Journal of Animal Ecology 2014

doi: 10.1111/1365-2656.12303

Linking niche theory to ecological impacts of successful invaders: insights from resource fluctuation-specialist herbivore interactions Cindy Gidoin1, Lionel Roques2 and Thomas Boivin1*
UR 629 Ecologie des Fore^ts Me
diterrane
ennes, INRA, F-84914 Avignon, France; and 2UR 546 Biostatistique et Processus Spatiaux, INRA, F-84914 Avignon, France 1

Summary 1. Theories of species coexistence and invasion ecology are fundamentally connected and provide a common theoretical framework for studying the mechanisms underlying successful invasions and their ecological impacts. Temporal fluctuations in resource availability and differences in life-history traits between invasive and resident species are considered as likely drivers of the dynamics of invaded communities. Current critical issues in invasion ecology thus relate to the extent to which such mechanisms influence coexistence between invasive and resident species and to the ability of resident species to persist in an invasive-dominated ecosystem. 2. We tested how a fluctuating resource, and species trait differences may explain and help predict long-term impacts of biological invasions in forest specialist insect communities. We used a simple invasion system comprising closely related invasive and resident seed-specialized wasps (Hymenoptera: Torymidae) competing for a well-known fluctuating resource and displaying divergent diapause, reproductive and phenological traits. 3. Based on extensive long-term field observations (1977–2010), we developed a combination of mechanistic and statistical models aiming to (i) obtain a realistic description of the population dynamics of these interacting species over time, and (ii) clarify the respective contributions of fluctuation-dependent and fluctuation-independent mechanisms to long-term impact of invasion on the population dynamics of the resident wasp species. 4. We showed that a fluctuation-dependent mechanism was unable to promote coexistence of the resident and invasive species. Earlier phenology of the invasive species was the main driver of invasion success, enabling the invader to exploit an empty niche. Phenology also had the greatest power to explain the long-term negative impact of the invasive on the resident species, through resource pre-emption. 5. This study provides strong support for the critical role of species differences in interspecific competition outcomes within animal communities. Our mechanistic-statistical approach disentangles the critical drivers of novel species assemblages resulting from intentional and nonintentional introductions of non-native species. Key-words: community theory, ecological specialization, interspecific competition, mechanistic– statistical modelling, Megastigmus

Introduction Communities invaded by alien invasive species experience new species assemblages leading to novel interspecific interactions with potentially strong effects on the dynam*Correspondence author. E-mail: [email protected]

ics of both communities and invaders. For instance, the outcome when an invasive species faces a resident competitor may be coexistence, competitive displacement of the resident species or extinction (Sakai et al. 2001; Reitz & Trumble 2002). Interestingly, the theories of invasion ecology and species coexistence are fundamentally connected (Melbourne et al. 2007). Invasions provide

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society

2 C. Gidoin, L. Roques & T. Boivin unplanned opportunities to investigate the fundamental ecological processes underlying interspecific competition and coexistence across large spatial and temporal scales (Sax et al. 2007). In turn, coexistence theory provides a powerful framework for the study of the processes underlying invasion success and impacts (Shea & Chesson 2002; Ricciardi et al. 2013). As used by Shea & Chesson (2002), ‘invasion success’ refers to the ability of a species to increase from low density, whereas ‘invasion impact’ refers to the effect of an invasive species, once established, on a community. Because invasions generally proceed in spatially and temporally heterogeneous environments, current critical issues in invasion ecology relate to the extent to which environmental heterogeneity influences coexistence between invasive and resident species and to the ability of resident species to persist in an invasive-dominated ecosystem (Melbourne et al. 2007; Ricciardi et al. 2013). Recent developments in invasion ecology suggest that environmental heterogeneity is likely to both favour invasibility (i.e. the susceptibility of the environment to invasion) and reduce invasion impacts by promoting coexistence mechanisms that cannot occur in homogeneous environments (Melbourne et al. 2007). A temporally fluctuating resource produces a type of heterogeneity during which episodes of increased resource availability are assumed to increase niche opportunities and relax interspecific competition, favouring both invasion success and species coexistence (Davis, Grime & Thompson 2000; Shea & Chesson 2002). Chesson (2000) showed that longterm coexistence in a fluctuating environment should involve fluctuation-dependent (e.g. storage effect and nonlinearity of competition) or fluctuation-independent (e.g. niche partitioning) mechanisms that each rely on species divergence in resource use. In particular, time partitioning of resource use is a likely driver of coexistence as diverging times of maximal resource use between species increases the access to a resource when a competitor is relatively inactive (Chesson et al. 2004). Such a coexistence mechanism can operate at different time-scales if species are phenologically asynchronous (within-year partitioning) or if they display different dormancy (or diapause) durations (between-year partitioning) (Venner et al. 2011). Fluctuation-dependent mechanisms have successfully explained the positive effects of a fluctuating resource on species coexistence and the absence of competitive displacement of a resident species by an invasive (Levine & Rees 2004; Descamps-Julien & Gonzalez 2005). In the absence of fluctuation-dependent mechanisms, the establishment and persistence of an invasive species depend on its ability to maintain a positive balance in population growth between periods of resource excess and suppression (Melbourne et al. 2007). Niche partitioning is an important fluctuation-independent mechanism that may increase invasion success and promote coexistence, provided that invasive and resident species differ in their resource use (Byers & Noonburg 2003). Niche-based

hypotheses to explain invasion success also rely on species differences that allow invaders to either access unused resources (empty niche hypothesis) or be more competitive in exploiting shared and limited resources (niche replacement hypothesis) (Ricciardi et al. 2013). In the latter case, niche overlap with a superior invasive competitor is likely to have strong direct effects on the resident species (e.g. competitive displacement). The goal of this study was to test how temporal fluctuations in resource availability and species trait differences may explain and help predict long-term impacts of biological invasions in communities. Ecologically and phylogenetically related invasive and resident species, brought successively into a common fluctuating environment, are ideal study systems for this purpose. Stronger interspecific competition is indeed expected among more closely related species than among more distantly related species due to a higher potential of niche overlap, but also when closely related species did not evolve strategies avoiding competition due a lack of co-evolutionary history (Reitz & Trumble 2002). Accordingly, we focused on the sequential invasion of French true cedar (Cedrus atlantica) forests by the two seed-specialized wasps, Megastigmus pinsapinis and Megastigmus schimitscheki (Hymenoptera: Torymidae). Wasps of the genus Megastigmus are highly invasive pre-dispersal seed predators, for which the global and largely unregulated seed trade has generated many recent and unique species assemblages in both urban and forest ecosystems (Auger-Rozenberg & Roques 2012). All Megastigmus species share the same cryptic life cycle, reflecting their strong specialization on seed resources (Roques & Skrzypczynska 2003). M. pinsapinis and M. schimitscheki exploit a fluctuating resource, as seed production in C. atlantica shows strong annual variation that is synchronized at the local scale (Krouchi, Derridj & Lefevre 2004). Sympatric populations of these two wasps share a narrow ecological niche, but display highly divergent phenologies (early vs. late), diapause strategies (short vs. prolonged) and modes of reproduction (sexuality vs. asexuality) (Boivin et al. 2008; Suez et al. 2013; Boivin et al. 2014), and from which one might expect divergences in resource use favouring invasion success and coexistence. In addition, in contrast to their respective native populations, these two species have neither other competitors for this seed resource nor indigenous or introduced specialist natural enemies in France (Fabre et al. 1999). Altogether, these features provide an interesting and simple invasion system comprising only two closely related species competing for a well-identified fluctuating resource. Here, we combine long-term data from an extensive survey of the dynamics of this system (1977– 2010) with mechanistic–statistical modelling. As such, we develop a particularly promising approach for addressing current critical issues in invasion impacts (Melbourne et al. 2007). We aimed in particular at (i) obtaining an accurate description of the population dynamics over time of these wasp species, which are interacting over the fluc-

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

Niche theory and invasion impacts competitor can exploit more efficiently the resource than its competitor in years of high abundance. This has been shown in a community of weevils (Coleoptera: Curculionidae) feeding on mast-seeding oak trees (Quercus sp.) (Venner et al. 2011). But from both modelling and biological perspectives, the dynamical results of our system rather describe a fairly straightforward scramble competition in favour of the invasive wasp. Thus, we postulate that our case study provides a strong support to the body of niche theory stating that species differences leading to ecological fitness differences (the relative average competitive ability of species, as used by Lankau 2011) can lead to competitive exclusion in invaded communities (Chesson 2000; MacDougall, Gilbert & Levine 2009). The temporal offsets observed between M. schimitscheki and M. pinsapinis were unlikely to favour distinct temporal niches allowing coexistence between both wasp species. While accessing the seeds earlier undoubtedly gives an advantage to the invasive, there is no clear evidence that the resident species gains greater access to the seeds later in the season, whatever the level of availability of the resource. The temporal offsets between wasps lead more likely to the resource pre-emption by M. schimitscheki, that is the exploitation of the resource before it becomes available to its competitor (Reitz & Trumble 2002). First, these highly seed-specialized wasps oviposit during a narrow period of development of cones, whose continuous development in both size and internal structure (e.g. intense lignification) rapidly limit oviposition possibilities (Rouault et al. 2004). Secondly, favourable oviposition sites represent a finite resource for wasp populations, as only a limited fraction of seeds can allow larval development (Boivin et al. 2008; T. Boivin, unpublished data). Moreover, when several eggs are laid within a seed, there is evidence for interference competition through larval cannibalism (Boivin et al. 2008), as generally observed when a larva has a larger body size than its competitor due to earlier phenology (Fincke 1999). The observed displacement of the resident wasp may thus be driven by a positive relationship between the increasing number of earlier competitors compared with the number of exploitable resources and the difficulty of accessing such resources later. This relationship was explicitly included in the competition function used in our modelling approach, which probably explained the interesting congruence between field data and model simulations. Consequently, we suggested that niche replacement through resource preemption is the main mechanism driving the negative impact of M. schimitscheki on M. pinsapinis (Shea & Chesson 2002). Much of the invasion literature links advantageous traits of invasive species to invasion success, but less is known regarding why and how such advantageous traits influence the impacts of invasions (Ricciardi et al. 2013). Species coexistence studies have primarily focused on a single mechanism, but multiple mechanisms are likely involved in community dynamics (Snyder, Borer &

9

Chesson 2005; Adler, Ellner & Levine 2010). In this study, we illustrate the benefit of accounting for multiple mechanisms to identify the main driver(s) of post-invasion coexistence or exclusion dynamics in communities. While the role of phenological difference between invasive and resident species in invasion ecological impact has been widely acknowledged in the plant invasion literature (Wolkovich & Cleland 2011), empirical evidence of similar processes within the animal kingdom remains scarce to date (Louda et al. 2011). Here, we provide one striking example in insects of the role that phenology can play in post-invasion competitive displacement at a relatively short timescale (i.e. in 10 generations) and despite a fluctuating resource environmental context. Some of the reliable predictors of competition in insect communities relate to shared ecological specialization on a discrete resource (Denno, McClure & Ott 1995; Reitz & Trumble 2002). By limiting the possibility of refugia, ecological specialization is a mechanism particularly prone to enhance species niche overlap and consequently to strengthen interspecific competition. However, coexistence between specialists with overlapping niches is possible when diverse ecological factors (e.g. antagonists, mutualists and host plant effects) can modulate the strength of competition, which is likely to occur when competitors share a co-evolutionary history (Smith, Mooney & Agrawal 2008). Yet, biological invasions generally produce novel species assemblages of specialists that share the same host but that did not co-evolve (Holway et al. 2002; Ness & Bronstein 2004). Our work suggests that this may dampen opportunities for such ecological competitive trade-offs and then favour competitive displacement rather than stable coexistence in specialist invaded communities. This prediction likely applies to many forest ecosystems invaded by insects (Kenis et al. 2009), but also to classical biological control programs for which rates of establishment of introduced natural enemies closely depend on interspecific interactions within the introduction area (Reitz & Trumble 2002). To conclude, our combination of extensive field observations with a mechanistic modelling approach allowed a better understanding of the life-history-based processes involved in successful invasions, as well as longer-term predictions of their ecological impacts. One further promising prospect is the consideration of the possible dynamic nature of such ecological systems, because temporal and spatial heterogeneity of environmental conditions are likely causes of context-dependent impacts of invasion (Ricciardi et al. 2013). Independently of the temporal heterogeneity (e.g. a temporally fluctuating resource), emerging disturbances (e.g. new competitors, predators and/or climate effects) may affect the population dynamics of invasive and resident species and differentially shape the outcome of their competition. In this regard, our modelling approach has the advantage of being flexible enough to further address what remains a key step in fully understanding how communities resist the impact of invasions

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

4 C. Gidoin, L. Roques & T. Boivin Table 1. Relevant characteristics of the invasion system used in this study Species History Native area Native host plant Introduction in France Current status in France Relevant diverging traits Diapause Phenology Reproduction (sex ratio)

Megastigmus pinsapinis

Megastigmus schimitscheki

North Africa Cedrus atlantica Early 20th century Resident, in decline

Middle East C. brevifolia/C. libani Early 1990s Invasive, in expansion

2-year (obligatory) Late spring Asexual (strongly biased towards females)

2- to 5-year (obligatory and prolonged) Early spring Sexual (balanced)

Table 2. Data types and their uses: (1) observations of seed infestation rates, referred to as SIRs, which included the global SIR (G*n;i ), SIR by Megastigmus schimitscheki (S*n;i ) and SIR by Megastigmus pinsapinis (P*n;i ), with G*n;i ¼ S*n;i þ P*n;i during the period n = 1977– 2010 at some sites i 2 I1=[1, 32] and (2) observations of cedar cone production (F*n;i ) during the period n = 2006–2012 at some sites i 2 I2 = [1, 27]. N: minimum–maximum yearly number of sampled seeds on which SIRs were estimated in the 32 sites (i in I1). Over the 32 sites (in I1) used to describe changes in wasp demography over time, 27 sites (set I2) are used for parameter estimation (seed production data are unavailable on the other 5 sites) and 11 sites (set I3) are used to model the variability of seed production (corresponding to sites where seed production has been monitored continuously from 2006 to 2012) Variable

Noted

Type

Uses

Year (n)

Site (i)

Global SIR (N: 9800
105) M. schimitscheki SIR

G*n;i S*n;i

Quantitative 2 [0, 1] Quantitative 2 [0, 1]

M. pinsapinis SIR

P*n;i

Quantitative 2 [0, 1]

Cedar cone production

F*n;i

Qualitative 5 modalities

Descriptive analysis Descriptive analysis Parameter estimation Descriptive analysis Parameter estimation Seed production model Parameter estimation

1977–2010 1994–2010 2001–2010 1977–2010 2001–2010 2006–2012 2006–2010

i i i i i i i

P5

gPk ¼ qP \1. This means that the rates gSk and gPk take into account the overall larval mortality rates during the 5 years following egg laying of M. schimitscheki and M. pinsapinis. These mortality rates take the values of (1
qS)=042 and (1
qP)=026 for M. schimitscheki and M. pinsapinis, respectively (see Appendix S2, Supporting information). Finally, we computed Sn and Pn: k¼1

8   S > S S Un > > < Sn ¼ f a s F ; n   S > U UP > S S > : Pn ¼ f a s n þ aP sP n
Sn ; Fn Fn

eqn 2

where aS and aP are the mean numbers of parasitized seeds per female, and sS and sP are the sex ratios at emergence. The function f is chosen so that it is an increasing function of the number of individuals satisfying the conditions: (i) f(x) is close to x whenever x is small, corresponding to an absence of competition when there is a small number of individuals compared with the quantity of resource, and (ii) f(x) is close to 1 whenever x is large, corresponding to a saturation of the resource when the number of individuals is large compared with the quantity of resource. We make the biological assumption that early-emerging individuals exploit the most easily available resource. This is consistent with the assumption that favourable oviposition sites are limiting for both wasp species on cedar cones. Such sites are mainly located in the middle height of a cone. Wasps indeed avoid ovipositing at both the apex and the base of a cone due to the physical constraints exerted by cone shape on seed development (Boivin et al. 2008; T. Boivin, unpublished data). As only one larva is yielded

2 2 2 2 2 2 2

I1 I1 I2 I1 I2 I3 I2

= = = = = = =

[1, [1, [1, [1, [1, [1, [1,

32] 32] 27] 32] 27] 11] 27]

per seed, our assumption implies that the higher the number of individuals compared with the number of seeds, the more difficult is the access to an exploitable resource. This is described by the 00 concavity of the function f (i.e. f < 0). Expression (2) specifies that M. pinsapinis can only exploit the part of the resource that remains following egg laying by M. schimitscheki, in accordance with the earlier emergence of M. schimitscheki (additional details in Appendix S2, Supporting information). In our computations, we took f(x) = x/(1 + x), which is one of the simplest functions satisfying the above conditions. A main advantage of the model is that the computation of the SIRs does not require knowledge of the value of Fn, only a relative notion of seed production, through Fn
k/Fn. Thus, without a loss of generality, Fn can be standardized to take on values in [0, 1].

Stochastic model of seed production For a realistic description of the invasion system, we built a stochastic model to describe the interannual fluctuation of cedar seed production. First, we assumed that the level of seed production in year n can be ‘mast’ (i.e. high), ‘intermediate’ or ‘nonmast’ (i.e. almost no cones). In C. atlantica, a mast year occurs every 3 or 4 years and is followed by a non-mast year (Krouchi, Derridj & Lefevre 2004). Thus, the interannual dynamic of seed production level was described by the following transition laws: mast ? non-mast (probability = 1), non-mast ? mast (probability = 05), non-mast ? intermediate (probability = 05), intermediate ? intermediate (probability = 05), intermediate ? mast (probability = 025) and intermediate ? non-mast (proba-

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

Niche theory and invasion impacts bility = 025). Secondly, Fn was drawn from an interval J⊂[0, 1] at a probability depending on the seed production level of year n (i.e. Fn is more likely to be drawn at an interval close to one during a mast year than during a non-mast year). These probabilities were estimated using field data (Table 2) and a logistic regression performed with the ‘VGAM’ package in R 2.15.2. For more details, see Appendix S1 (Supporting information).

Statistical model of the observations of SIRs To account for uncertainty in the field observations, we built a statistical model for the observations of S*n and P*n (the index i has been dropped here for the sake of clarity). The main assumption of this model is that the observed SIRs are drawn independently from normal zero-one inflated distributions with means given by the actual Sn and Pn. In other words, we had: S*n  N f0;1g ðSn ; r2 Þ and P*n  N f0;1g ðPn ; r2 Þ;

eqn 3

where N f0;1g ðl; r Þ corresponds to the normal zero-one inflated distribution with mean l and variance r2; see Appendix S3 (Supporting information). 2

Bayesian parameter estimation By fitting the mechanistic model and verifying the goodness-offit of the subsequent mechanistic–statistical model, we obtained a realistic model for the simulations. The estimation of the fertility parameters aS and aP was also independently of interest. These values had been measured in previous laboratory and field experiments (Boivin et al. 2008); however, they may vary greatly in real-world situations. Estimation of these parameters enabled us to compare the fertilities of M. schimitscheki and M. pinsapinis. The remaining parameters, which corresponded to the diapause rates and sex ratios at emergence, were known from previous laboratory experiments and are highly consistent among years and sites (Suez et al. 2013; Boivin et al. 2014). For each site, the statistical model described in (3) enabled us to compute the joint distribution of the observations   O* ¼ S*n ; P*n ; n ¼ 2006; . . .; 2010 ; conditioned on the actual SIRs, with the SIRs governed by the mechanistic model described by equations (2). The SIRs depended deterministically on the unknown parameters aS, and aP, on the seed production F between 2001 and 2010 and on the initial condition (S0,P0) corresponding to the SIRs during the first 5 years. Thus, the conditional distribution of the observation process was equal to the likelihood LO* ½aS ; aP ; F; ðS0 ; P0 Þ of the model parameterized by aS, aP, F, and (S0,P0). See Appendix S3 (Supporting information) for the computation of this likelihood. The prior distributions of the parameters p1(aS), p2(aP), p3(F), and p4(S0,P0) are detailed in Appendix S3 (Supporting information). The posterior distribution of the parameters was obtained using Bayes theorem: PðaS ; aP ; F; ðS0 ; P0 ÞjO* Þ / LðaS ; aP ; F; ðS0 ; P0 ÞÞp1 ðaS Þ p2 ðaP Þp3 ðFÞp4 ðS0 ; P0 Þ:

5

density functions for the pair (aS,aP) in all sites. We then computed a joint distribution Γ for the pair (aS,aP) among all study sites by taking the average of all posterior density functions (see Appendix S3, Supporting information for details).

realistic description of the invasion system and working hypotheses In a simulation study, we built a mechanistic approach on comparisons between a realistic description of the invasion system and four alternative working hypotheses, each differing by only one key characteristic of the system (resource supply, diapause, reproduction or phenology). This approach allowed us to investigate the effect of modifying these four characteristics on wasp dynamics and coexistence.

Realistic system It was assumed that (i) the resource is fluctuating (Fn is given by the above-mentioned stochastic model of seed production), (ii) the two species have divergent reproduction strategies; specifically, the sex ratios at emergence are sS = 1/2 and sP = 1, (iii) the diapause patterns differ such that gS2 =qS ¼ 0  843, gS3 =qS ¼ 0  014, gS4 =qS ¼ 0  143, gS5 =qS ¼ 0 and gP2 =qP ¼ 0  963, gP3 =qP ¼ 0  021, gP4 =qP ¼ 0  016, gP5 =qP ¼ 0 from data in Suez et al. (2013), and (iv) the seasonal emergence phenologies differ, with an earlier emergence of M. schimitscheki than M. pinsapinis and no overlap (Table 1). Note that, relative to M. pinsapinis, a larger proportion of M. schimitscheki enters a 4-year prolonged diapause, resulting in a higher mortality rate in M. schimitscheki (1
qS = 042) than in M. pinsapinis (1
qP = 026).

Constant resource hypothesis Assumption (i) is modified such that the seed production is constant over time (Fn = 1 during all years).

Similar reproduction hypothesis Assumption (ii) is modified such that both species share a similar asexual reproduction strategy (that of M. pinsapinis); that is sS = sP = 1.

Similar diapause hypothesis Assumption (iii) is modified such that both species share a similar diapause pattern with a low propensity to prolonged diapause (that of M. pinsapinis). Therefore, gS2 =qS ¼ gP2 =qP ¼ 0  963, gS3 =qS ¼ gP3 =qP ¼ 0  021, gS4 =qS ¼ gP4 =qP ¼ 0  016 and gS5 =qS ¼ gP5 =qP ¼ 0. In this case, the larval mortality rates of the two species are assumed to be identical and equal to the mortality rate of M. pinsapinis (1
qS=1
qP=026).

eqn 4

The posterior inference was performed by constructing a Markov chain with a stationary distribution that matches the posterior distribution using a classical Metropolis–Hastings algorithm. All sites were analysed separately, providing different joint posterior

Simultaneous seasonal emergences hypothesis Assumption (iv) is modified. To test the effect of differences in seasonal emergence phenologies, the mechanistic model described by equations (2) was modified. In the modified version, it was assumed that both species emerge simultaneously. At year n, the

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

0·5 0·3 0·0

where Gn = Sn + Pn is the global SIR (for more details, please see Appendix S2, Supporting information).

0·2

eqn 5

0·1

8   US UP > > > Gn ¼ f aS sS n þ aP sP n ; > > Fn Fn > > > < aS sS USn ; Sn ¼ Gn S S S > a s Un þ aP sP UPn > > > > P P P > a s Un > > : : Pn ¼ Gn S S S a s Un þ aP sP UPn

0·4

number of M schimitscheki and of M pinsapinis which emerge are still given by (1). However, as all the individuals enter into competition, we can assume that the SIR by each species is proportional to the ratios (number of eggs carried by one species)/(total number of eggs carried by both species), that is:

0·6

6 C. Gidoin, L. Roques & T. Boivin

Sensitivity analysis of the model outcomes We tested the sensitivity of the model outcomes to a range of alternative hypotheses (Appendix S5, Supporting information): (i) both wasp species share a similar sexual reproduction strategy (that of M. schimitscheki), (ii) both wasp species share a similar propensity to prolonged diapause (that of M. schimitscheki), and (iii) prolonged diapause is removed from the model for both wasp species. In all simulations, M. pinsapinis was introduced in the system during the first 4 years with SIRs equal to 005. M. schimitscheki was introduced later, during years n = 30 to 33, with SIRs equal to 0001. These simulation settings were based on the histories of the introductions of each species in France (Auger-Rozenberg et al. 2012). Two thousand simulations of the mechanistic model were carried out with different values of (aS,aP) drawn from the estimated distribution Γ and with the values of Fn either taken as constant or obtained from the above-mentioned stochastic model of seed production. The simulations were performed over a 100year period.

Results long-term field survey of wasp population dynamics During the period preceding the introduction of M. schimitscheki (1977–1993), the annual SIRs by M. pinsapinis were highly variable, but no detectable temporal trend could be observed (Fig. 1). Upon the invasion of M. schimitscheki (1994–2010), the annual SIRs of M. pinsapinis declined drastically in areas of sympatry, reaching the lowest value of 0002  0001 in 2009. From 2004 and beyond, the annual SIRs of M. pinsapinis did not exceed 003, suggesting that the declining populations of the resident were unable to recover their former population levels. Meanwhile, the annual SIRs of M. schimitscheki in areas of sympatry displayed an increasing trend between 1994 and 2003. From its introduction, the annual SIRs of M. schimitscheki were variable but displayed stationary pattern. In allopatry, the annual SIRs of M. pinsapinis

1977 1980 1983 1986

1994 1999 2002 2005 2008

Fig. 1. Temporal pattern of the mean seed infestation rates (  SE) by Megastigmus pinsapinis (MP) and Megastigmus schimitscheki (MS) from 1977 to 2010. The first detection of M. schimitscheki in 1994 is represented by the vertical dot-dash line.

varied (from 001  0005 to 021  006) but did not exhibit an obvious declining trend from 1994 to 2010. Finally, the annual SIRs of M. pinsapinis were significantly lower in the presence (i.e. areas of sympatry) than in the absence (i.e. areas of allopatry) of M. schimitscheki (F = 1072, d.f. = 1, P < 0001, ANOVA on arcsine square root transformed data).

parameter distributions and model fit The distribution Γ of the fertility parameters aS and aP clearly differs from the uniform prior distribution (see Appendix S3, Supporting information). This indicates that the observations carried information about the distribution of these parameters. The posterior medians of the parameters aS and aP are 57 and 28, respectively. The median number of females produced by a female M. pinsapinis, 28sP = 28, is therefore very similar to that produced by a female M. schimitscheki, 57sS = 285. Based on these fertility parameters alone, no clear advantage can be demonstrated for either species. More details of the Bayesian parameter estimation are provided in Appendix S3 (Supporting information). To assess the model’s goodness-of-fit, 95% rectangular confidence regions were computed for the pair of SIRs (Sn,Pn) for each year and site. The confidence regions were computed by simulating N = 105 times the observation process, at each site, on the computed SIRs Sn,i and Pn,i (n = 2006,. . .,2010) corresponding to the output of the mechanistic model using the parameters that maximized the posterior likelihood. All of the observations (102/102) fell within the confidence regions, indicating that the model fits the data well.

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

Niche theory and invasion impacts

7

Fluctuating resource constraints on wasp population dynamics

model predictions The outputs of the mechanistic–statistical model are presented in two different ways. First, the changes over time of the annual median and the interquartile range of the SIRs of M. pinsapinis and M. schimitscheki are presented (Fig. 2a). Secondly, the proportion of simulations where Sn/(Sn+Pn) < 10
6, which were assumed to reflect the exclusion of M. schimitscheki by M. pinsapinis; and the proportion of simulations where Sn/(Sn+Pn) > 1
10
6, which were assumed to reflect the exclusion of M. pinsapinis by M. schimitscheki are presented for each year (Fig. 2b). The complementary proportion reflects coexistence (Fig. 2b).

Under the hypothesis of a constant resource supply for both wasp species, there were noticeable increases in SIRs for both M. pinsapinis (by 75%, before M. schimitscheki introduction) and M. schimitscheki (by 100%), relative to the SIRs obtained from the realistic description of the system (Fig. 2a). This result suggests that the fluctuating resource constrained the population dynamics of both wasp species. Additionally, the exclusion/coexistence probabilities did not differ significantly between the realistic description and the constant resource hypothesis (Fig. 2b).

Realistic description of the invasion system When alone in the system (n 2 [0, 30]), M. pinsapinis exhibited a rapid increase of the SIRs that stabilized at a median approximately equal to 030 (Fig. 2a). Accordingly, the exclusion probability of the invasive species was equal to 1 (Fig. 2b). The progressive invasion of the system by M. schimitscheki resulted in a rapid and drastic decline in the SIRs of its resident competitor (Fig. 2a). In the same period, the exclusion probabilities of the resident by the invasive strongly increase over time (Fig. 2b). These results indicate coexistence strongly dominated by the invasive species, M. schimitscheki.

Seed infestation rate

Under the hypothesis of asexual reproduction in M. schimitscheki, the SIRs of the invasive species increased by 200% while those of the resident decreased more rapidly and to a greater extent than in the realistic description of the system (Fig. 2a). There was a strong increase in the probability that M. schimitscheki excludes M. pinsapinis (+50%, Fig. 2b). This suggests that the invasive may have a stronger impact on the resident if it has a reproductive strategy similar to that of its resident competitor. Under

Realistic

(a)

Resource

Diapause

Phenology

1

1

1

1

0·8

0·8

0·8

0·8

0·8

0·6

0·6

0·6

0·6

0·6

0·4

0·4

0·4

0·4

0·4

0·2

0·2

0·2

0·2

0·2

0

50

100

0

0

50

100

0

0

t

t (b)

Reproduction

1

0

Exclusion/coexistence probability

Fig. 2. Model-based analysis of the intrinsic effects of environmental and life-history features on the long-term coexistence dynamics of the resident Megastigmus pinsapinis (in blue) and the invasive Megastigmus schimitscheki (in red) seed wasps. Model output comparisons were conducted between a realistic description of the invasion system and each of four working hypotheses, each varying in only one characteristic of the system (i.e. resource, reproduction, diapause or phenology). (a) distribution of simulated seed infestation rates (SIRs) by M. pinsapinis and by M. schimitscheki between the first and the third quartiles (coloured envelopes); black dotted and solid lines show the median SIR of M. pinsapinis and M. schimitscheki, respectively. (b) probabilities of competitive exclusion of M. pinsapinis and M. schimitscheki (in red and blue, respectively) and of coexistence (in purple) in the system.

Reproduction and diapause constraints on wasp population dynamics

50

100

0

0

t

50

100

0

0

t

1

1

1

1

0·8

0·8

0·8

0·8

0·8

0·6

0·6

0·6

0·6

0·6

0·4

0·4

0·4

0·4

0·4

0·2

0·2

0·2

0·2

0·2

0

50

t

100

0

0

50

t

100

0

0

50

100

0

0

t

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

50

t

100

t

1

0

50

100

0

0

50

t

100

8 C. Gidoin, L. Roques & T. Boivin the hypothesis of sexual reproduction in M. pinsapinis, we found an increased exclusion probability of the resident by the invasive (Fig. S1, Appendix S5, Supporting information). This was primarily due to the lower SIRs by M. pinsapinis, suggesting that asexual reproduction may confer a demographic advantage to this species and may slow down the negative impact of M. schimitscheki. Under the hypothesis of a reduced prolonged diapause in M. schimitscheki, SIRs in this species increased by 70% (Fig. 2a) and its establishment in the system was more rapid than in the realistic description. There was an increase in the intensity of dominance of the invasive over the resident wasp, as depicted by the higher exclusion probabilities of M. pinsapinis (Fig. 2b). These results suggest a constraining effect of prolonged diapause on the population dynamics of M. schimitscheki, which results from the strong mortality cost of prolonged diapause in the model and which can affect both species similarly (Fig. S2, Appendix S5, Supporting information). Moreover, removing prolonged diapause from the model in both wasp species did not result in a strong decrease in their SIRs or a deterministic extinction of M. pinsapinis in a fluctuating environment (Fig. S2, Appendix S5, Supporting information). This may reflect the characteristics of our system, in which a 2-year obligatory diapause prevents the collapse of wasp populations as two consecutive non-mast years are unlikely to occur (see Appendix S1, Supporting information). A key role of earlier phenology in both invasion success and impact Under the hypothesis of similar phenologies between wasp species, the median SIR by M. schimitscheki approached 0, whereas that of M. pinsapinis remained positive (Fig. 2a). These results represent a pattern opposite to that obtained from the realistic description of the system. Interestingly, similar phenologies led to the exclusive and particularly strong dominance of the resident on the invasive. Indeed, there was a strong increase in the probability that M. pinsapinis excludes M. schimitscheki (+150%, Fig. 2b). The probabilities of both coexistence and exclusion of the resident in the system were noticeably reduced (
10% and
40%, respectively, Fig. 2b).

Discussion One central finding of our study is that the early phenology of M. schimitscheki was the main driver of its successful invasion, rapid population growth and numerical dominance over M. pinsapinis. Based on niche theory, the mechanism by which the invasion of M. schimitscheki was promoted through phenology is likely empty niche exploitation (Shea & Chesson 2002). As cedar seeds are as yet unused by M. pinsapinis upon the adult emergence and oviposition of M. schimitscheki, this resource availability

may be equivalent to an empty niche for M. schimitscheki and may have favoured its establishment. This scenario was strongly supported by the ‘Phenology’ simulations of the model (Fig. 2a), which demonstrated that, in case of simultaneous emergences with its competitor, M. schimitscheki was generally unable to establish sustainably in the system and M. pinsapinis remained dominant. Moreover, using these simulations, we found that a population of M. schimitscheki can grow from low density and invade the system only if the ratio between the M. pinsapinis population level and the seed production level is low (Appendix S4, Supporting information). This condition and the high SIRs by M. pinsapinis reached before M. schimitscheki introduction explained the high invasion failure probability of M. schimitscheki in case of simultaneous emergence with its competitor. This study provides strong evidence that species trait differences favouring empty niche exploitation can be important drivers of invasion success. We suggest that such a mechanism may especially operate by allowing population growth when abundance is low, as is commonly observed at the early stage of an invasion or when new favourable sites are colonized during post-establishment expansion (Sakai et al. 2001). Our computations (Appendix S4, Supporting information) demonstrate that a population of M. schimitscheki can grow from low density as soon as its fertility is sufficiently high (aS > 29), independently of the resource level and of the population level of M. pinsapinis. Interestingly, this shows the absence of an Allee effect for M. schimitscheki (i.e. fertility reaches its maximum at low population density), which likely contributed to its successful establishment and spread (Yamanaka & Liebhold 2009). Although invasion success is generally thought to be positively correlated with the number of founders, this result provides additional support for alternative genetic and ecological mechanisms allowing invaders to circumvent the cost of severe founder effects at introduction (Tsutsui et al. 2000; Auger-Rozenberg et al., 2012). Our case study may also illustrate how empty niche exploitation can operate in favour of enemy escape opportunities provided by the invaded environment (sensu Shea & Chesson 2002), as no indigenous or introduced specialist natural enemies of Megastigmus species have been detected in French cedar forests so far (T. Boivin, unpublished data). Determining the extent to which temporal environmental heterogeneity can influence the long-term impact of an ongoing invasion was another key issue in this study. We showed that a fluctuating seed resource could not promote coexistence between the resident and the invasive wasps, despite possibly important discrepancies in their use of resource at both within- and between-year scales (phenology and diapause duration, respectively). According to niche theory, time partitioning of the use of a fluctuating resource can be a key of the coexistence of competing species (Chesson et al. 2004), but only if there is an advantage for a species to be earlier than a competitor in years of limiting resource, and if the later

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

Niche theory and invasion impacts competitor can exploit more efficiently the resource than its competitor in years of high abundance. This has been shown in a community of weevils (Coleoptera: Curculionidae) feeding on mast-seeding oak trees (Quercus sp.) (Venner et al. 2011). But from both modelling and biological perspectives, the dynamical results of our system rather describe a fairly straightforward scramble competition in favour of the invasive wasp. Thus, we postulate that our case study provides a strong support to the body of niche theory stating that species differences leading to ecological fitness differences (the relative average competitive ability of species, as used by Lankau 2011) can lead to competitive exclusion in invaded communities (Chesson 2000; MacDougall, Gilbert & Levine 2009). The temporal offsets observed between M. schimitscheki and M. pinsapinis were unlikely to favour distinct temporal niches allowing coexistence between both wasp species. While accessing the seeds earlier undoubtedly gives an advantage to the invasive, there is no clear evidence that the resident species gains greater access to the seeds later in the season, whatever the level of availability of the resource. The temporal offsets between wasps lead more likely to the resource pre-emption by M. schimitscheki, that is the exploitation of the resource before it becomes available to its competitor (Reitz & Trumble 2002). First, these highly seed-specialized wasps oviposit during a narrow period of development of cones, whose continuous development in both size and internal structure (e.g. intense lignification) rapidly limit oviposition possibilities (Rouault et al. 2004). Secondly, favourable oviposition sites represent a finite resource for wasp populations, as only a limited fraction of seeds can allow larval development (Boivin et al. 2008; T. Boivin, unpublished data). Moreover, when several eggs are laid within a seed, there is evidence for interference competition through larval cannibalism (Boivin et al. 2008), as generally observed when a larva has a larger body size than its competitor due to earlier phenology (Fincke 1999). The observed displacement of the resident wasp may thus be driven by a positive relationship between the increasing number of earlier competitors compared with the number of exploitable resources and the difficulty of accessing such resources later. This relationship was explicitly included in the competition function used in our modelling approach, which probably explained the interesting congruence between field data and model simulations. Consequently, we suggested that niche replacement through resource preemption is the main mechanism driving the negative impact of M. schimitscheki on M. pinsapinis (Shea & Chesson 2002). Much of the invasion literature links advantageous traits of invasive species to invasion success, but less is known regarding why and how such advantageous traits influence the impacts of invasions (Ricciardi et al. 2013). Species coexistence studies have primarily focused on a single mechanism, but multiple mechanisms are likely involved in community dynamics (Snyder, Borer &

9

Chesson 2005; Adler, Ellner & Levine 2010). In this study, we illustrate the benefit of accounting for multiple mechanisms to identify the main driver(s) of post-invasion coexistence or exclusion dynamics in communities. While the role of phenological difference between invasive and resident species in invasion ecological impact has been widely acknowledged in the plant invasion literature (Wolkovich & Cleland 2011), empirical evidence of similar processes within the animal kingdom remains scarce to date (Louda et al. 2011). Here, we provide one striking example in insects of the role that phenology can play in post-invasion competitive displacement at a relatively short timescale (i.e. in 10 generations) and despite a fluctuating resource environmental context. Some of the reliable predictors of competition in insect communities relate to shared ecological specialization on a discrete resource (Denno, McClure & Ott 1995; Reitz & Trumble 2002). By limiting the possibility of refugia, ecological specialization is a mechanism particularly prone to enhance species niche overlap and consequently to strengthen interspecific competition. However, coexistence between specialists with overlapping niches is possible when diverse ecological factors (e.g. antagonists, mutualists and host plant effects) can modulate the strength of competition, which is likely to occur when competitors share a co-evolutionary history (Smith, Mooney & Agrawal 2008). Yet, biological invasions generally produce novel species assemblages of specialists that share the same host but that did not co-evolve (Holway et al. 2002; Ness & Bronstein 2004). Our work suggests that this may dampen opportunities for such ecological competitive trade-offs and then favour competitive displacement rather than stable coexistence in specialist invaded communities. This prediction likely applies to many forest ecosystems invaded by insects (Kenis et al. 2009), but also to classical biological control programs for which rates of establishment of introduced natural enemies closely depend on interspecific interactions within the introduction area (Reitz & Trumble 2002). To conclude, our combination of extensive field observations with a mechanistic modelling approach allowed a better understanding of the life-history-based processes involved in successful invasions, as well as longer-term predictions of their ecological impacts. One further promising prospect is the consideration of the possible dynamic nature of such ecological systems, because temporal and spatial heterogeneity of environmental conditions are likely causes of context-dependent impacts of invasion (Ricciardi et al. 2013). Independently of the temporal heterogeneity (e.g. a temporally fluctuating resource), emerging disturbances (e.g. new competitors, predators and/or climate effects) may affect the population dynamics of invasive and resident species and differentially shape the outcome of their competition. In this regard, our modelling approach has the advantage of being flexible enough to further address what remains a key step in fully understanding how communities resist the impact of invasions

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

10 C. Gidoin, L. Roques & T. Boivin and to develop effective control strategies against invasive species (Levine et al. 2003).

Acknowledgements We are grateful to Vincent Garreta, Etienne Klein, Samuel Soubeyrand for valuable discussions on mechanistic–statistical modelling, Ricco Rakotomalala for his help in logistic regression development, and Doyle McKey and two anonymous reviewers for insightful comments on a previous version of this manuscript. We acknowledge Alain Chalon and Marion Sondo for their help in wasp sampling and the observations of cedar seed production. This work was part of the ANR-10-INTB-1705-04-MACBI project funded by the French Agence Nationale pour la Recherche (http://www. agence-nationale-recherche.fr).

Data accessibility Spatio-temporal data sets of both cedar cone production and cedar seed infestation rates by Megastigmus spp. are available from the Dryad Digital Repository: http://doi.org/10.5061/dryad.54p64 (Gidoin, Roques & Boivin 2014).

References Adler, P., Ellner, S. & Levine, J. (2010) Coexistence of perennial plants: an embarrassment of niches. Ecology Letters, 13, 1019–1029. Auger-Rozenberg, M.A. & Roques, A. (2012) Seed wasp invasions promoted by unregulated seed trade affect vegetal and animal biodiversity. Integrative Zoology, 7, 228–246. Auger-Rozenberg, M.A., Boivin, T., Magnoux, E., Courtin, C., Roques, A. & Kerdelhue, C. (2012) Inferences on population history of a seed chalcid wasp: invasion success despite a severe founder effect from an unexpected source population. Molecular Ecology, 21, 6086–6103. Boivin, T. & Candau, J.N. (2007) Flow cytometric analysis of ploidy levels in two seed-infesting Megastigmus species: applications to sex ratio and species determination at the larval stage. Entomologia Experimentalis et Applicata, 124, 125–131. Boivin, T., Rouault, G., Chalon, A. & Candau, J.N. (2008) Differences in life history strategies between an invasive and a competing resident seed predator. Biological Invasions, 10, 1013–1025. Boivin, T., Henri, H., Vavre, F., Gidoin, C., Veber, P., Candau, J.N. et al. (2014) Epidemiology of asexuality induced by the endosymbiotic Wolbachia across phytophagous wasp species: host plant specialization matters. Molecular Ecology, 23, 2362–2375. Byers, J. & Noonburg, E. (2003) Scale dependent effects of biotic resistance to biological invasion. Ecology, 84, 1428–1433. Chesson, P. (2000) Mechanisms of maintenance of species diversity. Annual Review of Ecology and Systematics, 31, 343–366. Chesson, P., Gebauer, R., Schwinning, S., Huntly, N., Wiegand, K., Ernest, M. et al. (2004) Resource pulses, species interactions, and diversity maintenance in arid and semi-arid environments. Oecologia, 141, 236– 253. Davis, M., Grime, J. & Thompson, K. (2000) Fluctuating resources in plant communities: a general theory of invasibility. African Journal of Ecology, 88, 528–534. Denno, R., McClure, M. & Ott, J. (1995) Interspecific interactions in phytophagous insects: competition reexamined and resurrected. Annual Review of Entomology, 40, 297–331. Descamps-Julien, B. & Gonzalez, A. (2005) Stable coexistence in a fluctuating environment: an experimental demonstration. Ecology, 86, 2815– 2824. Eremko, R., Edwards, D. & Wallinger, D. (1989) A Guide to Collecting Cones of British Columbia Conifers. British Columbia Ministry of Forests, FRDA Report 055, Victoria, Canada. Fabre, J., Mouna, M., Du Merle, P. & Benhalima, S. (1999) Le point sur certains ravageurs du Cedre de l’Atlas en Afrique du Nord, en France et en Europe. For^ et M
editerran
eenne, 20, 203–221. Fincke, O. (1999) Organization of predator assemblages in Neotropical tree holes: effects of abiotic factors and priority. Ecological Entomology, 24, 13–23. Gidoin, C., Roques, L. & Boivin, T. (2014) Data from: Linking niche theory to ecological impacts of successful invaders: insights from resource

fluctuation-specialist herbivore interactions. Dryad Digital Repository. doi: doi.org/10.5061/dryad.54p64. Holway, D., Lach, L., Suarez, A., Tsutsui, N. & Case, T. (2002) The causes and consequences of ant invasions. Annual Review of Ecology and Systematics, 33, 181–233. Kenis, M., Auger-Rozenberg, M.A., Roques, A., Timms, L., Pere, C., Cock, M.W. et al. (2009) Ecological effects of invasive alien insects. Biological Invasions, 11, 21–45. Krouchi, F., Derridj, A. & Lefevre, F. (2004) Year and tree effect on reproductive organisation of Cedrus atlantica in a natural forest. Forest Ecology and Management, 197, 181–189. Lankau, R. (2011) Rapid evolutionary change and the coexistence of species. Annual Review of Ecology Evolution and Systematics, 42, 335–354. Levine, J. & Rees, M. (2004) Effects of temporal variability on rare plant persistence in annual systems. The American Naturalist, 164, 350–363. Levine, J., Vila, M., D’Antonio, C., Dukes, J., Grigulis, K. & Lavorel, S. (2003) Mechanisms underlying the impacts of exotic plant invasions. Proceedings of the Royal Society of London. Series B: Biological Sciences, 270, 775–781. Louda, S., Rand, T., Kula, A., Arnett, A., West, N. & Tenhumberg, B. (2011) Priority resource access mediates competitive intensity between an invasive weevil and native floral herbivores. Biological Invasions, 13, 2233–2248. MacDougall, A., Gilbert, B. & Levine, J. (2009) Plant invasions and the niche. African Journal of Ecology, 97, 609–615. Melbourne, B., Cornell, H., Davies, K., Dugaw, C., Elmendorf, S., Freestone, A. et al. (2007) Invasion in a heterogeneous world: resistance, coexistence or hostile takeover? Ecology Letters, 10, 77–94. Ness, J. & Bronstein, J. (2004) The effects of invasive ants on prospective ant mutualists. Biological Invasions, 6, 445–461. Reitz, S. & Trumble, J. (2002) Competitive displacement among Insects and Arachnids. Annual Review of Entomology, 47, 435–465. Ricciardi, A., Hoopes, M., Marchetti, M. & Lockwood, J. (2013) Progress toward understanding the ecological impacts of nonnative species. Ecological Monographs, 83, 263–282. Roques, A. & Skrzypczynska, M. (2003) Seed-infesting chalcids of the genus Megastigmus Dalman, 1820 (Hymenoptera : Torymidae) native and introduced to the West Palearctic region: taxonomy, host specificity and distribution. Journal of Natural History, 37, 127–238. Rouault, G., Turgeon, J., Candau, J.N., Roques, A. & von Aderkas, P. (2004) Oviposition strategies of conifer seed chalcids in relation to host phenology. Naturwissenschaften, 91, 472–480. Sakai, A., Allendorf, F., Holt, J., Lodge, D., Molofsky, J., With, K. et al. (2001) The population biology of invasive species. Annual Review of Ecology and Systematics, 32, 305–332. Sax, D., Stachowicz, J., Brown, J., Bruno, J., Dawson, M., Gaines, S. et al. (2007) Ecological and evolutionary insights from species invasions. Trends in Ecology & Evolution, 22, 465–471. Shea, K. & Chesson, P. (2002) Community ecology theory as a framework for biological invasions. Trends in Ecology & Evolution, 17, 170–176. Smith, R., Mooney, K. & Agrawal, A. (2008) Coexistence of three specialist aphids on common milkweed, Asclepias syriaca. Ecology, 89, 2187–2196. Snyder, R., Borer, E. & Chesson, P. (2005) Examining the relative importance of spatial and nonspatial coexistence mechanisms. The American Naturalist, 166, 75–94. Suez, M., Gidoin, C., Lefevre, F., Candau, J.N., Chalon, A. & Boivin, T. (2013) Temporal population genetics of time travelling insects: a long term study in a seed-specialized wasp. PLoS One, 8, e70818 Toth, J. (1978) Contribution a l’
etude de la fructification et de la r
eg
en
eration naturelle du c edre de l’Atlas (Cedrus atlantica Manetti) dans le sud de la France. PhD thesis, Facult
e des Sciences et Techniques de Marseille Saint-J
erome. Tsutsui, N., Suarez, A., Holway, D. & Case, T. (2000) Reduced genetic variation and the success of an invasive species. Proceedings of the National Academy of Sciences of the United States of America, 97, 5948–5953. Venner, S., Pelisson, P.F., Bel-Venner, M.C., Debias, F., Rajon, E. & Menu, F. (2011) Coexistence of insect species competing for a pulsed resource: toward a unified theory of biodiversity in fluctuating environments. PLoS One, 6, e18039 Wolkovich, E. & Cleland, E. (2011) The phenology of plant invasions: a community ecology perspective. Frontiers in Ecology and the Environment, 9, 287–294.

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology

Niche theory and invasion impacts Yamanaka, T. & Liebhold, A.M. (2009) Mate-location failure, the Allee effect, and the establishment of invading populations. Population Ecology, 51, 337–340.

11

Appendix S3. Methodological details for the statistical model and the Bayesian estimation of parameters. Appendix S4. Per capita growth rates.

Received 4 February 2014; accepted 25 September 2014 Handling Editor: Priyanga Amarasekare

Supporting Information Additional Supporting Information may be found in the online version of this article. Appendix S1. A stochastic model of seed production based on field data. Appendix S2. Construction of the mechanistic model of the seed infestation rates.

Appendix S5. Sensitivity analysis of the model outcomes to alternative hypotheses. Fig. S1. Model-based analysis of the effect of reproductive strategy on the long-term coexistence dynamics of the resident (Megastigmus pinsapinis, in blue) and invasive (Megastigmus schimitscheki, in red) seed wasps. Fig. S2. Model-based analysis of the effect of prolonged diapause on the long-term coexistence dynamics of the resident (Megastigmus insapinis, in blue) and invasive (Megastigmus schimitscheki, in red) seed wasps.

© 2014 The Authors. Journal of Animal Ecology © 2014 British Ecological Society, Journal of Animal Ecology