THE INTERMEDIATE DISTURBANCE HYPOTHESIS: THE ROLE OF VOLTINISM AND DISPERSAL IN AQUATIC MACROINVERTEBRATES

ABSTRACT

Species diversity is predicted to peak at intermediate levels of disturbance according to the intermediate disturbance hypothesis (IDH). We used a data set about floodplain aquatic macroinvertebrates to test the IDH along a gradient of flood disturbance. We examined the underlying mechanisms responsible for the observed pattern using functional diversity related to the voltinism and dispersal strategies in the invertebrate communities. We found that the rarefied taxonomic richness decreased when flood disturbance increased, but unimodal pattern expected by the IDH was not observed.

Consistent with predictions of the IDH, proportions of individuals capable of quickly reaching maturity, increased with disturbance, whereas individuals with poor colonization ability dominated undisturbed environments. Flood disturbance limited macroinvertebrate dispersal and we observed an opposite pattern between aerial and aquatic dispersal modes. Rarefied taxonomic richness was maximum when functional groups were evenly represented, suggesting higher species diversity when competition and colonization or dispersal modes are equilibrated. We conclude that habitat modifications (especially those induced by river restoration) are likely to alter both the balance of functional groups and species diversity.

Keywords: IDH, hydrological connectivity, disturbance gradient, functional diversity, competition, colonization, dispersal, species richness, floodplain, Rhône River.

INTRODUCTION

According to the intermediate disturbance hypothesis (IDH), species diversity is expected to be maximum at intermediate levels of disturbance (Connell, 1978).

Competitive ability and ability to cope with disturbance were used by Connell (1978) as mechanisms explaining the peaked pattern. To exemplify his theory, Connell (1978) used two examples (aquatic and terrestrial) with only a few species for the aquatic example, whereas the terrestrial example encompassed a higher number of species. In both cases, disturbances created niches where communities can settle and reach an equilibrium between competition and colonization processes. Connell (1978) suggested that high disturbance levels would favor colonization, whereas competition would be maximal under low levels of disturbance. The fact that species can trade-off between competition and colonization seems crucial for explaining patterns of coexistence at large spatial scales (Cadotte, 2007; Kneitel and Chase, 2004). Moreover, other less-studied mechanisms involving dispersal ability may contribute to diversity maintenance given appropriate habitat heterogeneity, e.g. by facilitating habitat settlements (Cadotte and Fukami, 2005).

The IDH was mainly used as an hypothesis to explain the diversity-disturbance relationship, but the published results remain contrasted, with only few studies supporting the predictions (Mackey and Currie, 2001; Shea et al., 2004). According to these authors, species diversity peaked in less than 20% of the analysed studies. Concerning the mechanisms involved in this pattern, most studies on competition and colonization were carried out under laboratory conditions with controlled disturbances and a small number of species (Cadotte et al., 2006; Haddad et al., 2008; Jiang and Patel, 2008). Recently, using a large-scale tropical forest dataset, Bongers et al. (2009) found that the number of

pioneer species (colonizers) increased with disturbance, whereas shade-tolerant (competitors) decreased. Bongers et al. (2009) results supported the IDH as an explanation of species diversity. A large number of studies were conducted in tropical forests, by contrast, only a few studies used aquatic environments to test or exemplify the IDH on a large number of species (Townsend et al., 1997; Usseglio-Polatera, 1994).

(Bongers et al., 2009)

Large river floodplains provide appropriate environments for testing the role of disturbance in structuring the species diversity at different spatial scales (Ward and Tockner, 2001) as well as the mechanisms responsible for the maintenance of species diversity. Indeed, floods represent a disturbance, that largely control floodplain communities through their intensity and frequency (Lake, 2000). Hydrological connectivity, which refers to the transfer of energy, matter and organisms within and between components of the floodplain, is the main process structuring the spatial heterogeneity of floodplains (Amoros and Bornette, 2002). The lateral dimension of a riverine floodplain defines a gradient of lateral hydrological connectivity, from the main river channel to less connected channels (Paillex et al., 2007). This gradient is, in turn, linked to hydrological disturbance; high hydrological connectivity implies frequent disturbance by floods, while low connectivity implies the absence of, or reduced disturbance. In floodplain systems, aquatic macroinvertebrates represent a taxonomically and functionally diversified group (Robinson et al., 2002). They can colonize the entire range of habitat conditions, from running waters, to stagnant and temporary water-bodies and can adopt different strategies to maintain in all types of habitats (Townsend and Hildrew, 1994; Verberk et al., 2008a; Verberk et al., 2008b). Some species tend to be first colonizers in disturbed environment using strategies such as a quick reproduction (e.g.

plurivoltine species), others in crowed habitat tend to be strong competitors with slowly reproducing animals (e.g. semi and univoltine species) (Macarthur and Wilson, 1963;

Townsend, 1989). Independently from this concept, dispersal is important also for establishing a population (Jakobsson and Eriksson, 2003; Yu et al., 2001) and as Cadotte

(2006) underlined, more research is required to address spatial heterogeneity and ask where immigrants come from. Aquatic macroinvertebrates exhibit several mechanisms of dispersal such as adult flight, larval drift or passive dispersion by a vector (Brittain and Eikeland, 1988; Robinson et al., 2002). In this context, species richness will depend on the capacity of species to recover from a disturbance through their dispersal modes.

Spatio-temporal variability of floodplain channels has been shown to shape macroinvertebrate life-history traits (Townsend and Hildrew, 1994) and diversity patterns (Tockner et al., 1999b), so we sought to relate species diversity with macroinvertebrate strategies. To define such strategies in a large river floodplain, we proposed the use of traits as surrogates for competition, colonization and dispersal abilities in communities.

Indeed, several authors proposed that species traits and their categories, may supplement classical approaches in explaining and predicting the community responses to biotic and abiotic factors (Haddad et al., 2008; Lavorel and Garnier, 2002; Statzner and Beche, 2010; Verberk et al., 2008b). Categories of life-history traits may be used to classify species into functional groups (i.e. groups of species that share common morphological, physiological or environmental responses) (Usseglio-Polatera et al., 2000), which in turn can be interpreted in terms of strategy (e.g. plurivoltinism used as a surrogate for colonization ability). Within a community, the equilibrium between alternative strategies may be assessed through the diversity of functional groups, a form of functional diversity (FD) (Lavorel and Garnier, 2002; Petchey and Gaston, 2006).

When the functional groups are equally represented, the functional diversity is maximal.

In several cases, functional diversity was calculated using trait databases that encompassed a number of species traits (Flynn et al., 2009; Petchey and Gaston, 2006).

As a consequence, contrasted responses of FD were highlighted along diverse disturbance gradients. To our knowledge, no studies carried out under field conditions tested if FD, based exclusively upon competition-colonization or dispersal traits, could be related to species diversity.

Following Shea et al. (2004), we addressed the following questions: i) does the unimodal diversity pattern proposed by the IDH exist in large river floodplain?, and ii) what mechanisms generate the observed pattern? Therefore, the goals of our study were (i) to test whether the IDH could be validated under our field conditions, and (ii) to test whether species diversity and macroinvertebrate density were related to two expressions of FD, based upon competition-colonization on the one hand, upon dispersal types on the other hand. We postulated that the lateral hydrological connectivity may be seen as a gradient of hydrological disturbance. According to Connell’s hypothesis (Connell, 1978) and the spatially structured components of a floodplain (Ward and Tockner, 2001), we firstly hypothesized a relationship of species diversity with a gradient of hydrological disturbance. We predicted a maximum of species diversity in intermediately disturbed sites. Secondly, according to Connell (1978), we hypothesized that the FD, based upon voltinism, allowed the prediction of species diversity. We predicted a maximum of species richness for an equal representation of voltinism categories (i.e. for a maximal FD based on voltinism). Indeed, if flood disturbance is frequent, the community consists only of those few species capable of quickly reaching maturity, therefore, colonization characteristics are predominant. At intermediate disturbance levels, colonizers and competitors are at equilibrium, the functional diversity and the species diversity are maximal. As the disturbance frequency declines, the functional diversity then declines, with a maximum of competition characteristics predicted in undisturbed environments (Connell, 1978). Finally, according to Cadotte (2006) we hypothesized that the functional diversity, based upon dispersal modes, would also allow the prediction of species diversity. Cadotte (2006) showed that immigration had a strong positive effect on local diversity, therefore, we predicted a maximum of species diversity where dispersal modes were equally represented (i.e. for a maximal FD based upon dispersal). (Cadotte, 2006) (Shea et al., 2004)