- Top of page
- Methods and materials
Many studies have reported that plants play an influential role in mediating a suite of behavioral and physiological interactions amongst the herbivores feeding on them and natural enemies of the herbivores. Plants possess a range of characteristics that may directly affect herbivore–enemy interactions. For example, morphological traits such as hairs, trichomes, or adhesive glands on the leaf surface may inhibit herbivore colonization or movement (Ribeiro, Pimento & Fernandes 1994; van Dam & Hare 1998) whilst simultaneously impeding the searching efficiency of predators or parasitoids (Romeis, Shanower & Zebitz 1998; Sutterlin & van Lenteren 2000; Fordyce & Agrawal 2001).
In addition to morphological defences, many plants produce a range of toxic secondary compounds (= allelochemicals) which may be constitutively expressed, or induced in response to herbivore damage (Karban & Baldwin 1997). These compounds can act as feeding deterrents or significantly alter the physiology and development of some herbivores, through reduced rates of growth, smaller adult size and increased mortality (Giamoustaris & Mithen 1995; van Dam, Hadwich & Baldwin 2000). However, many specialist herbivores are known to exhibit some degree of adaptation and may even require the compounds to be present as feeding stimulants (Renwick & Lopez 1999; Roessingh et al. 2000; Stadler 2000). Plant toxins may be sequestered in the haemolymph or body tissues of resistant herbivores, thus providing them with some degree of protection from their natural enemy complex (Tullberg & Hunter 1996; Wink et al. 2000; Muller et al. 2001).
Because arthropod herbivores derive their nutrition from plants, natural enemies using herbivores as prey or hosts will also obtain their nutrition indirectly from plants (Bottrell, Barbosa & Gould 1998). Several studies have reported that allelochemicals in host or prey diet negatively affects the growth, development, survival and morphology of their predators and parasitoids (Duffey, Bloem & Campbell 1986; Gunasena, Vinson & Williams 1990; Paradise & Stamp 1993; Havill & Raffa 2000). However, in many instances, these effects are also (as in generalist herbivores) more pronounced in generalist natural enemies, whereas many specialists are apparently adapted to plant toxins (Barbosa, Gross & Kemper 1991; Paradise & Stamp 1993; Vinson 1999).
Ultimately, the net effect of secondary plant compounds on plant fitness depends on how each trophic level separately responds to them. Although much attention has been paid to organisms in the third trophic level (e.g. Price et al. 1980; Turlings & Benrey 1998; Dicke 1999; van Loon, de Boer & Dicke 1999; Turlings et al. 2002), foodwebs of course do not stop there (Sullivan & Volkl 1999; Brodeur 2000). Many parasitoids (in the third trophic level) are attacked by one or more species of obligate hyperparasitoids (in the fourth trophic level) which may in turn be attacked by facultative (or tertiary) hyperparasitoids (Brodeur 2000). As pointed out by Sullivan & Volkl (1999), the dynamics of tri-trophic interactions involving plants, herbivores and parasitoids may be profoundly affected by hyperparasitoids. They may exert a significant negative effect on plant-fitness by removing parasitoids or predators of the herbivores, (top-down regulation) or else plant allelochemicals may be transferred vertically through herbivores feeding on plants to the third trophic level and perhaps higher (bottom-up regulation). However, thus far there has been no evidence that plants may be able to affect the fourth trophic level in the latter way (Brodeur 2000).
This study examines the effects of life-history variation in two closely related plant species on the development of an endoparasitoid (in the third trophic level) and its secondary (= pseudo) hyperparasitoid (in the fourth trophic level), when reared from a primary herbivore host. As far as we know, this is the first study to link the performance of an organism in the fourth trophic level to interspecific variation in foodplant quality, as mediated through a herbivore and its primary parasitoid. Conspicuously, hyperparasitoids have been mostly excluded from studies examining multitrophic interactions, even though they may play a profoundly important role in mediating community processes (Sullivan & Volkl 1999).
Cotesia glomerata L. (Hymenoptera: Braconidae) is a fairly host-specific gregarious endoparasitoid that attacks young larvae of several species of pierid butterflies, with its preferred host being Pieris brassicae L. (Lepidoptera: Pieridae). Lysibia nana Gravenhorst (Hymenoptera: Ichneumonidae) is a solitary hyperparasitoid of newly cocooned pre-pupae and pupae of several microgastrine braconids, including C. glomerata. Cotesia glomerata and L. nana are synovigenic parasitoids; adult females emerge with only a small fraction of their eggs fully mature and ready to be laid (Jervis et al. 2001). More eggs are matured over the days immediately following eclosion, although in the case of L. nana potential fecundity is limited because female wasps produce large, yolk-rich anhydropic eggs (as opposed to C. glomerata, which produces smaller, yolk-poor hydropic eggs).
Pieris brassicae is a specialist herbivore that feeds exclusively on plants producing inducible glycoside toxins known as glucosinolates (Renwick & Lopez 1999). Larvae of P. brassicae feed gregariously and are pests of cultivated crucifer species such as cabbage (Brassica oleracea L.), but also readily feed on wild species such as the black mustard (B. nigra L.). Like many gregarious folivores, older larvae of P. brassicae attain brightly contrasting colors which may serve to advertise their possession of repellant chemical defences that are sequestered from the plant (Aplin, d’Arcy-Ward & Rothschild 1975; Hunter 2000).
Although glucosinolates are known to be synthesized by plants in several plant families, they are best studied in the family Brassicaceae. Glucosinolates and their breakdown products play a role in mediating plant–phytophage interactions. (Rask et al. 2000). For example, they have been shown to act as feeding deterrents or to exhibit detrimental effects on the growth and development of herbivores, pathogens and nematodes (Potter et al. 1999; Li et al. 2000; Buskov et al. 2002). Alternatively, they are known to act as oviposition and feeding stimulants for specialist herbivores such as P. brassicae and a related species, P. rapae (van Loon & Schoonhoven 1999; Renwick & Lopez 1999; Stadler 2000). Artificial selection in B. oleracea has resulted in significant morphological variations in the shape and structure of different parts of the plant (Hawkes 1983; Benrey et al. 1998). Several species of cultivated and wild crucifers have also been shown to produce significantly different concentrations of nitrogen (Slansky & Feeny 1977) and glucosinolates (Simmonds 1979).
Thus far, few investigations have correlated the performance of herbivores and their natural enemies to interspecific differences in plant nutritional quality and allelochemistry (but see Karowe & Schoonhoeven 1992, and Benrey et al. 1998). The two crucifer species studied here also have contrasting life-histories: B. nigra is a summer annual whereas B. oleracea is a spring perennial. Different life-histories presumably reflect adaptations to environments with different herbivore and pathogen pressures. Therefore, we may expect these species to exhibit different ecological strategies with regard to induced responses (Karban & Baldwin 1997).
In this study we first compare levels of glucosinolates in young shoots of B. oleracea obtained from seeds of a feral (= established) population, and from a seeds of a naturally occurring population of B. nigra. We then examine the clutch size and pupal mass of C. glomerata when reared from cohorts of P. brassicae that were separately reared on B. oleracea and B. nigra. We lastly compare the survival, adult wasp body mass and development time of L. nana when reared from young (c. 24 h) and old (c. 72 h) pupae of C. glomerata. The aim is to determine if differences in the quality of the host foodplant separately affects the performance of the primary parasitoid and its secondary hyperparasitoid, as mediated through the herbivore host. Lastly, we assess the influence of plant defensive strategies, focusing on the role of allelochemicals, as they affect the development and fitness of herbivores and their natural enemies. The role that plant secondary compounds play in mediating multi-trophic interactions is discussed.
- Top of page
- Methods and materials
The results of this investigation provide the most striking evidence yet that interspecific differences in primary and/or secondary chemistry amongst closely related plant species manifest effects on the performance of organisms several trophic levels higher. Levels of glucosinolates in the shoots of B. nigra were significantly higher than in corresponding shoots of B. oleracea. Several studies have reported an even greater difference in foliar glucosinlate concentrations between the two species (Bradburne & Mithen 2000; Moyes et al. 2000; Francis et al. 2001). However, the differing effects of plant quality in the two crucifers were much more apparent in L. nana than in either C. glomerata or P. brassicae. Hyperparasitoids developing in young cocoons of C. glomerata originating from B. nigra suffered higher mortality and were much smaller than conspecifics developing in B. oleracea. By contrast, in older (c. 72 h) hosts, mortality and body size in L. nana were similar in both Brassica species, although size was depressed and mortality higher compared with younger hosts.
In L. nana, adult wasp size also increased markedly with host size, whereas development time was also extended in larger and older hosts. For many idiobiont species host size determines the size of the resultant parasitoid adult(s), as has been demonstrated in both solitary and gregarious species (Salt 1940; Corrigan & Lashomb 1990; Harvey & Gols 1998; Otto & Mackauer 1998). Since the host represents an ostensibly ‘fixed’ package of resources, larger hosts are expected to provide more resources than smaller hosts, and parasitoids developing on larger hosts are thus expected to benefit in terms of increased size and presumably fitness (Visser 1994). However, larger (and/or older) hosts may take longer to consume (and assimilate), accounting for the increased development time (Mackauer & Sequeira 1993).
As suggested by Benrey et al. (1998) cultivated species and their wild relatives offer considerable promise for examining interactions over three (or more) trophic levels. Artificially selected plants are known to exhibit allometric changes in nutrient allocation and in the potency of allelochemicals which deter herbivores (Evans 1993; Stowe 1998). By contrast, wild plants have potentially evolved for millions of years under natural selection against a range of natural enemies, such as herbivores and pathogens. In line with these observations, some workers have reported that domesticated plants contain lower concentrations of defensive compounds than their wild relatives (Rhoades 1979; Evans 1993). This suggests that artificial selection is frequently incompatible with defence and could help to explain why many herbivores, especially generalists, thrive in agricultural systems (Benrey et al. 1998).
Domestication has potentially reduced the effectiveness of direct plant defences in the Brassicaceae. Sznajder & Harvey (unpublished) found that the development of a generalist herbivore, Spodoptera exigua Hubner, and its solitary endoparasitoid, C. marginiventris Cresson, were much more negatively affected on a wild population of B. nigra, compared with an escaped culitvar of B. oleracea. Both species suffered higher mortality, attained lower body weights, and took longer to complete development when reared on B. nigra. Similarly, Benrey et al. (1998) reported that C. glomerata and its host, P. rapae, developed less successfully on the wild cruficer species Lunaria annua L. than on a cultivar of B. oleracea. By contrast, we found little difference in the performance of P. brassicae and C. glomerata on either B. nigra and B. oleracea. This could be because both plants are much more closely related than those used by Benrey et al. (1998), or because of differences in the degree of adaptation in the two pierids to variation in the phytochemistry (and potential toxicity) of the two plants.
Several other studies have confirmed that artificial selection for reduced levels of allelochemicals, often demonstrated in agricultural systems, can profoundly affect plant–herbivore interactions. For instance, Giamoustaris & Mithen (1995) and Stowe (1998) developed lines in the crucifers Brassica napus L. and Brassica rapa L., respectively, containing quantitative differences in foliar glucosinolate concentration. Both studies revealed that increases in foliar glucosinolates significantly reduced the rate of feeding and damage incurred by generalist herbivores, although the effects on specialists were idiosyncratic. Based on his results, Stowe (1998) argued that herbivores should impose strong selection for increased glucosinolate production in crucifers, something frequently incompatible with the directed selection for the increased production of specific types of crops. Furthermore, if artificial selection can alter the chemical profiles of domesticated plants compared with their wild relatives, it is likely that cultivars will also be more palatable for third and higher trophic levels as well, as has been demonstrated here and by Benrey et al. (1998).
Although it is very little studied, the host range of L. nana is known to include several primary parasitoid species in the higher Microgastrinae. In turn, these parasitoids attack hosts that exhibit differing degrees of dietary specialization. C. glomerata has a very narrow host range, and includes only larvae of pierid butterflies feeding on plants containing glucosinolates (van Loon & Schoonhoven 1999). By contrast, C. congregata Say parasitizes oligophages including Manduca sexta L., a species which feeds on a wide variety of plants containing alkaloid toxins (Yang, Stamp & Osier 1996). Consequently, L. nana must itself be adapted to potentially dramatic differences in host quality related ultimately to the diet of the primary herbivore host.
On the basis of its trophic status and comparatively broad host range, we conclude that L. nana is likely to be less well adapted to the higher toxicity of B. nigra than are the more specialized P. brassicae and C. glomerata. This may be because C. glomerata possesses an efficient regulatory mechanism that diminishes the potential hazard of high chemical loads in the host, or else because both species are able to store allelochemicals in their body tissues, presumably in a defensive capacity. Aplin et al. (1975) investigated the possibility for sequestration of glucosinolates and isothiocyanates in P. brassicae and P. rapae reared from B. nigra. They found evidence for presence of the glucosinolate sinigrin (allylglucosinolate) and allylisothiocyanate in the pupae of both pierid species. Moreover, allyl isothiocyanate was detected in whole-body extracts of larvae and adults of P. brassicae (Aplin et al. 1975). The presence of glucosinolates in haemolymph of the larval sawfly, Athalia rosae L. has been recently demonstrated by Muller et al. (2001). The authors suggest that sequestration also appears to have a defensive function in this species (see also a similar study by Aliabadi, Renwick & Whitman 2002).
Thus far, few studies have found evidence for the transfer of secondary plant compounds from second to higher trophic levels, although it has been inferred (Meiners et al. 1997). In particular, endoparasitoids make excellent candidates for examining the sequestration of secondary plant compounds from the host to the third trophic level. In endoparasitoids, the anal end of the larval alimentary tract is not connected externally until after pupation, which prevents the excretion of faeces that would ultimately lead to contamination and subsequent bacterial infection of internal host tissues and thus reduce host quality (Quicke 1997). Consequently, parasitoid larvae are expected to store plant compounds obtained from feeding on host tissues until at least 24–36 hours after cocoon formation and pupation, when voiding of the meconium (= excretion of wastes) occurs. In the case of C. glomerata, it is therefore likely that allelochemicals are stored in the body tissues during development but may be excreted with frass after pupation. This could explain why the detrimental effects of primary host diet on hyperparasitoid size and survival diminished in older hosts.
Interspecific differences in levels of allelochemicals amongst closely related plants (as mediated through artificial selection) may have a number of ecological consequences. First, specialist herbivores like P. brassicae, which sequester plant compounds and advertise their presence through aposematic coloration, may be at a competitive disadvantage when feeding on less well-defended plants compared with cryptic herbivores. The former species are effectively advertising a toxicity which may be diluted or absent entirely. However, this apparency may benefit the efficacy of generalist predators (and hyperparasitoids like L. nana) which would otherwise avoid them. Second, reduced levels of allelochemicals in cultivated plants enable generalist herbivores and their natural enemies to persist in agroecosystems whereas in natural ecosystems they may be comparatively rare.
It is known that, in agroecosystems, larvae and pupae of C. glomerata suffer heavily from hyperparasitism, often reducing parasitoid survival by 90% or more (J.A. Harvey, pers. obs.). If this is a general rule, then these systems probably represent population ‘sinks’, and parasitoid survival may rest on the availability of hosts feeding on more toxic wild plants which are less attractive to generalist predators and hyperparasitoids. Furthermore, the dynamics (and persistence) of tightly linked multitrophic interactions, such as the system studied here, will be influenced by the ability of herbivores and their antagonists to exploit natural plant communities which are considerably more heterogeneous than simple monocultures that frequently characterize agroecosystems.
Many idiobionts, and in particular idiobiont hyperparasitoids, are known to exhibit reduced fecundity compared with lower trophic levels (Brodeur 2000). In the case of L. nana, lifetime reproductive success under optimal conditions of constant food and unlimited hosts is between 60 and 100 (J.A. Harvey, unpubl.). Moreover, female wasps are able to sequentially parasitize more than 20 hosts in the laboratory, provided these hosts are available. Because L. nana typically exploits gregarious parasitoid hosts (e.g. C. glomerata and C. congregata) that may produce secondary broods of > 20 in nature, it is highly likely that a foraging female hyperparasitoid may encounter and exploit only one or two cocoon ‘clusters’ during her lifetime. This means that her progeny may be generated from a single primary parasitoid mother, a single secondary host and a single foodplant. This raises questions as to the effects of specific or restricted genotypes of lower trophic levels on hyperparasitoid performance, given that most other primary parasitoids may attack many hosts feeding on many individual plants.
In summary, we have reported that plant quality can affect the development of organisms several trophic levels higher, even when the performance of intermediate trophic levels is mostly unaffected. Understanding the many factors influencing the dynamics of multitrophic interactions involving plants, herbivores, parasitoids and hyperparasitoids is a fertile area for future research. Perhaps one of the most challenging areas in contemporary ecology concerns the relative importance of ‘top-down’ and ‘bottom-up’ effects mediating trophic interactions and regulating the structure of communities and food webs. It is known that hyperparasitoids may have a considerable negative influence on plant fitness through reducing the abundance of parasitoids that attack the herbivores feeding on the plants (Rosenheim 1998; Sullivan & Volkl 1999). However, we have demonstrated that plant fitness can be enhanced if secondary compounds more detrimentally affect the fourth trophic level than intermediate levels in food webs. For this to happen, these compounds need to be vertically transferred through herbivores and their parasitoids. It is hoped that further studies will elucidate the relative importance of these forces incorporating measures of plant defence as it affects life-history characteristics of herbivores and their antagonists.