Using river discharge to model and deconstruct the latitudinal diversity gradient for fishes of the Western Hemisphere

作者：McGarvey Daniel J; Terra Bianca de Freitas

来源：Journal of Biogeography, 2016, 43(7): 1436-1449.

DOI：10.1111/jbi.12618

• fish richness
• multiple regression
• power-law model
• residual analysis
• simultaneous autoregressive model
• structural equation model
• subtropical rivers
• temperate rivers
• tropical rivers

AimTo model the species-discharge relationship (SDR) for native freshwater fishes of the Western Hemisphere, to test whether the SDR is itself a function of latitude, and to create a conceptual framework to integrate the SDR with the species-area relationship and the latitudinal diversity gradient.
LocationWestern Hemisphere rivers between 70 degrees N and 50 degrees S latitude.
MethodsDischarge and fish richness data were compiled for 107 rivers. Ordinary least squares and simultaneous autoregressive models were built for a hemispheric-scale SDR using one of three discharge (Q) measures: mean annual Q, annual low flow Q or annual high flow Q. Hemispheric-scale SDR residuals were used to test for distinct low, mid and high latitude groups and to deconstruct the SDR into separate models for each latitudinal group. Structural equation modelling (SEM) was then used to examine the combined effects of Q, river basin area, and latitude on fish richness.
ResultsHemispheric-scale SDRs were significant (P0.001) for each discharge measure. Model residuals showed that the SDR differs among tropical, subtropical and temperate realms. All deconstructed SDRs remained significant (P0.001) and minimized the residual effect of latitude. The tropical SDR was significantly steeper (non-overlapping 95% confidence intervals) than the subtropical and temperate SDRs. For both hemispheric-scale and deconstructed SDRs, annual low flow Q was the best individual predictor of native fish richness. SEM analysis showed that Q may be an integrative measure of climate and physical habitat effects.
Main conclusionsA single, hemispheric-scale SDR will underestimate fish richness in low latitude, tropical rivers and overestimate richness in high latitude, temperate rivers. Deconstructing the SDR into separate tropical, subtropical, and temperate models can account for much of this bias. Deconstructed SDRs also show that fish richness increases much more rapidly per-unit Q in tropical rivers than in subtropical or temperate rivers. Low flow Q is the strongest correlate of fish richness. And a preliminary, system-level model suggests that Q may play a central, integrative role in the regulation of freshwater diversity.

• 出版日期2016-7