Structure of aquatic insect communities in tank-bromeliads in a East-Amazonian rainforest in French Guiana
Introduction
Human impact on the environment obviously causes a decline in biological diversity from a local to a global scale (Brooks and Kennedy, 2004). In this context, a major challenge for ecologists is to understand the structure and dynamics of biological communities in relation to environmental variability (Gaston, 2000). Whilst they only represent 6–7% of continental surfaces, tropical rainforests shelter more than half of the Earth’s species (Raven, 1988, Wilson, 1988). Thus, deforestation rates in the humid tropics provide compelling reasons to conduct detailed investigations of community diversity (Bowles et al., 1998, DeVries et al., 1997) for evidence of biological responses to habitat characteristics and/or degradation (Yanoviak et al., 2006). For instance, classifying sites of high biological quality is likely to provide system-specific predictions of the fauna and/or flora to be expected under undisturbed conditions. However, studying community patterns in tropical ecosystems is complicated by two factors. First, taxonomic information is limited, and most species remain undescribed (Godfray et al., 1999). Second, the very high species densities within continuous habitats makes it difficult, if not impossible, to exhaustively sample and, thus, quantify animal communities. To overcome this problem, some authors have focussed on natural microcosms with discrete boundaries such as small water tanks (Richardson, 1999, Kitching, 2001, Srivastava et al., 2004, Yanoviak, 2001).
Many species of Bromeliads (Bromeliaceae) have tightly interlocking leaves forming wells (or phytotelmata) that hold rainwater. These monocotyledon plants are represented by 1500–3000 species, and are mostly restricted to the neotropics and Florida (there is also one naturally occurring African species, namely Pitcairnia feliciana). Most of them are epiphytes. Because phytotelmata collect leaf litter and dead invertebrates, they provide a habitat for aquatic species (Carrias et al., 2001) among which insect larvae are well represented (Picado, 1913, Laessle, 1961, Kitching, 2000). The detritus provides a source of nutrients for the insect community, as well as the bromeliad itself (Richardson et al., 2000, Ngai and Srivastava, 2006). Tank-bromeliads are thus discrete habitats which contain distinct aquatic communities. The main advantage of studying these natural microcosms is their abundance in any given area. Moreover, owing to their small size, the entire aquatic community from each plant can be quantified with a degree of accuracy not possible in larger systems (Richardson, 1999). Because many aquatic insects complete their larval development in tank-bromeliads, differences in plant morphology may play an important role in habitat selection by ovipositing adults. More specifically, co-occurring bromeliads show a high variability in terms of colour patterns, size, reservoir shape and volume. Bromeliads with large central pools (e.g., Neoregelia sp.) were found to differ in their species assemblages from those with many separate, smaller pools (e.g., Guzmania sp.) (Harris, 1993). Armbruster et al. (2002) found that both biological interactions (i.e., competition and predation) and abiotic factors (the physical structure of the plant, water volume) influenced the species richness of bromeliad-inhabiting communities. However, they did not consider the different bromeliad species separately. To the best of our knowledge, there is nothing in the published literature to demonstrate that there have been previous attempts to classify individual tank-bromeliads according to their aquatic biota (see review of key studies on bromeliad communities in Kitching (2000)), and there is no evidence of species-specific associations between tank-bromeliad species and particular sets of aquatic animals (Benzing, 1990). Therefore, the existence of such associations must be tested by classifying different species of bromeliads according to the structure of their aquatic communities. Furthermore, classifying tank-bromeliads based on their biota is a crucial step towards optimizing the design of new surveys (e.g., biomonitoring) and/or experiments (hypothesis testing on some targeted insect communities), while improving our understanding of the dynamics of biodiversity in tropical habitats.
The aim of our study was to assess the relationships between tank-bromeliad species and the quantitative structure of aquatic insect communities in a primary rainforest in French Guiana. To this end, 158 plants belonging to five species were sampled at four distinct sites. The aquatic insects were mostly keyed to morphotypes and enumerated. Ecological data such as species abundances often vary and covary in a nonlinear fashion (Lek and Guégan, 2000), depending on the distribution of sampling locations. Thus, non-linear modelling methods such as artificial neural networks (ANNs) should theoretically be preferred for dealing with such data (Blayo and Demartines, 1991). Combining clustering and ordering abilities, the self-organizing map algorithm (SOM, unsupervised neural network; Kohonen, 2001) is a powerful tool used to visualize high-dimensional data. This technique has had particular relevance in revealing patterns of biological communities in relation to environmental characteristics because the gradient distribution of the biological variables can be visualized in a two-dimensional map (Park et al., 2003), and because it is particularly appropriate for dealing with outlying and rare species. We used the SOM algorithm to interpret the variability of aquatic insect communities with respect to bromeliad species. Self-organizing maps and other ANN techniques have been successfully implemented in various aspects of ecological modelling such as classifying groups (Levine et al., 1996), patterning complex relationships (Tuma et al., 1996), predicting population and community development (Recknagel et al., 1997), modelling habitat suitability (Özesmi and Özesmi, 1999), and assessing water quality (Walley et al., 2000, Céréghino et al., 2003). General linear modelling (GLM) was used to specify the influence of variables related to plant characteristics (species, volume of water and fine particulate organic matter in the tank, elevation above the ground) and site location (elevation a.s.l.) on the taxonomic richness and number of individuals per plant.
Section snippets
Study area
This study was conducted in a primary rainforest characteristic of the eastern Amazon, around the Nouragues Tropical Forest Research Station (4°5′N, 52°41′W, French Guiana, Fig. 1). The area is totally uninhabited and anthropogenic disturbance is almost completely absent. The Station is located in the Nouragues Natural Reserve, 100 km away from Cayenne and 40 km from the nearest village (Regina). The area is delineated by hills (elevation <120 m a.s.l.), and by the Balenfois mountains (maximum
Community composition and sampling adequacy
Insect abundance and taxonomic richness ranged from 1 to 112 individuals and from 1 to 9 taxa per plant, respectively. The list of taxa found in each bromeliad species is provided in Table 2. Overall, the five bromeliads studied hosted 44 aquatic insect taxa. Most taxa (30) belonged to the Diptera. Other taxa were the Heteroptera (7 morphotypes), Coleoptera (6 morphotypes), and Odonata (1 morphotype). In terms of abundance, seven taxa (Chironomini sp1, Tanypodinae, Chironomidae sp2, Wyeomyia,
Discussion
Many studies have described the diversity and/or taxonomy of bromeliad-dwelling organisms (e.g., mosquitoes, midges, crustaceans; Frank, 1983, Little and Hebert, 1996, Grogan and Hribar, 2006), but relatively few authors have actually used tank-bromeliads to analyse the diversity of simplified communities in tropical ecosystems (but see Cotgreave et al., 1993, Richardson, 1999, Armbruster et al., 2002). It should be noted, however, that research focused on other phytotelm species (notably the
Acknowledgements
We are grateful to Andrea Dejean and Peter Davies for proofreading the manuscript. This work was supported by the Programme Amazonie II of the French Centre National de la Recherche Scientifique (Project 2ID), and the Programme Convergence 2007–2013 (Région Guyane) from the European Community (Project DEGA). We wish to thank Alain G.B. Thomas for his taxonomic assistance, Marc Gibernau for his help and advice on the GLM analyses, Jérôme Orivel for providing us with Fig. 1. Two anonymous
References (62)
Separating the elements of habitat structure: independent effects of habitat complexity and structural components on rocky intertidal gastropods
Journal of Experimental Marine Biology and Ecology
(2000)- et al.
Spatial analysis of stream invertebrates distribution in the adour-garonne drainage basin (France), using Kohonen self-organizing maps
Ecological Modelling
(2001) - et al.
The relationship between body size and population size in bromeliad tank fauna
Biological Journal of the Linnean Society
(1993) - et al.
Species diversity in vertical, horizontal, and temporal dimensions of a fruit-feeding butterfly community in an Ecuadorian rainforest
Biological Journal of the Linnean Society
(1997) - et al.
A comparison of self-organizing map algorithm and some conventional statistical methods for ecological community ordination
Ecological Modelling
(2001) - et al.
Classifying soil structure using neural networks
Ecological Modelling
(1996) - et al.
An artificial neural network approach to spatial habitat modelling with interspecific interaction
Ecological Modelling
(1999) - et al.
Applications of artificial neural networks for patterning and predicting aquatic insect species richness in running waters
Ecological Modelling
(2003) - et al.
Artificial neural network approach for modelling and prediction of algal blooms
Ecological Modelling
(1997) - et al.
Are natural microcosms useful model systems for ecology?
Trends in Ecology and Evolution
(2004)