Using codispersion analysis to quantify and understand spatial patterns in species–environment relationships

Abstract

SummaryThe analysis of spatial patterns in species-environment relationships can provide new insights into the niche requirements and potential co-occurrence of species, but species abundance and environmental data are routinely collected at different spatial scales. Here, we investigate the use of codispersion analysis to measure and assess the scale, directionality and significance of complex relationships between plants and their environment in large forest plots.We applied codispersion analysis to both simulated and field data on spatially located tree species basal area and environmental variables. The significance of the observed bivariate spatial associations between the basal area of key species and underlying environmental variables was tested using three null models.Codispersion analysis reliably detected directionality (anisotropy) in bivariate species-environment relationships and identified relevant scales of effects. Null model-based significance tests applied to codispersion analyses of forest plot data enabled us to infer the extent to which environmental conditions, tree sizes and/or tree spatial positions underpinned the observed basal area-environment relationships, or whether relationships were a result of other unmeasured factors.Codispersion analysis, combined with appropriate null models, can be used to infer hypothesized ecological processes from spatial patterns, allowing us to start disentangling the possible drivers of plant species-environment relationships.

Tree mycorrhizal associations mediate soil fertility effects on forest community structure in a temperate forest

“…Codispersion analysis has been used in previous studies to quantify the spatial patterns of species-environment relationships by measuring the spatial covariation of two or more spatially explicit datasets (Buckley et al, 2016). We used the paired datasets…”

“…The magnitude of the codispersion values and the context of the final codispersion maps could provide information about the strength and the directionality of covariation for each of the paired variables at increasing distances (see Buckley et al, 2016 for details). The R code for the codispersion analysis was acquired from the supporting information of Buckley et al (2016).…”

“…This scale-dependence may derive from the clumping of trees for two reasons. First, Wang et al (2015) suggested that dispersal limitation, which could influence the clumping of trees (Buckley et al, 2016), accounted for the observed community structure at scales below 150 m in the Changbaishan (CBS) forest dynamics plot; however, the TSM null model could not detect these clumping patterns. Second, after removing any spatial patterns, including clumping, using complete spatial randomness (CSRM) null models, we found that the observed codispersion values were significantly different from the expected values for the AM and EM trees at both the 20-m and 50-m scales (columns 'CSRMam' and 'CSRMem' in Figs 4,S3).…”

Foundation Species Across a Latitudinal Gradient in China

“…Identifying foundation species is difficult because it can take many years—often decades— to collect enough data to distinguish foundation species from other species that also are common, abundant, or dominant ( sensu Grime, 1987) but lack “foundational” characteristics (Baiser et al, 2013; Ellison, 2014, 2019). Rather than investigating one common or dominant species at a time in myriad ecosystems, Ellison and his colleagues have worked with data from individual and multiple large forest dynamics plots within the ForestGEO network 1 to develop statistical criteria that can suggest which tree species might merit further attention as candidate foundation species in forests (Buckley et al, 2016 a , b ; Case et al, 2016; Ellison et al, 2019). Specifically, Ellison et al (2019) proposed two statistical criteria for candidate foundation tree species based on their size-frequency and abundance-diameter distributions, and on their spatial effects of on the alpha diversity (as Hill numbers: Chao et al, 2014) and beta diversity (e.g., Bray-Curtis dissimilarity) of co-occurring species.…”

“…Indeed, we derived these two statistical criteria after more than a decade of observational and experimental studies of Tsuga canadensis -dominated forests in New England, USA that lend strong support for the hypothesis that T. canadensis is a foundation species (Orwig et al, 2013; Ellison, 2014). These criteria subsequently were applied to five additional forest dynamics plots in the western hemisphere (Buckley et al, 2016 b ; Ellison et al, 2019) with encouraging results. Here, we apply these criteria to 12 large forest dynamics plots in China that range from cold-temperate boreal forests to tropical rain forests.…”