Skip to main content

Scale-dependence of ecological assembly rules: insights from empirical datasets and joint species distribution modelling

Cite this dataset

Mod, Heidi; Chevalier, Mathieu; Luoto, Miska; Guisan, Antoine (2020). Scale-dependence of ecological assembly rules: insights from empirical datasets and joint species distribution modelling [Dataset]. Dryad.


1. A comprehensive understanding of the scale-dependency of environmental filtering and biotic interactions influencing the local assembly of species is paramount to derive realistic forecasts of the future of biodiversity and efficiently manage ecological communities. A classical assumption is that environmental filters are more prevalent at larger scales with diminishing effects towards the finest scales where biotic interactions become more decisive. Recently, a refinement was proposed stipulating that the scale-dependency of biotic interactions should relate to the type of interaction. Specifically, the effect of negative interactions (e.g. competition) should diminish with coarsening scale, whereas positive interactions (i.e. facilitation) should be detected irrespective of the scale.

2. We use multiple vascular plant species datasets sampled at nested spatial scales (plot size varying from 0.04 to 64 m2) and recently developed joint species distribution models (JSDM) to test the hypotheses.

3. Our analyses indicate slightly stronger environmental filtering with increasing plot size. While the overall strength of biotic interactions did not vary consistently across scales, we found a tendency for negative interactions to fade away with increasing plot size slightly more than positive interactions. Synthesis: We provide partial, but not unambiguous, evidence of the scale-dependency of ecological assembly rules. However, our correlative methodology only allows us to interpret the findings as indication of environmental filtering and biotic interactions.


The datasets are collected from nested plots (sizes varying from 20 × 20 cm = 0.04 cm2 to 8 × 8 m = 64 m2) located in non-forested sites in two study areas. Each dataset comprises presence-absence observations of vascular plant species with prevalence of 0.05-0.95 at all plot sizes and spatially related information of the abiotic environment. Species were recorded as present even if only a part of aboveground vegetative growth was within the plot.

The Alpine datasets was collected in the western Swiss Alps (46°23’ N, 7°5’ E). AlpineM consists of 434 sites, each with four nested square plots of 1, 4, 16 and 64 m2. AlpineCM consists of 298 square sites of 4 m2, with additional five square subplots of 20 × 20 cm within each site The environmental variables (growing degree days, topographic position index, soil pH and solar radiation) for each site were derived from raster layers with a spatial resolution of 25 × 25 m.

The Arctic datasets were collected in mount Saana in northern Finland (69°2' N, 20°51' E). The plots are organized in 21 grids of 8 m × 20 m, each grid consisting of 160 plots of 1 m2. ArcticCM contains 12 grids with species sampled within plots of sizes 20 × 20, 40 × 40 and 100 × 100 cm. Under this design, plots of 20 and 40 cm resolutions are nested and randomly located inside the 100 cm-plots. ArcticM comprises all 21 grids, and 1 m2 is used as the smallest plot size, and larger plot sizes (4 and 16 m2) were formed by dividing the grids in 2 m × 2 m and 4 m × 4 m plots. Few plots of 1 and 4 m2 containing no species were removed. For arctic datasets, the abiotic variables (soil temperature, moisture and pH, annual solar radiation, and rock cover) were derived at a 1 × 1 m.

Usage notes

These datasets (partly or completely) have been used in many other publications from the groups ECOSPAT (University of Lausanne; led by Prof. Antoine Guisan) and BioGeoClimate Modelling Lab (University of Helsinki; led by Prof. Miska Luoto). Thus, subsets of sites and species of these datasets can already be found in DryAd. Sites can be connected based on siteID and species by their names (or abbreviations). 


Swiss National Science Foundation, Award: CR23I2_162754

Academy of Finland, Award: 1140873