A trade-off between quantity and quality of offspring in haematophagous ectoparasites: the effect of the level of specialization

Authors

  • Irina S. Khokhlova,

    1. Wyler Department of Dryland Agriculture, French Associates Institute for Agriculture and Biotechnology of Drylands, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede-Boqer Campus, Midreshet Ben-Gurion, Israel
    Search for more papers by this author
  • Shai Pilosof,

    1. Mitrani Department of Desert Ecology, Swiss Institute for Dryland Environmental and Energy Research, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede-Boqer Campus, Midreshet Ben-Gurion, Israel
    Search for more papers by this author
  • Laura J. Fielden,

    1. School of Science and Math, Truman State University, Kirksville, MO, USA
    Search for more papers by this author
  • A. Allan Degen,

    1. Wyler Department of Dryland Agriculture, French Associates Institute for Agriculture and Biotechnology of Drylands, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede-Boqer Campus, Midreshet Ben-Gurion, Israel
    Search for more papers by this author
  • Boris R. Krasnov

    Corresponding author
    1. Mitrani Department of Desert Ecology, Swiss Institute for Dryland Environmental and Energy Research, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede-Boqer Campus, Midreshet Ben-Gurion, Israel
    Search for more papers by this author

Summary

  1. Theory predicts an adaptive trade-off between quantity and quality of offspring if mothers can reliably predict the offspring environment.
  2. We studied egg production and quality of offspring in two flea species (host-specialist Parapulex chephrenis and host-generalist Xenopsylla ramesis) exploiting eight rodent species. We evaluated quality of new imagoes via their developmental time, size (length of a femur as a proxy) and resistance to starvation without a blood meal.
  3. We predicted that the offspring quality would increase with (i) a decrease in the number of eggs produced by mothers and (ii) an increase in phylogenetic distance between maternal host and principal host of a flea. We also predicted that negative relationships between offspring quality and either maternal egg production effort or phylogenetic distance between maternal host and the principal host or both would be manifested stronger in host-opportunistic than in host-specific fleas.
  4. The highest number of eggs produced per female flea was accompanied by the longest duration of development and the smallest offspring in X. ramesis, while P. chephrenis that hatched from larger clutches survived for less time under starvation. Although there was no significant effect of host species on any dependent variable, association between offspring quality and phylogenetic distance of the maternal host from the principal host of a flea was found in X. ramesis (but not P. chephrenis) with new imagoes being larger if their maternal hosts were phylogenetically distant from the principal host.
  5. Our results demonstrated stronger trade-off between quantity and quality of offspring in a generalist than in a specialist flea, supporting the association between life-history plasticity and generalist feeding strategy.

Introduction

As early as in the middle of the last century, Lack (1947) proposed the hypothesis of a trade-off between quantity and quality of offspring. Initially, this hypothesis was based on studies of birds, but later was accepted as a common rule for many species (Smith & Fretwell 1974). Assuming resource limitation for each individual female, an increased investment in the number of offspring inevitably leads to a decreased investment into each individual offspring. Investment into an individual offspring affects its subsequent survival and performance. As a result, fitness of a female is determined not by the mere number of offspring produced but, rather, by how many of the offspring will breed and what will be their lifetime reproductive success, that is, quality (Roff 2002). The trade-off between quantity and quality of offspring has been demonstrated for numerous animal species both free-living (e.g. Gustafsson & Sutherland 1988; Koskela 1998; Allen, Buckley & Marshall 2008) and parasitic (e.g. Poulin 1995), as well as for pre-industrial humans (Gillespie, Russel & Lummaa 2008).

The offspring quantity vs. quality hypothesis, as formulated by Lack (1947), suggested that the fecundity of a species in its natural population reflects a balance between offspring number and quality that (i) maximizes fitness and (ii) as such has been favoured by natural selection (Stearns 1992). However, Parker & Begon (1986) argued that environmental conditions should strongly affect optimal offspring quality (e.g. size), so condition-dependent adjustment of, for example, the size of the offspring is expected. This idea was supported by several empirical studies that have demonstrated intraspecific variation in the number and size of offspring, with different optimal offspring size favoured by selection in different environments (Fleming & Gross 1990; Braby 1994; Fox, Thakar & Mousseau 1997; see review in Fox & Czesak 2000).

Natural selection is expected to favour females that can anticipate the offspring environment and adaptively adjust their quality (Parker & Begon 1986; Mousseau & Dingle 1991). For example, female seed beetles were able to adjust the size of their eggs according to the host plant (Fox, Thakar & Mousseau 1997). This adaptive response of a female seed beetle was possible because the offspring environment could be predicted by the laying female as the offspring feed only on seeds of plant species on which eggs are laid by the female. Similar patterns can be expected in nidicolous, nest or permanent body parasites with direct life cycles such as fleas, lice or gamasid mites (Marshall 1981; Krasnov 2008). In these parasites, the environment of a mother (i.e. host species) can be a reliable predictor for the environment of the offspring because the probability that they will feed on the same host species as the mother is high. However, the effect of the maternal host species on the offspring quality in parasites is poorly (if at all) known.

Fleas (Siphonaptera) represent a convenient model taxon for investigation of maternal effects with respect to adjustment of quality and quantity of the offspring. These insects are holometabolous haematophages that are most abundant and diverse on small and medium burrowing mammalian species. The majority of flea species alternate periods of on-host occupancy and time spent off the host in the burrow or nest. Time of an uninterrupted stay on the host body varies among flea species being longer in the so-called body or fur fleas and shorter in the so-called nest fleas (see review in Krasnov 2008). The female flea oviposits while on or off the host but, even in the former case, the eggs typically drop off into the burrow/nest of the host. In nearly all cases, larval and pupal development is entirely off-host (Krasnov 2008). After emergence from the pupa and cocoon, new adult fleas must locate a suitable vertebrate host. Conspecific fleas as any other parasites are unequally distributed among different host species. Traditionally, a host in which a parasite attains the highest abundance and/or prevalence is considered to be its principal host, while other hosts exploited by the parasite are considered to be auxiliary hosts (Dogiel, Petrushevski & Polyanski 1961; Dogiel 1964).

In this study, we focused on the quantity (number of eggs) and the quality of new imago offspring produced by the fleas (Parapulex chephrenis and Xenopsylla ramesis) that exploited different host species. We evaluated the quality of newly emerged fleas via their developmental time, size and resistance to starvation without a single blood meal. Larvae that hatch earlier may have a certain competitive advantage over larvae that hatch later due to limited amount of larval food (organic debris including partially digested blood in the faeces of the adult fleas) (Day & Benton 1980; Krasnov et al. 2005) and/or cannibalism of younger larvae by older larvae (Lawrence & Foil 2002), while an imago that emerges earlier has a better chance of locating a host and producing more new generations during the breeding season than a late-emerging imago. Larger body size is intraspecifically associated with higher fecundity in insects (Honek 1993). The ability of a new imago to survive unpredictable and lengthy periods without a blood meal reflects the amount of energetic storage in the fat tissue accumulated during pre-imaginal development (Krasnov 2008).

Both flea species in this study are common rodent parasites in the central Negev Desert. Parapulex chephrenis is a typical ‘body’ flea (Krasnov 2008). It is a host specialist found mainly on the Egyptian spiny mouse Acomys cahirinus and, to a much lesser extent, on golden spiny mouse Acomys russatus (Krasnov et al. 1997). In contrast, X. ramesis is a ‘nest’ flea (Krasnov 2008). It is a host generalist and parasitizes many rodent species but attains the highest prevalence and abundance on the gerbil Meriones crassus (Krasnov et al. 1997). Earlier (Khokhlova et al. 2012a), we studied reproductive performance of these two flea species when they exploited nine rodent hosts and found that the number of new imagoes produced by female fleas differed among host species. Furthermore, production of either eggs or new imagoes or both when feeding on an auxiliary host decreased significantly with an increase in phylogenetic distance between this host and the principal host. The number of new imagoes produced by a female flea is a net result of both quantity and quality of eggs produced by this female. For example, flea eggs may differ in their survival or resistance to microclimatic fluctuations (Krasnov et al. 2001). Based on the results of Khokhlova et al. (2012a) and aiming to disentangle quantitative and qualitative components of the effect of host species of flea fitness, we tested two alternative hypotheses regarding (i) the ability of female fleas to anticipate the offspring environment and (ii) the occurrence of the trade-off between quantity and quality of offspring. If selection favours female fleas that anticipate the environment of their future offspring and adjust the quality of the offspring in accordance with this environment, then, across host species, duration of development of new imagoes would decrease, while their size and/or time to death under starvation would increase with a decrease in the number of eggs produced by their mother. In addition, we predicted that if the trade-off between quantity and quality of the offspring occurs, then the quality of the new fleas would increase with an increase in phylogenetic distance between the maternal host and principal host of a flea because their quantity has been shown to decrease (Khokhlova et al. 2012a). Finally, given the difference in the degree of host specificity between the two fleas, we predicted that the negative relationships between the quality of the new imagos and either maternal egg production effort or phylogenetic distance between the maternal host and the principal host or both would be manifested stronger in the host-opportunistic X. ramesis than in the host-specific P. chephrenis. This is because the ability to adjust the offspring quality as a response to a heterogeneous environment (e.g. host species) may account for the generalist feeding strategy (see Fox, Thakar & Mousseau 1997).

Alternatively, if female fleas are unable to adjust the quality of the offspring in accordance with their future environment, then a decrease in quantity of offspring in unfavourable hosts due to physical, physiological or immunological incompatibility will be accompanied by a decrease in their quality. Consequently, quality of new fleas would decrease with (i) a decrease in the number of eggs produced by their mother and (ii) an increase in phylogenetic distance between the maternal host and principal host of a flea. This pattern will not depend on the degree of host specialization of a flea.

Materials and methods

Experimental animals

We used fleas (X. ramesis and P. chephrenis) and eight murine rodents from our laboratory colonies. Among the latter, there were five gerbils [M. crassus (mean body mass = 138·8 g), Gerbillus dasyurus (mean body mass = 29·9 g), Gerbillus andersoni (mean body mass = 33·5 g), Gerbillus pyramidum (mean body mass = 53·2 g), and Gerbillus nanus (mean body mass = 34·9 g)], two spiny mice [A. cahirinus (mean body mass = 50·5 g) and Acomys russatus (mean body mass = 70·0 g)] and a house mouse [Mus musculus (mean body mass = 16·5 g)]. Details on breeding and maintaining the colonies have been described in earlier publications (e.g. Khokhlova et al. 2012a,b,c). All fleas and rodents used in experiments were selected randomly from colonies. The fleas were 24–48 h old and were never fed (and, consequently, bred) prior to experiments. We used male rodents 6–8 months old. Each treatment with each flea species and each host species was replicated 5–9 times.

Experimental design and procedures

The procedure for obtaining flea eggs was fully described in Khokhlova et al. (2012a). In brief, an individual rodent was placed in a plastic cage (60 cm by 50 cm by 40 cm) with a floor of 3–5 mm of clean sand covered by a wire mesh (5 mm by 5 mm). Then, 20–50 newly emerged (24–48 h old) female and 10–30 male fleas (P. chephrenis or X. ramesis) were released into a cage with a rodent and allowed to feed for 3 days. Variation in the number of fleas applied to a host animal was related to difference in body mass among rodent species. The lowest number of fleas was released on the smallest species (M. musculus), while the highest number of fleas was released on the largest species (M. crassus). Under these conditions, fleas start to oviposit no sooner than on the second day of the stay on a host (preliminary observations). In majority of flea species, blood feeding is necessary for dissolution of a testicular plug in newly emerged males and development of ovaries in newly emerged females (see Krasnov 2008 for review). Consequently, fleas that stayed on a host during 3 days were undoubtedly able to copulate and produce eggs. In other words, the first phase of the experiment served to make fleas ready for oviposition.

To estimate egg production effort per female flea per day while guaranteed equal feeding chances of fleas, we collected fleas from both the rodent's body (over a white plastic pan using a toothbrush until no flea was recovered) and cage substrate, placed fleas recovered from the same rodent individual in 100 mL glass vials and transferred them into an incubator (FOC225E; Velp Scientifica srl, Milano, Italy) at 25 °C air temperature and 90% relative humidity (RH) for 24 h. On the next day, we placed each individual rodent in a wire mesh (5 mm by 5 mm) tube (15 cm length and 5 cm diameter or 10 cm in length and 2 cm depending on rodent's size; see Khokhlova et al. 2012a) that limited movement and did not allow self-grooming. Tubes with rodents were placed in individual white plastic baths. Fleas collected earlier from this rodent individual were then released into its hair. After feeding on a host for 2 (X. ramesis) or 6 (P. chephrenis) h (time necessary for satiation of the majority of individuals of the two species; preliminary observations), fleas were collected and examined for blood in the midguts under light microscopy (40× magnification). Fleas of each group that took a blood meal were returned to a new vial and transferred again into an incubator for 24 h, after which time we checked the vials, counted newly laid eggs and calculated the mean number of eggs produced per female. We used these values as estimates of egg production effort by a mother flea.

Each vial with eggs was filled with a 3 mm layer of sand and larval food medium (94% dry bovine blood, 5% millet flour and 1% grinded excrements of the host on which fleas fed) and covered by perforated lids. Vials were then maintained at 25 °C air temperature and 90% RH.

Starting from the 18th day after eggs were produced (approximately a week less than minimal duration of metamorphosis; Krasnov et al. 2001), we checked each vial daily either until all adults emerged (i.e. the number of emerged adults was equal to the number of eggs) or for 60 consecutive days. We recorded the day of emergence of each new imago and counted new imagos produced by each group of parent fleas. After emergence, each adult was placed in an individual eppendorf vial and left into the incubator at the same air temperature and RH. Vials with newly emerged adults were checked daily until all adults died. After the death of each imago, we identified its sex by examining its genitalia under light microscopy.

After the death of an adult, we estimated its body size by measuring the maximal length of its right hind femur. The length of a femur was shown to be a reliable indicator of body size in fleas because these traits were strongly correlated (Krasnov et al. 2003; Khokhlova et al. 2010a). Length of femur was measured on a screen (using a digital microscope camera Moticam 2000 with the Motic Images Plus 2.0ML program; Motic, Speed Fair Cp., Ltd., Causeway Bay, Hong Kong) to the nearest 0·01 mm under magnification of ×40 and with calibration using an object-micrometer. In total, we used data on 460 newly emerged P. chephrenis and 1215 newly emerged X. ramesis.

Data analyses

Data were analysed separately for P. chephrenis and X. ramesis. We analysed duration of development, time of survival under starvation and femur length (dependent variables) using linear mixed-effects models (LME; Zuur et al. 2009) with rodent host species, body mass of an individual host, initial flea number, offspring flea sex and egg production effort of a parent flea as fixed explanatory variables. The individual number of each rodent was included as a random factor in each model allowing us to control for the individual variation among rodents within species. This was because we fed a group of parent fleas on the same host individual and thus needed to account for within individual host non-independence of flea offspring. Inclusion of flea sex as a fixed variable in the models was necessary because all dependent variables have been shown to differ substantially between male and female fleas. Flea females are (i) larger (Krasnov et al. 2003); (ii) usually develop faster (Khokhlova et al. 2010a); and (iii) usually survive longer under starvation (Khokhlova et al. 2009) than males, all else being equal. For each flea offspring, maternal egg production effort was taken as the mean number of eggs produced per mother flea in the respective group of fleas on the day when the egg from which this offspring hatched. We fitted the models using the lme function as implemented in ‘nlme’ package (version 3.1-106; Pinheiro et al. 2013) in R (version 2.14; R Development Core Team 2012). After running the models, we used Tukey's HSD tests for multiple comparisons adjusted for mixed-effects models to test for differences between rodent species in their effect on dependent variables using the glht function in ‘multcomp’ R package (version 1-2-15; Hothorn, Bretz & Westfall 2008). Reference levels for ordinal explanatory variables of rodent species and flea sex were established arbitrarily as A. cahirinus and female, respectively.

To test for the association between offspring quality in an auxiliary host and its phylogenetic relatedness to the principal host, we calculated average values of dependent variables within flea sex and host species and regressed them against phylogenetic distance of a host from the principal host separately for male and female fleas and for P. chephrenis and X. ramesis. We calculated phylogenetic distances between a principal host (A. cahirinus for P. chephrenis and M. crassus for X. ramesis) and each of the remaining seven host species from branch lengths of a phylogenetic tree based on Jansa & Weksler (2004), Chevret & Dobigny (2005), Bininda-Emonds et al. (2007) and Abiadh et al. (2010). The tree and details of calculation of phylogenetic distance can be found in Khokhlova et al. (2012b). The explanatory variable in the regression analyses included phylogenetic information and, consequently, there was no need for further phylogenetic correction.

Results

In both flea species, host species did not affect any of the three dependent variables as demonstrated by LME (Table 1). This was supported by lack of differences between host species in their effect on any dependent variable indicated by multiple comparisons (Tukey's HSD tests; P = 0·49–0·99 for duration of development, P = 0·83–0·99 for time of survival under starvation and P = 0·42–0·99 for femur length).

Table 1. Summary of linear mixed-effects models of duration of development (DD), time of survival under starvation (SS) and length of right hind femur (FL) in offspring of fleas (Parapulex chephrenis and Xenopsylla ramesis) as affected by host species (HS), flea sex (FS), egg production effort (EPE), rodent body mass (RBM) and flea density (FD). Reference levels for categorical independent variables were Acomys cahirinus for HS and female for FS
Flea speciesDependent variableExplanatory variableCoefficient estimate ± SE P
Parapulex chephrenis DDHS−2·22 ± 2·77 to 6·89 ± 4·77>0·15
FS4·11 ± 0·43<0·01
EPE−0·52 ± 0·470·27
RBM−0·08 ± 0·050·12
FD0·04 ± 0·070·51
SSHS−3·02 ± 4·61 to 1·52 ± 2·18>0·48
FS2·11 ± 0·44<0·01
EPE−1·20 ± 0·470·01
RBM0·05 ± 0·050·36
FD−0·05 ± 0·100·66
FLHS−0·008 ± 0·01 to 0·01 ± 0·01>0·11
FS−0·08 ± 0·002<0·01
EPE−0·003 ± 0·0030·26
RBM0·00009 ± 0·00030·81
FD0·00008 ± 0·00050·87
Xenopsylla ramesis DDHS−5·31 ± 3·38 to 0·51 ± 2·29>0·12
FS5·11 ± 0·24<0·01
EPE1·70 ± 0·36<0·01
RBM0·02 ± 0·030·57
FD0·09 ± 0·080·24
SSHS−4·38 ± 5·60 to 4·43 ± 11·63>0·44
FS1·28 ± 0·650·04
EPE1·26 ± 1·000·21
RBM−0·06 ± 0·110·57
FD0·54 ± 0·370·10
FLHS−0·02 ± 0·02 to −0·003 ± 0·01>0·18
FS−0·07 ± 0·001<0·01
EPE−0·01 ± 0·001<0·01
RBM0·0002 ± 0·00010·14
FD−0·0004 ± 0·00040·24

The effect of flea sex was significant for all dependent variables and for both fleas (Table 1). The signs of the estimated coefficients suggested that, in general, male fleas had longer development times, died under starvation faster, and were smaller than female fleas.

Egg production effort of mother fleas affected two variables of offspring quality (duration of development and femur length) in X. ramesis and one variable (survival under starvation) in P. chephrenis (Table 1). The signs of the estimated coefficients indicated that a higher number of eggs produced per female flea was accompanied by a longer duration of development and a smaller size of offspring in X. ramesis, while P. chephrenis that hatched from larger clutches survived under starvation shorter than conspecifics that hatched from smaller clutches (Table 1). Illustrative example with length of right hind femur in X. ramesis is presented in Fig. 1.

Figure 1.

Relationship between length of right hind femur of male (closed circles, solid line) and female (open circles, dashed line) offspring of Xenopsylla ramesis and maternal egg production effort.

No effect of either rodent body mass or flea density was found for any dependent variable (Table 1).

A significant association between average values of the variables describing the quality of offspring produced by fleas fed on a given host species and phylogenetic distance of this host from the principal host of a flea was found for the length of the right hind femur (r2 = 0·62, F = 9·8, P = 0·02 for males and r2 = 0·63, F = 10·1; P = 0·02 for females) but not for the other two dependent variables in X. ramesis (r2 = 0·02–0·16, F = 0·1–1·2, > 0·32 for all). Both male and female new imagos were larger if their maternal hosts were phylogenetically distant from the principal host (M. crassus) (standardized slope ± SE = 0·78 ± 0·25 for males and standardized slope ± SE = 0·79 ± 0·24 for females) (Fig. 2). There was no association between average duration of development, time of survival under starvation or length of the right hind femur of the offspring and the distance of the maternal host from A. cahirinus (the principal host) in P. chephrenis (r2 = 0·001–0·42, F = 0·01–3·6, > 0·11 for all).

Figure 2.

Relationship between length of right hind femur of male (closed circles, solid line) and female (open circles, dashed line) offspring of Xenopsylla ramesis fed on eight murine hosts and phylogenetic distance of these hosts from the principal host (Meriones crassus).

Discussion

This study supported our main hypothesis. First, we found evidence of a trade-off between quantity and quality of flea offspring, although only in some but not other indicators. Secondly, this trade-off was manifested more strongly in the host-generalist X. ramesis than in the host-specialist P. chephrenis. The occurrence of a trade-off between quantity and quality of flea offspring suggested also that the alternative hypothesis about inability of female fleas to adjust the quality of the offspring in accordance with their future environment is unlikely. Below, we will discuss our results from physiological, ecological and evolutionary perspectives.

Physiological mechanisms

In holometabolous insects, the quality of a new imago depends strongly on the nutritional status of a larva (Emlen 1994; Karino, Seki & Chiba 2004). In the majority of flea species (including those used in this study), the larvae are not parasitic and feed on organic debris and materials in the nest of the host. Under these circumstances, maternal control on the larval nutritional environment is limited. However, in some fleas, females void their gut content (i.e. non-digested or partly digested blood) and expel faecal pellets near the clutch which may be used by the larvae (Hinkle, Koehler & Kern 1991; Silverman & Appel 1994; Larsen 1995; Cox, Stewart & Macdonald 1999). Moreover, the protein content of female faeces has been found to be higher than the host blood upon which they fed (Hinkle, Koehler & Kern 1991). If mother fleas can adjust protein content of their faeces or gut voids in response to host species, then this can be considered as an indirect investment into offspring and an adaptive mechanism aimed to compensate for their low number in unfavourable hosts. This, however, has never been studied and warrants further investigation.

Direct maternal investment into offspring is related not only to egg size but also to egg provisioning (e.g. concentration of proteins and lipids; McIntyre & Gooding 2000; Giron & Casas 2003). Although larger egg size is generally considered as an indicator of higher offspring quality (e.g. McLain & Mallard 1991), it is not always the case (Karlsson & Wiklund 1984). In fact, variation in the nutrient composition of the egg may also occur independently of its size (Giron & Casas 2003). In an earlier study (Khokhlova et al. 2013), we found that P. chephrenis laid fewer but larger eggs when it exploited an auxiliary (i.e. non-preferable; M. crassus) than a principal (i.e. preferable; A. cahirinus) host. However, we found in this study that the size of new imagos of P. chephrenis did not differ between these host species. In other words, the trade-off between the number of offspring and their size occurred at an earlier (eggs), but it disappeared at a later stage (newly emerged imago) of the offspring development. The reasons for this could be trade offs among quality-related offspring traits (Benton et al. 2005). For example, offspring may trade off their growth at larval stage against survival (Plaistow, Lapsley & Benton 2006). In addition, smaller eggs could have relatively higher energy content (Geister et al. 2008a), so that the new imago from smaller and larger eggs would not differ in size.

Both egg size and egg composition depend on a variety of intrinsic (e.g. maternal age; Benton et al. 2005) and extrinsic factors (e.g. temperature; Geister et al. 2008b). One of the factors strongly influencing egg size and composition in insects is maternal diet (Kyneb & Toft 2006; Geister et al. 2008a). Blood composition and its physical properties differ among hosts, so that different hosts represent different diets for haematophagous arthropods (Lehane 2005). Consequently, physiological responses to different diets could result in differential maternal investment which may be manifested further in host-dependent viability of newly emerged fleas in our study.

Ecological causes and consequences

One of the key assumptions underlying the theory of adaptive trade-off between quantity and quality of the offspring is that the offspring environment can be reliably predicted by mothers (Parker & Begon 1986; Moran 1992). In other words, if the offspring environment can be assessed by mothers, they could adjust the quality of offspring accordingly. In majority of flea species, newly emerged individuals stay in the burrow of the maternal host waiting for a host. Therefore, the maternal environment (in terms of host species) is likely a good predictor of the offspring environment.

Ability to adjust offspring quality according to environmental conditions and, in particular, to compensate for low number of offspring by their higher quality may allow a species to persist under unfavourable conditions. Plasticity in response to environmental heterogeneity is well documented for both plants and animals (e.g. Bradshaw 1965; Newman 1994; Zhivotovsky, Feldman & Bergman 1996; Madec, Desbuquois & Coutellec-Vreto 2000; Finston 2004; Galloway & Etterson 2007; Baythavong & Stanton 2010). From a parasite's perspective, heterogeneity of environment is variation among co-occurring host species. In fact, plasticity in response to host species has been reported for parasites (Poulin & Hamilton 2000; Vignoles et al. 2004). Regarding fleas and despite their ‘philopatry’, interspecific host-to-host transfer of these insects may sometimes occur either via body contact or via hosts visiting burrows of other species (see review in Krasnov 2008). As a result, some individuals may turn up on an unexpected host and the same flea species may be found on multiple host species within a locality (e.g. Krasnov et al. 1997). These hosts represent environments of different quality because of difference in blood composition (for flea imagos) and in microclimate, substrate texture and/or amount of organic matter (for pre-imaginal fleas) with one host providing the best overall conditions (the principal host; see Krasnov et al. 2004a; Khokhlova et al. 2012a). At the population level, flea distribution among host species is characterized by a decrease in abundance with a decrease in phylogenetic relatedness (and thus ecological, physiological and/or immunological similarity) of these hosts to the principal host (Krasnov et al. 2004a). Our results suggest that the occurrence of fleas even on the least preferable host (the farthest relative of the principal host) might, at least in part, be ensured by an adaptive adjustment of the offspring quality (that is, fewer offspring but of high quality). Allocation of resources by a female varies in a manner that allows increasing of individual fitness when parasitizing a non-preferable host by producing faster-developing and larger (X. ramesis) or more starvation-resistant (P. chephrenis) offspring.

Evolutionary implications

From the evolutionary perspective, a plastic strategy seems to be favoured mainly in a generalist, but not in a specialist parasite. According to recent theoretical studies (e.g. Fischer, Taborsky & Kokko 2011), plastic strategies of offspring quality are expected to be superior to fixed strategies when (i) environmental cues are reliable; (ii) resource availability is similar between environments; and (iii) offspring survival varies greatly between environmental states. While the former condition (i) is likely similar for both flea species (see above), the two latter conditions (ii and iii) seem to hold for X. ramesis, but not for P. chephrenis. Although the main resource provided by a host to fleas (i.e. blood) is the same for both fleas, the pattern of use of this resource differs substantially between X. ramesis and P. chephrenis. Xenopsylla ramesis spends similar amounts of energy expenditure for digestion of blood consumed from different hosts, while energy cost of blood digestion depends strongly on host identity in P. chephrenis (Khokhlova et al. 2012c). Survival of pre-imaginal X. ramesis until emergence was found to depend on the maternal host species, whereas this was not the case for P. chephrenis (Khokhlova et al. 2010b).

Host-associated adjustment of offspring quality allows X. ramesis to exploit different host species and may underlie its host opportunism. Alternatively, a generalist feeding strategy may promote the evolution of plasticity in the offspring quality (Fox, Thakar & Mousseau 1997). Although it is difficult to distinguish between cause and consequence in the evolution of phenotypic plasticity because it is complex and environment- and context-dependent (Czesak, Fox & Wolf 2006), evidence suggests that both scenarios are possible. For example, Nosil (2002) reported that generalist-to-specialist transitions were generally more frequent than specialist-to-generalist transitions among phytophagous insects, but the opposite was true in some groups of butterflies and beetles. Although host opportunism in many flea lineages tends to increase on a macro-evolutionary scale, this trend was manifested mainly in basal than in derived lineages (Poulin et al. 2006). Furthermore, evolutionary changes in the level of host specificity in fleas were found to be not constrained in a single direction, but were clearly reversible (Poulin et al. 2006).

The ability to adjust offspring quality in a host-dependent manner in a generalist flea but not a specialist flea, as found in this study, may explain, at least in part, the positive relationship between the level of host specificity of a flea and average abundance achieved in different hosts (Krasnov et al. 2004b). Among flea species, generalists exploiting many host species consistently achieve higher abundance than specialists using only one or very few host species (Krasnov et al. 2004b). However, no generality among parasite and host taxa was found for the pattern of the relationship between the number of hosts exploited and average abundance achieved in this hosts. Even in the same parasite–host association (metazoan parasites of freshwater fish), both positive (Poulin 1998) and negative (Barger & Esch 2002) correlations between the number of hosts used and average abundance of parasites were reported.

An additional reason for the lower ability of P. chephrenis to adjust their offspring quality in a host-dependent manner can be an evolutionary trade-off between investment into offspring quality adjustment and investment into the tools of attachment to the host's body. Anatomical features of fleas such as helmets, ctenidia and modifications of shape and size of spines and setae are correlated with particular characteristics of the host's fur (Traub 1980). The main hosts of P. chephrenis (Acomys) have thick but short hairs and widely spaced long rigid keratin spines from which fleas can be easily dislodged. The body of P. chephrenis is covered with sclerotized bristles that anchor the flea within the host hair and resist the host's grooming. We recognize that this explanation is highly speculative and is not supported by any hard evidence. However, many flea species possessing specialized attachment organs are characterized by high host specificity (Traub 1980).

In conclusion, our study demonstrated stronger trade-off between quantity and quality of offspring in a generalist than in a specialist flea. This supports the notion of an association between life-history plasticity and generalist feeding strategy.

Acknowledgements

The study was conducted under permits from the Israel Nature and National Parks Protection Authority (license 2012/38402) and Ben-Gurion University Committee for the Ethical Care and Use of Animals in Experiments (license IL-52-07-2009). This study was supported by the Bi-National Science Foundation (Grant no. 2008142 to BRK, ISK and LJF). The experimental procedures complied with the laws of the State of Israel. This is publication no. 813 of the Mitrani Department of Desert Ecology.

Ancillary