Fig 1.
Locations of ‘ōpe‘ape‘a (Lasiurus semotus) roost trees and stands identified May 2018–August 2021.
Fig 2.
‘Ōpe‘ape‘a (Lasiurus semotus) roost perch canopy cover by sex.
Cover classes used as maternity roosts are indicated with asterisks.
Table 1.
Statistics of predictor variables measured at the perches of roosts used by female (n = 19) and male (n = 36) ‘ōpe‘ape‘a (Lasiurus semotus).
Fig 3.
‘Ōpe‘ape‘a (Lasiurus semotus) roost perch height (relative to tree height) by sex.
Height classes used as maternity roosts are indicated with asterisks.
Fig 4.
Mean (arrow) aspect of roost perches (points) used by male and female ‘ōpe‘ape‘a (Lasiurus semotus), by season and by maternity groups.
Fig 5.
Roost tree species used by ‘ōpe‘ape‘a (Lasiurus semotus) by sex.
‘Ōhi‘a (Metrosideros polymorpha) and lama (Diospyros sandwicensis) are endemic tree species, and uluhe (Dicranopteris linearis) is an indigenous matted fern. All other species listed are non-native. Tree species observed used as maternity roosts are indicated with asterisks.
Fig 6.
Height classes of roost trees used by ‘ōpe‘ape‘a (Lasiurus semotus) by sex.
Height classes of trees used as maternity roosts are indicated with asterisks.
Fig 7.
Diameter at breast height (DBH) of roost trees used by ‘ōpe‘ape‘a (Lasiurus semotus) by sex.
Diameter classes of trees used as maternity roosts are indicated with asterisks. We excluded a single roost in low-stature matted fern (uluhe or false staghorn fern, Dicranopteris linearis) as it did not allow for standard measures of tree physiognomy.
Table 2.
Descriptive statistics of predictor variables used in discrete choice models of tree-level roosts used by ‘ōpe‘ape‘a (Lasiurus semotus) (n = 63) and associated random trees (n = 182).
Categorical variables are described as the proportion (Prop) of observations for each of used and random tree groups.
Table 3.
Top-ranked tree-level roost selection models for ‘ōpe‘ape‘a (Lasiurus semotus).
See Table 2 for description of model variables and Table 4 for output of the top-ranked models. Only models with ΔAICc ≤ 4.0 are reported here (see S1 Appendix for full listing of all tree-level models). LogLik = loglikelihood; ΔAICc = difference of AICc between a model and the model with the smallest AICc (Akaike’s information criterion adjusted for small sample size); DF = degrees of freedom; Weight = relative model weight. Performance is the percentage of model validation tests that correctly predicted roost use.
Fig 8.
Relative probability of ‘ōpe‘ape‘a (Lasiurus semotus) roost selection for the top-ranked tree-level model given the observed range in tree height and tree diameter at breast height, canopy cover (closed and open), and whether roost trees were located adjacent to open areas (edge—present; edge—not present).
Table 4.
Coefficients, standard errors (SE), p-value, and confidence limits [lower—LCL 95%; upper—UCL 95%] of predictor variables in the four top-ranked (ΔAICc ≤ 4.0) tree-level discrete choice models (S1 Appendix) for ‘ōpe‘ape‘a (Lasiurus semotus).
Reference category for canopy cover was “closed,” and for edge was “not present.” See Methods and Table 2 for parameter descriptions.
Table 5.
Descriptive statistics of predictor variables used in discrete choice models of stand-level roosts used by ‘ōpe‘ape‘a (Lasiurus semotus) (n = 128) and associated random stands (n = 410).
Categorical variables are described as the proportion of observations in used and random stand groups.
Fig 9.
Eight of stands by sex used as roosts by ‘ōpe‘ape‘a (Lasiurus semotus).
Height classes observed used as maternity roosts are indicated with asterisks.
Fig 10.
Slope of stands by sex used as roosts by ‘ōpe‘ape‘a (Lasiurus semotus).
Slope classes observed used as maternity roosts are indicated with asterisks.
Table 6.
Top-ranked stand-level roost selection models for ‘ōpe‘ape‘a (Lasiurus semotus).
See Table 5 for description of model variables and Table 7 for output of the top-ranked models. Only models with ΔAICc ≤ 4.0 are reported here (see S1 Appendix for full listing of all stand-level models). LogLik = loglikelihood; ΔAICc = difference of AICc between a model and the model with the smallest AICc (Akaike’s information criterion adjusted for small sample size); DF = degrees of freedom; Weight = relative model weight. Performance is the percentage of model validation tests that correctly predicted roost use.
Fig 11.
Relative probability of roost selection by ‘ōpe‘ape‘a (Lasiurus semotus) predicted by the top-ranked stand-level model given the observed range in canopy height, slope (degrees), and canopy cover (closed, open and scattered).
Table 7.
Coefficients, standard errors (SE), p-value, and confidence limits [lower—LCL 95%; upper—UCL 95%] of the top-ranked (ΔAICc ≤ 4.0) stand-level discrete choice models (S1 Appendix) for ‘ōpe‘ape‘a (Lasiurus semotus).
Reference category for canopy cover is “closed.” See Table 5 for parameter descriptions.