SEARCH

SEARCH BY CITATION

Keywords:

  • adaptive evolution;
  • admixture;
  • biocontrol;
  • biological invasion;
  • Europe;
  • harlequin ladybird;
  • Harmonia axyridis;
  • Hybridization;
  • life history;
  • phenotype

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Hybridization can fuel evolutionary processes during biological invasions. The harlequin ladybird Harmonia axyridis has long been used as a biocontrol agent before the species became invasive worldwide. Previous analysis based on microsatellite data has shown that European invasive populations bear traces of admixture between an eastern North American source, which is at the origin of the worldwide invasion, and biocontrol strains used in Europe. In this study, we tested the hypothesis that this early admixture event may have fostered the European invasion by impacting on the phenotypes of wild European populations. Mean life history traits of experimental F1 hybrids are compared with pure parental sources and wild European crosses. Our results reveal a biased impact whereby North American beetles benefitted from being admixed with European biocontrol strains. Resemblance between experimental hybrids and wild European invasive crosses further suggests a long-lasting effect of admixture that may still be at work and fostering invasiveness.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Biological invasions offer prime examples of rapid, contemporary adaptive evolution (e.g. Reznick & Ghalambor, 2001; Lee, 2002; Facon et al., 2006; Carroll et al., 2007; Dlugosch & Parker, 2008; Prentis et al., 2008; Suarez & Tsutsui, 2008). In the introduced range, new selective regimes can cause genetically based shifts in phenotypes that provide a greater fitness in the new environment. Examples are quickly accumulating and include changes in tolerance to the abiotic environment and/or in major life history traits (e.g. Lee et al., 2003; Bohn et al., 2004; Bossdorf et al., 2005; Xu et al., 2009).

Hybridization is one way to foster such adaptive evolution during invasion (Ellstrand & Schierenbeck, 2000; Rieseberg et al., 2007; Schierenbeck & Ellstrand, 2009). Interspecific hybridization leads to new allelic composition, and evolutionary novelties may become fixed in allopolyploids or clonally reproducing lineages (e.g. Thompson, 1991; Abbott et al., 2003). At the intraspecific level, multiple introductions and admixture of genetically differentiated source populations increase genetic diversity and often result in novel genotypes in invasive populations (e.g. Kolbe et al., 2004; Darling et al., 2008). Hybrid vigour may favour heterozygotes in early generations (Lynch & Walsh, 1998) and change mean population phenotypes. Importantly, admixture may also increase evolutionary potential when higher genetic variance, involving novel, recombinant and potentially fitter phenotypes, translate in heritable phenotypic variation, hence facilitating prolonged response to selection. Through these processes, invasive hybrid populations may outperform parental sources, strongly indicating that admixture can promote invasiveness (Facon et al., 2005; Lavergne & Molofsky, 2007; Kolbe et al., 2007; Facon et al., 2008; Keller & Taylor, 2010; but see Wolfe et al., 2007). Certainly, hybridization and admixture may also have negative effects by disrupting coadapted gene complexes and weakening local adaptations (e.g. Barton & Hewitt, 1985; Keller et al., 2000, Burke & Arnold, 2001; Bailey & McCauley, 2006).

Demonstrating that admixture resulted in adaptive evolution and so enabled a species to become invasive is not an easy task. It first requires that the identity of the ancestral populations at the origin of admixed invasive population be known (Keller & Taylor, 2008; Estoup & Guillemaud, 2010). Second, differences between derived and parental populations must confer higher fitness to admixed individuals and should not be because of chance events (Wolfe et al., 2007; Xu et al., 2009). Ideally, the fitness advantages should be matched with the new selective challenge imposed by the new environment in the introduced range and/or their impact on population growth, survival and expansion should be quantified.

Native to Asia, the coccinellid Harmonia axyridis (Pallas) (HA) has been introduced repeatedly in North America as a biocontrol agent against aphids since 1916 (Tedders & Schaefer, 1994; Krafsur et al., 1997) and in Europe and South America since 1980s (Ongagna et al., 1993; Poutsma et al., 2008). These biocontrol strains were developed from small samples originating from various regions of the vast native area. Despite recurrent intentional releases, the species did not establish for decades. However, for unknown reasons, it recently and suddenly became invasive in eastern and western North America in 1988 and 1991 (USA, Chapin & Brou, 1991; LaMana & Miller, 1996), Europe in 2001 (Belgium, Adriaens et al., 2003), South America in 2001 (Argentina, Saini, 2004) and Africa in 2004 (South Africa, Stals & Prinsloo, 2007). The species has spread widely in these areas where it consumes nontarget arthropods, invades households and is a pest of fruit production (Koch, 2003; Koch & Galvan, 2008).

On the basis of analysis of neutral genetic variation, Lombaert et al. (2010) recently retraced the routes of all five worldwide HA invasions. Eastern and western North American invasive populations originate from two independent introductions from the native Asian range. Surprisingly, eastern North America is the source of colonists for all other successfully invaded areas. In South America and South Africa, invasive populations bear no trace of genetic admixture with other sources. In Europe, however, there is clear evidence of admixture between eastern North American and the local biocontrol strain (with a contribution of biocontrol genes estimated at 43%, 95% CI: 18–83%; Lombaert et al., 2010). The admixture scenario in Europe is strongly supported by quantitative comparisons with alternative invasion scenarios not involving admixture. Moreover, the microsatellite allele distribution in the European invasive population taken as reference sample (Gent, Belgium) is better explained by invoking contributions from both eastern North American populations and biocontrol strain; at several loci, the few European biocontrol strain alleles, of which some are not observed in America, are overrepresented and co-occur with alleles common in America.

Given the success of the colonists from eastern North America at invading several remote areas, parsimony suggests that the most important evolutionary shift enabling HA invasion has occurred in eastern North America following the introduction from the native range. The nature of this shift remains unknown. Moreover, it appears that admixture may not be necessary for invasiveness to develop in other areas colonized by eastern North American propagules. Indeed, there are no traces of admixture in South America, and HA was never used for biocontrol in South Africa prior to the recent invasion. Nevertheless, the admixture between eastern North American HA and the European biocontrol strain evidenced in Lombaert et al. (2010) may have played a role in impeding or facilitating the first outburst in Europe. The positive influence of admixture is classically associated with heterosis, i.e. admixed individuals display higher fitness than the mean of parental sources. In the context of the European invasion by HA, however, we are especially interested by the positive or negative consequences of admixture on the fitness of American propagules.

This study hence follows from our previous knowledge of the global H. axyridis invasion routes indicating that invasive European HA derive from the admixture between wild invasive populations from eastern North America and the European biocontrol strain (Lombaert et al., 2010). We specifically tested the hypothesis that this admixture event affected HA life history traits early during the European invasion. Experimental crosses between biocontrol and American harlequin ladybird were performed to obtain admixed individuals. The impact on mean life history trait values was examined in the first hybrid generation and reveal that American HA benefited from admixture. Also, phenotypic resemblance between experimentally admixed and wild invasive European HA offers support for a long-term impact of admixture in the invasion process in Europe.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Experimental procedures

We used invasive populations from eastern North America and European biocontrol strains as the two parental types (American, Biocontrol). One American population was sampled in the late summer of 2009 in Quebec City, Quebec, Canada (hereafter ‘Q’). The other was sampled in October 2007 in Brookings, South Dakota, USA (hereafter ‘D’) and was kept in the laboratory for two generations before this experiment started in the fall of 2009. South Dakota is located in central North America, but close monitoring of the spatial expansion of the invasion as well as microsatellite data (E. Lombaert & A. Estoup, unpublished data) indicate that H. axyridis from this state derived from the eastern North American invasion originating from Louisiana (Koch et al., 2006). In Europe, three commercial H. axyridis strains have been used for biocontrol, all being derived from the INRA strain (Institut National de Recherche Agronomique, France) first released in France in 1982. Microsatellite genotyping of HA samples collected in these biocontrol strains confirmed that they are derived from the original INRA strain (unpublished results). Here, we used two strains that were in use in Europe when the first invasive population was reported in 2001 (Adriaens et al., 2003); the third strain no longer exists. The first strain originates from the company Biobest NV (hereafter ‘B’) and was maintained in a laboratory at Ghent University (Belgium) at low population size for over 60 generations. The second strain was commercialized by the firm Biotop SA (hereafter ‘T’) until 2000 and was also maintained in the laboratory for many generations at INRA and then at Biotop rearing facilities. It is worth noting that this is not the Biotop flightless strain (Tourniaire et al., 2000a,b) first released in France in 2000 and which is the only biocontrol strain used in Europe since 2002. Finally, we used a wild invasive European population sampled in 2009 in Belgium (Ghent, hereafter ‘G’), the area where the European invasion began in the early 2000s (Adriaens et al., 2003, 2008; Brown et al., 2008). Based on data at 18 microsatellite loci, genetic diversity and levels of differentiation vary sharply among these populations and strains (unpubl. data). The two biocontrol strains have relatively low genetic diversity (expected heterozygosity: 0.31 and 0.38 in B and T, respectively), and they are strongly differentiated from one another (Fst = 0.38) as well as from every other wild invasive population (Fst = 0.16–0.33, mean: 0.26). In contrast, wild populations are genetically more diverse (expected heterozygosity: 0.57, 0.58 and 0.61 in D, Q and G, respectively), with the American populations not differentiated from one another (Fst = 0) and only moderately differentiated from Ghent (Fst = 0.034–0.046).

To reduce maternal effect, 50–60 individuals from each available population or strain (experimental G0) were kept separately in the laboratory for one generation prior to the experimental crosses. For each population or strain, 20 couples were bred separately to produce the next generation (G1). Upon emergence, adult males and females (G1) were kept separated until all individuals were at least 1 week old. Rearing conditions remained constant for the entire experiment (23 °C; 65% RH; L : D 14 : 10). Individuals were fed ad libitum with irradiated eggs of Ephestia kuehniella Zeller (Lepidoptera: Pyralidae).

Each American population (Q and D) was crossed with each biocontrol strain (B and T), including reciprocal crosses between the sexes, totalling eight crosses; intrapopulation/strain crosses were also performed including that of the wild invasive European population (G), totalling five additional crosses (Table 1). These 13 crosses are grouped into four Types, namely American, Biocontrol, Admixed and Europe. For each cross, 20 G1-females were placed with 20 G1-males in a large box and allowed to mate (Fig. 1). Seven to ten days elapsed before females were isolated in Petri dish and data collection started.

Table 1.   Description of Harmonia axyridis experimental crosses. Admixed crosses (QB, QT, DB, DT, underlined) involve reciprocal crosses between sexes (e.g. QB = QF × BM and QM × BF).
StatusProvenancePop/StrainCrosses
QDBTG
InvasiveNorth AmericaQuebec (Q)QQQBQT
S. Dakota (D) DDDBDT
BiocontrolEuropean biocontrolBiobest (B)  BB
Biotop (T)   TT
InvasiveEuropeGhent (G)    GG
image

Figure 1. Harmonia axyridis life history traits estimated for each experimental cross (see Table 1), along with total sample sizes. G2-larvae and G2-adults were kept individually in separate Petri dishes. Variables followed by a star were used to calculate the composite fitness index for 147 G1-females (Fitness Index = Hatching Rate × Larval Survival × Fecundity).

Download figure to PowerPoint

For each cross, several life history traits were recorded for the G2-individuals (Fig. 1). Hatching rate (HatchRate) was estimated as the proportion of eggs that developed into larvae. This was estimated as the mean hatching rate of two clutches from each of 11–20 females that laid eggs per cross. Larvae were then kept individually, and larval survival (LarvSurv) to successful pupation was recorded (0 or 1) for a mean of 50 individuals per cross (mean of 2.90 larvae per female for 12–20 females per cross). Development time (DevoTime, days) from egg to pupation was recorded for a mean of 45 individuals per cross (mean of 3.26 individuals per female for 12–20 females per cross).

G2-adults were then separated in two groups (see Fig. 1). A subset of females were kept individually and fed upon emergence. These were used to record the age at laying of the first egg clutch (AgeClutch), fecundity (Fecund) and a composite fitness estimate (FitIndex). At age 7–10 days, each female was presented with a male for 24 h. Males were of similar age and randomly selected from a pool of males containing a balanced mix of males from each cross. The next day, they were offered another such male, again for 24 h. In cases where the females did not lay eggs within the next week, this procedure was repeated. AgeClutch was estimated as the number of days elapsed from pupation to the first clutch. The number of eggs laid over eight consecutive days following the first clutch were counted and averaged to estimate Fecund. The composite fitness estimate (FitIndex) was calculated for 147 G2-females by multiplying HatchRate, LarvSurv (expressed in %) and Fecund. Individuals (both males and females) not used for AgeClutch and Fecund were kept individually without any food upon emergence. The number of days they survived was recorded as the starvation survival period (StarvSurv).

Statistical analyses

For each variable, conformance to Normal and Poisson distributions was appraised with Shapiro Wilk W and Kolmogorov’s D tests, respectively. Larval survival coded as 0 (death) and 1 (survival) was treated as binomial. All time variables (estimated in days) were Poisson-distributed, whereas Fecund and FitIndex were normally distributed. HatchRate was arcsin-transformed to approach normality. For each variable, differences between reciprocal crosses (e.g. cross DB: DF × BM vs. DM × BF) were assessed by means of Tukey–Kramer HSD tests as well as hierarchical anovas with female origin nested within female Type (B or T within Biocontrol; D or Q within American). The vast majority of reciprocal crosses displayed no significant differences (results not shown). The only significant difference was between DevoTime for DM × BF vs. DF × BM (= 0.013) when using HSD tests. Therefore, all subsequent analyses were performed with pooled reciprocal crosses (DB, DT, QB, QT).

First, we tested the hypothesis that admixed individuals are different from parental source(s). To do so, we tested for differences among the three corresponding Types of crosses, i.e. Admixed, Biocontrol and American. We used hierarchical general linear models with Cross nested within Type while specifying the appropriate distribution (normal, Poisson or binomial). When a Type effect was detected, we performed pairwise contrasts between Types to determine which Type differed and whether differences indicated lower or higher fitness. Higher values for HatchRate, LarvSurv, Fecund and StarvSurv, as well as lower values for AgeClutch and DevoTime, were considered indicative of higher fitness. For these variables, we also tested whether there was evidence of heterosis, i.e. if hybrids had mean trait values suggesting higher fitness than the mean traits of parents. This was performed by testing whether the mean of each admixed cross was higher (LarvSurv, Fecund, FitIndex) or lower (AgeClutch) than the mean of pure parental crosses [e.g. DB vs. mean (DD, BB)] using one-sided Tukey–Kramer HSD tests for Fecund and FitIndex, Wilcoxon rank-sum test for AgeClutch and logistical regression (modelling error variance with binomial distribution) for LarvSurv.

Second, we compared wild invasive European H. axyridis (Europe, represented by cross GG) with the other three Types to determine whether European invasive were different from Admixed and/or most similar to either alleged parental source (Biocontrol and American). To do so, we performed GLM analysis as above, using only the Type effect, followed by contrasts between Europe and the other three Types. Tukey–Kramer HSD tests provided highly similar results (not shown). All analyses were performed with jmp 8.01 (SAS Institute 2009, JMP, release 8.01: SAS Institute, Cary, NC, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

For each variable, Table 2 summarizes results for Types and Crosses within Types. Variables were not correlated globally, within Type or within Cross. While a few odd correlations were detected, they were generally only slightly significant (0.01 < < 0.09), and the same variables were not correlated in more than two Cross comparisons.

Table 2.   Mean values (and SEM) for life history traits estimated in Harmonia axyridis experimental crosses. Values for reciprocal admixed crosses (QB = QF × BM and QM × BF) are pooled. See Table 1 for codes and cross description.
Type crossAmericanBiocontrolAdmixedEuropeGrand total
DDQQTotalBBTTTotalDBDTQBQTTotalGG
Hatching rate
 Mean73.8373.8873.8566.4178.8773.1475.4375.4476.2176.9776.0277.2875.29
 SEM4.354.242.993.562.452.323.032.542.172.791.333.091.02
 N1514291720373434303313116213
Larval survival
 %67.978.873.188.094.091.091.091.188.591.290.492.287.84
 N5652108505010010010110410240751666
Development time (days)
 Mean20.4719.2019.8120.1420.0920.1120.1519.5219.4919.2219.5919.4319.69
 SEM0.140.170.130.130.070.070.120.080.120.100.060.140.04
 N3841794447919192929336847585
Age first clutch (days)
 Mean12.7913.9213.3111.4710.8011.1310.6911.6810.6811.5711.1612.1411.52
 SEM0.991.480.860.690.780.520.260.480.280.430.190.910.20
 N1412261515302931313012114191
Fecundity (egg/day)
 Mean22.9128.3325.4131.2646.4838.8735.9634.8039.0033.4835.8229.5434.43
 SEM2.091.481.403.251.512.261.571.882.091.820.942.300.79
 N1412261515302931313012114191
Composite fitness index (egg/day)
 Mean14.2616.4715.3020.1333.8627.2326.3622.8327.4925.9125.5921.2824.15
 SEM2.542.011.622.622.162.102.461.632.372.041.062.350.85
 N10919141529202321228613147
Survival in starvation (days)
 Mean8.218.078.148.417.537.977.377.397.948.037.698.007.82
 SEM0.610.380.360.260.240.190.260.340.300.280.150.330.11
 N1414281717343031323112416202

GLM statistical results are reported in Table 3. For HatchRate, StarvSurv and DevoTime, there were no significant Type effect detected among Admixed, American and Biocontrol (> 0.39). For LarvSurv, AgeClutch, Fecund and FitIndex, highly significant statistical Type effects were detected (< 0.01). In these cases, pairwise contrasts indicated that American was significantly different from Admixed and Biocontrol, but the two latter types were not different (Fig. 2a–d). American crosses had lower larval survival and laid their first clutch when they were older than Admixed and Biocontrol crosses. Fecundity was also lower in American than in Admixed and Biocontrol crosses. The two Biocontrol crosses were variable, but nevertheless averaged higher than American crosses (Table 2). FitIndex varied as fecundity, one of its component variables.

Table 3.   Results of nested GLM analyses (and pairwise contrasts) testing for differences in H. axyridis life history variables among parental (American: AME and Biocontrol: BIO) and Admixed (ADM) cross types. Significant P-values are shown in bold.
Larval survivalType (d.f. = 2)Cross (Type) (d.f. = 5)Type Contrasts
χ2P-valueχ2P-valueAME-BIOAME-ADMBIO-ADMPattern
Hatching rate1.840.396.280.27
Larval survival19.21< 0.0013.390.64< 0.001< 0.0010.77AME[DOWNWARDS ARROW]
Development time0.300.591.460.57
Age at first clutch8.940.0113.330.650.0180.0030.98AME[UPWARDS ARROW]
Fecundity28.49< 0.0012.09< 0.001< 0.001< 0.0010.11AME[DOWNWARDS ARROW]
Composite fitness21.07< 0.00118.610.002< 0.001< 0.0010.49AME[DOWNWARDS ARROW]
Survival (no food)0.770.682.350.79
image

Figure 2.  Mean Harmonia axyridis trait values by Type (American, Biocontrol, Admixed and Europe) for variables showing Type effect (see Table 3). Error bars indicate SEM. Types with same letter are not significantly different. Small letters refer to tests for differences between Admixed and parental types (American and Biocontrol). Capital letters refer to tests comparing all four cross Types (American, Biocontrol, Admixed and Europe).

Download figure to PowerPoint

Heterosis was apparent, but affected crosses differently for different life history traits. There was no evidence for heterosis for AgeClutch (for DB vs. (DD, BB), DT vs. (DD, TT), QB vs. (QQ, BB) and QT vs. (QQ, TT); = 1.37, 1.33, −0.76, −0.49 and one-sided = 0.084, 0.091, 0.222, 0.310, respectively). There was heterosis for LarvSurv only when the Dakota population was involved, but not when Quebec served as one parent (for DB vs. (DD, BB); DT vs. (DD, TT); QB vs. (QQ, BB) and QT vs. (QQ, TT): inline image = 7.36, 1.12, 5.09, 1.23 and one-sided = 0.003, 0.010, 0.133, 0.144, respectively). Evidence for heterosis was also found for Fecund; mean parental values were lower than those of the admixed cross only when the Biobest strain served as one parent (DB < mean (DD, BB) and QB < mean (QQ, BB); HSD one-sided = 0.002). This same pattern was detected for FitIndex, which involves fecundity (HSD one-sided = 0.002).

Comparisons among all four cross types indicated a strong Type effect (LarvSurv: inline image = 21.75, < 0.001; AgeClutch: inline image = 9.12, = 0.0277; Fecund: inline image = 29.71, < 0.001; FitIndex: inline image = 20.63, P = 0.001). Comparisons between European invasive and the three other Types revealed that Europe was never different from Admixed (> 0.13, Fig. 2a–d). For LarvSurv, Europe was similar to Admixed and Biocontrol and clearly significantly higher than American (= 0.003, Fig. 2a). For AgeClutch, Fecund and FitIndex, Europe was intermediate between American and the other two Types. American was not different from Europe for these traits (Fig. 2b–d; > 0.32), but nevertheless remained significantly different from Biocontrol and Admixed (< 0.02, Fig. 2b; < 0.001, Fig. 2c; < 0.003, Fig. 2d).

.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Our precise knowledge of invasion routes of H. axyridis (Lombaert et al., 2010) allowed generating the hypothesis that hybridization between eastern North American populations and the European biocontrol strain might have had an impact early during the European outburst. On one hand, the eastern North American propagules, having spread worldwide, were probably already invasive when they reached Europe. On the other hand, the European biocontrol strains seem to have been unable to establish sustained populations (Ferran et al., 1997) and are thus likely to have low overall fitness in the wild. It was hence hypothesized that such an admixture could either enhance or restrict the invasive process in Europe.

Our experimental results show that admixed HA were often different from at least one parental type. The strongest trend was that admixed individuals possessed mean trait values different from American HA. Admixed individuals survived better in the larval period, females laid more eggs and at an earlier age. Changes in larval survival and fecundity coincided with egg hatching rate and so resulted in a higher composite fitness estimate of admixed relative to American individuals. Moreover, differences between admixed and American HA were all biased towards values contributing to faster population increase rate in the admixed crosses. In contrast, admixed individuals were similar to the biocontrol type for these traits and never possessed values suggesting lower fitness. This biased impact suggests that American HA can benefit from admixture. Despite recurrent use in agriculture, historical and genetic data indicate that, at least in Europe, the biocontrol strains never established and spread in natura. The biocontrol strains used in Europe seemingly possess unfavourable characteristics at some other important fitness-related traits not investigated in this study. For example, the Biotop strain has been shown to have lower hatching rate and very poor survival rate at low temperature (5–15 °C) compared with invasive European populations (Lombaert et al., 2008); these debilitating traits may have prevented its establishment in Europe. Nevertheless, both biocontrol strains can, via admixture, enhance fitness-related trait value of American populations.

It could be argued that the observed bias in the effect of admixture, which mostly changed traits of the wild American type, simply reflects the acquisition of alleles conferring trait values favourable in the laboratory environment, for which biocontrol strains have long been indirectly selected for. However, the wild invasive European beetles were generally similar to the admixed and biocontrol individuals, indicating that the laboratory environment is not, per se, the sole factor at work to explain the observed phenotypic effects. Likewise, heterosis does not appear to be a common cause for the change in life history traits of the admixed individuals. For fecundity (and the composite fitness index including fecundity), heterosis was observed only when crosses involved the biocontrol strain Biobest. This strain is characterized by a low fecundity relative to the biocontrol strain Biotop, as well as a suite of trait values suggesting lower general fitness (lower egg hatching and larval survival rate, greater age at first clutch, Table 2) We suspect that the Biobest laboratory strain we used was subjected to long-term low effective population size, and our experimental admixture per se may indeed have been beneficial for this strain. In any cases, crosses between both American sources and either biocontrol strains resulted in similar fecundity levels in Admixed individuals (GLM: χ2 = 4.94, d.f. = 3, = 0.17), suggesting that crossing these two specific HA types does cause increased fitness in admixed individuals relative to the American parents.

General resemblance between experimentally admixed and wild invasive European HA bring support to the hypothesis that the admixture process affected important phenotypic characteristics of the resulting invasive populations. For larval survival and the age at first clutch, admixed and European crosses were comparable with the biocontrol strains, possessing higher mean trait values compatible with higher fitness. For these traits, it thus appears that admixed European invasive populations have retained the biocontrol genetic background associated to higher fitness. In contrast, the European cross displayed lower fecundity than the biocontrol strains. Higher fecundity in these two biocontrol strains relative to other invasive European populations has already been documented (Lombaert et al., 2008), suggesting that our results are representative of a real difference. It is difficult to envision how lower fecundity may be advantageous for the invasive European beetles. The intermediate fecundity of European beetles may simply reflect their intermediate (i.e. hybrid) ancestry and segregation of additive genetic effects. However, when fecundity is combined with larval survival and hatching rate into the composite fitness index, invasive European HA resemble both admixed and biocontrol types. Given that fecundity is often related to fitness, there may also exist a trade-off between fecundity and unknown trait(s) not considered in that study.

Differences in values between experimentally admixed and wild invasive European HA relative to the parental sources may partly results from the fact that this comparison involves two types of hybrids. In our experiment, we measured phenotypic traits in F1 hybrids raised in the laboratory. The invasive European population used for comparison is likely not composed of F1 hybrids. The invasion was detected in 2001, and our Ghent sample is from 2009. Given that HA can produce 2–3 generations per year (Koch et al., 2006), ca. 20 generations of evolution in natural settings might have elapsed. This time lag between experimentally admixed individuals (F1) and wild invasive European HA (Fn) may, in fact, reveal the action of natural selection in nature on F1 hybrids. In this case, our results showing that F1 admixed HA are not significantly different from wild invasive European HA, while differing from the American parental source, would strongly suggest that phenotypic changes operating in the early admixture stage are, to a large extent, maintained in further generations. Alternatively, the intermediate values of European invasive relative to representative of American and biocontrol parental types may only result from additive effects. Nonetheless, these values would confer higher population increase rate (fitness) to the admixed individuals.

Our experimental design was inspired from the inferred invasion route indicating that European invasive genotypes are admixed HA between biocontrol and American sources at neutral genetic loci. Here, we show that life history traits of experimentally admixed individuals were also affected. Having used only two populations for the American parental type, and a single population to represent invasive European HA, it obviously cannot be strictly affirmed that our results are fully representative of what happened in the wild early during the European invasion. Nevertheless, despite the variation present between American populations (e.g. LarvSurv, Fecund, Table 2), the latter were clearly affected by admixture. Also, the extant Ghent population was the best choice for comparison with F1 hybrids probably formed in this area where the European invasion began in the early 2000s. Overall, our experiment may not be an exact reproduction of the admixture event, but our results show quite clearly that biocontrol strains can favourably affect wild population via admixture. It is worth noting that genetic differentiation at neutral loci was not always related to phenotypic resemblance. For example, American populations were phenotypically very different from wild European HA, yet they were only slightly genetically differentiated (Fst = 0.034–0.046, unpublished data). In contrast, the strong genetic differentiation of each biocontrol strain with wild populations (Fst = 0.26–0.42, unpublished data) was paralleled by either strong (with American) or generally weak (with Europe) phenotypic differences. As per other studies (Dlugosch & Parker, 2008; Keller & Taylor, 2008), these comparisons stress the fact that neutral genetic characteristics, while crucial for reconstructing invasion routes, are not sufficient to inform on the adaptive processes at work during invasions.

Experimental evidence is accumulating that admixture can effectively fuel both early and late HA invasion stages in Europe. In this study, we reproduced the initial European admixture event previously evidenced by neutral genetic markers. We show that American propagules benefit from contacts with biocontrol strain, leading to phenotypes resembling established invasive European populations. Recently, Facon et al. (2011) showed that admixture with another biocontrol strain still used in Europe (i.e. the flightless strain, also derived from the initial INRA biocontrol strain; Tourniaire et al., 2000a,b) can further affect the phenotypic characteristics of contemporary invasive European populations. These two genetically differentiated HA types breed readily in the laboratory, and admixed offspring differ from parental types in terms of development time and their ability to withstand starvation periods. Moreover, mate choice experiments revealed that males of the biocontrol flightless strain sired more offspring, suggesting that admixture may be fostered by invasive female preferences and/or biocontrol male superiority. Altogether, it thus appears that both the initial propagule and the ensuing admixed wild invasive HA can benefit from genetic introgression with biocontrol individuals. Given the invasive success of propagules from eastern North America in South America and South Africa (Lombaert et al., 2010), admixture may not have been necessary for the spread of HA in Europe. Nevertheless, biocontrol strains can effectively contribute to phenotypic changes compatible with higher invasion potential. A simple precautionary principle calls for ceasing to release HA strains for biocontrol control in Europe, irrespective of their flying ability.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This work was supported by grants from the Agence Nationale de la Recherche (ANR-06-BDIV-008-01) and from the Agropolis Fondation (RTRA – Montpellier, BIOFIS Project). We are grateful to A. Loiseau for help in the laboratory during rush periods.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References