Community structure along environmental gradients
Given differences in regional sub-faunas on the two islands we sought to understand the consequences of these sub-faunas for local diversity. To do so, we again examined alpha-diversity patterns on each island by elevation, but eliminated from analysis the five species on Hispaniola and single species on Jamaica that do not occur at sea level. With these highland species absent, abundance trajectories are no longer statistically indistinguishable between islands: total community size declines much more rapidly on Hispaniola than Jamaica (Island*Elevation interaction, P = 0.021;Island*Elevation2 interaction P = 0.025, R2 = 0.80; Comparison to model with no island effects, P = 0.009). Specifically Hispaniola shows a lizard deficit at middle and high elevations with respect to Jamaica, while the presence of some Jamaican lowland species that still occur in middle elevations means that eliminating highland species had relatively little effect on Jamaica’s lizard abundances (Fig. 4a).
In contrast, when high-elevation specialists are eliminated species richness patterns along the elevational gradient become equivalent, with the best-fit model including only a linear elevation term (Elevation term P < 0.001, marginal R2 = 0.88). Without the highland species Hispaniola loses its middle-elevation hump in local species richness, and lowland species disappear from communities at roughly the same rate with increasing elevation on both islands (Fig. 4b, All island effects non-significant:Island*Elevation2 interaction P = 0.51,Island*Elevation interaction P = 0.43, Island effect P = 0.13).