Little association of biological trait values with environmental variables in invasive alien round goby (Neogobius melanostomus)

Abstract The relative importance of species‐specific biological trait characteristics and environmental factors in invasions of nonindigenous species remains controversial because both have mostly been studied independently. Thus, the main objective of this study was to examine the correlation of biological traits with environmental variation in the globally invasive round goby Neogobius melanostomus from the upper Danube River. Based on a sample of 653 specimens along a continuous 200 km river pathway, links between nine environmental factors (substrate‐type, six water measurements, and the communities of fishes and macroinvertebrates) and seven biological traits (nutritional and energetic status, trade‐offs of parasite resistance and resource allocation, and three growth proxies) were analyzed. Biological trait values of N. melanostomus hardly correlated with the environment, could not explain invasion progress and imply a general low overall importance for invasion success. Instead, alternative individual life‐history trajectories appear to determine invasion success. This is in line with up to 15% of all specimens having outlying biological trait values of potential adaptive value, suggesting a considerable importance of adaptive trait variation among single individuals for the whole invasion progress. This “individual trait utility hypothesis” gives an alternative explanation for success of invasive species by single individuals carrying particular traits, and it should be specifically targeted and analyzed at currently invaded sites.

revealed that several hypotheses have little power to explain and predict biological invasions, that their applicability is sometimes limited, and that hypotheses taking into account the invaded environment have higher empirical support than hypotheses focusing on single species-specific criteria or single environmental factors (Jeschke et al., 2012). The low applicability of generalizing hypotheses may indicate that invasion success is multifactorial and that it rather depends on a combination of site-and habitat-specific settings. In addition, varying characteristics and traits of single individuals may also be crucial (Nilsson, Brönmark, Hansson, & Chapman, 2014).
Nonindigenous species are simultaneously exposed to several new and multifactorial stressors in the invaded systems (Strayer, Eviner, Jeschke, & Pace, 2006). Local impacts may leave both, individual and site-specific signatures in the genotype and the phenotype. This study attempts to identify such potential signatures of environmental variables by analyzing biological trait variation of invasive alien Neogobius melanostomus (Pallas 1814) (Teleostei: Gobiidae) individuals along a chain of sampling sites along a recently invaded river stretch.
An ongoing invasion along a fluvial ecosystem is a particularly appropriate model system, because species characteristics may vary with particular environmental factors along two dimensions only (Sakai et al., 2001). In such a spatially restricted ecosystem, crucial characteristics for invasion success could be identified along small geographic scales considering environmental factors and time since invasion. Such candidate species characteristics may relate either to life-history (van Kleunen, Dawson, Schlaepfer, Jeschke, & Fischer, 2010), phenotypic plasticity (Cerwenka, Alibert, Brandner, Geist, & Schliewen, 2014), or to tolerance for anthropogenic disturbance and habitat degradation (Allendorf & Lundquist, 2003). Besides, the genomic constitution (Lambrinos, 2004) appears important, too : Matesanz, Horgan-Kobelski, and Sultan (2014) showed quantitative variation of genetically determined traits and linked genetic variability with local adaptation and invasion progress. However, the genetic variability of invasive populations often varies substantially and depends on propagule pressure, that is, number of introduced specimens and frequency of introduction events (Wilson, Dormontt, Prentis, Lowe, & Richardson, 2009).
Propagule pressure is strongly influenced by factors of passive translocation and of primary dispersal (e.g., transportation vessels), whereas active migration propensity of IAS may particularly influence secondary dispersal (area expansion) and population admixture after introduction.
After range expansion, additional traits may become important for successful establishment of novel populations. Particularly for this phase, which undoubtedly is crucial for long-term success, the importance of environmental factors is poorly understood and still understudied.
If success of IAS is indeed controlled by specific environmental variables (van Kleunen et al., 2010), specimens' biological trait values should vary along an environmentally heterogeneous fluvial gradient, but should vary little within single sites within this gradient.
Alternatively, success of IAS may be determined by single individuals (see Nilsson et al., 2014). Such "high-performance" individuals would be characterized by biological trait values that strongly deviate from population average and individuals would have outlying biological trait values, not related to particular environmental factors (Biro & Stamps, 2008;Cote, Clobert, Brodin, Fogarty, & Sih, 2010).
To explore these alternatives, we addressed the following two research questions: (1) Are there significant differences in variability of biological trait values, particularly in outlier frequency, of N. melanostomus between sampling localities, and (2) are these differences of biological trait variation correlated with specific variables? If significantly more individuals with biological trait outlier values were detected at expanding frontend populations and if variation of biological traits was not correlated with environmental variation, this would support the hypothesis that high-performance individuals contributed more to invasion success than environmental variation. Finally, potential associations and correlations with environmental variables were investigated.

| Field sampling
Sampling covered two entire growth seasons of N. melanostomus (see Brandner, Auerswald, et al., 2013;Cerwenka, Alibert, et al., 2014) with a maximal time period between samplings of 1 month. Sampling areas ( Figure 2, Table 1) comprised both river sides and two mesohabitat structures (artificial stone blocks and natural gravel banks).

| Biological traits
Adult and mature specimens with a proximal L T of 8-12 cm (range = 2.9-15.2 cm, mean = 9.6 cm, SD = 1.5 cm) were anaesthe-   (6) all sites together and analyzed independently (n boxplots = 49). Further, specimens that deviated more than 1.5 times from the interquartile range of the all sites together dataset were identified as outliers in PAST 3.06 (Hammer, Harper, & Ryan, 2001).

| Data analyses
To validate the number of specimens carrying outlying trait values, 1,000 potential trait allocations were estimated at each sampling site, following Dawson (2011). Therefore, the extreme values of each trait in the all sites together real-world dataset were used as cutoffs.
Subsequently, the mean number of extreme outliers of both datasets (real-world vs. estimated) were compared using a parametric t-test.
The number of specimens carrying outlying biological traits at each river area was compared using pairwise Bonferroni corrected Mann-Whitney U tests.

| RESULTS
Only few significant (p < .05) pairwise correlations of biological traits with ecological variables and covariables were detected; that is, only few biological traits were linked to feeding resources, anthropogenic stressors, water temperature, migration barriers, and to covariables

| Environmental factors and biological trait values
Lipid content (lipids) Lipids were used to quantify the ecological differentiation. The isotopic C/N ratio was calculated from muscle tissues.
Growth rate (annuli) [μm] Distance between the first and the second annulus as a correlate for different resource allocation. Heino and Kaitala (1999) and Martin (2012) Growth rate (# circuli) Number of circuli between the first and the second annulus, as a correlate for different resource allocation. Heino and Kaitala (1999) and Martin (2012) Growth heterogeneity (var circuli).
Variance in circuli spacing between the first and the second annulus as a correlate for different resource allocation. Heino and Kaitala (1999) and Martin (2012)

| DISCUSSION
Invasion progress is known to be triggered by both, species-specific traits (Colautti et al., 2014;Nathan et al., 2008)  The Danube River has been heavily altered in both, morphology and temperature. Früh et al. (2012a,b) showed that anthropogenic caused environmental degradation correlates with high abundance of IAS. A potential explanation may be a generally higher tolerance of invaders against stressors (Früh et al., 2012a,b;N. melanostomus: Ondračková et al., 2014). However, such anthropogenic alterations, for example, migration barriers, can limit species movement and admixture of genetically differentiated populations and may thus differentially change species fitness (Crispo, Moore, Lee-Yaw, Gray, & Haller, 2011). In N. melanostomus, migration barriers may be important for rapid differentiation (Cerwenka, Alibert, et al., 2014;. However, no coherence of biological trait variation and genetic and morphometric population structure was detected in the present study. Temperature regimes of the upper Danube River increased almost 2°C over the last 100 years (Bloesch, 2011). In this study, water tem- For environmental variables, all significant PCs, their variance, and the variable with the highest loading are indicated (variable 1). Biological trait allocation in dependence of the environment was compared by subdividing sampling sites into sites with N. melanostomus having lower and higher biological trait values than indicated by the median value (pooled for all sites; CF = condition factor, parasites = number of acanthocephalan parasites, lipids = isotopic ratio, HSI = hepato-somatic index, annuli = distance between first and second annulus, # Cir = number of circuli between first and second annulus, VarCir = variance of circuli distances between circuli of first and second annulus). The number of sampling sites having lower and higher environmental PC values is denoted for each biological trait and significant differences are indicated by stars (***p < .001, **p < .01, *p < .05).

| Dynamics of biological traits in Neogobius melanostomus
About 15% of all analyzed N. melanostomus specimens had at least one biological trait value deviating from the average variation. Similarly, a subset of all specimens from a population are thought to contribute disproportionately high to the species expansion in the Gulf of Gdansk, Poland (Thorlacius, Hellström, & Brodin, 2015), and the Great Lake tributaries (Bronnenhuber et al., 2011). Thorlacius et al. (2015) found "bolder" N. melanostomus individuals to be more aggressive, more explorative, and being more capable to cover greater migration distances. The authors suggest boldness to be beneficial (providing easier access to new resources), but also to involve a trait-off in terms of higher costs (risk of predation). However, bold (in our case: largesized and high conditioned) individuals may even benefit from a lower predation risk at (expanding) fronted populations (Brownscombe & Fox, 2012) which might further support their fast dispersal Lindström et al., 2013).
Selection operates predominantly at low levels of the hierarchical organization of species, that is, genes and individuals (Lewontin, 1970). Individuals with outlying biological trait values may indeed contribute disproportionally to population fitness (Coulson et al., 2006) (lower in the case of elevated parasitic load (Milinski, 2009)). Such individuals may not dramatically alter average biological trait variation at the population level ad hoc, but they nevertheless may significantly influence invasion progress, because they may act beneficially for all members of a rapidly expanding population by increasing their fitness during a critical phase of the invasion process (Smith & Blumstein, 2008). Genome scans revealed that contemporary site-specific differentiation of N. melanostomus (in the same river stretch) can be correlated with just three out of 189 analyzed AFLP-loci: 178 individuals (31%) carried the one of those, 81 (14%) another one, and 13 (2%) the third one. In the context of the present study, it is noteworthy that three specimens (0.5%) carried all three alleles and that each of these three specimens originated from the expanding frontend population . Interestingly, the occurrence of all three individuals was not independent from area and time since invasion (χ 2 -test, p = 1). These population genomic results in combination with the results presented herein indicate that single individuals with an exceptionally high fitness may indeed have an effect on invasiveness of a whole population. In this context,  already pointed to a contribution to invasion success of individuals being comparatively large and having a high condition, that is, possibly being bolder.
Surprisingly, and in contrast to Cerwenka, Alibert, et al. (2014),  and , the present study did not find strong trait distribution differences between recently invaded and longer established sites. A possible explanation for this paradox may lie in fast personality-dependent dispersal (Thorlacius et al., 2015), because high migration rates may rapidly level out small scale trait distribution differences even across large distances (Brandner, Auerswald, et al., 2013;Brandner et al., 2015). In addition, multiple inoculations may have the same leveling effect (Colautti et al., 2004), possibly by increasing the likelihood of introducing locally fit (i.e., well adapted) individuals.
In conclusion, our study demonstrates that biological traits on the level of single individuals appear to determine the success of invasive alien gobies. This "individual trait utility hypothesis" should be specifically targeted and tested in future studies of nonindigenous and native expanding populations.

ACKNOWLEDGMENTS
We thank all the friendly and helpful owners of the local fishing rights, the "Fischereifachberatungen" and Dr. G. Zauner who supported our sampling in Austria. Furthermore, we thank K. Auerswald, R. Schäufele, N. Straube, and K. Lindner for helpful assistance in the laboratory and practical advice.