3.3 Alpha Diversity (α)
The mean number of species per plot was 6.70 (+ 0.02 1SE) before and 5.36 (+ 0.03 1SE) after the fire. The Shannon-Weiner, Simpson, and Unbiased Simpson indices did not change from pre- to post-fire, and temporal change for these metrics was not associated with vegetation type, fire severity, or topographic position (P >0.05; results not shown). Figures 2-6 and Table 1 display the results for species richness, evenness, and the inverse Simpson index, all of which exhibited significant patterns across space and time.
Mean species richness per plot decreased significantly from the first to the second sample period for all plots combined (Figure 2) and for each vegetation type separately (Figure 3; P >0.05). If this temporal change were caused by the Horseshoe Two Fire, we would expect shifts only for burned plots—a prediction supported by the results, with mean species richness decreasing significantly at low and high severity and marginally insignificantly at moderate severity (Figure 4). Richness exhibited a hump-shaped relationship with dNBR, but this was, unexpectedly, true for both the pre-fire (i.e., before any fire at all) and the post-fire data, with no significant interaction with timestep (Table 1, Figure 5). The change in the number of species for each plot over time (pre-fire minus post-fire) was not significantly related to dNBR for all plots combined (P>0.10 data not shown) or for any of the vegetation types analyzed separately, although a marginally insignificant linear decline was found for piñon woodlands (Figure 6). Species richness was significantly shaped by elevation (humped-shaped polynomial) and TRI (curvilinear increase) but not by TRMI (Table 1, Figure 5)—a relationship that was true for both pre-fire and post-fire, with no interaction of timestep and topography.
Mean evenness increased from the first to the second sample period for all plots combined (Figure 3). This increase occurred for pine-oak forest and piñon woodland but not for juniper woodland (Figure 4). Evenness increased for all fire severity levels, including unburned plots, although the relationship was marginally insignificant for moderate severity (Figure 5). The relationship of evenness to dNBR changed significantly over the two sample periods, exhibiting a hump-shaped relationship for pre-fire and a curvilinear increase for post-fire (Table 1, Figure 6). For both sample periods, evenness was significantly shaped in a curvilinear fashion by elevation and TRI but not by TRMI (Table 1, Figure 6).
Mean inverse Simpson index increased from the first to the second sample period for all plots combined (Figure 3). This increase occurred for pine-oak forest and juniper woodland but not for piñon woodland (Figure 4). This index increased significantly only for high severity plots, although an increase for moderate severity was marginally insignificant (Figure 5). The Inverse Simpson index exhibited a hump-shaped relationship with dNBR, but this was unexpectedly true both for the pre-fire (i.e., before any fire at all) and the post-fire data, with no significant interaction with timestep (Table 1, Figure 6). The Simpson Index changed in a curvilinear manner with elevation (hump-shaped) and TRI (increasing) but was not significantly related to TRMI (Table 1, Figure 6).