Habitat heterogeneity drives the geographical distribution of beta diversity: the case of New Zealand stream invertebrates

Abstract

To define whether the beta diversity of stream invertebrate communities in New Zealand exhibits geographical variation unexplained by variation in gamma diversity and, if so, what mechanisms (productivity, habitat heterogeneity, dispersal limitation, disturbance) best explain the observed broad-scale beta diversity patterns. We sampled 120 streams across eight regions (stream catchments), spanning a north-south gradient of 12° of latitude, and calculated beta diversity (with both species richness and abundance data) for each region. We explored through a null model if beta diversity deviates from the expectation of stochastic assembly processes and whether the magnitude of the deviation varies geographically. We then performed multimodel inference analysis on the key environmental drivers of beta diversity, using Akaike's information criterion and model and predictor weights to select the best model(s) explaining beta diversity. Beta diversity was, unexpectedly, highest in the South Island. The null model analysis revealed that beta diversity was greater than expected by chance in all eight regions, but the magnitude of beta deviation was higher in the South Island, suggesting differences in environmental filtering and/or dispersal limitation between North and South Island. Habitat heterogeneity was the predominant driver of beta diversity of stream macroinvertebrates, with productivity having a secondary, and negative, contribution. This is one of the first studies accounting for stochastic effects while examining the ecological drivers of beta diversity. Our results suggest that local environmental heterogeneity may be the strongest determinant of beta diversity of stream invertebrates, more so than regional- or landscape-scale variables.

Keywords: Beta diversity; habitat heterogeneity; macroinvertebrates; null models; productivity; streams.

Figure 1

Gamma (γ), alpha (α), and beta (β_w) diversity of stream macroinvertebrates across eight regions (catchments) in New Zealand, separately for species-richness-based data (A–C) and for effective diversity (exponential of Shannon diversity) (D–F). Regions are presented in north-to-south order (NL, Northland; UR, Urewera; EG, Egmont; TA, Tararua; KA, Kahurangi; AP, Arthur's Pass; WE, Westland; and FI, Fiordland). Red lines: North Island regions, blue lines: South Island regions. Plots represent total richness (or total effective diversity) across the 15 streams in a region (gamma diversity), median species richness (or median effective diversity) for a region and the regional-to-local diversity ratio for a region (β_w = γ/α).

Figure 2

(A) Patterns in observed and expected multiplicative β_w diversity of stream macroinvertebrate communities in eight regions of New Zealand, organized along a latitudinal gradient from lower to higher latitudes. (B) Patterns in β_w deviation across the eight regions. Red symbols: North Island regions, blue symbols: South Island regions. β_w deviation represents a standard effect size for the deviation of beta diversity from a null model that corrects for dependence of beta on gamma diversity (see text for further details). Positive values indicate higher β_w diversity than expected by chance.

Figure 3

Occupancy–frequency distributions of stream macroinvertebrate communities in the North and South Island of New Zealand. The proportion of sites occupied is divided into occupancy–frequency classes (i.e., the first class represents 0.2–5% of sites occupied, the second one represents 5–10% sites occupied).

Figure 4

Beta diversity–environment relationships for the New Zealand stream macroinvertebrate communities. Red symbols: North Island regions, blue symbols: South Island regions mean (±1 SE) distance of a site (n = 15 sites per region) to the respective group (region) centroid in relation to mean (±1 SE) for habitat heterogeneity (Euclidean distance) (y = 2.9 + 11.9x, P < 0.0001) (A), chlorophyll-a concentration (μg·cm⁻²) (B), bottom Pfankuch (stream disturbance index) (C) and topographic steepness (D).