Plant diversity spatial layers for Australia

Brief description

These data are spatial layers of predicted patterns in plant community diversity across Australia, as presented and described in full in Mokany et al. (2022). The layers represent spatial predictions from models of plant community species richness and plant community compositional dissimilarity. The plant diversity models utilised data from 115,086 plant community survey plots across Australia, and environmental predictors from fine resolution (3 arc-second; ≈90m) spatially complete layers. The models were projected across Australia and the resulting layers provided here. The original spatial layers were also aggregated to courser spatial resolutions (0.0025° (≈250m) and 0.01° (≈1km)) which are also provided here, for applications that require smaller data files. Two additional spatial layers are also provided, synthesising the diversity model projections: (i) compositional uniqueness; (ii) diversity importance (both only provided at 0.01° (≈1km) resolution).

Lineage

Methods
The methods used to generate these data are described in full in Mokany et al. (2022). In summary, 115,083 plant community survey plots from the HAVPlot dataset (Mokany et al. 2022b) were selected and used as the basis for modelling native plant species richness and compositional dissimilarity.

Observed species richness values in each plot were scaled to a common area of 400 m2 using the species-area power model (S = cAz) with a scalar (z) value of 0.25 used for all plots. The scaled species richness was then modelled as a function of nine environmental predictor variables using Generalized Additive Modelling, with the model explaining 33.0 % of the deviance in species richness. The model was then projected across Australia at 3 arc-second resolution (≈90m) using spatially complete predictor layers. The predicted species richness values were aggregated to 0.0025° (≈250m) and 0.01° (≈1km) spatial grids, using the mean of the finer resolution grid cell values.

Observed compositional dissimilarities (Sorensen’s) between pairs of plots were scaled to a common area of 400 m2 using the species-area power model, with a scalar (z) value of 0.25 used for the species richness of each plot, and a scalar (z) value of 0.4 used for the number of shared species between two plots (Mokany et al . 2013). The scaled compositional dissimilarity was then modelled as a function of eight environmental predictor variables and geographic distance, using Generalized Dissimilarity Modelling (GDM), with the model explaining 32.7 % of the deviance in compositional dissimilarity (intercept = 1.3). The model was then projected across Australia at 3 arc-second resolution (≈90m) using spatially complete predictor layers. The GDM transformed spatial layers for each predictor were aggregated to 0.0025° (≈250m) and 0.01° (≈1km) spatial grids, using the mean of the finer resolution grid cell values.

The spatial predictions of species richness and compositional dissimilarity were used to derive two summary layers, provided here at 0.01° (≈1km) resolution. The first summary layer is the predicted compositional uniqueness, being the mean compositional dissimilarity of each grid cell to the rest of Australia (implemented using a random sample of 1 % of all grid cells) (Mokany et al. 2022). The second summary layer is an estimate of diversity importance, which combines both the species richness predictions and the compositional uniqueness predictions. To derive this layer, we normalised the compositional uniqueness values to a 0–1 range, we normalised the predicted species richness values to a 0–1 range, and took the mean of these two normalised values for each grid cell across Australia.

Data products
The spatial layers are provided in three separate folders, one for each of the spatial resolutions: ‘90m’ (3 arc-second, 0.000833°); ‘250m’ (9 arc-second, 0.0025°); ‘1km’ (0.01°).

Within each folder, a species richness prediction grid is provided (‘Richness…’), and ten GDM transformed predictor layers are provided (‘Dissimilarity_GDMtran…’) (Mokany et al. 2022c).

Withing the ‘1km’ folder, the compositional uniqueness layer (‘CompositionUniqueness_1km’) and the diversity importance layer (‘DiversityImportance_1km’) are also provided.

All spatial layers are in GDA94 geographic projection (EPSG:4283) and geotiff format, with no-data values set to -9999.

References
Mokany, K., et al. 2013. Scaling pairwise β-diversity and α-diversity with area. - Journal of Biogeography 40: 2299-2309.
Mokany, K., et al. 2022. Patterns and drivers of plant diversity across Australia. Ecography (in press)
Mokany, K., et al. 2022b. Harmonised Australian Vegetation Plot dataset (HAVPlot). - CSIRO Data Access Portal, https://doi.org/10.25919/5cex-4s70
Mokany, K., et al. 2022c. A working guide to harnessing generalized dissimilarity modelling for biodiversity analysis and conservation assessment. - Global Ecology and Biogeography 31: 802-821.