Results
We directly assessed occupation by synanthropic free-tailed bats for 1,014 buildings (checked for signs, sighting and/or audio confirmation, or discussed occupation with owner) (Figure 1). Of these 1,014 buildings, 93 showed evidence of free-tailed bat occupation and were included in building- and landscape-level analyses of bat building use. We recorded the building and roof structure for an additional 95 buildings which could not be accessed directly (1,109 buildings total) (Figure 2). With evidence weighted to better reflect active roosts (Appendix S1), 84 out of the total 1,109 buildings surveyed were considered presently active (occupation category >=4) and included in landscape-level analyses of bat-human exposure risk. Lines of evidence for free-tailed and other bat occupation are given in Appendix S2.
Modern-build style and triangular roofing were identified as building-level attributes of synanthropic free-tailed bat occupancy (modern build style: coef= 1.36 ± 0.29, p<0.0001; triangular roofing: coef= 2.17 ± 0.37, p<0.0001). At the landscape level, building availability and the proportion that were of modern-build – but not the proportion with triangular roofing – drove the number of buildings occupied by free-tailed bats (total number of buildings: coef= 0.007 ± 0.002, p=0.0008; proportion of modern-builds: coef= 0.012 ± 0.006, p=0.030; proportion with triangular roofing: coef= -0.004 ± 0.007, p= 0.540).
This positive association remained when the more stringent assessment of building occupancy was applied to capture roosts very likely to be actively occupied (occupation category >=4) (total number of buildings: coef= 0.005 ± 0.002, p=0.001; proportion of modern-builds: coef= 0.015 ± 0.006, p=0.017; proportion with triangular roofing: coef= -0.005 ± 0.008, p= 0.521, Figure 3). In both datasets, the effect of building number was less than half of the effect from the availability of modern buildings.
The average building-roost density of roosts considered to be actively occupied, was 8.4 occupied buildings per km2 (range 2-24 per km2), estimated using a traditional measure of density, or 8.2 occupied buildings per km2 (range 2-21.7 per km2) estimated as a kernel density (Figure 4).
Given specific interest in the transition from traditional- to modern-style housing, all analyses were repeated on a subset of data that included traditional- and modern-style houses only, and houses with triangle and flat roofing only (i.e., with structures including latrines, livestock coupes, and greenhouses removed, leaving 806 buildings total). Model results were not substantially or directionally different in the sensitivity analysis (Appendix S3), though proportionally more houses were occupied (one-in-seven with evidence of occupation, 13.1%, and with nearly one-in-ten being active roosts, 10.4%).