Biological Invasions (2006) Ó Springer 2006 DOI 10.1007/s10530-006-9001-0 Biological and ecological characteristics of invasive species: a gammarid study Simon Devin & Jean-Nicolas Beisel* Laboratoire Biodiversite´ et Fonctionnement des Ecosyste`mes, Universite´ Paul Verlaine – Metz, LBFE – Campus Bridoux, Avenue du Ge´ne´ral Delestraint, 57070, Metz, France; *Author for correspondence (e-mail: beisel@ univ-metz.fr; fax: +33-0-3-87-37-84-23) Received 3 November 2005; accepted in revised form 9 February 2006 Key words: biological invasions, biological traits, ecological traits, Gammaridae, profiling invader Abstract Knowledge of characteristics helpful in screening potential invaders and in elaborating strategies to limit their success is highly desirable. We focused on gammarid amphipods from Western Europe and North America to discover biological and/or ecological traits that may explain successful invasion by these species. Two typologies were considered: an analytical one, with groups built on the basis of biological or ecological similarities, and an empirical one, with groups constituted a priori according to a species’ invasive status and its fresh or brackish water origin. The results obtained are discussed in the light of three hypotheses that may influence invasiveness: biotic potential, species size and euryoeciousness. The analysis revealed a particular ecological profile for invaders, with a strong influence of salinity tolerance, but no typology was found based on biological characteristics. Invasiveness cannot be predicted from a limited number of criteria, and is the result of a combination of several characteristics. Invasive species therefore exhibit a particular ecological profile rather than a biological one, contrary to most classical explanations. Introduction without the introduction of new individuals is called an exotic (stage III according to Colautti Research on biological invasions has largely fo- and MacIsaac 2004). If it reaches high densities cused on the impacts of introduced species and and spreads further, it becomes invasive (stage V, on methods of their control. Recently, focus has Colautti and MacIsaac 2004), and impacts the turned to the development of tools to prevent recipient ecosystem. Here, we use this definition invasions. In addition to technical prevention of an invasive species and, following Kolar and methods, such as ballast water treatment, atten- Lodge (2001), we define indigenous (autochtho- tion is given to identify potential future invaders nous) species as taxa restricted to their native from their biological and ecological characteris- ranges. tics (e.g. Ricciardi and Rasmussen 1998; Hayes A species’ traits determine its success or failure and Sliwa 2003). in the transition between the different stages of The invasion process itself may be divided into the invasion process, and only particular trait three to five distinct stages (Vermeij 1996; Kolar combinations are assumed to make a species and Lodge 2001; Colautti and MacIsaac 2004). invasive. Some studies have adopted a quantita- After transport and introduction into a recipient tive approach for determining what these traits ecosystem, any species that maintains a popula- are (Kolar and Lodge 2001). Most deal with tion through local reproduction and recruitment plant or bird invaders, with particular attention for island ecosystems (Kolar and Lodge 2001, Material and methods Lloret et al. 2005). For invertebrates, their role is more difficult to analyze due to a lack of data on Bibliographic analysis: traits and species studied failed introductions (Vasquez and Simberloff 2001). To the best of our knowledge, studies using We selected 10 biological and 7 ecological traits a quantitative approach are lacking for aquatic (Table 1) and documented features related to the invertebrate species, but a few empirical descrip- life cycle of taxa (‘maximal size,’ ‘size distribu- tions of potential traits have been proposed tion,’ ‘hatching length,’ ‘minimal size at sexual (Ricciardi and Rasmussen 1998; van der Velde maturity,’ ‘mean fecundity,’ ‘maximal egg num- et al. 2000; bij de Vaate et al. 2002). Biological ber,’ ‘sex-ratio,’ ‘number of generations per year,’ (autoecological) characteristics are suspected to ‘life-span’), aspects of nutrition (‘diet’), different play a major role and have been more frequently habitat conditions (‘salinity,’ ‘temperature toler- investigated than ecological characteristics. All ance,’ ‘altitude,’ ‘longitudinal distribution,’ studies have considered traits individually, poten- ‘microhabitat,’ ‘current velocity’). Traits common tially underestimating a possible cumulative/ to all species, such as respiration mode or loco- antagonistic effect of traits on invasiveness. motion, were not taken into account. The size In this study, we use a quantitative method to distribution variable reflects the distribution usu- investigate a combination of biological and eco- ally observed in field of individual among size logical traits involved in invasion success. We as- class. The mean fecundity is the mean number of sume that an organism’s biological traits reflect eggs observed considering all size classes. The the biotic characteristics and the ecological func- microhabitat variable reflects the distribution at tions of the species, whereas its ecological traits the scale of the bottom substrate. In addition, describe its potential environmental tolerance. ecological traits include the mean density This study focused on gammarid amphipods, fre- observed where the species occurred. quent in biological invasions, paying attention to In order to document the biological and eco- Western Europe and North America, two well logical traits of the highest possible number of documented recipient areas. Gammarids are am- species, all accessible references dating from 1911 phibiotic and constitute a taxonomic group of to 2003 were consulted. In all, eighteen fresh and phylogenetically related species with variable bio- brackish water species were included (Table 2), ecological features (Illies 1978; Sainte Marie viz. almost all gammarids common in Western 1991). We considered species from fresh and Europe (Eastern distribution limit: Germany– brackish water origin. Switzerland–Italy; Illies 1978; Leppa¨koski et al. We investigate two questions: (1) could a 2002), as well as the most common North Ameri- typology based on a combination of biological or can freshwater species (Barnard and Barnard ecological traits be related to the invasive charac- 1983). Among 18 species considered, seven are ter of the species, and (2) does a given trait con- known to be invasive (Dick and Elwood 1993; sidered alone discriminate between invasive and Devin et al. 2005; Table 2). non-invasive species? The results are discussed in the framework of three conceptual hypotheses Fuzzy coding and statistical analysis about invasiveness: Data were retained in an array where lines repre- H1: Invasive species exhibit a higher biotic po- sent the 18 species and columns represent the 17 tential (typically an r-strategy) than non-inva- biological and ecological variables. Each variable sive ones (Lodge 1993; Williamson 1996; van was subdivided into several modalities (Table 1) der Velde et al. 2000; bij de Vaate et al. 2002). with limits allowing an equal distribution of spe- H2: Body size influences invasion success (Roy cies score among them. The affinity (score) of a et al. 2002). species for each modality was assigned using a H3: Invasive species tolerate a wider range of fuzzy coding approach (Chevenet et al. 1994). environmental conditions (Ricciardi and Ras- This procedure standardizes the description of mussen 1998). each species by assigning it a score from 0 (no Table 1. Biological and ecological variables taken into ac- affinity) to 3 or 5 (high affinity) (Table 1) and is count and the modalities defined for each variable. particularly effective for the description of vari- Variables Modalities ables for which several modalities might be simultaneously involved. Bicological traits Two multiple correspondence analyses (MCA) Maximal size (mm) [0–16[ [16–20[ adapted to fuzzy coding were performed on this [20)+[ data set, one for the biological and one for the Size distribution [0–10[ ecological traits (Chevenet et al. 1994). From the (juveniles excluded, mm) [10–13[ MCA performed on this data set, axes represent- ) [13 +[ ing about 70% of the total inertia were kept. A Hatching length (mm) [0–1.5[ [1.5–2[ cluster analysis was then performed on the facto- [2)+[ rial coordinates of the species on the MCA axes Minimal size at sexual maturity [0–5.5[ to define whether invasive species grouped to- (Females, mm) [5.5–6.5[ gether. The multivariate analyses were performed ) [6.5 +[ using ADE-4 software, a package for multivari- Mean fecundity [0–20[ [20–25[ ate analysis and graphic display (Thioulouse [25)+[ et al. 1997). Species classification was done using Maximal egg number [0–30[ Ward’s method applied to Euclidean distances [30–60[ (Statsoft, Statistica 7). ) [60 +[ Inferential analyses were performed to assess Sex-ratio F>M F=M which traits best discriminate (1) groups obtained F<M from the species traits typology or (2) groups Number of generations per year 1 constituted a priori of invasive or non-invasive 2 species. Kruskal–Wallis tests were first performed >2 on the species’ traits among groups obtained with Life span <1 year 1 year the cluster analysis, for each modality of all traits. >1 year The differences were tested modality by modality Diet Detritivorous