Prioritising conservation areas using species surrogate measures: consistent with ecological theory?
1. Surrogate species measures of biodiversity (SSB) are used worldwide in conservation prioritisations. We address the important question whether the ideas behind SSB are consistent with current knowledge on distribution patterns of species, as reflected in theories of community assembly.
2. We investigated whether assumptions necessary for successful functioning of SSB (nested species assemblages, cross-taxon congruence, spatio-temporal consistency) were supported by predictions from either niche or neutral community models.
3. We found a general mismatch between ideas behind SSB and ecological community theory, except that SSB based on complementarity may be consistent with niche-based theory when gradients in species composition are strong.
4. Synthesis and applications. The lack of a necessary scientific foundation may explain the disappointing results of empirical tests of SSB. We argue that site selection should be based on costs and opportunities within complementary environmental/land units, rather than expensive inventories of unfounded surrogate species.
The numerous attempts to identify and test surrogate measures of biodiversity based on indicator species or taxa have mainly produced disappointing results of little consistency (e.g. Hess et al. 2006; Wolters, Bengtsson & Zaitsev 2006; Larsen, Bladt & Rahbeck 2009). Nevertheless, surrogate species are used in practical conservation work ranging from selection of nature reserves and woodland key habitats in Scandinavia to the use of ‘biodiversity intactness indices’ (e.g. Scholes & Biggs 2005). At small to medium scales, congruence in the distribution of different species groups is often low; hence, surrogate measures are ambiguous at these scales (e.g. Lindenmayer et al. 2002; van Weerd & de Haes 2010). Generally, surrogate measures tend to perform better at larger scales, or along strong gradients (e.g. Howard et al. 1998). However, the overall impression is that the success of surrogates varies unpredictably with scale (extent and grain size), region and taxa (e.g. Hess et al. 2006). Even when statistically significant correlations between taxonomic groups have been found, these relationships have not provided reliable predictions of biodiversity patterns between surrogate groups and overall biodiversity (e.g. Heino 2010). In this forum paper, we investigate whether the general lack of consistent positive results may be caused by a mismatch between the assumptions behind ideas of surrogate measures and fundamental patterns of species distributions reflected in ecological theory.
Conservation planning is the process of identifying spatially explicit priorities and actions for the conservation of biodiversity (Margules & Pressey 2000). As such, it should be based on sound ecological theory. Investing in surrogate species data is expensive and time-consuming.
If assumptions behind the use of such surrogate species data are contradicted by theory, it will fail to fulfil the goal of preserving biodiversity in an efficient way. If so, the costs of species surrogate inventories are not justified. Selecting sites based on inappropriate surrogates could give a false impression of scientific foundation. In the face of increasing habitat destruction, limited resources for conservation may then rather be devoted to conservation action (Cowling et al. 2010). Furthermore, a lack of theoretical support for surrogate species measures may indicate that conservation biologists should devote more time to studying other aspects of conservation planning, such as transdisciplinary processes (Reyers et al. 2010), environmental surrogates (Grantham et al. 2010) or more generally, the relationship between ecological theory and conservation (Clark 2009).
At present, there are two opposing groups of theoretical models in community ecology to explain the distribution and abundance patterns of species. The first group is the niche-based models (e.g. MacArthur & Levins 1967; May 1973; Silvertown 2004), which are based on the idea that different species have different niches (sensuHutchinson 1957). The niche describes the set of abiotic and biotic conditions where a species is able to persist. The driving force regulating community structure is competition and competitive adaptations, i.e. deterministic ecological differences between species. This implies that the distribution of species is mainly determined by the distribution of niches, i.e. environmental variation. The second group of models assumes that neutral mechanisms (stochastic processes of birth, death and dispersal) are regulating community structure. For a given habitat type and guild of species, neutral models imply that local species assemblages are determined by metacommunity abundance, random drift and dispersal limitations (e.g. Hubbell 2001). As pointed out by Clark (2009), conservation biologists seem to have ignored neutral models to explain biodiversity, possibly because it significantly reduces the chances of predicting the distribution and abundance of threatened species. Here, we discuss to what degree assumptions for the successful use of different surrogate species are supported by predictions from niche and neutral models, respectively.
Basic assumptions behind the use of surrogate measures for biodiversity
There are many types of surrogate species. These include, for example, indicators of environmental conditions, such as pollution or ecological processes (e.g. Lindenmayer, Margules & Botkin 2000). Here, we will focus discussion on the use of three types of surrogates of biological diversity: (i) focal species (Lambeck 1997), (ii) species richness indicators (e.g. Hess et al. 2006) and (iii) complementary species or taxa (Margules & Pressey 2000).
Different assumptions underlie different types of surrogate measures of biodiversity (Table 1):
Table 1. Four basic assumptions behind the use of surrogate species or taxa compared with the predictions from two types of community models
|Assumptions – surrogate species|
|1. Focal species||No||Yes||Yes||Yes|
|2. Species richness||No||Yes||No||Yes|
|Predictions from models|
- 1Focal species. This approach centres on species thought to indicate threatening processes by being be the most sensitive to such processes. An important assumption behind focal species indicators is that all species are thought to respond in varying degrees to the same threatening processes (Lindenmayer, Margules & Botkin 2000), and species assemblages therefore are nested. Furthermore, it is assumed that different taxa show congruence in species richness patterns, and there is spatial and temporal consistency in indicator ability.
- 2Species richness indicators. It is assumed that there is congruence in the distribution of species richness between indicator species group and target species groups. If so, the most species-rich sites (the hotspots) for the indicator group will also be hotspots for the target species groups (e.g. Hess et al. 2006). With this measure, there is no assumption of nestedness, but it is assumed that there is a spatio-temporal consistency in cross-taxon congruence. This implies that for instance, a correlation between species richness of birds and fungi in one area is also found in other areas or at other scales.
- 3Complementarity indicators. For the complementarity indicators, it is assumed that different species, and indeed higher taxa, show congruence in complementarity. In other words, that the complementary sites for the indicator species group also are complementary sites for the target species groups. As a result, a network of sites selected to maximise the representation of an indicator species group will successfully represent the target species groups as well (Margules & Pressey 2000). An important point about complementarity is that taxa may be uncorrelated, or even negatively correlated with respect to species richness, but at the same time show cross-taxon concordance in complementarity. Finally, there is an assumption of spatio-temporal consistency in cross-taxon congruence in complementarity.
Are the assumptions behind surrogate species supported by community theories?
Nested species assemblages
Focal species indicators are based on the assumption of nestedness (Lindenmayer, Margules & Botkin 2000; Lindenmayer et al. 2002). A nested species biota is one in which the species assemblages at depauperate sites are comprised of a subset of species found at successively richer sites. This assumption is generally not supported by ecological theory (Table 1). Niche theory predicts competitive exclusion, a process that counteracts nested species assemblages, and neutral models imply unpredictable variation in abundance in time and space. However, both niche and neutral models predict that nested species distributions may arise from dispersal limitation and variation in isolation (e.g. Greve et al. 2005). In fact, a certain degree of nestedness is frequently documented. However, a high level of nestedness is necessary for focal species to function, and this is rarely the case in real ecosystems (Fischer & Lindenmayer 2005).
Concordance in species richness
As discussed earlier, the niche-based models predict that different species respond differently to environmental conditions because of evolutionary adaptations. Moreover, clades of closely related species are thought to share some key evolutionary adaptations (e.g. Harvey & Pagel 1991). For example, a higher proportion of liverworts is generally associated with more humid conditions than vascular plants. Partly because of this, liverwort species richness generally peaks at higher latitudes and altitudes (Grau, Grytnes & Birks 2007) than vascular plants. Thus, as a result of patterns of shared ancestry, hotspots of selected indicator species, or indicator taxa, are not likely to be hotspots of biodiversity in general (Table 1).
Cross-taxon congruence in complementarity
The niche-based models assume that competition through the competitive exclusion principle will result in different species having different niches. As a result, the distribution of different species within a community is usually independent of each other (Gleason 1926; Hutchinson 1957). An important point with respect to the use of indicators of complementarity is that if species belonging to different taxa do not differentiate along the same ecological variation then concordance in complementarity is not to be expected (Table 1). Generally, different taxa differ in habitat specialisations and environmental adaptations. However, particularly in situations with strong and persistent gradients, different taxa may show a high degree of complementarity because of cross-taxon congruence in evolution and adaptation to very different environments (Pimm & Lawton 1998). Examples of such strong gradients may be latitudinal and altitudinal gradients (Warman et al. 2004) or a shift in biomes from tropical rainforests to deserts (Larsen & Rahbeck 2003).
Neutral models also assume that species are independent of each other in the sense that presence of one or more species cannot be an indicator of the presence of other species. In these models, species presence, absence and abundance are driven by dispersal, ecological drift, extinction and speciation (e.g. Hubbell 2001). The abundances of each species within communities fluctuate then because of random chance resulting in a dynamic random walk. It follows from this that species are not highly co-adapted or co-dependent, rendering consistent patterns in individual indicator species futile.
Spatio-temporal consistent patterns
For all three types of surrogate species, an important assumption is that there is a high degree of consistency of indicator patterns in time and space (Table 1). For a species, a taxon or a species list to properly function as an efficient surrogate, it should function at other places (within the same region and biotope type) and at other times than the time and place where it was tested (Lawton 1999; Sætersdal, Gjerde & Blom 2005; Hess et al. 2006). Again, both the niche-based models and the neutral models predict that abundance and occurrence of individual species differ from one place to another; niche-based models, because as one moves in space the environment changes, and so does the extent to which the niche requirements are met for the various species (Nekola & White 1999). The neutral models predict that the abundance of the species varies stochastically in space and time under restrictions of dispersal limitation, and therefore, unpredictable differences in communities arise from one place to another.
The use of surrogates of complementary species composition is not based on individual species but on the use of specific taxa, such as vascular plants or birds. The assumption is that if complementarity based on, for example, vascular plants is a good representation of complementarity in other taxa in boreal forests of northern Europe, then they will give a good representation of other taxa in other parts of the northern boreal forests too. There is no expectation of spatio-temporal consistency in individual species, but of consistency in distribution patterns of taxa along major gradients, such as soil fertility and humidity in boreal forests. As such, niche-based models support the use of surrogate taxa of complementarity.
To sum up the limitations of surrogate assumptions (Table 1), both focal species and surrogates of species richness seem to be based on assumptions that are not supported by niche-based models or neutral models. Surrogates of complementarity may be supported by niche-based models but mainly along strong gradients.
A number of studies have investigated the spatial congruence of species similarity and richness between taxonomic groups (e.g. Howard et al. 1998; Lawler & White 2008). Some studies found that species richness of certain taxonomic groups indicate overall species richness in the same geographic and spatial scale of analyses (e.g. Tognelli 2005). Some studies indicate the contrary (e.g. Howard et al. 1998; Heino et al. 2005), and others report mixed results (e.g. Larsen & Rahbeck 2003; Moore et al. 2003).
As for species richness indicators, the effectiveness of indicator taxa in a complementary selection seems to vary substantially (e.g. Howard et al. 1998; Lund & Rahbeck 2002; Moore et al. 2003; Juutinen & Mönkkönen 2004; Cabeza, Arponen & van Teefelen 2008; Larsen, Bladt & Rahbeck 2009). As a conclusion, the effectiveness of species richness indicators seems to vary unpredictably (Hess et al. 2006). It may be possible to find cross-taxon congruence in some situations and notably in situations where clear differences in communities along major gradients are involved. It is shown that in areas with a strong and persistent gradient, complementary selection may be relatively successful (e.g. Howard et al. 1998). Likewise, a hotspot selection may do reasonably well in regions with a strong gradient in species richness, for example at continental scales with clear north–south gradients in species richness in most taxa (Warman et al. 2004). However, as variation between sites becomes more obvious, for example between different forest types along a biogeographical gradient as in Howard et al. (1998), the need for surrogate species or taxa, to indicate this variation, decreases. Surrogate species are most urgently needed in situations where differences in species composition are not easily detected using cheaper surrogates such as broad habitat units.
The main problem with the use of surrogate species is that it involves assumptions about species-specific patterns of community composition. Furthermore, it requires that such species-specific patterns are similar in different taxa or groups of species. These are assumptions that many theoreticians would question (e.g. Lawton 1999; McGill 2010). Neutral models make no predictions about species-specific composition of communities. However, as shown in Table 1, even niche-based models may give little support in this regard. Recent work to reconcile niche and neutrality by Adler, HilleRisLambers & Levine (2007) using classical coexistence theory (Chesson 2000) argues that both niche and neutral processes simultaneously influence the dynamics of species communities. McGill (2010) summarises and compares six different unified theories of biodiversity and concludes that all have three assertions or rules in common. One of these rules is that individuals from different species can be considered as independent and placed without regard to other species, which clearly contradicts the assumptions behind the use of species surrogate measures of biodiversity.
Based on the arguments we present above, it is our assertion that neither neutral models nor niche models, or any reconciled version, are likely to support sufficient congruence in species distribution patterns. The only exception we found, according to niche models, is the greater the environmental differences between sites, the more likely congruent complementarity will be found between different taxonomic groups. Therefore, complementarity may still be useful at larger scales or strong gradients. We suggest that complementary strategies be utilised as a coarse filter approach (Noss 2004) to ensure that a broad representation of environments are included in the set of selected sites. This requires a general understanding of the main environmental gradients that affect species composition, and that sites are classified according to these gradients in a simple and manageable way (Faith & Walker 1994; Margules & Pressey 2000; Gjerde, Sætersdal & Blom 2007). The use of environmental surrogates has the potential of being particularly useful because continuous environmental data are more readily accessible at low cost, whereas biological surveys are not. The empirical evidence for the success of this approach is currently under debate (e.g. Faith, Ferrier & Walker 2004; Hortal, Araujo & Lobo 2009). However, methodological developments, such as those presented by Leathwick et al. (2010), may indicate that this is an interesting approach to pursue.
Ecological theory predicts that assumptions concerning congruence are not to be expected along weaker gradients. We therefore suggest that the selection of sites within each type or class should primarily be based on factors such as costs and opportunities of protecting different sites (Polasky, Camm & Garber-Yonts 2001; Knight & Cowling 2007; Pressey & Bottrill 2008; Cowling et al. 2010), instead of unsupported surrogate species measures.
We thank Tim Caro and two anonymous referees for valuable comments on an earlier version of this paper. The study was funded by the Norwegian Ministry of Food and Agriculture.