A latent threat to biodiversity: consequences of small-scale heterogeneity loss
- First Online:
- Cite this article as:
- Hewitt, J., Thrush, S., Lohrer, A. et al. Biodivers Conserv (2010) 19: 1315. doi:10.1007/s10531-009-9763-7
- 347 Views
The threat of homogenisation to biodiversity is generally considered to occur at broad scales or in response to high-intensity impacts. Therefore, most biodiversity studies estimate local average or total species richness rather than local heterogeneity. Here we consider the potential for relative shifts between these different aspects of biodiversity at small spatial scales to be an early warning signal for biodiversity loss. In response to chronic, very low-level pollution, we observed a disjunctive response with gamma diversity (total species richness) and beta diversity (heterogeneity) decreasing while alpha diversity (average species richness) was still increasing. Homogenisation may, therefore, affect biodiversity through thresholds that alter the relationship between the average species richness and its heterogeneity, leading to the potential for regime shifts. Our stressor also had a strong negative effect on rare species, meaning that the purported importance of rare species as “insurance” in the face of environmental change may be overstated.
In response to the growing threat of biodiversity loss in nearly every ecological system (Pimm et al. 1995) many ecologists are seeking better understanding of, and improved ability to predict, the consequences of anthropogenic disturbances (Loreau et al. 2001; Naeem 2002). As biodiversity is inextricably linked to habitat diversity (Huston 1994; Guégan et al. 1998; Ricklefs and Lovette 1999; Hewitt et al. 2005; Thrush et al. 2006), habitat homogenisation is expected to be a major threat to biodiversity. Habitat homogenisation particularly endangers the rare, often habitat-specific, species that represent a large proportion of species richness (Gaston 1994). Homogenisation is generally considered to occur as a result of broad-scale, high-intensity impacts that remove habitats (e.g., conversion of forests to pasture) or alter them (e.g., sedimentation in estuaries resulting in extended mud flats). Therefore, most biodiversity studies do not focus on estimating local heterogeneity in diversity, instead only estimating average or total diversity.
Biodiversity is a multidimensional concept (Purvis and Hector 2000) and encompasses many scales of variation in biological organisation (from genes-to ecosystems). Despite the broad nature of this term, species diversity is one of the more commonly used attributes. Even the simplest of its measures, species richness, has three aspects, generally referred to as γ (total number of species), α (average number of species) and β (variability, heterogeneity or species turnover (Loreau 2000)). Although γ-diversity usually refers to a region and α-diversity to the scale of the sampling unit in any particular study, there is no explicit spatial scale for any of these variables. Different ecological theories also frequently emphasise different aspects of biodiversity at different scales, e.g., eutrophication and habitat homogenisation affecting γ-diversity (Velland et al. 2007), biodiversity-ecological functioning experiments focussing on α-diversity (Tilman et al. 1996; Emmerson et al. 2001; Solan et al. 2004; Hooper et al. 2005).
The relationship between stress (whether driven by physical disturbance or contaminants) and ecological responses are not necessarily monotonic, especially when stress is imposed unevenly across landscapes. At low levels of stress, asynchronous dynamics of individual patches should produce heterogeneity, with mosaics of patches of differing successional ages emerging, increasing overall species richness (Connell 1978; Huston 1979). This leads to the potential for relative shifts in average, total and heterogeneity of the number of species found at small spatial scales to be an early warning signal for biodiversity loss as a result of low levels of anthropogenic stress.
Here we investigate the relationships between α-, β- and γ-diversity along a relatively weak gradient of heavy metal contamination (copper, lead and zinc), where even the most contaminated site was well below levels assumed to have no effects (see methods). Macrofaunal communities found below the mean tide mark in intertidal sandflats were used as they are generally species rich and contain multiple trophic levels.
Intertidal sandy sites were selected along a known contamination gradient (concentration of heavy metals produced by urbanisation) within a large harbour (Waitemata, New Zealand). Previous studies at 95 sampling stations along this same contaminant gradient established strong community and single species responses to contaminants (Thrush et al. 2008b; Hewitt et al. 2009). The four sites selected for the current study occupied the lower half of this gradient (low contamination), and represented a single substrate type (fine sand). Positioning of the sites was carefully chosen to prevent correlation with other possible gradients (e.g., inundation times, wave exposure and distance to channels). The four sites varied from domination by a patchy mix of suspension- and deposit-feeding bivalves and large tube-dwelling polychaetes (the least contaminated), through domination by two bivalve species with patches of varying densities of these and shell hash, to domination by a mix of species, mainly deposit feeders and often small.
At each site, 48 macrofaunal cores (13 cm diam. × 15 cm deep) were collected along two parallel transects (80 m length, approximately 20 m apart). Core positions were randomly allocated every 3–4 m within a 1 m strip either side of the transect lines. Between transects, smaller cores of surface sediment (2.5 cm diam. × 2 cm deep) were collected approximately every 15 m for contaminant analysis and to confirm that sediment grain size at the sites was similar.
Macrofaunal cores were sieved across a 0.5 mm mesh screen prior to identification and enumeration to as low a taxonomic resolution as possible (predominantly species level). Species accumulation curves were derived for each site using the jacknife technique available in Estimate S (Colwell 2006). γ-diversity for each site was determined as that predicted to be achieved by the species accumulation curve at 100 samples. Note that 100 samples was chosen as the degree of separation between the curves had largely stabilised by this point. β diversity was defined as the difference between gamma diversity and average species richness (within-site heterogeneity Lande 1996; Crist et al. 2003; Gering et al. 2003; Klimek et al. 2008). This definition, rather than γ/α (Whittaker 1960; Legendre et al. 2005; Ricotta 2008), was used as the latter more accurately represents species turnover rather than within-site heterogeneity. Species evenness, Shannon–Weiner and Simpson’s Index were calculated using Primer E (Clarke and Gorley 2006). Rare species (those occurring in only 1 or 2 of the 40 replicates at a site) were identified at each site. Those species that were rare at the least contaminated site were examined at each of the other sites to determine whether they continued to be rare, disappeared, or became more common.
Total contamination index at each site together with measured sediment concentrations of copper, lead and zinc (mg kg−1)
Copper (mg kg−1)
Lead (mg kg−1)
Zinc (mg kg−1)
Interestingly, while there was a clear negative relationship for γ-diversity with stress, no such relationships were observed for the more typically used expressions of biodiversity in impact assessments e.g., average species richness (α-diversity), evenness, Shannon–Weiner H′ or Simpson’s diversity indices (Fig. 2). α-diversity exhibited a unimodal curve; initially increasing with increasing contamination then decreasing as contamination continued to increase (Fig. 2b). These different relationships with stress between γ- and α-diversity resulted in high β-diversity at less contaminated sites and a strong negative linear relationship with the contaminant gradient (Fig. 2a, Pearson’s R = −0.99, P < 0.0001). No relationships were obvious for evenness, Shannon–Weiner H′ or Simpson’s diversity indices (Fig. 2c).
Effect of the contaminant gradient on rare species
Not found at next site
Common at next site
Rare at next site
Studies on the response of biodiversity to stress/disturbance generally focus on species richness following a unimodal response to increasing stress (e.g., Connell 1978; Hacker and Gaines 1997), and do not incorporate effects on β-diversity. However, patch theory suggests that massive stress events should override patch dynamics by re-setting all patches to the same (early) successional state (Connell and Slatyer 1977; Pearson and Rosenberg 1978; Denslow 1980), while at lower levels of stress, asynchronous dynamics of individual patches should produce heterogeneity (Connell 1978; Huston 1979). While we observed indications of a unimodal response of average species richness, we did not observe an initial increase in β- and γ-diversity, both of which were decreasing while α-diversity was still increasing. α-diversity did not begin to decrease until the ratio of γ to β was greater than 1.46.
The implications of habitat heterogeneity and habitat selectivity to biodiversity preservation are accepted (Olson et al. 2001; Hoekstra et al. 2005) and the protection of habitats and connectivity between them is a focus of many conservation networks. However, Dornelas et al. (2006) recently demonstrated that similar habitats adjacent to each other can have markedly different communities, thereby decreasing the scale at which we should consider heterogeneity to be important to biodiversity. Similarly, our results suggest that, even within a soft-sediment habitat apparently homogeneous at the 100 m scale, enough heterogeneity in species distributions can exist such that homogenisation of these can still pose a threat to gamma diversity.
The interplay observed between α-, β- and γ-diversity has a major implication for resilience. Rare species have been implicated as providing insurance and functional resilience against change (Walker 1992; Lawton and Brown 1993; Naeem and Li 1997; Yachi and Loreau 1999). The percentage of rare species (those found in only one or two samples) we observed at our least contaminated site (42%) was similar to other figures reported for marine systems (Schlacher et al. 1998; Shin and Ellingsen 2004; Ellingsen et al. 2007). Moreover, our rare species appeared more sensitive to our stressor than many of the more common species as the percentage of rare species dropped to 18% at the most contaminated site, similar to findings by Hewitt et al. (2009), and most of the rare species found at the least contaminated site did not become more common over the contaminant gradient. This suggests that communities with a high proportion of rare species may prove less resilient, as indicated by early studies of species abundance distributions under stress (e.g., Whittaker 1975; Gray and Mirza 1979). In this case the stressor is likely to have an increasing effect, rapidly approaching a threshold, over which a regime shift occurs. As many rare species are large and often functionally important (Loreau et al. 2001; Thrush and Dayton 2002), their response to stressors is crucial for functional resilience. Our empirical results therefore give new impetus to the need to include the response of γ- and β-diversity and rare species into theoretical models predicting resilience and regime shifts and to empirical studies trying to understand the role of rare species in different systems.
The authors thank the New Zealand Foundation for Science and Technology (C01X0504) for provided funding for this work. Thanks also to David Hawksworth, who as editor encouraged us to revise the paper, and to 2 anonymous reviewers, whose suggestions improved the paper.