Summary

  1. A linear discriminant analysis showed that the sites had a unique soil composition.

  2. Most species didn’t occur evenly across sites.

  3. Nonetheless, models of different spatial configuration predicted similarly well.

Site soil composition

A linear discriminant analysis based on soil variables showed that the plots clearly grouped by site. Site 1 had some overlap with sites 2 and 3, which makes sense as it was the middle plot.

Evenness of species occurrence across sites

Species tended to be widespread in only one or two sites. Possibly could test for evenness somehow. Chi squared?

Site 1 Site 2 Site 3 Total
R_burtoniae 9 10 2 21
R_comptonii 7 0 15 22
D_diversifolium 6 12 16 34
A_delaetii 21 16 18 55
A_fissum 2 1 9 12
A_framesii 8 4 0 12
C_spissum 24 13 26 63
C_staminodiosum 1 0 9 10
Dicrocaulon_sp 10 16 3 29
Oophytum_sp 1 11 2 14

Spatial structure

Created seven models with different spatial configurations. Red represents plots used in the models, blue shows plots used as test data. Model 7 is an average of 20 randomisations

AUCs

All models predicted almost equally well.

Species Mod1 Mod2 Mod3 Mod4 Mod5 Mod6 Mod7
R_burtoniae 0.78 0.89 0.88 1.00 0.88 0.94 0.92
R_comptonii 0.81 NA 0.95 0.96 0.96 0.92 0.91
D_diversifolium 0.81 0.92 0.76 0.83 0.79 0.91 0.83
A_delaetii 0.64 0.78 0.70 0.80 0.76 0.75 0.79
A_fissum 0.49 0.27 0.72 0.65 0.63 0.42 0.61
A_framesii NA 0.63 0.57 0.55 NA 0.84 0.76
C_spissum 0.62 0.66 0.69 0.67 0.66 0.86 0.69
C_staminodiosum 0.72 NA 1.00 0.66 0.72 0.93 0.79
Dicrocaulon_sp 0.73 0.85 0.59 0.72 0.77 0.79 0.85
Oophytum_sp 0.94 0.77 0.96 0.92 0.94 0.87 0.87

There didn’t particularly seem to be a strong pattern with sample size.