Limits...
Exploring Regional Variation in Roost Selection by Bats: Evidence from a Meta-Analysis.

Fabianek F, Simard MA, Desrochers A - PLoS ONE (2015)

Bottom Line: We tested whether effect sizes from 34 studies were consistent enough to reject the hypothesis that trees selected by bats did not significantly differ in their characteristics from randomly selected trees.Size and direction of effects varied greatly among studies with respect to distance to water, tree density, slope, and bark remaining on trunks.Inclusion of mean summer temperature and sex in meta-regressions further explained heterogeneity in tree diameter effect sizes.

View Article: PubMed Central - PubMed

Affiliation: Centre d'Étude de la Forêt (CEF), and Faculté de Foresterie, de Géographie et de Géomatique, Université Laval, Québec, Quebec, Canada.

ABSTRACT

Background and aims: Tree diameter, tree height and canopy closure have been described by previous meta-analyses as being important characteristics in roost selection by cavity-roosting bats. However, size and direction of effects for these characteristics varied greatly among studies, also referred to as heterogeneity. Potential sources of heterogeneity have not been investigated in previous meta-analyses, which are explored by correlating additional covariates (moderator variables). We tested whether effect sizes from 34 studies were consistent enough to reject the hypothesis that trees selected by bats did not significantly differ in their characteristics from randomly selected trees. We also examined whether heterogeneity in tree diameter effect sizes was correlated to moderator variables such as sex, bat species, habitat type, elevation and mean summer temperature.

Methods: We used Hedges' g standardized mean difference as the effect size for the most common characteristics that were encountered in the literature. We estimated heterogeneity indices, potential publication bias, and spatial autocorrelation of our meta-data. We relied upon meta-regression and multi-model inference approaches to evaluate the effects of moderator variables on heterogeneity in tree diameter effect sizes.

Results: Tree diameter, tree height, snag density, elevation, and canopy closure were significant characteristics of roost selection by cavity-roosting bats. Size and direction of effects varied greatly among studies with respect to distance to water, tree density, slope, and bark remaining on trunks. Inclusion of mean summer temperature and sex in meta-regressions further explained heterogeneity in tree diameter effect sizes.

Conclusions: Regional differences in roost selection for tree diameter were related to mean summer temperature. Large diameter trees play a central role in roost selection by bats, especially in colder regions, where they are likely to provide a warm and stable microclimate for reproductive females. Records of summer temperature fluctuations inside and outside tree cavities that are used by bats should be included in future research.

No MeSH data available.


Related in: MedlinePlus

L’Abbé plots of the tree characteristics selected by bats (experimental groups) against the random tree characteristics (control group) with the 95% CI (black dashed lines) for each dataset, and for each characteristic (tree diameter, tree height, snag density, bark remaining on trunks, distance to water, canopy closure, elevation, slope, and stand density).The size of the circle varies according to the assigned random weight (inverse variance of the standardized mean differences) of each dataset. The diagonal (x = y) grey dotted line is the equality line (1:1) between both means (i.e., the zero effect line, for which the mean difference = 0). Above the x = y line, the experimental group mean is higher than the control group mean. Below the x = y line, the experimental group mean is lower than the control group mean. Tau-squared (τ2) and Higgins’ I2 heterogeneity indices are shown in each plot. Higgins' I2 index is expressed in percentage and is used to interpret the severity of heterogeneity.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4587962&req=5

pone.0139126.g004: L’Abbé plots of the tree characteristics selected by bats (experimental groups) against the random tree characteristics (control group) with the 95% CI (black dashed lines) for each dataset, and for each characteristic (tree diameter, tree height, snag density, bark remaining on trunks, distance to water, canopy closure, elevation, slope, and stand density).The size of the circle varies according to the assigned random weight (inverse variance of the standardized mean differences) of each dataset. The diagonal (x = y) grey dotted line is the equality line (1:1) between both means (i.e., the zero effect line, for which the mean difference = 0). Above the x = y line, the experimental group mean is higher than the control group mean. Below the x = y line, the experimental group mean is lower than the control group mean. Tau-squared (τ2) and Higgins’ I2 heterogeneity indices are shown in each plot. Higgins' I2 index is expressed in percentage and is used to interpret the severity of heterogeneity.

Mentions: Funnel plots were well balanced (Fig 3); therefore, asymmetry tests did not reveal any significant publication bias (Table 1). Higgins’ I2 heterogeneity index indicated considerable levels of heterogeneity (i.e., I2 indices ranging from 50% to 100%) for each characteristic of roost selection by bats (Table 1; Fig 4).


Exploring Regional Variation in Roost Selection by Bats: Evidence from a Meta-Analysis.

Fabianek F, Simard MA, Desrochers A - PLoS ONE (2015)

L’Abbé plots of the tree characteristics selected by bats (experimental groups) against the random tree characteristics (control group) with the 95% CI (black dashed lines) for each dataset, and for each characteristic (tree diameter, tree height, snag density, bark remaining on trunks, distance to water, canopy closure, elevation, slope, and stand density).The size of the circle varies according to the assigned random weight (inverse variance of the standardized mean differences) of each dataset. The diagonal (x = y) grey dotted line is the equality line (1:1) between both means (i.e., the zero effect line, for which the mean difference = 0). Above the x = y line, the experimental group mean is higher than the control group mean. Below the x = y line, the experimental group mean is lower than the control group mean. Tau-squared (τ2) and Higgins’ I2 heterogeneity indices are shown in each plot. Higgins' I2 index is expressed in percentage and is used to interpret the severity of heterogeneity.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4587962&req=5

pone.0139126.g004: L’Abbé plots of the tree characteristics selected by bats (experimental groups) against the random tree characteristics (control group) with the 95% CI (black dashed lines) for each dataset, and for each characteristic (tree diameter, tree height, snag density, bark remaining on trunks, distance to water, canopy closure, elevation, slope, and stand density).The size of the circle varies according to the assigned random weight (inverse variance of the standardized mean differences) of each dataset. The diagonal (x = y) grey dotted line is the equality line (1:1) between both means (i.e., the zero effect line, for which the mean difference = 0). Above the x = y line, the experimental group mean is higher than the control group mean. Below the x = y line, the experimental group mean is lower than the control group mean. Tau-squared (τ2) and Higgins’ I2 heterogeneity indices are shown in each plot. Higgins' I2 index is expressed in percentage and is used to interpret the severity of heterogeneity.
Mentions: Funnel plots were well balanced (Fig 3); therefore, asymmetry tests did not reveal any significant publication bias (Table 1). Higgins’ I2 heterogeneity index indicated considerable levels of heterogeneity (i.e., I2 indices ranging from 50% to 100%) for each characteristic of roost selection by bats (Table 1; Fig 4).

Bottom Line: We tested whether effect sizes from 34 studies were consistent enough to reject the hypothesis that trees selected by bats did not significantly differ in their characteristics from randomly selected trees.Size and direction of effects varied greatly among studies with respect to distance to water, tree density, slope, and bark remaining on trunks.Inclusion of mean summer temperature and sex in meta-regressions further explained heterogeneity in tree diameter effect sizes.

View Article: PubMed Central - PubMed

Affiliation: Centre d'Étude de la Forêt (CEF), and Faculté de Foresterie, de Géographie et de Géomatique, Université Laval, Québec, Quebec, Canada.

ABSTRACT

Background and aims: Tree diameter, tree height and canopy closure have been described by previous meta-analyses as being important characteristics in roost selection by cavity-roosting bats. However, size and direction of effects for these characteristics varied greatly among studies, also referred to as heterogeneity. Potential sources of heterogeneity have not been investigated in previous meta-analyses, which are explored by correlating additional covariates (moderator variables). We tested whether effect sizes from 34 studies were consistent enough to reject the hypothesis that trees selected by bats did not significantly differ in their characteristics from randomly selected trees. We also examined whether heterogeneity in tree diameter effect sizes was correlated to moderator variables such as sex, bat species, habitat type, elevation and mean summer temperature.

Methods: We used Hedges' g standardized mean difference as the effect size for the most common characteristics that were encountered in the literature. We estimated heterogeneity indices, potential publication bias, and spatial autocorrelation of our meta-data. We relied upon meta-regression and multi-model inference approaches to evaluate the effects of moderator variables on heterogeneity in tree diameter effect sizes.

Results: Tree diameter, tree height, snag density, elevation, and canopy closure were significant characteristics of roost selection by cavity-roosting bats. Size and direction of effects varied greatly among studies with respect to distance to water, tree density, slope, and bark remaining on trunks. Inclusion of mean summer temperature and sex in meta-regressions further explained heterogeneity in tree diameter effect sizes.

Conclusions: Regional differences in roost selection for tree diameter were related to mean summer temperature. Large diameter trees play a central role in roost selection by bats, especially in colder regions, where they are likely to provide a warm and stable microclimate for reproductive females. Records of summer temperature fluctuations inside and outside tree cavities that are used by bats should be included in future research.

No MeSH data available.


Related in: MedlinePlus