian landscape.