DOI: 10.1111/j.1570-7458.2007.00565.x Blackwell Publishing Ltd

Plant phenology and seed predation: interactions between gorses and weevils in Brittany (France) Myriam Barat*, Michèle Tarayre & Anne Atlan ECOBIO, Université de Rennes 1, CNRS, Av. du Général Leclerc, 35042 Rennes, France Accepted: 22 February 2007

Key words: Coleoptera, Curculionoidea, Apionidae, Fabaceae, Exapion ulicis, Exapion lemovicinum, Ulex genera, host specificity, Pteromalus sequester, allochronic speciation, biological control

Abstract

Exapion ulicis (Forster) and Exapion lemovicinum (Hoffmann) (Coleoptera: Curculionoidea: Apionidae) are seed predators of the three gorse species occurring in Brittany (France): Ulex europaeus L., Ulex gallii Planch., and Ulex minor Roth.(Fabaceae). Host-plant phenology plays a major role in the relationship between apionid weevils and their gorse species, because larvae develop within gorse pods and adults have to wait for pod dehiscence to be released. We monitored flowering and fruiting phenology of gorse species, weevil reproductive behaviour, and egg-laying patterns in six natural populations in the native area of these gorse species. At each site, U. europaeus, which flowers mainly in spring, was sympatric with one of two autumn flowering gorse species, U. gallii and U. minor. We noticed that E. ulicis laid eggs in spring and was restricted to U. europaeus whereas E. lemovicinum laid eggs in autumn and was restricted to the two autumn-flowering species U. gallii and U. minor. Therefore, host specificity depended on gorse phenology, and not on geographic proximity. In addition, the infested pod content showed that E. ulicis laid several eggs per pod and suggested that females chose pods with the highest numbers of seeds. In contrast, E. lemovicinum laid a single egg per pod and showed no preference for pods with many seeds. Finally, the impact of seed predation by E. ulicis was higher than that of E. lemovicinum.

Introduction Most phytophagous insects have a high host specificity, using only one or a few closely related plants (Ehrlich & Murphy, 1988; Bernays & Chapman, 1994). As the availability and quality of plant resources vary with seasons, plant lifespan, and the environment (Schoonhoven et al., 1998), host lifehistory traits and especially plant phenology may strongly influence the life cycle and life-history traits of phytophagous insects. This appears particularly true for seed predators, whose immature stages develop inside the ripening seeds. However, cases reporting an influence of host phenology on the life cycle of seed predators are rare, although they may be particularly interesting. For example, Ishihara (1998) has shown that the growth rates of the bruchid Kytorhinus sharpianus varied in response to the changes in development of its host Sophora flavescens (Fabaceae). This result suggests that specialist seed predators are able to rapidly evolve new behavioural and physiological adaptations in response to variations in host phenology. *Correspondence: E-mail: [email protected]

© 2007 The Authors Entomologia Experimentalis et Applicata Journal compilation © 2007 The Netherlands Entomological Society

Therefore, host phenological factors can allow or limit host use and may even lead to allochronic speciation of insects in which the initial segregation of sympatric populations is in time rather than space. Indeed, if an insect species adapts to a plant species that flowers during one restricted period, it will lead to temporal constraints on the use of other plant species with different flowering periods (Bernays & Chapman, 1994). Some cases of close relationship between insects and plants have resulted in temporal isolation between herbivore populations feeding on different host species (Pratt, 1994; Berlocher & Feder, 2002; Drès & Mallet, 2002). Reciprocally, insects may also influence the phenology of their host. Indeed, some plants escape predators in time by blooming before (Mahoro, 2002) or after predator activity (Green & Palmbald, 1975; Carroll & Loye, 1987). Others produce many seeds in a short period to surpass seed predator activity (predator satiation, Kelly et al., 2000). Therefore, the close relationship between phytophagous insects and their host plants may lead to co-evolutionary interactions. Interactions between apionid weevils (Coleoptera: Curculionoidea: Apionidae) and gorse species may represent

1

2

Barat et al.

a suitable model for investigating the reciprocal influence of host-plant phenology on a phytophagous insect. In Brittany (western France), three species of gorse (Fabaceae, Genistae) are present and differ in their flowering period. In U. europaeus L., flowering peaks in spring (but some plants may also flower in other seasons, Tarayre et al., 2007), whereas in U. gallii Planch. and U. minor Roth the flowering peak is in autumn (Des Abbayes et al., 1971; Cubas, 1999). The seeds of all three gorse species are infested by several apionid weevils. Larvae develop within gorse pods until the adult stage, and the adults depend entirely on host-plant phenology for their release as they are not able to escape from the pod before its dehiscence. In addition, apionid weevils are often specialist insects, and are therefore frequently used in biological control programs (Heard & Forno, 1996). Exapion ulicis (Forster) is currently used against U. europaeus, one of the 30 most invasive plant species in the world (Lowe et al., 2000). The weevils were collected in Great Britain and were introduced first to New Zealand (Hill et al., 2000), then to several invaded areas including Chile (Norambuena et al., 1986) and Hawaii (Markin & Yoshioka, 1998). Previous work conducted in Brittany suggested that U. europaeus phenology evolved in response to the parasitic pressure exercised by E. ulicis (Tarayre et al., 2007). Indeed, in this study, U. europaeus exhibited a polymorphism of flowering phenology: long-flowering plants start flowering in autumn or winter and continue through spring, whereas short-flowering plants only flower in spring. Short-flowering plants probably reduce the proportion of infested pods by predator satiation, as they produce mass fruiting in spring. Long-flowering plants partly escape parasitism in time, as their autumn and winter pods are never infested. Autumn pods of U. gallii and U. minor are infested by apionid weevils, but much less is known about the relationship between these two gorse species and their seed predators. This study documents gorse–weevil interactions between spring- and autumnflowering species, which may help in understanding the reciprocal influence of gorse phenology and weevil life history. We monitored natural populations of gorses in Brittany, where autumn- and spring-flowering species occur sympatrically, to answer the following questions: what were the flowering and fruiting phenologies of U. gallii and U. minor? which weevil species infest gorse pods in Brittany and what is the specificity of their association? what is the influence of gorse phenology on the life cycles and life-history traits of weevils? what is the impact of weevils on natural populations of gorse species? The interactions between weevils and gorse species are discussed from a biological and evolutionary perspective.

Materials and methods Study organisms

Gorse species. Gorses are perennial spiny evergreen shrubs with yellow hermaphrodite flowers. The genus Ulex includes a dozen species whose native range is western Europe and North Africa. Three species also occur naturally in western France and Great Britain: U. europaeus (ssp. europaeus), U. gallii, and U. minor. Ulex europaeus is approximately 2 m high but its height can vary from 0.5 to 4 m. In Brittany, it is widespread in both coastal and inland locations, and its flowering peak is in spring. Each plant produces several hundred pods. Ulex gallii and U. minor are smaller (from 0.5 to 1 m) and flower in autumn (Des Abbayes et al., 1971; Cubas, 1999). They generally produce between 50 and 100 pods per individual. Ulex gallii is found in coastal areas and is often encountered in sympatry with U. europaeus. These two species may hybridize but their hybrids can be distinguished from the parent species (Benoit, 1962; Gloaguen, 1986) and were discarded from this study. Ulex minor is a woodland understorey species that sometimes occurs in sympatry with U. europaeus, but no hybrids have ever been reported. In Brittany, U. minor and U. gallii never occur sympatrically. Weevil species. The apionid weevil species observed in this study were identified as E. ulicis (Apion ulicis, Forster) and Exapion lemovicinum (Hoffmann). In order to determine host specificity, a sample of 49 adult weevils collected within pods from the three gorse species were identified by Miguel A. Alonso-Zarazaga (National Museum of Madrid, Spain), using morphological criteria based on Hoffmann’s (1958) and Ehret’s (1990) identification keys. We used the same criteria to identify all the other adult weevils collected in this study. Weevil larvae could not be identified, but ongoing molecular analyses of both adults and larvae have confirmed the presence of no more than two different species. Exapion ulicis is univoltine (Hoddle, 1991a). In its European native range, it has been recorded on U. europaeus and U. minor (Portevin, 1935; Hoffmann, 1958; Ehret, 1990). According to Hoffman (1958), it lays eggs in the pods of different gorse species, notably U. europaeus and U. minor. Most of the data about its life history in Europe comes from the observations of Davies (1928) on U. europaeus in Great Britain. Exapion ulicis mates and lays eggs in spring. Females dig a hole in young green pods with their rostrum and lay 6–8 eggs per pod. Weevil development lasts about 2 months. Adults emerge from the pod at dehiscence. Free adults feed on the vegetative parts and flowers of Ulex species, and hibernate on gorse branches. Much less is known about E. lemovicinum. It has been observed on U. europaeus and U. minor in western France

Gorse–weevil interactions 3

(Portevin, 1935; Hoffmann, 1958; Ehret, 1990) and northern Spain (MA Alonso-Zarazaga, pers. comm.). The only data about its life history in France come from Hoffmann (1958). This author reported that females lay one egg (rarely two) within U. europaeus pods in spring and have a life cycle similar to that of E. ulicis. Other insect species emerging from gorse pods. Parasitoid wasps of weevil larvae were observed in the pods of the three gorse species. In Great Britain, Davies (1928) identified all weevil parasites within U. europaeus pods as Pteromalus sequester (Walker) (Hymenoptera: Pteromalidae), which is described as a solitary ectoparasite of the mature larvae and pupae of E. ulicis (Parnell, 1964). We sent two specimens collected in U. minor and U. gallii pods to Jose L. Nieves-Aldrey (National Museum of Madrid) who also identified them as P. sequester. Gorse pods may also contain Lepidoptera larvae, mainly the gorse pod moth, Cydia succedana (Denis and Schiffermüller) (A Sheppard, pers. comm.). Population monitoring

Gorse phenology. To distinguish species differences from population differences, we compared species in populations where they were sympatric. Seven gorse populations were monitored (Figure 1, Table 1). Three (CF, IB, and PL) were

coastal populations where U. europaeus and U. gallii were sympatric. The four others (PA, RT, SJ, and LO) were inland populations where U. europaeus and U. minor were sympatric. All populations were observed once in autumn of 2004, once in winter of 2004, and every month from April to July 2005. At each visit, the presence of flowers and pods was recorded. The degree of maturation of gorse pods was determined from their colour and roughness: young immature pods are green and soft, whereas ripe pods are rough and brown. The general pattern of gorse phenology was confirmed by a long-term monitoring at IB and LO, which were observed every month from 2000 to 2005. However, the LO population was destroyed in February 2005 and could thus not be monitored after this month. Weevil life cycles. At each field visit for studying gorse phenology, the presence of apionid weevils was also recorded and their activity was divided into four categories: walking, courtship, copulation, and egg laying. The relative proportions of each weevil species on gorse branches was estimated from weevils collected in February and July 2005 and was identified in the laboratory under a stereomicroscope. The relative abundances of weevils on branches of the three gorse species were compared in July 2005 by counting the number of weevils falling on a white sheet after striking the plants three times with a stick. Variables estimated at pod maturation

Figure 1 Location of the seven Ulex populations studied in Brittany (France). Ulex europaeus is sympatric with Ulex gallii in the three coastal populations, Cap Fréhel (CF), Ile Besnard (IB), and Ploumanach (PL) (black dots). Ulex europaeus is sympatric with Ulex minor in the four inland populations, Lande d’Ouée (LO), Paimpont (PA), Saint-Just (SJ), and Rochefort en Terre (RT) (white dots).

Adult weevils were obtained from ripe pods. The pod contents were therefore observed at the peak of ripe pod production of each gorse species, when 10 plants per species were randomly chosen in each population. For U. europaeus, 50 ripe pods per plant were collected in July 2005. For U. gallii and U. minor, all ripe pods produced per plant (minimum = 50, maximum = 113) were collected in April and May 2005. Collected pods were opened in the laboratory, where the number of seeds and insects per pod were counted, and adult weevils were identified. To estimate seed production from uninfested pods, the mean number of seeds per uninfested pod was calculated for each plant. Rotten or flat seeds were not taken into account. Seed predation by weevils was studied by estimating three variables for each plant: 1 the proportion of infested pods, calculated by dividing the number of pods infested by weevils by the total number of pods opened; 2 the mean number of weevils per infested pod. When the pods contained parasitoid wasps, the total number of weevils was calculated by adding the number of wasps to the number of live weevils, as each wasp emerged from a weevil larva; and

4

Barat et al.

3 the mean number of seeds per infested pod. Rotten and flat seeds were not taken into account. To estimate the proportion of infested weevils, we divided the number of parasitoid wasps by the total number of weevils (wasps + live weevils). To estimate seed predation by lepidopterans, we divided the number of pods infested by lepidopterans by the total number of pods opened. Statistical analyses

Statistical analyses were performed with SAS software (SAS, 2005) and based on a GLM (general linear model) procedure. Exapion species (exap) and population (pop) were treated as crossed factors and Ulex species (ulex) was nested within the Exapion factor. A nested design was the most appropriate because Ulex species of each flowering type were infested by a distinct weevil species. As a result, the model was: µ = exap + ulex(exap) + pop + ε, where µ represents the average of the tested variable and ε represents the residual. Ulex species effects were compared on the basis of their least squares estimates (lsmeans) by the PDIFF option of SAS (giving a table of P-values for the three possible pairwise comparisons). Data were arcsine transformed for analyses of proportions (Sokal & Rohlf, 2000). The normal distribution of residuals was checked for each parameter. The CORR procedure of SAS estimated the significance of the correlation between mean proportion of infested pods and mean seed numbers per uninfested pod. All results are given with the standard error of the mean (SEM).

Results Population monitoring

Weevil species within gorse pods. Among the 5622 weevils counted within ripe U. europaeus pods, 4% were in a larval or in the pupal stage, 38% were replaced by parasitoid wasps, and 57% were adults. All adults belonged to E. ulicis. Among the 546 weevils counted within U. gallii ripe pods, 35% were immatures, 31% were replaced by parasitoid wasps, and 34% were adults. Among the 242 weevils counted within ripe U. minor pods, 36% were in a larval or in the pupal stage, 58% were replaced by parasitoid wasps, and 5% were adults. All adult weevils found in U. gallii and U. minor pods belonged to E. lemovicinum. As the identification of adults showed a 100% correlation between gorse and weevil species, we considered that, whatever their stage, weevils within U. europaeus pods belonged to E. ulicis and weevils within U. gallii or U. minor pods belonged to E. lemovicinum. Weevil species on gorse branches. Weevils were abundant on U. europaeus branches (between 10 and 50 weevils fell after

three strikes on a plant). The abundance was much lower on U. gallii (1–5 weevils) and U. minor (0–3 weevils). Adult weevils collected on U. europaeus branches and identified in the laboratory (n = 372) nearly always belonged to E. ulicis (99%) and only 1% to E. lemovicinum. The result was exactly the reverse on U. gallii branches (99% E. lemovicinum and 1% E. ulicis; n = 125). Finally, U. minor branches yielded 92% E. ulicis and 8% E. lemovicinum (n = 167). Weevil observations. The high density of weevils on U. europaeus allowed data on life history of E. ulicis to be deduced from direct observations of weevil behaviour in the field. Walking and feeding weevils were observed when the outside temperature was higher than 10 °C. Below this temperature, many immobile weevils could be collected by striking the branches. The mating period was deduced from the observations of dozens of instances of courtship behaviour and copulation. The egg laying period was deduced from observations of females boring a hole through the pod (a process taking several hours and therefore easy to observe). The emergence period was deduced from observations of adult weevils within gorse pods. Observations on U. gallii and U. minor were more difficult because the abundance of weevils on gorse branches was very low. Therefore, their life cycle could not be deduced from direct behavioural observations in the field. The reproductive period was deduced from gorse phenology, and from observations of weevils collected after striking the branches. The egg-laying period was considered to begin with the onset of pod production, and to end when no moving adults could be collected after striking the branches (in winter, a few immobile adults could be collected from U. gallii branches, but no adults were collected from U. minor branches after November). The development time and emergence periods were deduced from observations of pod contents. Gorse phenology and weevil life cycles. Figure 2 shows the flowering and fruiting period of U. europaeus, U. gallii, and U. minor in relation to parasitism by Exapion spp. in Brittany. Ulex europaeus produced flowers in both autumn and spring, but E. ulicis only infested U. europaeus pods resulting from spring flowers. It had a short development time and adults that emerged in early summer did not show any courtship or copulation behaviour before overwintering. In contrast, U. gallii and U. minor only produced flowers in autumn, and their pods were infested by E. lemovicinum. This species had a longer development time; individuals also emerged in spring but did not need to overwinter to show reproductive behaviour.

Gorse–weevil interactions 5

Figure 2 Flowering and fruiting periods of Ulex europaeus, Ulex gallii, and Ulex minor in relation to parasitism by Exapion spp. in Brittany (France). Most U. europaeus individuals flower only in spring but some individuals flower both in autumn/winter and in spring. Exapion ulicis lays eggs only in pods resulting from spring flowers and emergence occurs late spring/early summer. All U. gallii and U. minor individuals flower in autumn. Pods from these flowers are infested by Exapion lemovicinum and emergence occurs the following spring.

Variables estimated at pod maturation

The population effects were included in the statistical model and appeared to be significant for all the variables studied. However, this study aimed to compare within sympatric populations and hence population differences are not discussed further.

parison between least square means, P = 0.54). The population effect was significant (Table 2) and showed that coastal populations of U. europaeus produced significantly more seeds per uninfested pod than inland populations. The numbers of seeds per uninfested pod varied from 0 to 10 for U. europaeus (n = 769 pods), from 0 to 7 for U. gallii (n = 1915), and from 0 to 4 for U. minor (n = 1440).

Gorse seed production from uninfested pods. Figure 3 shows the mean numbers of seeds per uninfested pod for each gorse species and for both coastal and inland U. europaeus populations. The mean numbers of seeds per uninfested pod depended on the gorse species (Table 2): U. europaeus produced significantly more seeds per uninfested pod than U. gallii (comparison between least square means, P<0.001), but the difference between U. europaeus and U. minor in the same habitat was not significant (com-

Seed predation by weevils. The proportions of infested pods depended both on weevil and gorse species (Figure 4). The proportion of U. europaeus pods infested by E. ulicis was significantly higher than by E. lemovicinum on U. gallii or U. minor, and E. lemovicinum infested a significantly higher proportion of pods in U. gallii than in U. minor (Table 2). The population effect was also significant (Table 2) and showed that proportions of infested pods of inland

Figure 3 Numbers of seeds per uninfested pod for Ulex europaeus (UE), Ulex gallii (UG), and Ulex minor (UM) in Brittany (France). ‘Seaside’: mean + SEM for the three coastal populations (CF, IB, and PL); ‘Inland’: mean + SEM for three inland populations (PA, SJ, and RT). For abbreviations of population locations, see Table 1.

Figure 4 Parasitism rates of Exapion ulicis and Exapion lemovicinum on Ulex europaeus (UE), Ulex gallii (UG), and Ulex minor (UM) in Brittany (France). ‘Seaside’: mean + SEM for the three coastal populations (CF, IB, and PL); ‘Inland’: mean + SEM for three inland populations (PA, SJ, and RT). For abbreviations of population locations, see Table 1.

6

Barat et al.

Table 1 Mean parasitism rates by Exapion apionid weevils in six gorse (Ulex) populations in Brittany (France)

Location

Population

Gorse species

Number of plants sampled

Number of pods collected

Mean parasitism rate weevils (± SEM)

Seaside

Cap Fréhel (CF)

Ulex europaeus Ulex gallii Ulex europaeus Ulex gallii Ulex europaeus Ulex gallii

10 10 10 10 10 10

498 1035 491 806 496 554

0.31 ± 0.06 0.12 ± 0.03 0.33 ± 0.06 0.04 ± 0.01 0.39 ± 0.04 0.37 ± 0.06

Ulex europaeus Ulex minor Ulex europaeus Ulex minor Ulex europaeus Ulex minor

9 10 8 10 8 10

439 664 399 797 396 756

0.78 ± 0.04 0.06 ± 0.02 0.55 ± 0.04 0.10 ± 0.03 0.66 ± 0.03 0.06 ± 0.01

Ile Besnard (IB) Ploumanach (PL) Inland

Paimpont (PA) Rochefort en Terre (RT) Saint-Just (SJ)

U. europaeus populations were significantly higher than for coastal populations (Figure 4, Table 1). Depending on the plant, the percentage of infested pods varied from 20 to 98% for U. europaeus (n = 55 plants), from 0 to 66% for U. gallii (n = 30), and from 0 to 36% for U. minor (n = 30). The numbers of seeds per U. europaeus uninfested pod were negatively correlated with the proportions of pods infested by E. ulicis (R = –0.35, P = 0.01; n = 53; Figure 5). In contrast, the numbers of seeds per U. gallii or U. minor uninfested pods were not significantly correlated with the proportions of pods infested by E. lemovicinum (R = 0.1396, P = 0.30; n = 60; Figure 5). The mean number of weevils per infested pod was significantly higher for E. ulicis in U. europaeus than for E. lemovicinum in U. gallii or U. minor, and E. lemovicinum

Figure 5 Correlation between mean parasitism rates on Ulex europaeus by Exapion ulicis (black diamonds) or on Ulex gallii and Ulex minor by Exapion lemovicinum (white diamonds) and mean numbers of seeds per uninfested pod (means ± SEM for each population).

infested a significantly higher proportion of pods in U. gallii than in U. minor (Table 2). The population effect was also significant (Table 2), but no clear tendency was detected. The number of E. ulicis found in U. europaeus infested pods was about four (it varied from 1 to 16), whereas a U. gallii or a U. minor pod infested by E. lemovicinum almost always contained a single weevil: 1224 of the 1249 infested pods opened contained only one weevil, 20 contained two weevils, and only five contained three weevils. The mean number of seeds per infested pod depended both on weevil and gorse species (Figure 6). Ulex gallii and U. minor pods infested by E. lemovicinum contained significantly more seeds than U. europaeus pods infested by E. ulicis and the number of viable seeds in infested pods was

Figure 6 Numbers of seeds and numbers of Exapion ulicis and Exapion lemovicinum weevils per infested Ulex europaeus (UE), Ulex gallii (UG), or Ulex minor (UM) pod in Brittany (France). ‘Seaside’: mean + SEM for the three coastal populations (CF, IB, and PL); ‘Inland’: mean for three inland populations (PA, SJ, and RT). For abbreviations of population locations, see Table 1.

Gorse–weevil interactions 7

Table 2 Analysis of variance (ANOVA) table for the effects of weevil species (exap), gorse species (ulex) and gorse population (pop) on the content of infested and uninfested pods Variable

Source

d.f.

MS

F

P

Seeds per uninfested pod

ulex pop error R2 = 0.35

2 5 107 P<0.001

7.10 5.01

10.83 7.04

<0.001 <0.001

Proportion of infested pods

exap ulex(exap) pop error R2 = 0.74

1 1 5 107 P<0.001

6.40 1.69 0.52

219.37 57.81 17.70

<0.001 <0.001 <0.001

Weevils per infested pod

exap ulex(exap) pop error R2 = 0.84

1 1 5 54 P<0.001

66.97 9.03 12.09

134.88 18.19 24.35

<0.001 <0.001 <0.001

Seeds per infested pod

exap ulex(exap) pop error R2 = 0.60

1 1 5 105 P<0.001

26.47 2.21 1.16

97.46 8.15 4.29

<0.001 0.005 0.001

Parasitoids per weevil

exap ulex(exap) pop error R2 = 0.25

1 1 5 105 P<0.001

0.37 1.68 0.33

3.40 15.33 3.02

0.068 <0.001 0.014

d.f., degrees of freedom; MS, mean squares. Results with P<0.05 were considered to be statistically significant.

significantly higher in U. gallii than in U. minor (Table 2). The population effect of individuals was also significant (Table 2), but no clear tendency was detected. Depending on the plant, the number of seeds per infested pod varied from 0 to 5 in U. europaeus (n = 1333 pods), from 0 to 4 in U. gallii (n = 407), and from 0 to 3 in U. minor (n = 155). Weevil parasitoidism and seed predation by Lepidoptera. Parasitoid wasps were observed in pods of the three gorse species where they parasitized an average of 40% of weevils, and this rate did not vary significantly between the two weevil species (Table 2). In contrast, Lepidoptera larvae were only observed within U. europaeus pods, where they infested 25% of pods. Such infested pods always harboured a single Lepidotera larva but no weevils.

Discussion Weevil specificity and gorse phenologies

In Brittany, gorse pods are infested by two weevil species, E. ulicis and E. lemovicinum, which exhibit strong host specificity. All adults emerging from U. europaeus belonged

to E. ulicis whereas all adults emerging from U. gallii and U. minor belonged to E. lemovicinum. The two weevil species have very different life cycles: E. ulicis only lays eggs in spring whereas E. lemovicinum only lays eggs in autumn. Indeed, E. ulicis is unable to infest autumn pods of U. europaeus as female ovaries are immature at this moment (Davies, 1928) and U. europaeus autumnal pods are never infested by any weevils in Brittany (Tarayre et al., 2007). Similarly, E. lemovicinum did not infest spring pods as its host species only produces pods in autumn. Spring and autumn flowering gorses occur sympatrically, therefore host specificity depends on gorse phenology, and not on geographical proximity. However, phenology alone does not explain the strict host specificity for oviposition we observed, as there is an overlap of green pod production by the three gorse species in autumn and early spring. Indeed, in autumn, E. lemovicinum did not infest the green U. europaeus pods that were present. Reciprocally, in early spring, E. ulicis did not infest the small number of the green U. gallii and U. minor pods that were still present. This could be due to physical factors such as toughness of pods (Lucas et al.,

8

Barat et al.

2000), but also to chemical factors, such as oviposition deterrents (Karban & Baldwin, 1997) or toxic repellents (Panda & Khush, 1995) especially as Ulex species are rich in flavonoids (Máximo et al., 2000). The presence of E. ulicis within U. minor pods and that of E. lemovicinum within U. europaeus pods were reported in Hoffmann’s fauna (1958). Although we investigated sympatric populations of both Ulex species, we cannot confirm these earlier observations. Nevertheless, our results must be taken with caution: seed predation was much lower in U. gallii and U. minor than in U. europaeus. In addition, a lower proportion of weevils reached the adult stage in these two gorse species. As a result, far fewer adult weevils were identified within U. gallii and U. minor pods compared to U. europaeus. In addition, because morphological determination keys can only be applied to adults, some uncertainty remained on the status of larvae, and we cannot exclude the possibility that E. ulicis individuals were present among the immature stages that we could not identify. Habitat specificity was not as restricted as oviposition specificity, and did not depend on host phenology. In the field, E. ulicis was observed on the gorse species it emerged from, U. europaeus, but was also frequently observed on U. minor, and more rarely on U. gallii. This observation may be related to the distribution of gorse plants in the populations studied. Populations with U. europaeus and U. minor consisted of mixed small patches of both gorse species, whereas populations with U. europaeus and U. gallii only occurred in separate areas for each species. In contrast, E. lemovicinum was only observed on gorse species it emerged from, which is probably due to its low abundance. Pod choice

Exapion ulicis and E. lemovicinum exhibited different podchoice strategies. Pod choice by female insects contributes to maximizing offspring performance (preference/performance theory, Thompson, 1988), which depends on food quality and quantity, as well as on the presence of congeners or predators. Comparison between infested and uninfested pods of U. europaeus suggests that the egg-laying behaviour by females is influenced by the content of pods. Indeed, infested pods produced more weevils than uninfested pods produced seeds. As each weevil generally develops from one seed, this result indicates that infested pods contained about twice as many seeds as uninfested pods. In addition, the higher the proportion of pods infested by E. ulicis, the lower the mean seed numbers per remaining uninfested pods. Therefore, the most likely explanation for these patterns is that E. ulicis females preferentially laid eggs in pods with the highest number of seeds. This pattern was not found in E. lemovicinum. Indeed, the numbers of seeds per uninfested pod corresponded

roughly to the numbers of weevils and seeds per infested pod, and the proportions of pods infested by E. lemovicinum were not correlated with the mean seed numbers per uninfested pod in U. gallii and U. minor. The tendency of E. ulicis to choose pods with a higher number of seeds can be compared to the tendency of other seed predators to select the largest or the most concentrated resources as oviposition sites (e.g., the bean weevil, Callosobruchus maculatus; Cope & Fox, 2003 and the broom seed beetle, Bruchidius villosus; Redmon et al., 2000). Therefore, podchoice differences between E. ulicis and E. lemovicinum could be explained by the numbers of eggs laid per pod. Indeed, E. ulicis females lay 6–8 eggs at a time (Davies, 1928, M Barat, M Tarayre and A Atlan pers. obs.), which means that their fitness could be strongly reduced if they chose a pod at random, because more than half of U. europaeus pods contain four seeds or less. In contrast, E. lemovicinum females lay only one egg per pod (Hoffmann, 1958, our observations) and each U. gallii or U. minor pod contains at least one seed, consequently, E. lemovicinum females would not benefit from spending time and energy selecting only the largest pods. Exapion ulicis females were also able to discriminate a previously infested pod, as previously shown by Hoddle (1991b). This study provides indirect arguments suggesting that E. lemovicinum females also prefer uninfested pods. Indeed, the rarity of pods yielding more than one weevil, even in relatively highly infested populations, suggests that very few pods were attacked more than once. Avoidance of superparasitism is common among parasitoids (van Lenteren, 1981; van Alphen & Visser, 1990) but has rarely been reported among seed predators [e.g., the apple maggot, Rhagoletis pomonella (Roitberg & Prokopy, 1983) and the seed beetle, Stator limitatus (Fox et al., 1996)]. Laying eggs in uninfested pods avoids larval competition for food and thus increases the survival probabilities of each larva (Messina & Tinney, 1991). Whatever the oviposition strategy adopted by apionid weevils, offspring survival was altered by hymenopteran predation, mainly by P. sequester. Hymenopterans infested the two weevil species indifferently and they killed nearly 40% of the larvae. Impact of seed predators on gorses

Exapion ulicis had a greater impact on U. europaeus seed production than E. lemovicinum on U. gallii or U. minor. Indeed, E. ulicis infested 49% U. europaeus pods, whereas E. lemovicinum infested 18 and 7% U. gallii and U. minor pods, respectively. Moreover, E. ulicis destroyed 97% of the seeds in infested U. europaeus pods, whereas E. lemovicinum destroyed 42 and 51% of the seeds in infested U. gallii and U. minor pods, respectively. Furthermore, the inland popu-

Gorse–weevil interactions 9

lations of U. europaeus studied here were more heavily infested by weevils and produced fewer seeds per uninfested pod than coastal populations. This may be explained by the fact that U. europaeus was at its limit of ecological range in populations where it was sympatric with U. minor. Exapion ulicis adults were more abundant on gorse branches than E. lemovicinum adults, and therefore caused more foliage damage. Finally, gorse pod moths, C. succedana, were observed only in U. europaeus pods; they infested approximately a quarter of pods in which they ate almost all the seeds. The great impact of E. ulicis on the reproduction success of U. europaeus supports its use as biological agent in Hawaii (Markin & Yoshioka, 1998), New Zealand (Hill et al., 2000), and Chile (Norambuena & Piper, 2000). However, in these regions the high proportions of pods resulting from autumn flowering reduce the impact of this weevil (Hill et al., 1991). The introduction of an autumn-laying weevil would be interesting. However, E. lemovicinum has a low impact on gorse pod production and, at least in situations where there is a choice, does not infest the autumn pods of U. europaeus in Brittany. It therefore cannot be considered as a potentially successful biological control agent. Conclusion and perspectives

Gorse phenology plays a major role in the life cycle, host choice, and performance of apionid weevils in Brittany. The close relationship between weevil and gorse species might have involved mutual adaptations between insects and plants. The high seed predation exercised by weevils has probably led to the flowering shift of some U. europaeus plants towards autumn to temporarily escape parasitism (Tarayre et al., 2007). Host specificity by weevils depends on gorse phenology, as the gorse species are sympatric. Divergence between E. ulicis and E. lemovicinum could thus have occurred through allochronic speciation. Studying gorse–weevil interactions with more species in a phylogenetic context would certainly provide interesting data for the understanding of speciation processes within these groups.

Acknowledgements We are grateful to L. Parize and C. Antoine for help in the field and in the laboratory, Y. Rantier for drawing Figure 1, and J.-S. Pierre for help with the statistics. We thank J.-S. Pierre, D. Poinsot, J. van Baaren, and P. Vernon for comments on our article, and R. H. Britton for improvement of English.

References van Alphen J & Visser ME (1990) Superparasitism as an adaptive strategy for insect parasitoids. Annual Review of Entomology 35: 59–79.

Benoit PM (1962) Ulex europaeus × Ulex gallii. Proceedings of the Botanical Society of the British Isles 4: 414–415. Berlocher SH & Feder JL (2002) Sympatric speciation in phytophagous insects: moving beyond controversy? Annual Review of Entomology 47: 773–815. Bernays EA & Chapman RF (1994) Host-Plant Selection by Phytophagous Insects. Chapman & Hall, London, UK. Carroll SP & Loye JE (1987) Specialization of Jadera species (Hemiptera: Rhopalidae) on the seeds of Sapindaceae (Sapindales), and coevolutionary responses of defence and infest. Annals of the Entomological Society of America 80: 373–378. Cope JM & Fox CW (2003) Oviposition decisions in the seed beetle, Callosobruchus maculatus (Coleoptera: Bruchidae): effects of seed size on superparasitism. Journal of Stored Products Research 39: 355–365. Cubas P (1999) Ulex L. Plantas Vasculares de la Península Ibérica e Islas Baleares, vol 7. (ed. by S Talavera, C Aedo, S Castroviejo, C Romero Zarco, L Sáez, FJ Salguerio & M Velayos), pp. 212– 239. Real Jardín Botánico, CSIC Madrid, Spain. Davies WM (1928) The bionomics of Apion ulicis Först. (gorse weevil), with special reference to its role in the control of Ulex europaeus in New Zealand. Annals of Applied Biology 15: 263– 285. Des Abbayes H, Claustres G, Corillion R & Dupont P (1971) Flore et végétation du Massif Armoricain. I Flore vasculaire. Presses Universitaires de Bretagne, Saint-Brieuc, France. Drès M & Mallet J (2002) Host races in plant-feeding insects and their importance in sympatric speciation. Philosophical Transactions of the Royal Society of London B 357: 471–492. Ehret J-M (1990) Les Apions de France. Clés d’identification commentées (Coleoptera Curculionidae Apioninae). Bulletin Mensuel de la Société Linnéenne de Lyon 59: 209–292. Ehrlich PR & Murphy DD (1988) Plant chemistry and host range in insect herbivores. Ecology 69: 908–909. Fox CW, Martin JD, Thakar MS & Mousseau TA (1996) Clutch size manipulations in two seed beetles: consequences for progeny fitness. Oecologia 108: 88–94. Gloaguen JC (1986) Les ajoncs de Bretagne. Bulletin de la Société Botanique de France – Lettres 133: 363–385. Green TW & Palmbald IG (1975) Effects of insect seed predators on Astragalus cibarius and Astragalus utahensis (Leguminosae). Ecology 56: 1435–1440. Heard TA & Forno IW (1996) Host selection and host range of the flower-feeding weevil, Coelocephalapion pigrae, a potential biological control agent of Mimosa pigra. Biological Control 6: 83–95. Hill RL, Gourlay AH & Fowler SV (2000) The Biological Control Program Against Gorse in New Zealand 909. Proceedings of the X International Symposium on Biological Control of Weeds, pp. 909–917. Montana State University, Bozeman, MT, USA. Hill R, Gourlay AH & Martin L (1991) Seasonal and geographic variation in the predation of gorse seed, Ulex europaeus L., by the seed weevil Apion ulicis Forst. New Zealand Journal of Zoology 18: 37–43. Hoddle MS (1991a) Lifetable construction for the gorse seed

10

Barat et al.

weevil, Apion ulicis (Forster) (Coleoptera: Apionidae) before gorse pod dehiscence, and life history strategies of the weevil. New Zealand Journal of Zoology 18: 399–404. Hoddle MS (1991b) Gorse pod susceptibility and oviposition preference to the gorse seed weevil, Apion ulicis (Forster) (Coleoptera: Apionidae). New Zealand Journal of Zoology 18: 31–35. Hoffmann A (1958) Coléoptères Curculionides. 62 Coléoptères Curculionides (Troisième Partie). (ed. by Naturelles FFdS), pp. 1209–1839. Faune de France, Paris eds, Paris, France. Ishihara M (1998) Geographical variation in insect developmental period: effect of host plant phenology on the life cycle of the bruchid seed feeder Kytorhinus sharpianus. Entomologia Experimentalis et Applicata 87: 311–319. Karban R & Baldwin IT (1997) Induced Responses to Herbivory. University of Chicago Press, Chicago, IL, USA. Kelly D, Harrison AL, Lee WG, Payton IJ, Wilson PR & Schauber EM (2000) Predator satiation and extreme mast seeding in 11 species of Chionochloa (Poaceae). Oikos 90: 477–488. van Lenteren JC (1981) Host discrimination by parasitoids. Semiochemicals, their Role in Pest Control (ed. by DA Nordlund, RL Jones & WJ Lewis), pp. 153–179. Wiley, New York, NY, USA. Lowe S, Browne M, Boudjelas S & De Poorter M (2000) 100 of the World’s Worst Invasive Alien Species. A Selection from the Global Invasive Species Database. Published by The Invasive Species Specialist Group (ISSG) a specialist group of the Species Survival Commission (SSC) of the World Conservation Union (IUCN). Auckland, New Zealand. Lucas PW, Turner IM, Dominy NJ & Yamashita N (2000) Mechanical defences to herbivory. Annals of Botany 86: 913– 920. Mahoro S (2002) Individual flowering schedule, fruit set, and flower and seed predation in Vaccinium hirtum Thunb. (Ericaceae). Canadian Journal of Botany 80: 82–92. Markin GP & Yoshioka ER (1998) Introduction and establishment of the biological control agent Apion ulicis (Forster) (Coleoptera: Apionidae) for control of the weed gorse (Ulex europaeus L.) in Hawai’i. Proceedings of the Hawaiian Entomological Society 33: 35–42. Máximo P, Lourenço A, Feio SS & Roseiro JC (2000) Flavonoids from Ulex species. Zeitung Naturforschung 55: 506–510. Messina FJ & Tinney TR (1991) Discrimination between self and non self eggs by egg-laying seed beetles: a reassesment. Ecological Entomology 16: 509–512.

Norambuena H, Carrillo R & Neira M (1986) Introducción, establecimiento y potencial de Apion ulicis como antagonista de Ulex europaeus en el sur de Chile. Entomophaga 31: 3–10. Norambuena H & Piper GL (2000) Impact of Apion ulicis Forster on Ulex europaeus L. seed dispersal. Biological Control 17: 267–271. Panda N & Khush GS (1995) Host Plant Resistance to Insects. CAB International, Wallingford, UK. Parnell JR (1964) The parasite complex of two seed beetles Bruchidius ater (Marsham) (Coleoptera: Bruchidae) and Apion fuscirostre Fabricius (Coleoptera: Curculionidae). Transactions of the Royal Entomological Society of London 116: 73–88. Portevin G (1935) Histoire Naturelle des Coléoptères de France – Tome IV. Paris (VIe), France. Pratt GF (1994) Evolution of Euphilotes (Lepidoptera, Lycaenidae) by seasonal and host shifts. Biological Journal of the Linnean Society 51: 387–416. Redmon SG, Forrest TG & Markin GP (2000) Biology of Bruchidius villosus (Coleoptera: Bruchidae) on scotch broom in North Carolina. Florida Entomologist 83: 242–253. Roitberg BD & Prokopy RJ (1983) Host deprivation influence on response of Rhagoletis pomonella to its oviposition-deterring pheromone. Physiological Entomology 8: 69–72. SAS (2005) SAS Software: version 9.1. SAS institute, Cary, NC, USA. Schoonhoven LM, Jermy T & van Loon JJA (1998) Insect-Plant Biology: From Physiology to Evolution. Chapman & Hall, London, UK. Sokal RR & Rohlf FJ (2000) Biometry: The Principle and Practice of Statistics in Biological Research, 3rd edn. W.H. Freeman, New York, NY, USA. Tarayre M, Vergnerie G, Schermann-Legionnet A, Barat M & Atlan A (2007) Flowering phenology of Ulex europaeus: ecological consequences of variation within and among populations. Evolutionary Ecology in press. Thompson JN (1988) Evolutionary ecology of the relationship between oviposition performance and preference of offspring in phytophagous insects. Entomologia Experimentalis et Applicata 47: 3–14. Tikkanen O-P & Lyytikaïnen-Saarenmaa P (2002) Adaptation of a generalist moth, Operophtera brumata, to variable budburst phenology of host plants. Entomologia Experimentalis et Applicata 103: 123–133.