SEARCH

SEARCH BY CITATION

Over the last two decades, a number of studies in different locations have identified a positive correlation between human population size or density and the species richness of various taxonomic groups across space (e.g. Hunter & Yonzon, 1993; Chown et al., 2003; Gaston & Evans, 2004; Fjeldså & Burgess, 2008). This correlation is surprising given the undoubted negative impact of human development on the persistence of many species. At closer inspection, the spatial congruence between people and species richness is generally weaker when the size of the sampling unit (grain size) is small (Luck, 2007; Pautasso, 2007). For example, strong positive correlations are often reported for studies across broad extents (e.g. continents or countries) using a sampling grain of 0.5–1° grid cells, but these correlations tend to weaken or become negative when grain size is much smaller (e.g. 10 km2). The latter makes intuitive sense, as increasing urbanization can reduce species richness at local scales (McKinney, 2008), but the broad-scale positive correlation is nevertheless unexpected and warrants greater attention.

While some have speculated about what drives this broad-scale correlation, there have been few comprehensive tests of potential drivers (although see, e.g. Hugo & van Rensburg, 2008; Steck & Pautasso, 2008; Luck et al., 2010). Barbosa et al. (2010) make an important contribution to this research field in three ways. First, they use Carabid beetles as their focal taxonomic group. The human population density (HPD)–species richness correlation has generally been poorly studied among invertebrates. Second, they examine relationships at multiple spatial scales (see also Chown et al., 2003). Not surprisingly they record a positive correlation between HPD and Carabid richness across Italian regions (the largest sampling grain size), no correlation across provinces, and a negative correlation across 10 × 10 km grid cells. Finally, and most importantly, they examine how their results change when accounting for variation in sampling effort across survey sites.

As Barbosa et al. (2010) notes, variation in sampling effort has largely been ignored by researchers working on the HPD–species richness correlation, although it could reasonably be described as ‘the elephant in the room’. Indeed, the positive correlation between HPD and species richness could simply be an artefact of greater sampling in areas with more people. Sampling effort is often positively correlated with both HPD and species richness, confounding attempts to identify the existence of potential environmental drivers of the HPD–richness correlation. Others have raised this issue (e.g. Balmford et al., 2001; Luck et al., 2004), but generally not dealt with it in an explicit way. Yet, accounting for variation in sampling effort is a crucial first step in determining if the spatial congruence between human populations and species richness is driven by underlying environmental factors or simply a result of uneven sampling effort. If the former, the conservation implications are substantial. If the latter, we need to proceed more cautiously and attempt to increase sampling effort in areas of low HPD to get a better estimate of true species richness.

After accounting for variation in sampling effort and area, Barbosa et al. (2010) find that the positive relationship between Carabid species richness and HPD across regions largely disappears. However, and most surprisingly, they find that the negative correlation between HPD and richness across 10 × 10 km grid cells reverts to a positive correlation (although the partial r2 is very small). This result appears to be driven by widespread species, which show a positive association with people, whereas endemic species are negatively correlated with human density.

The results of Barbosa et al. (2010) at the largest sampling grain contrast with the findings of Luck et al. (2010), who examined the congruence between HPD and bird species richness across Australia at a grain size of 1° grid cells. Luck et al. (2010) found that the positive correlation between HPD and species richness remained even after controlling for variation in sampling effort. Moreover, they identified net primary productivity as a possible key driver of this spatial congruence, and found that the strength of the correlation between people and richness was mediated by the presence of land designated for conservation (i.e. the correlation between HPD and species richness was stronger in grid cells with a higher proportion of conservation land).

These contrasting results demonstrate that much greater attention should be assigned to the impact of sampling effort on the HPD–richness correlation. Moreover, the influence of variation in sampling effort on well-established biogeographic patterns warrants more study. Both Barbosa et al. (2010) and Luck et al. (2010) note that the presence or strength of environmental gradients in species richness (e.g. latitudinal gradients or relationships with primary productivity) may be influenced by interrelationships with sampling effort and HPD.

Future work should also examine variation in sampling completeness. For example, fewer samples in species poor areas may record the same proportion of the total species pool as more samples in species rich regions. Ultimately, we want the most accurate estimate possible of the total number of species in a given location, and calculating sample completeness (e.g. by using estimator procedures; Colwell & Coddington, 1994) indicates how close we have come to achieving this objective. We should also determine if variation in spatial patterns of HPD from one location to another influences the HPD–species richness correlation. For example, do we find different relationships in heavily populated countries (or those with a more evenly distributed population) versus those with lower HPD? The HPD–species richness correlation is becoming a well-established spatial pattern. Identifying the drivers of this pattern has ramifications for species conservation and our understanding of broader biogeographic relationships.

References

  1. Top of page
  2. References