The biotic resistance hypothesis (Elton 1958), and its later formulation as the diversity-invasibility hypothesis (Tilman 1999), posits that invasive species may fail or suffer reduced success in more diverse communities owing to lack of niche space, increased competition and other antagonistic interactions with native and other resident species (eg previous invaders; Levine et al. 2004). Overall, this hypothesis has received equivocal support, because even though many studies show a negative correlation between invasibility and diversity (Maron and Marler 2007) many others report no association (Havel et al. 2005a, Havel et al. 2005b) or even a positive correlation (Dunstan and Johnson 2004). Further, prey naïveté, rather than reduced diversity in insular systems likely drives higher invasibility and impacts (Cox and Lima 2006).

Allied with the concept of biotic resistance, the enemy release hypothesis posits that escape from enemies (such as predators, parasites, pathogens) might explain higher competitive ability and hence the heightened invasiveness of some introduced species (Torchin and Mitchell 2004). Again, the general utility of this hypothesis has been debated (Colautti et al. 2004, Hatcher et al. 2006, Dick et al. 2010). However, it is well known that native generalist predators can readily consume introduced species (Rodriguez 2006, Carlsson et al. 2009); this has been linked to the reduction in spread, distribution or abundance of invaders, although evidence ranges from circumstantial to clearly demonstrating cause and effect (see Reusch 1998, Monserrat et al. 2005, deRivera et al. 2005, Jensen et al. 2007, Carlsson et al. 2010, Carlsson et al. 2011, Twardochleb et al. 2012, Takizawa and Snyder 2012). Indeed, the presence of resident predators is often invoked as an explanation for geographic variation in the establishment success and post-establishment spread and impact of many introduced aquatic species (Baltz and Moyle 1993, Ward et al. 2008, Marsh-Matthews et al. 2011).

Many such hypotheses in invasion ecology have, however, been recently criticized as being vague, poorly defined and their tests therefore not robust (see Jeschke et al. 2012, Heger et al. 2013, Ricciardi et al. 2013). Heger et al. (2013) in particular show that invasion biology could be improved by its hypotheses being “branched” into “specific and testable hypotheses”, such that robust conclusions can emerge. In this context, we argue that tests of the “biotic resistance hypothesis” (BRH) with respect to predation cannot simply rely on demonstrations that introduced species are preyed on by residents, because predator-prey relationships can clearly be stable (Juliano 2001) or so weak as to have a negligible effect on invasive prey abundance (e.g. Twardochleb et al. 2012). We thus refine the BRH here to our “predator restriction hypothesis”, support for which requires negative associations between invaders and resident predators in the field and, further, evidence of a predator-prey interaction that can regulate or de-stabilise prey populations. Thus, the utilization of the “functional response” of a predator (the relationship between prey density and consumption rate; Holling 1959, Holling 1966) may allow more rigorous tests of this hypothesis, as functional responses can explain and predict the impact of predators on prey populations (see Juliano 2001, Dick et al. 2013). In particular, it is important to distinguish between the various forms of this relationship owing to differences in potential contributions to prey population stability (Juliano 2001, Murdoch and Oaten 1975, Hassell et al. 1977). The Type III functional response is typified by relatively low prey consumption at low prey densities, and hence may provide prey with a refuge from predation pressure and thus promote population stability (but Type III responses can still strongly regulate prey populations and provide biotic resistance; Twardochleb et al. 2012). This is somewhat contrasted with the potentially population de-stabilising Type II functional response, whereby high proportions of prey are consumed at low prey densities, hence potentially de-stabilising those prey populations (although weak Type II functional responses of resident predators can still allow invader prey to proliferate; Twardochleb et al. 2012). Such an approach can therefore demonstrate that the predatory behaviour of resident species translates into a significant reduction in the invader prey population (such as the local extinction of the invader and disjunct distributions of invader and natives), driven by a clearly limiting or potentially de-stabilising predator-prey relationship (e.g. see also Kushner and Hovel 2006). Further, invasive species encounter many potential resident predators and we require methods to establish the relative role of each in biotic resistance.

Here, we test our hypothesis and compare biotic resistance between two residents in a system where field data indicate that an invader might be restricted in distribution and abundance by the two resident species that are known to prey on the invader. We use comparative functional responses, a methodology that has successfully elucidated the impacts of various predatory invaders on native prey (Bollache et al. 2008, Dick et al. 2010, Kestrup et al. 2011, Dick et al. 2013). Specifically, the North American amphipod crustacean Crangonyx pseudogracilis has invaded western European freshwaters, but typically shows a mutually exclusive distribution relative to Gammarus spp. amphipods over a range of spatial scales – from large geographical areas of N. America and Europe (Barton and Hynes 1976, Pinkster et al. 1992, J. Holsinger personal communication) to between and within individual river basins (Holland 1976, Cao et al. 1996, Dick 1996, MacNeil et al. 2000). This pattern in Europe is often assumed to be driven by environmental factors (see MacNeil and Dick 2011), but may also be linked to predation of the smaller, mainly herbivorous invader Crangonyx pseudogracilis by the larger European natives Gammarus pulex and Gammarus duebeni celticus (Dick 1996, MacNeil et al. 1999, 2000). However, support for our ‘predator restriction’ hypothesis requires demonstration that the interaction between natives and invaders can strongly regulate or drive local extinction of the latter. Demonstration of a strong Type II predatory functional response of the resident predators towards the invasive prey, consistent with field patterns of invasion, contact and overlap of the invader and residents, would be a powerful test of our hypothesis. Further, however, since functional responses are sensitive to environmental heterogeneity, whereby Type II responses often become Type III responses when moving from homogenous to heterogeneous habitats (Lipcius and Hines 1986, Anderson 2003, Alexander et al. 2012), support for our hypothesis would be enhanced if we find consistency in the type and magnitude of functional responses under varying environmental conditions. In addition to this, we expect field distributions to reflect interspecific differences in the abilities of resident predators to exert biotic resistance towards invading prey; specifically, we predict greater functional responses by those resident species that are associated with greater impacts on invader populations in the field.

We thus tested our ‘predator restriction hypothesis’ by: (1) determining field patterns of negative associations between the invasive N. American amphipod Crangonyx pseudogracilis and the European amphipods Gammarus pulex and Gammarus duebeni celticus, by re-analysing several published field survey data sets; (2) examining these data sets for any apparent differential in biotic resistance of the two resident predators on the invader; and (3) experimentally deriving the type and magnitudes of functional responses of the two resident predators towards the invasive prey, in both homo- and heterogeneous habitats.

The ‘biotic resistance hypothesis’ (BRH; Elton 1958, Levine et al. 2004) posits that resident species (natives and previous invaders) can constrain the success of invasive species, but the hypothesis does not specify at what stage of the invasion process (see Blackburn et al. 2011) this might occur: do natives resist the introduction, establishment, or spread of the invader? Further, biotic resistance may occur through many mechanisms; although most emphasis has been placed on competition as the mode of resistance (Levine et al. 2004), consumption of invaders by resident species has also been invoked (Maron and Vila 2001, Monserrat et al. 2005, Rivera et al. 2005, Jensen et al. 2007, Carlsson et al. 2009, Carlsson et al. 2010, Carlsson et al. 2011, Twardochleb et al. 2012). However, broad hypotheses such as the BRH require refinement if they are to be specific and testable (Heger et al. (2013)); simply demonstrating the existence of a predator-prey relationship between residents and invaders does not in itself provide evidence of biotic resistance, as predator-prey interactions can be stable (Juliano 2001), predators may not affect the abundance of their prey (e.g. Twardochleb et al. 2012) and predators and their prey can clearly co-exist. More compelling support requires demonstration that resident predators have population-level effects on the invader – as evidenced by, for example, local extinctions of invaders and disjunct resident/invader distributions. Further, demonstrating that the predatory impact of residents can limit an invader, such as by characterising the functional response of residents to invaders (see Twardochleb et al. 2012, Dick et al. 2013), would be compelling evidence of cause and effect. We thus refine the BRH for the purposes of this study to our ‘predator restriction hypothesis’ and test this in the field and laboratory in a system of two resident predatory amphipods and their invasive prey.

Our survey data sets of the European residents Gammarus pulex and Gammarus duebeni celticus and the N. American invader Crangonyx psuedogracilis reveal that the latter species has strong negative associations with the two former species. All rivers and lakes considered in our analyses (see Dick et al. 1994, Dick 1996, Dick et al. 1997, MacNeil et al. 2009, 2001) have all three species present and the invader has had several decades in which to spread throughout the systems. We can thus assume that Crangonyx pseudogracilis has had ample opportunity to colonise all sites, despite being absent from the majority of sites where the aforementioned residents are found. There may be some abiotic component of such patterns, whereby Gammarus spp. are less likely to be found in heavily organically polluted areas where Crangonyx pseudogracilis can survive (MacNeil and Dick 2011). However, this cannot explain the disjunct distributions of invader and residents in, for example, stretches of the same lake shoreline or river where water quality is consistent among sample sites, as with our sampling sites selected and analysed here. Indeed, the current study excluded any grossly organically polluted sites and we thus eliminated as far as possible this potential major environmental driver of differing species distributions and associations. Further, during transplantation experiments, Crangonyx pseudogracilis was able to survive at Gammarus sites when the invader was in bioassay tubes that allowed water exchange but protected the invader from contact with the residents (MacNeil et al. 2000), strongly suggesting that biotic rather than abiotic factors limit the spread of the invader.

Our field data also revealed a significantly greater negative association of the invasive Crangonyx psuedogracilis with the previous invader Gammarus pulex as compared to the native Gammarus duebeni celticus. This is fully consistent with our experimental findings of Type II functional responses of both resident predators towards this invader prey, and with the functional responses of Gammarus pulex being significantly greater in magnitude, and its higher attack rates, lower handling times and greater maximum feeding rates than Gammarus duebeni celticus. In addition, for both predators, the functional response was clearly and consistently of Type II even in heterogeneous habitat conditions, where prey may often have refuge from predators, leading to a change to more stabilising Type III functional responses (Lipcius and Hines 1986, Anderson 2003, Alexander et al. 2012, in press). We saw no such change in functional response Type from II to III when predator and prey were housed in heterogeneous conditions, as was noted in our similar experimental systems for the native marine amphipod Echinogammarus marinus (Alexander et al. 2012, in press). Although predation rate was somewhat reduced overall in heterogeneous conditions (with lower attack rates, increased handling times and hence lower maximum feeding rates), this latter reduction, although statistically significant, was actually rather trivial biologically; for example, the maximum feeding rate for Gammarus duebeni celticus in heterogeneous conditions only dropped to 0.58 prey per hour (p/hr) from 0.59p/hr in homogeneous conditions, and for Gammarus pulex the decline was only to 0.70p/hr from 0.75p/hr. A further consideration of the effect of Gammarus as predators is that they themselves are subject to predation, for example by fish, and this could alter their functional responses towards prey through trait-mediated indirect interactions (TMIIs). We have shown that TMIIs can alter both the shape and magnitude of amphipod functional responses in sometimes counter-intuitive ways (eg heterogeneity increased the magnitude of Type III responses; Alexander et al. in press). We thus encourage more studies of the community context within which the biotic resistance hypothesis, and its refinements, are tested and conducted.

Given that Type II predatory functional responses are considered as potentially de-stabilising towards prey populations owing to the increased risk of mortality at low prey densities (Hassell 1978), and have been suggested as potential drivers in local extinctions of prey (Taylor and Collie 2003, Rindone and Eggleston 2011), our experimental and field results are congruent, with the resident predators showing disjunct distributions with the invader, which mostly fails to establish where the residents are present. This is most extreme with regards to the more efficient predator Gammarus pulex, which exhibits a significantly greater magnitude Type II functional response and relatively rarely co-exists with the invader. However, we must not only consider the type of functional response, but also the magnitude, because relatively ‘low’ Type II functional responses – where predation rate is more than offset by prey reproductive rate – could lead to exponential growth of the prey and hence, in the present context, successful invasion (see Twardochleb et al. 2012). On the other hand, relatively ‘high’ Type II functional responses, where predation rate outstrips prey reproductive rate, may drive invasive prey to extinction (Twardochleb et al. 2012). Thus, the balance of resident predation rate and invader reproductive rate must be examined to further elucidate the likely population consequences of the resident/invader interaction. Female Crangonyx pseudogracilis produce on average 33 offspring every 22 days at our experimental temperature (see Hynes 1955, Sutcliffe and Carrick 1981), that is, approximately 1.5 offspring per day. Gammarus pulex and Gammarus duebeni celticus, even under heterogeneous conditions, can consume 17 and 14 Crangonyx pseudogracilis per day, respectively. All else being equal, therefore, the resident predator predation rate can clearly outstrip the invader prey reproductive rate, even when the ratio of predator:prey abundance is 1:1 or substantially in favour of the invader. Such ratios are, however, unlikely in an invasion scenario whereby Crangonyx pseudogracilis attempts to colonise areas where Gammarus populations are established at densities measured in the 100s to 1000s per square metre (Kelly and Dick 2005); indeed, Crangonyx pseudogracilis was only present at sites where Gammarus densities were measured only in the 10s of individuals or were absent (see above and Dick 1996). Our field and laboratory results are thus in agreement with the theoretical biotic resistance framework provided by Twardochleb et al. (2012); that is, we have shown a relatively high Type II functional response that is likely to drive invasive prey extinct, and we find no evidence, even in heterogeneous habitat, of a Type III functional response that is more likely to allow invasion. Our negative field associations of predator and prey, with the complete absence of the invader commonly associated with the presence of resident predators, strongly corroborates the laboratory results. Further, it is clear that comparing two resident predators with respect to their functional responses can reveal the mechanism of differential biotic resistance, again corroborated by our field results whereby the predator with the higher functional response exerts more biotic resistance.

Predator exclusion experiments provide compelling evidence for biotic resistance (Robinson and Wellborn 1988, Carlsson et al. 2011); however, it is often unfeasible to manipulate small invertebrate predators for such experiments in the field. Laboratory microcosm studies, although sometimes criticised as to their ecological realism and problems with edge effects are, however, a fruitful route to identifying mechanisms behind the success or otherwise of invasions (see discussion in Dick and Platvoet 2000). The experimental derivation of the types and magnitudes of functional responses of resident predators towards invasive prey can offer a practical alternative to field based studies and still allow understanding and perhaps prediction of the population- and community-level outcomes of invasions in the face of resident predators.We encourage the use of functional response experiments more broadly in invasion biology, where hypotheses require tests of ecological impacts. Thus, for example, we have shown that the invasive ‘killer shrimp’ Dikerogammarus villosus has a higher functional response than native gammarid species, consistent with its field patterns of impact on prey (Bollache et al. 2008), while the magnitude of difference in the functional responses of invader and native mysids in the laboratory explained and predicted impacts on prey species in the field (Dick et al. 2013). This method has been used to test the enemy release hypothesis and discovered that, counter to the predictions of this hypothesis, parasitized invasive amphipods had higher functional responses than those unparasitized (Dick et al. 2010). Differential functional responses of natives and invaders towards juvenile heterospecifics have been used to explain invader/native coexistence in another amphipod system in N. America (Kestrup et al. 2011). In addition, some authors, whilst not using comparisons among species, still elucidate the impacts of invaders by deriving their functional responses (Hoof and Bollens 2004, Jones et al. 2011). It should also be noted that the functional response methodology is not restricted trophically or taxonomically, as all consumers utilise resources in quantifiable ways that can be assessed by their functional responses. Finally, methods of assessing functional responses are not restricted to small-scale experiments, since they can be measured in the field (Schenk and Bacher 2002, Smout et al. 2010) and by scat and gut contents analyses (Hoof and Bollens 2004, Middlemas et al. 2006, Jones et al. 2011), making the comparative functional response method widely applicable and achievable.

The success of invading species may be restricted if biotic resistance occurs as a result of predation by resident species. Support for this requires evidence of negative associations between invaders and resident species in the field, in addition to a strongly regulating or de-stabilising predator-prey interaction. We show that two resident predatory amphipods, one native and the other a previous introduction, have strong negative associations with an invasive amphipod prey in the field. Further, our experiments indicate that the resident species both exhibit potentially de-stabilising Type II functional responses towards the invasive prey, in both homo- and heterogeneous environments. Furthermore, however, the resident predator exhibiting the greater biotic resistance in the field also had the higher functional response in the laboratory. In addition, resident amphipod predation rates are considerably greater than the reproduction rate of the invader, suggesting biotic resistance is likely as predation can clearly outstrip reproduction. We therefore recommend the use of comparative functional response methodologies as an effective way of understanding, as well as potentially predicting, the success and failure of invasions and testing invasion ecology hypotheses.