Implications of scale dependence for cross‐study syntheses of biodiversity differences

Abstract

Biodiversity studies are sensitive to well‐recognised temporal and spatial scale dependencies. Cross‐study syntheses may inflate these influences by collating studies that vary widely in the numbers and sizes of sampling plots. Here we evaluate sources of inaccuracy and imprecision in study‐level and cross‐study estimates of biodiversity differences, caused by within‐study grain and sample sizes, biodiversity measure, and choice of effect‐size metric. Samples from simulated communities of old‐growth and secondary forests demonstrated influences of all these parameters on the accuracy and precision of cross‐study effect sizes. In cross‐study synthesis by formal meta‐analysis, the metric of log response ratio applied to measures of species richness yielded better accuracy than the commonly used Hedges' g metric on species density, which dangerously combined higher precision with persistent bias. Full‐data analyses of the raw plot‐scale data using multilevel models were also susceptible to scale‐dependent bias. We demonstrate the challenge of detecting scale dependence in cross‐study synthesis, due to ubiquitous covariation between replication, variance and plot size. We propose solutions for diagnosing and minimising bias. We urge that empirical studies publish raw data to allow evaluation of covariation in cross‐study syntheses, and we recommend against using Hedges' g in biodiversity meta‐analyses.

Effects of planted tree species on biodiversity of conifer plantations in Japan: a systematic review and meta-analysis

“…fungi and bacteria (three individual studies only for cypress family plantations), mammals (four studies for cypress family plantations [one on the Japanese Macaque Macaca fuscata, three on the Japanese Hare Lepus brachyurus], three studies for pine family plantations [each study on bats, rodents, and the Sika Deer Cervus nippon, respectively]), soil animals (three studies for cypress family plantations, and three studies for pine family plantations), aquatic animals (three studies only for cypress family plantations). In addition, results of preliminary analyses suggested negligible differences among summary effects using different weighting methods, small effects of pseudo-replication, no serious bias derived from differences in plot size, and validity of using Hedges' g (Appendix S1.2, S3; Lajeunesse 2015; Spake and Doncaster 2017;Hamman et al 2018;Spake et al 2020). Moreover, we also examined whether summary effects differed among natural forest types, stand age classes of plantations, and those of natural forests.…”

Land-use change from food to energy: meta-analysis unravels effects of bioenergy on biodiversity and amenity

“…A log response ratio was calculated to represent the effect size: the natural log of the ratio of the mean value of the treatment (bioenergy) to the mean value of the control (arable or grassland). The log response was considered a more appropriate response metric than calculating the effect size using the standardised difference between group means (e.g Hedges’ g) because it does not use within-group variance in its calculation 37 . This is important since variance can vary notably between the studies owing to differences in study design such as geographic location, distribution, and taxonomic group 37 .…”

Novel Insights to Be Gained From Applying Metacommunity Theory to Long-Term, Spatially Replicated Biodiversity Data

“…In light of these challenges, research questions focused on understanding the underlying processes that structure metacommunity assembly (i.e., species interactions, environmental filtering, dispersal limitation) must be aware of heterogeneity in sampling effort and spatial grain (i.e., plot size) across studies, which biodiversity estimates and variability among samples are sensitive to [Chase and Knight (2013), Spake et al (2020)]. Spake et al (2020) suggest that in formal meta-analyses scale dependence in effect sizes may be assessed using meta-regressions exploring relationships between either spatial (i.e., plot size) or temporal (i.e., sampling interval) grain and effect sizes across studies. They also illustrate with simulated community data how effect sizes calculated with the log response ratio metric applied to biodiversity estimates (i.e., species richness) were more accurate than those calculated with the common Hedge's g metric.…”

“…Given the challenges of synthesizing biodiversity data with varying grains of sampling in space and time, we have two recommendations for monitoring programs. First, ensure that raw data are published with ample metadata, so that synthesis researchers can extract relevant information on grains of sampling (Spake et al, 2020). Second, we recommend that programs coordinate efforts to agree upon standardized sampling protocols for particular taxa to promote synthesis (see more specifics on such coordination in the Challenge Three subsection below).…”