Assessing the Risk of Freshwater Fish Introductions Into the Iberian Peninsula

Assessing the risk of freshwater fish introductions into the Iberian Peninsula

MI G U EL CL AV ERO* , †

*Don˜ ana Biological Station, Department of Conservation Biology, CSIC, Sevilla, Spain

†Grup d’Ecologia del Paisatge, A` rea de Biodiversitat, Centre Tecnolo`gic Forestal de Catalunya, Carretera vella de Sant

Llorenc¸ de Morunys, Solsona, Spain

SUMMARY

1. Preventing the introduction of species likely to become invaders is the best management option to deal with biological invasions. A data set consisting of native, introduced and species not currently present in Iberian Peninsula (n = 167 species) was used to identify freshwater fish species that are likely to be introduced and become successful invaders in the near future.

2. Principal component analysis (PCA) of species traits was used to determine species likely to be introduced, assuming that the traits of species introduced in the future will resemble those of previously introduced species. The likelihood of introduction was calculated as the proportion of neighbour species (in the space defined by the PCA) that have been introduced to the Iberian Peninsula and, together with metrics related to different stages of invasion, was used to construct a region-specific risk index (Iberian risk index).

3. Introduced species had higher index values compared with native species or species currently absent from the region. The Iberian risk index was positively related to the results of an independent risk analysis for freshwater fish as well as to the geographical spread of species previously introduced to the Iberian Peninsula.

4. Iberian risk index values were used to establish a cut-off value for estimating the probability of a successful invasion. This threshold value was used to construct a list of 20 species to be included in a ‘watch list’ to prevent freshwater fish invasions in the Iberian Peninsula.

Introduction

Keywords: environmental management, invasive species, Mediterranean streams, risk analysis, watch lists

human activity continues to facilitate the transport of species across biogeographical barriers. Some of these

The introduction of non-native species into new environments and their spread over natural areas pose important threats for the conservation of biodi- versity (Mack et al., 2000; Clavero Garc´ıa-Berthou,

2005). Despite growing concern of the negative impacts of many non-native species on biodiversity,

non-native species cannot establish self-sustaining populations in their new habitats and eventually disappear, whilst others thrive, become abundant and expand their ranges, subsequently acquiring the status of invasive species (Lockwood, Hoopes & Marchetti, 2007). Recent effort has focussed on trying

to identify the characteristics of invasive species to

Correspondence: Miguel Clavero, Don˜ ana Biological Station, Department of Conservation Biology, CSIC, Americo Vespucio s⁄n, 41092 Sevilla, Spain. E-mail:

better understand why some species are more suc- cessful than others (Kolar & Lodge, 2001). Indeed, species profiling may be useful to identify species that

are likely to become invaders, potentially providing a valuable tool to prevent future invasions or mitigate their effects. The most commonly applied output of invasive species profiling is risk analysis, which aims to quantify the likelihood that a species will become established and⁄or will result in environmental im- pacts (Andersen et al., 2004; Baker et al., 2008).

Since the eradication or control of established introduced species is often costly, difficult and gen- erally impracticable management actions (e.g. Mack et al., 2000), preventing an introduction is considered as the best alternative to avoid new invasions (Keller, Lodge Finnoff, 2007; Hulme et al., 2008). However, since it is not possible to stop international move- ments of all live organisms or to forbid the tenancy of every non-native species (e.g. Keller Lodge, 2007), legislators dealing with biological invasions need to decide which non-native species should be subjected to regulation. Two alternative approaches are often considered in developing regulatory procedures. Fol- lowing a precautionary principle (‘guilty until proven innocent’), the first option is to require the assessment of the invasiveness of any imported species. Species classified as having low invasive potential are included on a ‘white list’ (e.g. Wittenberg Cock,

2001). The second approach is to construct ‘watch lists’ or ‘black lists’ that include non-native species that should not be imported or maintained in a given territory. Black listing is commonly used to denote invasive species that have become established in a given territory and are subjected to regulation, while species not yet present but considered as likely future invaders are included on watch lists (EEA, 2010). However, management authorities often have diffi- culty in determining which species should be included on watch lists. For example, information concerning the characteristics of a potential invader and potential donor regions or species pools is often lacking. Moreover, since many of these factors may be region specific, decisions adopted in one country or territory may not always be applicable to other areas (e.g. Moyle Marchetti, 2006).

Freshwater ecosystems are especially sensitive to biological invasions, and not surprisingly, some of the most striking examples of the impacts that invasive species have on ecosystem structure and function come from aquatic environments (e.g. Darwall et al.,

2008). The Mediterranean Basin is a global hot spot for freshwater fish invasions (Leprieur et al., 2008), with

many endemics threatened by invasive species (Cla- vero et al., 2010). For example, the main river basins of the Iberian Peninsula currently have more exotic than native fish species (Clavero & Garc´ıa-Berthou, 2006). Although the mechanisms involved in the interactions between native and exotic species are often unclear (e.g. Leunda, 2010), it has been shown that invasive fish species are an important and probably the main threat for Iberian native icthyofauna (Maceda-Veiga et al., 2010; Hermoso et al., 2011). Moreover, new fish species are being introduced to the Iberian Peninsula (e.g. Franch et al., 2008; Gante et al., 2008), while previously established species are expanding their ranges (e.g. Vinyoles et al., 2007; Ribeiro et al., 2009a). In this context, the identification of fish species that could be introduced in the near future is important to prevent further invasions.

Here, I apply a region-specific procedure to identify likely future introductions into the Iberian Peninsula and the risks associated with new invasions using freshwater fish as case study. When trying to identify successful invaders, a series of sequential stages (transport, introduction, establishment, spread and impact) and traits that may influence the success of species need to be considered (Marchetti, Moyle & Levine, 2004; Ribeiro et al., 2008). As a first step, I used a multivariate approach to quantify the likelihood of introduction of freshwater fish that are not yet present in the Iberian Peninsula. I compared the traits of fish species that are native to or have been introduced into the Iberian Peninsula with species that could poten- tially be introduced. In a second step, I complemented the likelihood of introduction with metrics related to the different stages of the invasion process to generate a region-specific risk index for freshwater fish inva- sions. My main aims were to provide an objective account of plausible future fish introductions and subsequent invasions and to provide a useful tool for biodiversity managers both in the Iberian Peninsula and elsewhere.

Methods

Species included in the analyses

The pool of fish species considered in the analyses was limited to (i) species that are present, either as native or as introduced, in the Iberian Peninsula and⁄or neighbouring European countries (France,

Italy, Switzerland, Germany, Belgium, the Nether- lands, England and Wales) or (ii) species that have been introduced in Europe, independently of whether they are known to have established self-sustained wild populations. North African species were ex- cluded from the analysis because of the paucity of ecological information (e.g. Smith Darwall, 2006) and because almost all fish introductions in the Iberian Peninsula come from Europe, and none thus far has been documented from northern Africa (Gar- c´ıa-Berthou et al., 2005). The analysis was limited to bony fish, thus excluding lampreys. Kottelat Frey- hof (2007) were used to select the pool of species analysed, supplemented with species used by Copp et al. (2009) to calibrate a risk analysis for freshwater fish [the fish invasiveness scoring kit, (FISK)] and with other fish species that have been introduced to the Iberian Peninsula (Gante et al., 2008; Ribeiro et al.,

2008). I excluded some range of restricted salmonid species that are endemic to single or small groups of lakes (mainly in the genera Coregonus, Salvelinus and Salmo). The final list included 227 freshwater fish species and followed the taxonomic nomenclature of Fishbase (Froese Pauly, 2010).

Likelihood of introduction

Twelve variables characterising biogeographical fea- tures and human uses were used to quantify the

likelihood of introduction into the Iberian Peninsula (Table 1). Unless stated otherwise, variables were obtained from Fishbase, a database including infor- mation on about 32 000 fish species (Froese Pauly,

2010). The final data set, excluding species for which information was incomplete, comprised 167 species (Table S1).

I collected data on the number of countries or territories (in Fishbase) where each species is native, excluding those in which the presence of the species was coded as ‘questionable’ or ‘misidentification’. The distribution of each species was further characterised by the average latitude of their native range (i.e. the mean between maximum and minimum latitudes) and the range of latitudes occupied (i.e. the difference between maximum and minimum latitudes). When distributions occurred on both sides of the equator, the latitudinal range was set to the maximum latitude occupied. The introduced range of each species was described by the number of countries or territories where the species had been introduced, independent of whether introductions resulted in established pop- ulations.

I included popularity in the analyses assuming that the most popular species would be more likely to be introduced into the region. Popularity for each species was evaluated using Google web browser. I intro- duced the Latin names in quotes and recorded the number of results of the search. The search procedure

Table 1 Variables used to describe the characteristics of fish species used in the calibration of the Iberian risk index. Data transfor- mations (Transf.) are indicated, with a dash denoting no transformation. For continuous variables, numbers in brackets indicate

the range of values (minimum and maximum). The inclusion of each variable in the principal component analysis (PCA, see Fig. 1)

is also stated

Code Values (min, max) Units Transf. PCA Total length SIZE Continuous (4.5, 800) cm log10X Yes

Native countries NATI Continuous (1, 55) Count X Yes

Average latitude avLAT Continuous (0.5, 60.5) Degrees X2 Yes Latitude range raLAT Continuous (1, 67) Degrees – Yes Introduced countries INTR Continuous (0, 121) Count X Yes Popularity POPU Continuous (406, 543000) Count log10X Yes Iberian popularity POPIBE Continuous ()1.9, 1.5) Count (residual) – Yes Game fish GAME 0, 1, 2 None – Yes Aquaculture AQRE 0, 1, 2 None – Yes Fisheries FISH 0, 1, 2 None – Yes Aquarium AQUM 0, 1, 2 None – Yes Bait BAIT 0, 1, 2 None – Yes Status in France Absent, nativeand introduced None – No Habitat breadth 1–5 Count – No Climatic mismatch Continuous (0, 39.5) Degrees X No Parental care Yes, no None – No Pest Yes, no None – No

was then restricted to Spanish and Portuguese pages to obtain an Iberian index of popularity. Residuals of the number of results of the Iberian search regressed against the number of results of the unrestricted search (log10-transformed in both cases; F1,165 = 78.1;

P 0.001; R2 = 0.32) were used as an indicator of the

relative popularity of each species in the Iberian Peninsula. Human uses of each species were charac- terised by coding game fish, aquaculture, fisheries, aquarium and bait from 0 to 2: 0 = ‘no use’,

1 = ‘minor use’ or ‘occasional use’ and 2 = ‘otherwise’ (according to Fishbase). Since introduced fish species have larger body size than expected by chance alone (Blanchet et al., 2010), I also included maximum length (in cm) of each species in the analysis of the likelihood of introduction.

Principal component analysis (PCA) with varimax normalised rotation (McGarigal, Cushman Sttaford,

2000) was used to summarise the variation in vari- ables related to the likelihood of introduction. Only the first three axes were used since preliminary analyses showed that these axes explained most of the total variance. Position in the three-dimensional space defined by the three axes (PC1, PC2 and PC3) was used to indicate fishes likely to be introduced in the Iberian Peninsula. Differences in the average position of absent, introduced and native Iberian species along the three axes were analysed using one- way A N O V A s. A dissimilarity square matrix from the scores of each species along the three principal components (PCs) using Euclidean distances was calculated and used to estimate the percentage of introduced neighbours for each species (counts of neighbours within a given radius). The radius was set as the average distance of the 25th nearest neighbour for the 167 species included in the PCA. In 18 cases, the number of neighbours within the fixed radius was

<10; here, I used the 10 nearest species to calculate the percentage of introduced neighbours. This index, which ranged from 0 to 69%, was used to estimate the likelihood of introduction, under the assumption that the characteristics of fish species introduced in the future will resemble those of current introduced species.

Risk index for the Iberian Peninsula

Together with the likelihood of introduction, addi- tional metrics related to different stages of the invasion