Covariation between intraspecific genetic diversity and species diversity within a plant functional group


  • Tianhua He,

    Corresponding author
    1. Centre for Ecosystem Diversity and Dynamics, Department of Environmental Biology, Curtin University of Technology, PO Box U1987, Perth, WA 6845, Australia;
    2. Botanic Gardens and Parks Authority, Kings Parks and Botanic Garden, West Perth, WA 6005, Australia;
      Correspondence author. E-mail:
    Search for more papers by this author
  • Byron B. Lamont,

    1. Centre for Ecosystem Diversity and Dynamics, Department of Environmental Biology, Curtin University of Technology, PO Box U1987, Perth, WA 6845, Australia;
    Search for more papers by this author
  • Siegfried L. Krauss,

    1. Botanic Gardens and Parks Authority, Kings Parks and Botanic Garden, West Perth, WA 6005, Australia;
    2. School of Plant Biology, University of Western Australia, Nedlands, WA 6009, Australia; and
    Search for more papers by this author
  • Neal J. Enright,

    1. Centre for Ecosystem Diversity and Dynamics, Department of Environmental Biology, Curtin University of Technology, PO Box U1987, Perth, WA 6845, Australia;
    2. School of Environmental Science, Murdoch University, Perth, WA 6150, Australia
    Search for more papers by this author
  • Ben P. Miller

    1. Centre for Ecosystem Diversity and Dynamics, Department of Environmental Biology, Curtin University of Technology, PO Box U1987, Perth, WA 6845, Australia;
    2. Botanic Gardens and Parks Authority, Kings Parks and Botanic Garden, West Perth, WA 6005, Australia;
    Search for more papers by this author

Correspondence author. E-mail:


  • 1Species diversity and genetic diversity are fundamental components of biodiversity. A primary goal of biodiversity studies is to explain the distribution of species and alleles in space and time. A new challenge is to cross discipline boundaries and explore the relationship between these two scales of diversity.
  • 2In the biodiverse northern sandplain shrublands of south-western Australia, the woody shrub Banksia attenuata occurs on patchily distributed sand dunes, and coexists with B. hookeriana, B. menziesii, and two small tree species, Eucalyptus todtiana and Xylomelum angustifolium, which together comprise a plant functional group of large shrubs/small trees.
  • 3Genetic variation (alleles per locus and heterozygosity) between 736 B. attenuata individuals on 27 discrete dunes was assessed using 11 polymorphic microsatellite markers. For each dune, the physical properties of area and height, and species diversity (richness and evenness) of the functional group, were measured.
  • 4Synthesis. Genetic diversity of B. attenuata covaried positively with species diversity, which in turn covaried strongly with dune height rather than dune area. The positive relationship between species and genetic diversity demonstrated here supports the theory of common environmental, rather than neutral, processes acting directly or indirectly on both scales of diversity, and suggests the possibility of predicting one component of diversity on the basis of the other.


Genetic diversity and species diversity are two fundamental components of biodiversity. Antonovics (1976, 2003) suggests that the environmental and evolutionary forces accounting for species diversity and genetic diversity are similar. The processes assumed to shape community assemblages in niche-oriented models, and species coexisting via heterogeneous patches and migration, also have evolutionary analogues in natural selection and gene flow (Vellend 2003, 2005; Alonso et al. 2006). A major goal of studies on species diversity is to explain the observed patterns of species abundance in space and time, and across scales (Ricklefs & Schluter 1993; Chave 2004). Similarly, studies on population genetic diversity also try to explain the observed patterns of allelic abundance in space and time, and across scales (Alonso et al. 2006). Research synthesizing the disciplines of community ecology and population genetics provides a unifying perspective at the levels of genes and species, helping to advance our understanding of the nature of biodiversity (Antonovics 1976; Bell 2001; Vellend 2005).

The idea that the forces accounting for given levels of species diversity and genetic diversity are similar has existed for decades (Antonovics 1976). Vellend (2003, 2005) showed that the species genetic–diversity correlation (SGDC) was generally positive, and such a correlation was consistent with island biogeography theory, depending on size of the reference area, immigration rates and environmental heterogeneity. In addition, Hubbell (2001) proposed the neutral theory of community ecology that is largely analogous to the theory of neutral genetic variation. Consequently, Vellend & Geber (2005) suggested that the parallel action of neutral processes on the two scales of diversity is also a parsimonious explanation for positive species–genetic diversity relationships if such pattern could not be correlated with environmental and selection processes. While this notion has been discussed in recent theoretical papers, supporting empirical evidence which takes into account possible effects of environmental properties is limited. For instance, Booth & Grime (2003) found that Shannon–Weiner species diversity increased, although not significantly, with the number of genotypes in experimental communities. Cleary et al. (2006) revealed a strong correlation between species and allelic richness of butterfly communities across rainforest habitats, some of which were affected by fire. They suggested that environmental ‘regimes’, habitat discontinuities and dispersal limitations may have accounted for the shared patterns although they did not seek any associated environmental gradients. A positive correlation was also found between trans-specific genetic diversity per species for three isozyme-gene systems and plant species diversity in forest tree communities, without suggesting an explanation (Wehenkel et al. 2006). However, Karlin et al. (1984) derived a negative correlation between the two levels of diversity that they related to the effects of differences in elevation.

It is usually impossible to study all species living in an ecosystem simultaneously, and a common practice in both theoretical and empirical ecology is to examine complex natural communities in terms of functional groups. This practice presumes that all species within a group are more similar in their effects on population, community, or ecosystem processes than they are to the species from another group (e.g. Symstad 2000). This approach is particularly pertinent for research in species-rich systems such as that reported on here.

We assessed covariations between intraspecific (population genetic) diversity in a metapopulation of Banksia attenuata with species richness of the coincident putative community functional group to which B. attenuata belongs, and associated habitat variation in the biodiverse northern sandplain shrublands of the south-western Australian Floristic Region. This region exhibits high levels of species diversity at the point, community and landscape scales (Hopper & Gioia 2004), and is recognized as an international biodiversity hotspot (Myers et al. 2000), with an estimated 7380 plant species. Our objective was to test the hypothesis that parallel processes acting on species diversity and genetic diversity could result in positive covariation between these two measures. We first tested the hypothesis of a positive SGDC (Antonovics 1976; Vellend 2003). Then we tested the relative importance of dune height (a surrogate for water availability; Enright & Lamont 1992) and habitat area, which has been hypothesized as one of the most important driving forces for positive SGDC (Vellend 2003). Finally, we explored the possible causative relationships between habitat properties, species diversity and genetic diversity.


The study area of 3 × 4 km2 is described in He et al. (2004) and Calviño-Cancela et al. (2008). Mean annual rainfall is 504 mm, mostly falling in winter–spring (May–November), with mean temperatures in the warmest month (February) of 29 °C, and temperatures > 45 °C often reached in summer–autumn. The soil substrate in the study and surrounding area is an unconsolidated sand overlying silt-clay at depths of 0.5 to 10 m, forming an undulating landscape of sand dune ‘islands’ and intervening swales. Sand depth is the main determinant of the distribution of most species (Hnatiuk & Hopkins 1981; Lamont et al. 1989). All Banksia species in the study area are restricted to the dunes (i.e. absent from the swales), with populations connected through gene flow, via seeds and pollen (He et al. 2004).

We assessed the distribution and abundance of the five largest nonclonal, shrub–tree species that formed a deep rooted, large shrub–small tree, plant functional group in the study area: B. attenuata, B. hookeriana, B. menziesii and Xylomelum angustifolium (Proteaceae), and Eucalyptus todtiana (Myrtaceae), Individuals of B. menziesii, X. angustifolium and E. todtiana can achieve tree size (3–5 m tall) and emerge above the general shrub layer while B. attenuata, B. menziesii and B. hookeriana dominate the shrub layer (1–2 m). The study species vary in their responses to fire, flowering phenology, and fecundity (Enright & Lamont 1989, 1992) but they are all confined to deep sands, have the highest leaf area index and the deepest and widest root systems of species in the community, enabling them to access subsurface-stored water and thus transpire and grow throughout the summer–autumn drought (Lamont & Bergl 1991; Pate & Bell 1999). They have large winged seeds (except for E. todtiana), which are released from woody fruits after fire. Their seedlings grow fast and develop a long taproot early (Enright & Lamont 1992; Milberg & Lamont 1997; Schütz et al. 2002), so that throughout their life cycle these species are drought-avoiders rather than drought-tolerators. Many other species in the community are also serotinous, but generally are sub-shrubs, with smaller seeds held at lower heights and shallower root systems that cannot access sufficient soil water to keep transpiring over the summer (Lamont et al. 1993, Pate & Bell 1999; Enright et al. 2007). Eucalyptus todtiana, by far the largest species in the area, has a massive lignotuber, and has a major influence on the ecosystem where it occurs; a number of species are found only under its crown (N.J. Enright, unpublished data).

The entire area of each dune was searched for the occurrence of each of the functional group species. The number of plants for each of B. attenuata and B. hookeriana on each dune was determined for six randomly chosen 10 × 10 m plots, while for the other three (less abundant) species, five plots with a diameter of 100 m were surveyed to estimate densities on each dune. Populations on small dunes were determined by counting all individual plants of each target species directly. The number of E. todtiana patches was cross-checked on satellite images from Google Earth <>. The area of each dune was calculated using imagej <available from> after adapting the image of dune distribution from Google Earth. Consequently, the size of larger populations of each species was determined from the average plant density in the plots and dune area, since none of the five species appeared patchily distributed within dunes. Dune heights were calculated by interpolating the local elevations of adjacent swales and crests from a digital map with 2 m contour intervals (Department of Environment and Conservation, Government of Western Australia, unpublished map).

Measures of species diversity, based on the five species noted above, were species richness (Ns) and evenness (E). Ns is the number of the possible five species occurring on each dune (community), and E is calculated using a version of Simpson's diversity index as inline image, where fi is the relative frequency of species i. Both measures of species diversity were calculated using primer v5.0 (Primer-E Ltd. Plymouth, UK).

Genetic variation for populations of B. attenuata sampled from the same 27 dunes was measured using 11 microsatellite DNA primers. The detailed protocol for microsatellite genotyping is described in He et al. (2007). Two parameters, allelic richness (Na) and expected heterozygosity (He), which are analogous to Ns and E (Etienne 2005), were calculated by GenAlEx 6 (Peakall & Smouse 2006). To eliminate the effect of uneven sample size on the measurements of number of alleles per locus, a rarefaction procedure was implemented using hp-rare 1.0 (Kalionwski 2005). The rarefaction procedure re-samples individuals from populations with sample size larger than the minimum to calculate allelic richness expected if the smallest samples were taken from each population (Gotelli & Colwell 2001).

statistical analysis

Dune area and population size were log transformed to meet normality requirements. The SGDC was tested by correlating Ns with its analogous Na, and E with He, respectively, using general linear models. The total effects of habitat properties (dune height and dune area) were also individually correlated with the four diversity parameters of diversity (Ns, E, Na and He) using general linear models. Stepwise multiple regressions were carried out to determine the relative contribution of dune properties and one level of diversity to variance of the other level of diversity. Path analysis was carried out to determine the possible pathways of effects between environmental properties (dune area and dune height), species diversity (Ns and E) and genetic diversity (Na and He), and a best-fit reduced model was built to show the pathway of significant correlations. General linear regressions were conducted in statistica (StatSoft Inc., Tulsa, OK, USA) and path analysis was implemented in amos 7.0 (SPSS Inc., Chicago, IL). Statistical significance was taken at the level of P < 0.05.


field survey

For the 27 surveyed sand dunes containing B. attenuata, dune area ranged from 1.5 to 54.3 ha (averaging 10 ha), while dune height ranged from 1 to 8 m with an average of 3.4 m (Table 1). The number of the five target species present on each dune ranged from one (only B. attenuata) to five, and population size of each species also varied greatly (Table 1). The population sizes of all five species were significantly correlated with each other (all Ps < 0.001) and the population size of all five species was positively correlated with dune height and dune area (Ps < 0.01). Species diversity of this functional group varied from dune to dune, in terms of species richness (Ns) and evenness index (E).

Table 1.  Summary of demography and genetic diversity of the 27 studied populations of Banksia attenuata and four other large shrub–small tree species, and species diversity measures for the five species plant functional group per dune
  1. Population size of B. attenuata (BA), B. hookeriana (BH), B. menziesii (BM), Eucalyptus todtiana (ET) and Xylomelum angustifolium (XA); Da, dune area; Dh, dune height; S, sample size; Na, number of alleles per locus, rarefacted to the smallest sample size; He, expected heterozygosity; Ns, number of species; E, evenness.

BA0122.44356.10.778  2900500050014219050.558
BA0254.37365.90.73812 00010 000110083043050.585
BA0320.75355.90.761  33003090092017050.556
BA04 3.71315.60.725   2000001220.107
BA05 4.73335.70.710  210010040155050.167
BA06 8.44326.00.75528 00050001501199050.273
BA0710.05355.80.735  660040003001248050.516
BA08 1.52115.80.729    15000010.000
BA09 5.05346.10.751  1300700451786550.576
BA1010.65355.90.742  29003002512040.109
BA11 2.95316.30.763   4002005234050.547
BA12 8.55355.80.746  2700201003540.046
BA13 3.63276.10.759   1000001320.204
BA14 8.32345.90.752  180030306540.097
BA1514.16325.90.739  18004001555612550.556
BA16 1.93135.80.716    400202530.504
BA1715.53306.00.738  170020002027619550.601
BA18 7.93326.10.761   9007901022810050.636
BA1918.04325.70.726  200018001082532550.674
BA2015.88346.20.763  470080015090517050.478
BA2112.82255.80.736  16000004020.048
BA22 7.01225.70.748   7400003020.075
BA23 2.32205.70.745   10000163030.477
BA24 2.21205.70.720  10000001520.227
BA25 8.31105.20.695    12000320.320
BA26 3.91155.70.725    400002020.444
BA27 1.61 74.50.581    11000010.000

correlations between genetic diversity, species diversity and dune dimensions

Genetic diversity in populations of B. attenuata was measured by microsatellite markers. The number of alleles per locus (Na) per population after rarefaction ranged from 4.5 (BA27) to 6.3 (BA20), with an average of 5.8, while the mean expected heterozygosity (He) varied from 0.581 (BA27) to 0.778 in BA01 (Table 1). Na was correlated with dune height (R2 = 0.310, P = 0.002), but was unaffected by dune area (R2 = 0.113, P = 0.085), while He was correlated with both dune area (R2 = 0.175, P = 0.029) and dune height (R2 = 0.193, P = 0.022). Stepwise multiple regression showed that Na was a significant function of Ns but not of dune height or area.

Measures of genetic diversity for B. attenuata (Na and He) were positively correlated with functional group species diversity (Ns and E) (Fig. 1). Stepwise multiple regression showed that Ns was a significant function of dune height, but not of area or Na. Path analysis revealed a network effect between dune dimensions (dune height, dune area), species diversity (Ns and E) and genetic diversity (Na and He) (Fig. 2). Dune height had greater total effects on species diversity (Ns and E) and genetic diversity (Na and He) than dune area (Table 2), suggesting that dune height is driving the positive SGDC. Dune height, rather than dune area, determined variation in Ns, while the effect of dune height on Na was largely via its effect on Ns: the direct effect of dune height on Na was 0.28 and non-significant, while the indirect effect (via Ns) was 0.50 and significant. E was largely a function of Ns and He essentially a function of Na (Fig. 2), and both responded to dune dimensions independently (Table 2).

Figure 1.

Plots of the relationship between functional group species diversity (species richness Ns and evenness E) and genetic diversity of B. attenuata (rarefied allelic richness Na and heterozygosity He) (all P < 0.05).

Figure 2.

Diagram showing the pathway of significant effects (P < 0.05) of dune dimensions on species diversity and genetic diversity. Numbers show standardized direct effects calculated from a reduced path model.

Table 2.  Standardized effects of dune height and dune area on species diversity and genetic diversity calculated from the saturated path model
  • *

    P < 0.05.

Dune heightTotal effects0.78* 0.50* 0.58* 0.44*
Direct effect0.61*–0.05 0.28–0.29*
Dune areaTotal effects0.28* 0.19*–0.01 0.25*
Direct effect0.28*–0.01–0.15 0.26*


covariation between species diversity and genetic diversity

We demonstrate a positive covariation between species diversity of the large shrub–small tree functional group and the genetic diversity of its most consistent member, B. attenuata. Further, this covariation was shown to be driven by common habitat properties. These results provide rarely tested empirical support for the hypothesis that species diversity within communities and genetic diversity within associated populations in widely varying but common habitats covary positively (Cleary et al. 2006; Wehenkel et al. 2006). Even rarer are studies that also attempt to determine an environmental explanation for any covariation (Karlin et al. 1984; Vellend 2003, 2005; Vellend & Geber 2005).

Unlike Vellend (2003) and subsequent papers that built on island biogeography theory, our results of possible environmental causes do not give a central place to ‘island’ (dune) area. Instead, correlations between species diversity of the plant functional group and genetic diversity of B. attenuata are the result of their common response to dune height, which is essentially an index of water availability. Dunes with deeper sands can store a greater volume of groundwater (Lamont et al. 1989; Enright & Lamont 1992), crucial for species survival during the hot, dry summers in this region.

Taller dunes are more likely to promote survival of all dune-restricted species, and therefore species diversity. Conversely, local extinction of some species, especially following unusually dry winters, decreases species diversity on the lower dunes, as shown by population extinction in B. hookeriana due to exceptional drought (T. He & B.B. Lamont, unpublished data). Lamont et al. (1989) demonstrated the death of high numbers of seedlings competing for water at drier sites independent of species composition, while on the sand dunes, density of seedlings was little affected by competition. Similarly, genetic diversity in B. attenuata populations is greater on the higher dunes because more individuals (total, and relative to seed supply), and thus more genotypes, can survive there. With the better growing conditions, individual plants are larger and more fecund, including greater percentage seed set, so that a greater range of genotypes can be present (Lamont et al. 1994, 2003).

Our study has also shown that species richness itself may have an important effect on the level of allelic richness in B. attenuata populations. Coexistence with other species may hold promise of better establishment and survival, and thus more alleles, of B. attenuata by providing shelter for seeds to avoid predation and for seedling establishment, for example, litter patches after wildfire (Lamont et al. 1993). Moreover, the coexistence of populations of functionally related species may be crucial for the maintenance of pollinators and wide outcrossing promoting genetic variation. Banksias in the study area flower out of phase and are pollinated mainly by nectar-feeding birds and small marsupials (Weins et al. 1979; Whelan & Burbidge 1980; Lamont et al. 2003). The presence of E. todtiana, one of the few tree species in the region, may also provide shelter and nesting places for these birds, while other Banksia species provide food for the pollinators when B. attenuata is not flowering.

Microsatellite DNA variation is generally assumed to have a neutral effect on phenotypes (Selkoe & Toonen 2006). The net biology of a species is considered equivalent in the neutral theory of community ecology (Hubbell 2001), which led Etienne & Olff (2004) to suggest that any positive correlation between genetic and species diversity may be interpreted as ‘evidence of neutral processes’, and ‘it provides a new test of the neutral model’. Vellend & Geber (2005) also suggested that the parallel action of neutral processes on the two levels of diversity is a likely explanation for positive species diversity–genetic diversity relationships. However, our analysis has revealed that covariation between these two components of diversity was driven by common habitat properties, rather than neutral processes. These results suggest that interactions between alleles and species within the shared physical environment are more important than neutral processes in this landscape.


Genetic diversity and species diversity are inseparable components of biodiversity. The present work revealed a positive covariation between genetic diversity within populations of B. attenuata and species diversity within a single plant functional group of which it is a member. These correlations are more likely to be the result of the parallel action of similar responses to common habitat conditions rather than the outcome of neutral processes. Establishing such a relationship provides the opportunity to predict one level of diversity on the basis of the other, with implications for rationalizing biodiversity conservation, monitoring and research efforts.


This is contribution CEDD21-2008 of the Centre for Ecosystem Diversity and Dynamics, Curtin University of Technology. We thank the Australian Research Council for funds (DP0556767 and DP0343511), Clinton van den Bergh, Mark Wallace and Katherine Baker for assistance in the field, and Matthew Williams for statistical advice. Roy Turkington and two referees provided constructive comments on an earlier version of this paper.