Oecologia (2006) 148: 153–161 DOI 10.1007/s00442-005-0332-9
B E H A V I O U R A L E C O LO G Y
Isabelle Amat Æ Marcela Castelo Emmanuel Desouhant Æ Carlos Bernstein
The influence of temperature and host availability on the host exploitation strategies of sexual and asexual parasitic wasps of the same species
Received: 26 January 2005 / Accepted: 2 December 2005 / Published online: 19 January 2006 Springer-Verlag 2006
Abstract In the hymenopteran parasitoid Venturia canescens, asexual (obligate thelytoky not induced by Wolbachia bacteria) and sexual (arrhenotokous) wasps coexist in field conditions despite the demographic cost incurred due to the production of males by sexual females. Arrhenotoky predominates in field conditions, whereas populations in indoor conditions (mills, granaries) are exclusively thelytokous. These differences in the relative abundance of the two modes of reproduction between environments suggest that the individuals of each reproductive mode may have developed strategies adapted to the conditions prevailing in each kind of habitat. The two environments contrast in temperature variability and in the spatial heterogeneity of host availability. In this study, we considered the combined effect of temperature and host availability on host patch exploitation by thelytokous and arrhenotokous V. canescens. As expected, arrhenotokous females were more sensitive to temperature changes. If the temperature decreased before foraging, they remained longer and exploited patches more thoroughly. This is consistent with the expected behaviour of parasitoids in response to signs of unfavourable conditions that entail increasing risk of time limitation or a reduced proba-
Communicated by Roland Brandl I. Amat (&) Æ E. Desouhant Æ C. Bernstein Laboratoire de Biome´trie et Biologie Evolutive (UMR 5558), CNRS, Universite´ Lyon 1, 43 boulevard 11 novembre, 69622 Villeubanne Cedex, France E-mail:
[email protected] Tel.: +33-4-72432929 Fax: +33-4-72431388 M. Castelo CONICET, Laboratorio de Ecologı´ a y Comportamiento Animal, Dpto. de Ecologı´ a, Gene´tica y Evolucio´n, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Ciudad Universitaria, Pabello´n II, C1428EHA Ciudad de Buenos Aires, Argentina
bility of attaining further patches. Both arrhenotokous and thelytokous females increased patch exploitation with host availability. However, unexpectedly, we found no difference in the way the two types of wasp responded to differences in host availability. Differences in the strategies adopted under different environmental conditions may indicate divergence of niche-specific life history traits between the two modes of reproduction. Niche displacement may partly account for the coexistence of these two modes of reproduction at a geographical scale. Keywords Reproduction modes’ coexistence Æ Habitat specialisation Æ Patch-leaving decision Æ Proportional hazards model Æ Venturia canescens
Introduction According to demographic considerations, sexual reproduction entails a competitive disadvantage when compared with asexual reproduction, due to the cost of producing males (Maynard Smith 1978). The predominance of sexual reproduction in nature therefore seems paradoxical. Two possible explanations for the success of sexual reproduction based on long-term processes have been proposed: sex accelerates the rate of evolutionary change (Mu¨ller 1932; Crow and Kimura 1965) and, by recombination, prevents the accumulation of deleterious mutations (Mu¨ller 1964; Haccou and Schneider 2004). However, arguments based on longterm processes cannot account for the coexistence of sexual individuals and asexually reproducing competitors observed in some systems. These conditions call for explanations based on short-term processes of how sexual individuals compensate for the immediate cost of producing males (Williams 1975). The relative fitness of sexual and asexual animals may depend on environmental conditions, so ecological mechanisms may account for the short-term advantages of sex (Hamilton 1980; Bell 1982; Case and Taper 1986;
154
Stearns 1990). For instance, according to the Red Queen hypothesis, sexual animals should perform better than asexual animals in the face of parasitism, due to the greater genetic variability of their offspring (Hamilton 1980). In the absence of parasitism, asexual animals should have a higher fitness than sexual animals due to their superior growth rate. Once the initial differences have been established, natural selection over time will favour one mode of reproduction or the other, depending on the environmental conditions. Animals would be subjected to different selective pressures, increasing the divergence of niche-specific life history traits, further reducing competition between reproductive modes and favouring the persistence of sexual individuals in the close vicinity of their asexual counterparts (Case and Taper 1986). Systems in which both reproductive modes coexist are known in various taxa [e.g. Anthoxanthum odoratum, a Cyperales grass (Kelley et al. 1988), the gastropod Potamopyrgus antipodarum (Lively 1987), amphibians of the genus Rana (Semlitsch et al. 1997), fish of the genus Poeciliopsis (Schultz 1969), members of the Homoptera (aphid species, see Simon et al. 2002 for a review), the cladoceran Daphnia pulex (Hebert 1981), the hymenopteran Venturia canescens (Beukeboom et al. 1999) and Lysiphlebus fabarum (Belshaw et al. 1999; Belshaw and Quicke 2003), etc.]. Evidence has been obtained for interactions between the relative fitness of each mode of reproduction and environmental conditions in several of these systems. In water frogs (Rana esculenta and R. lessonae; Negovetic et al. 2001) and brine shrimps (Artemia tunisiana and A. parthenogenetica; Barata et al. 1996), closely related sexual and asexual species have been shown to respond differently to temperature. In both cases, one mode of reproduction performed better (higher survival rate or competitive ability) at moderate temperature, whereas the other performed better at lower temperature. Similarly, coexisting clonal and sexual fishes of the genus Poeciliopsis display differences in survival in response to environmental stress (heat, cold and hypoxia stress; Vrijenhoek and Pfeiler 1997). The short-term maintenance of sexual forms may also depend on the association between sexual reproduction and a particular ecological function (Lloyd 1980; Gouyon 1999). For instance, in aphids, sex is the only means by which the cold-resistant form, eggs, can be produced (Leather 1992). The parasitoid V. canescens Gravenhorst (Hymenoptera: Ichneumonidae) reproduces either asexually (obligate thelytoky not induced by Wolbachia; Beukeboom and Pijnacker 2000) or sexually (obligate arrhenotoky, i.e. haplo-diploid reproduction; Beukeboom et al. 1999; Schneider et al. 2002). Work by Schneider et al. (2002) revealed that essentially arrhenotokous and thelytokous populations are distinct genetical entities but that occasional gene flow between both reproductive modes cannot be completely ruled out. Conversely, the rare matings between thelytokous females and arrhenotokous males in laboratory conditions do not have a
lasting influence on the genotype of the offspring (Schneider et al. 2003). Only thelytokous populations thrive in indoor conditions such as mills and granaries. In contrast, extensive sampling has shown that arrhenotokous reproduction predominates in field conditions, in which arrhenotokous females are found either alone (80% of the sampling locations and dates on which V. canescens was present) or together with thelytokous females (20% of the sampling locations and dates) (Schneider et al. 2002; I. Amat, unpublished data). In field conditions, V. canescens attacks pyralid moth larvae in desiccated fruits (Salt 1976). Host pyralid larvae are normally uniformly distributed and there are generally only one or two such larvae in fruits harbouring them, and up to a maximum of four hosts in 1.5% of the contaminated carob fruits (Driessen and Bernstein 1999). The larvae are concealed in or under fruits and are often outside the reach of the ovipositor of the foraging females (Schneider 2003). Thus, under field conditions, available hosts for oviposition are scarce and homogeneously distributed. Within buildings, the wasps attack host larvae feeding on the surface of stored products (Ahmad 1936; Waage 1979). The pyralidae moth species, especially Ephestia species and Plodia interpunctella are major pests worldwide of stored food (Rees 2004). In contrast to what is observed in field conditions, hosts may attain high densities, with a highly heterogeneous distribution, within buildings (Bowditch and Madden 1996). Hence, selective pressures for coping with differences in host availability should be weaker for arrhenotokous than thelytokous females. These two environments also differ in thermal conditions. Temperature generally remains almost constant within buildings but is subject to long-term (seasonal) and short-term fluctuations in the field. The foraging strategies of parasitoids are subject to strong selective pressures and wasps thriving in conditions of heterogeneous host availability would be expected to evolve mechanisms for coping with differences in host availability (van Alphen et al. 2003). Moreover, the thermal environment may profoundly affect morphology, physiology, and life history traits in ectotherms (Huey and Kingsolver 1989). Thus, ectotherms foraging in different thermal environments may have evolved adaptations both to prevailing temperatures and to temperature fluctuations. Thus, as thelytokous (mills, granaries) and arrhenotokous (field conditions) V. canescens wasps are found preferentially in different environments, they may require different foraging strategies. The aim of this study was to investigate whether the behaviour of thelytokous and arrhenotokous females differs according to the local environmental conditions prevailing in each kind of habitat. We hypothesised that, for each mode of reproduction, wasps would be most sensitive to factors with high levels of variability in their preferred environment. We investigated host availability and thermal conditions as these factors vary differently in the two environments and parasitoids are known to
155
respond to both factors (van Alphen et al. 2003). We expected thelytokous females to be more sensitive than arrhenotokous females to cues indicating high local host availability. More precisely, the patch residence time of thelytokous V. canescens is known to increase with the density of hosts in the patch and to level off at high host densities (Driessen and Bernstein 1999). We expected that in natural conditions, selective pressures for detecting high-density patches would be weaker, and thus arrhenotokous females would saturate their response to the presence of hosts at lower densities (see below for a more detailed description of the response of thelytokous females to host densities). Conversely, we thought that arrhenotokous females would be more sensitive than thelytokous females to changes in temperature. Deteriorating conditions (e.g. a decrease in temperature or barometric pressure) may indicate a decrease in future opportunities to lay eggs or an increased risk when migrating between patches, as might be the case in a storm provoked by a progressing cold front (Ahrens 1993). Theoretical and experimental studies have suggested that parasitoids increase patch exploitation under conditions of time limitation (Mangel 1989; Roitberg et al. 1992; Roitberg et al. 1993). We assessed parasitoid behaviour by considering patch residence time and the number of ovipositions. These variables were studied because they are known to be key aspects of foraging behaviour and because there is ample knowledge on the mechanisms involved (van Alphen et al. 2003).
Materials and methods Biological details and cultures Venturia canescens is a solitary koinobiont larval endoparasitoid of lepidopteran larvae (mainly Pyralidae, Salt 1976). It attacks hosts in their second to fifth instar (Harvey and Thompson 1995). While in a patch, V. canescens searches for hosts by probing the host-contaminated substrate with its ovipositor. Oviposition can be recognised by a movement of the abdomen, the ‘‘cocking’’ movement, which V. canescens uses to place a new egg in the tip of its ovipositor in preparation for the next oviposition (Rogers 1972). Host searching is elicited by a mandibular gland secretion (a kairomone) released by host larvae (Corbert 1971). In thelytokous V. canescens, patch residence time has been shown to increase with kairomone concentration (Waage 1979; Driessen et al. 1995; Driessen and Bernstein 1999) levelling off at higher concentrations (Driessen and Bernstein 1999). The parasitoids used originated from thelytokous and arrhenotokous strains established in 1999 from larvae parasitised in field conditions. The animals resulted from an extensive sampling where pieces of cardboard with kairomone-contaminated wheat semolina and healthy larvae were hung from trees and bushes for 3 days in a period when V. canescens populations are normally
abundant. Sampling points (n=12) were at least 100 m apart, and four pieces of cardboard were placed at each of them. The emerging females (0–57 females per day and sampling point) were kept individually with excess hosts and food until the emergence of the next generation, which allowed identification of their reproductive mode. Animals of the same reproductive mode were grouped to initiate the cultures. Arrhenotokous and thelytokous females were obtained from the same locations (often from the same piece of cardboard) near Antibes (south-east France). The sampling procedure does not allow one to estimate the original number of females that attacked the hosts in the pieces of cardboard, but the number of sampling points and their distance apart ensure an adequate number of foundress females. As Beck et al. (1999, 2001) and Li et al. (2003) suggested that a deletion in the sequence encoding a virus-like particle protein has pleiotropic effects on the reproductive biology of V. canescens, only wasps carrying the wild-type allele (called RP by Hellers et al. 1996) were used. However, it has subsequently been shown that the strain used in this experiment displays no such pleiotropic effects on reproductive biology (Amat et al. 2003). These RP lines were established in 2000 by regrouping wasps of known genotype (n50) from mixed cultures. To maintain the parasitoid strains, about 75 females taken from the rearing boxes (together with an excess of males for the arrhenotokous strain) were introduced into new boxes (generally two or three) with healthy hosts. In some rare occasions the number of wasps introduced was somewhat lower. The hosts used, Ephestia kuehniella Zeller (Lepidoptera: Pyralidae), were obtained from a mass-rearing facility (INRA) in Antibes. Hosts were kept in plastic boxes containing wheat semolina and, like parasitoids, were reared in a controlled, constant environment (25±1C, 75±5% relative humidity and 12 h:12 h light:dark). General experimental conditions The aim of this study was to investigate the effects of thermal conditions and host availability on patch exploitation (patch residence time and number of ovipositions) in females of the two modes of reproduction. As the volatilisation and, hence, perception of kairomone may depend on temperature, we considered the combined effects of both environmental factors and mode of reproduction. We used 48-hour-old wasps for the experiments. During the first 24 h after emergence, wasps of each reproductive mode were maintained in cages (30·30·30 cm) with an ample supply of honey diluted 1:2 in water. Newly emerged arrhenotokous females were kept with males (1 or 2 days old, two males per female) to allow mating. During the next 24 h, all females were kept individually in tubes (7·3 cm) with a drop of diluted honey. Just before the trial, females were provided with host-contaminated semolina (without the
156
host), and the rare females not actively foraging for hosts were discarded. Trials were initiated by introducing a single parasitoid into a host patch. During a patch visit, V. canescens regularly makes short excursions outside the patch. We defined patch residence time as the total time between entering the patch for the first time and the ‘‘last leaving of the patch’’—i.e. the last time the patch edge was crossed before the insect left the large dish surrounding the patch (see below for further details). Whilst the female was visiting the patch, we registered on a microcomputer the time when various types of behaviour (cocking, entering or leaving the patch, leaving the large outer dish) were performed. During the experiment, the females were not confined to a cage and were free to leave the experimental patch by flying or walking. Not confining the females to patches makes it possible for the wasps to unambiguously leave the experimental environment. If the wasp was still on the patch 40 min after entering it, the experiment was stopped (16 individuals out of 225 tested wasps) and, for the statistical analysis, the patch residence time was considered as censored (Kalbfleisch and Prentice 1980). After the trial, we checked that arrhenotokous females had indeed mated by offering them 30 hosts and diluted honey and allowing them to oviposit until death. Eighteen (22%) of the 81 arrhenotokous females used in experiment 1 and ten (24%) of the 41 arrhenotokous females used in experiment 2 had not mated. As virgin and mated arrhenotokous females differ in foraging behaviour (I. Amat, unpublished data), unmated wasps were discarded from the analysis. Patches and host availability A 3-mm-deep 34-mm-diameter dish containing host larvae in 2 g semolina was placed in the centre of a larger dish (185 mm diameter) filled to the rim with clean semolina. The small dish was embedded such that its surface was sunk flush with the surrounding clean semolina. We defined the patch as the small central dish. We kept two or 12 Ephestia kuehniella larvae in the patch for 7 days before the experiment. The age of the host larvae was such that, by the seventh day, they had developed to fifth instar. Patches were covered with a fine cloth to prevent larvae from escaping. At a host density of two larvae per patch (referred to hereafter as‘‘low host availability’’), thelytokous V. canescens responds to increases in host density by increasing its patch residence time; 12 larvae (referred to hereafter as‘‘high host availability’’) corresponds to the asymptote of the patch residence time–host density response of these females (their response levelling-off beyond eight larvae per patch, Driessen and Bernstein 1999). V. canescens females are likely to encounter conditions similar to the low-host-availability conditions of this experiment in the field (Driessen and Bernstein 1999). Both high and low host availability are likely to be found in buildings.
Temperature Ten years of meteorological observations close to the site at which the strains used were first collected (Me´te´o France, the French meteorological service) showed that during the months in which V. canescens is most active (July, August), daytime temperatures are most frequently in the 24–28C range. We therefore took the median of this interval (26C) to represent conditions favourable for flight and foraging by V. canescens. Temperatures of 20C or lower are rare and are only registered during the night or in unfavourable weather conditions. We therefore took 21C as a less favourable temperature for foraging. During the daytime, a sudden decrease to 21C or below may indicate a deterioration in weather conditions (for instance, the progression of a cold front; Ahrens 1993). This may correspond to a reduced probability to reach further patches in a wasp’s lifetime. Experiment 1: influence of foraging temperature and host availability on patch exploitation We assessed the combined effect of host availability and foraging temperature on patch exploitation by females of the two reproductive modes. The conditions previously experienced by the wasps were standardized by keeping the wasps at 26C (pre-trial temperature) for the 48 h preceding the trial. Trials were carried out at 21 or 26C (the temperature used during the trial is referred to hereafter as foraging temperature) on patches with low or high host availability. We assessed the effects of foraging temperature, reproductive mode and host availability using a factorial design with eight combinations of factors (Table 1). An alternating sequence of trials with low and high host availability was retained to minimise bias due to possible diurnal variations in activity patterns. All trials were run blind for the reproductive mode (i.e. the observer did not known the reproductive mode of the female during the experiment). The experimental facilities did not allow a full randomisation of foraging temperature. We therefore alternated foraging temperature on a daily base. Experiments were performed over the course of 5 weeks, with a total of 18 days of observation. One of the 78 wasps tested at low host availability oviposited 19 times whereas the maximum number of eggs laid per patch visit for the other 77 wasps was six (mean=1.7, SE=0.16). This outlier was discarded from the analyses, with no qualitative change in the results. Experiment 2: effect of a preceding temperature drop on patch exploitation In experiment 1, differences in the behaviour of the females foraging at 21 and 26C may be due to the lower foraging temperature itself or to a decrease in tempera-
157 Table 1 Summary of the experimental design. In experiment 1, the pre-trial temperature was 26C and the foraging behaviour of thelytokous and arrhenotokous females was compared at two temperatures (21 or 26C) and two levels of host availability (two or 12 host larvae). This factorial design led to eight combinations of factors. In experiment 2, all wasps exploited two-host patches at 21C and the effect of the pre-trial temperature (21 or 26C) on foraging behaviour of thelytokous and arrhenotokous females was assessed. Thelyt. thelytokous females, Arrhenot. arrhenotokous females, n number of tested animals per combination Pre-trial temperature (C)
Foraging temperature (C)
Host availability
Experiment 1 26
26
12 2
21
12 2
Experiment 2 26
21
21
2
Thelyt. n=22 Arrhenot. n=14 Thelyt. n=22 Arrhenot. n=19 Thelyt. n=24 Arrhenot. n=14 Thelyt. n=21 Arrhenot. n=16 Thelyt. n=22 Arrhenot. n=19 Thelyt. n=20 Arrhenot. n=12
ture (26–21C) before foraging. Experiment 2 was designed to identify which of the two cues was triggering a wasp’s response. Thelytokous and arrhenotokous females were observed while foraging at 21C in conditions of low host availability, but some of the wasps were kept at a pre-trial temperature of 21C and the others were kept at a pre-trial temperature of 26C (Table 1). Thus, all females were observed at a low foraging temperature, but only those kept at 26C before the trial had experienced a decrease in temperature. Wasps were kept in two different pre-trial rooms and the temperatures of these rooms were interchanged each week. Every day, wasps kept at a pre-trial temperature of 21 or 26C were observed alternately to minimise bias due to possible diurnal variation in activity patterns. All trials were run blind for the mode of reproduction of the wasp tested. Experiments were performed for 3 consecutive weeks, with a total of 14 days of observation.
model accurately fitted the data. Paired comparisons were performed by means of Student’s t-tests and by modifying the contrast matrix when required (Crawley 2002). In experiment 1, the effects on patch residence time of the reproductive mode of the foraging female, the number of eggs laid (a time-dependent covariate), host availability and foraging temperature, were analysed by means of Cox’s proportional hazards model (Cox 1972). Interactions between the reproductive mode and the other covariates were included to determine whether the patch residence times of thelytokous and arrhenotokous wasps differed under different environmental conditions. The Cox model assumes that the leaving tendency [or hazard rate = h(T), the probability per unit time of the wasp leaving the patch, provided that it is still on that patch] is the product of a baseline leaving tendency [h0(T)] and a factor the joint effect of the p Pprepresenting covariates exp b z : T is the time elapsed since i i¼1 i the wasp entered the patch. b values are the regression coefficients estimating the relative contribution of the p covariates (zi). These coefficients are interpreted through eb, the hazard ratio. A hazard ratio >1 indicates that the covariate concerned increases the tendency of the wasp to leave, thereby reducing patch residence time. Conversely, a hazard ratio
