Species-specific traits plus stabilizing processes best explain coexistence in biodiverse fire-prone plant communities

Abstract

Coexistence in fire-prone Mediterranean-type shrublands has been explored in the past using both neutral and niche-based models. However, distinct differences between plant functional types (PFTs), such as fire-killed vs resprouting responses to fire, and the relative similarity of species within a PFT, suggest that coexistence models might benefit from combining both neutral and niche-based (stabilizing) approaches. We developed a multispecies metacommunity model where species are grouped into two PFTs (fire-killed vs resprouting) to investigate the roles of neutral and stabilizing processes on species richness and rank-abundance distributions. Our results show that species richness can be maintained in two ways: i) strictly neutral species within each PFT, or ii) species within PFTs differing in key demographic properties, provided that additional stabilizing processes, such as negative density regulation, also operate. However, only simulations including stabilizing processes resulted in structurally realistic rank-abundance distributions over plausible time scales. This result underscores the importance of including both key species traits and stabilizing (niche) processes in explaining species coexistence and community structure.

Figure 1. Basic model assumptions.

a) The model works at two spatial scales: fire spread is modelled at a 1-ha resolution, whereas neighbourhoods of four contiguous grid cells all containing suitable habitat comprise a patch containing one local community. b) Number of seedlings per adult b_i, vary between mature resprouters (crosses) and non-sprouters (circles). We varied non-sprouter reproductive rates across a wide range of plausible values (shaded grey area in Fig. 1b). c) Fire survival probabilities for resprouters change with plant age; variation in the fire survival probabilities are indicated by the grey shaded area in Fig. 1c.

Figure 2. Species richness and RAD.

Impact of non-sprouter number of seedlings per adult (the number of non-sprouter seedlings at age 9 β_i), and resprouter maximum fire survival probability p_surv,max on final species richness (after 10 000 time steps a,d,g) and RAD after 10 000 (b,e,h) and 100 000 years (c,f,i, circles show relative abundance of non-sprouters and triangles relative abundance of resprouters) for three selected scenarios: a,b,c) all demographic parameters are identical for species within a PFT and there is no density regulation (neutral), d,e,f) demographic parameters vary between species within a PFT and there is no density regulation(species differ), and g,h,i) demographic parameters vary between species and density regulation is operating (niche).

Figure 3. Simulated rank-abundance curves.

Simulated rank-abundance curves for two scenarios: neutral scenario, and niche scenario after 10 000 time steps (a, b) and 100 000 time steps (c, d). Circles represent non-sprouting species and triangles resprouting species. The simulation started with 132 non-sprouting and 132 resprouting species.

Figure 4. Sensitivity analysis.

Partial dependence plots for the four most important predictor variables for: a) overall species richness S, b) non-sprouter species richness S_NS, and c) resprouter species richness S_RE based on a boosted regression tree analysis of 2 010 random parameter samples (see Table 1 for details). The fitted value on the y-axis shows the effect of a given variable on the response after accounting for the average effects of all other variables in the model. The relative influence (%) of each predictor variable on the response is given in brackets in the legend of the x-axis.