Effects of willow hybridisation and simulated browsing on the development and survival of the leaf beetle Phratora vitellinae

Background Interspecific hybridisation is common between many plant species and causes rapid changes in a variety of plant characters. This may pose problems for herbivores because changes in recognition characters may be poorly correlated with changes in quality characters. Many studies have examined different systems of hybrids and herbivores in attempts to understand the role of hybridisation in the evolution of plant resistance. The results from different systems are variable. Studies of hybrids between Salix caprea (L., Salicaceae) and S. repens show that they are intermediate between the two parental species in most resistence characters. However, a plants herbivore resistence depends also on its biotic and abiotic environment. Important biotic factors that may influence plant growth and plant chemistry include the interactions between different herbivores that occur through their exploitation of common host plants. Although the effects on plants of previous herbivory are likely to be strongly affected by environmental conditions, they are also species-specific. Damage may therefore have different effects on hybrids than on their parental species, and this could influence the performance of herbivores on pure and hybrid species of plants. To evaluate the effects of hybridisation on insect performance, the development and survival rates of Phratora vitellinae (L. 1758, Coleoptera: Chrysomelidae) larvae on pure S. repens, pure S. caprea and Fl hybrids of the two species was monitored. Further, to examine the effect of herbivorous mammals on the performance of the larvae, plants were damaged to simulate winter foraging by voles or spring leaf stripping by moose. Results The results show that development rates were highest on S. repens and equally low on S. caprea and the Fl hybrid. In addition, development of the plants treated to simulate mammalian herbivore damage was slower than that of corresponding controls. Conclusions The results of this experiment suggest that P. vitellinae has a higher development rate, and thus probably higher performance, on species with high concentrations of phenolic glucosides. Therefore, it would be of adaptive benefit for P. vitellinae females to have an ovipositional preference for S. repens, compared to S. caprea and intermediate preference for Fl hybrids. The faster development observed on S. repens supports the hypothesis that P. vitellinae obtains additional adaptive benefits from phenolic glucosides beyond protection against predators. Therefore, it is important to consider further factors, such as damage caused by other herbivores, when studying this hybrid complex.


Background
To understand the distribution of herbivore populations on different plants we need to consider both host selection and offspring performance on potential hosts [1]. The more similar plants are in recognition characters, and the more variable they are in quality, the more important it is for a herbivore to make appropriate host choices [2][3][4]. Herbivores' foraging and oviposition decisions are especially challenging when host plant species hybridise with non-host species. It may thus be instructive, especially from an evolutionary perspective, to consider not only distribution and host selection on hybrids but also offspring performance on hybrids, pure hosts and nonhosts that are expected to differ in quality [5].
Interspecific hybridisation is common between many plant species [6] and causes rapid changes in a variety of plant characters [7][8][9][10]. This may pose problems for some herbivores because changes in recognition characters may be poorly correlated with changes in quality characters. Many studies have examined different systems of hybrids and herbivores in attempts to understand the role of hybridisation in the evolution of plant resistance [7,11,12]. However, fewer studies have considered the performance of herbivore offspring on hybrid plants [1,[13][14][15][16].
The amount of resources allocated by a plant to herbivore defence depends on its biotic and abiotic environment, as well as its genotype. Important biotic factors that may influence plant growth and plant chemistry include the interactions between different herbivores that occur through their exploitation of common host plants [17][18][19][20][21]. Although the effects on plants of previous herbivory are likely to be strongly affected by environmental conditions, such as nutrient and water availability, they are also species-specific [18]. Damage may therefore have different effects on hybrids than on their parental species, and this could influence the performance of herbivores on pure and hybrid species of plants.
Willows are extensively used by many different herbivores [22], and willow species are known to hybridise frequently with other willows [23]. However, they can be divided into two different groups based on their leaf chemistry: one with high levels of phenolic glucosides and low levels of condensed tannins and one in which this relationship is reversed [24,25]. Both phenolic glucosides and condensed tannins are secondary chemicals that are known to affect herbivores [22,25,26]. Several different kinds of leaf beetle utilise willows as host species, some of which are adapted to live on willows that are low in phenolic glucosides, while others are adapted to species with high concentrations of these compounds [22,27,25]. Phratora vitellinae (L. 1758; Coleoptera: CHRYSOMELI-DAE Lat. 1802; insect nomenclature from [28]) is a leaf beetle that is oligophagous on willow species with high concentrations of phenolic glucosides [29,22,25,27]. In the area where the experiments were performed P. vitellinae is most often observed on S. myrsinifolia (L.) [30].
In the study reported here I examined the development and foraging of P. vitellinae larvae on pure Salix repens, pure S. caprea, and Fl hybrids between them. From previous studies it is known that S. repens has high concentrations of phenolic glucosides, while S. caprea has at most low concentrations of these compounds [24,31], and the hybrids are usually intermediate between the parental species in this respect [31]. To study the effect of other herbivores on insect herbivores I also simulated winter foraging by voles and leaf stripping by mammalian herbivores (moose). More specifically, I addressed the following questions:

Results
There were significant effects of both plant category and treatment on development, but not on survival (Figures, 1  and 2 and Tables 1 and 2). There was a higher probability for a surviving individual to develop to adult or pupa stage before the end of the experiment on S. repens than on S. caprea (p = 0.04), but no significant difference between the Fl hybrid and S. caprea in this respect (p = 0.86). There was also a slightly lower probability for individuals to reach pupa or adult stage on plants treated to simulate leaf stripping by a browser than on control plants or plants treated to simulate winter foraging by voles or hares (p = 0.07 for control and p = 0.05 for simulated vole damage).
There were no significant differences in consumed leaf area between plant categories (ANOVA F = 0.51, p = 0.60; Figure 3) or between treatments (ANOVA F = 1.77, p = 0.18; Figure 4), and no significant interaction between treatment and plant category.

Discussion
Development of P. vitellinae larvae is equally slow on Fl hybrid plants as on the non-salicylate parental species, S. caprea, but significantly faster on pure S. repens. Since the time spent as larvae affects several factors that, in turn, influence the risk of mortality (e.g. predation and the scope for choosing over-wintering sites), I believe that it is safe to argue that rapid development from egg to adult is indicative of high performance. The results are consistent with reports that the insect's preference for salicylic plants is of adaptive benefit [25,27]. It has been suggested that the preference of P. vitellinae for phenolic glucosides may be related to its ability to synthesise salicylaldehydes from phenolic glucosides for its own defence, although several studies have shown that other factors may be involved. P. vitellinae larvae convert salicyl glucosides from the plants they feed on into salicylaldehydes [25], which are excreted onto their backs when they are disturbed and are believed to act as deterrents against generalist predators [32,25]. However, it has also been shown that even in the absence of predators, survival and development rates tend to be higher when the insects feed on plants with high amounts of phenolic glucosides [25,27]. It has been suggested that this may be due to P. vitellinae metabolising the sugar from the sequestered phenolic glucosides [25]. In the present study I found no difference in survival between plant categories, but significant differences in the development time to the pupal stage.
In a previous study Fl hybrids proved to be intermediate between parental species in phenolic glucoside concentration [31]. If the phenolic glucoside concentration really does influence larval performance, more than any other character, the results indicate that P. vitellinae may require leaves with more than a specific threshold concentration to provide optimal nutrition. It is also interesting to note that although performance was greater on S. repens, a large proportion of the larvae on S. caprea and the Fl hybrid were still developing when I terminated the experiment. This shows that under enemy-free conditions, phenolic glucosides are not essential dietary components for P. vitellinae. Figure 1 Larvae survival Probability (y-axes) of a surviving P. vitellinae larva reaching adult or pupal stage by the end of the experiment for S. caprea (c), Fl hybrid (f) and S. repens (r), and the three treatments, browser -simulated herbivore damage through leaf stripping and removal of apical bud, to simulate the foraging of a browser, for example moose, control -no damage and vole = simulated herbivore damage through cutting shoots to simulate winter foraging of voles.   It has been suggested that P. vitellinae displays a compensatory feeding behaviour, eating more when forced to feed on leaves with low levels of phenolic glucosides [27]. Nevertheless, although there were differences in developmental stage between plant categories when the experiments were terminated, there were no differences in the leaf area consumed. This indicates that the larvae may need to forage less on S. repens than on the Fl hybrid or S. caprea to reach the same developmental stage. However, although phenolic glucosides are clearly important to these beetles, other plant characters that may differ between pure and hybrid individuals may also be important [25]. For instance, other studies of P. vitellinae have shown that leaf trichomes have a clear effect on both larval performance and female host selection [33,34]. However, all plant types tested in this study, both hybrids and the pure spe-cies, have hairs on their leaf surfaces, irrespective of whether their phenolic glucoside concentrations are low or high, so the differences observed in this case are unlikely to be due to the presence or absence of trichomes. Herbivore development is not influenced solely by secondary metabolites in the food plants, it is also clearly dependent on the amount of primary nutrients they contain, especially nitrogen and carbohydrates [35,36]. Even though the plants in the present study were treated equally, they may have differed in primary nutrient contents, and these differences may also have affected the development of P. vitellinae larvae.

Larvae survival
For the two plant categories on which larval development was reduced, S. caprea and the Fl hybrid, the simulated leaf stripping, which included removal of the leafs on the   [37,23]. S. caprea adopts an upright tree habit, and thus has fairly strong apical dominance, while S. repens has a creeping growth pattern and weak apical dominance [37,23]. The hybrid plants are commonly intermediate in growth pattern between the parental species, although there is also great variation in morphology, and individuals may be more similar to either one of the parental species. The secondary chemistry may be more affected by damage in plants that have strong apical dominance, than in plants with weak apical dominance [18]. This is because source-sink relationships, and thus intra-plant regulatory processes, are affected when apical dominance changes. If the plant has weak apical dominance, damage to a shoot will not change the source-sink structure as much as if the plant has strong apical dominance [18].

Conclusions
This study shows that there would be adaptive value for P. vitellinae females to show oviposition preference for S. repens, compared to S. caprea and Fl hybrids between S. caprea and S. repens, under the conditions of these experiments. The faster development observed on S. repens supports the hypothesis that P. vitellinae obtains additional adaptive benefits from phenolic glucosides, as shown by Rank et al. [27], beyond protection for the larvae against predators. It may also be important to consider damage caused by other agents, for example other herbivores, when studying this hybrid complex, as there are indications that plant responses to damage differ between S. repens, S. caprea and the Fl hybrid.

Plant material
Pure and hybrid plants were grown from seeds collected from naturally growing parents hand-pollinated with pollen from either individuals of the same species or from the other parental species used in this study. Individuals with typical characters for their respective species [37], indicating genetic purity, were selected for use as parents. For details about the hand pollination methods see [38] and There were no significant differences in the amount of material removed from the three types of plants when simulating moose leaf stripping, average amounts being 0.13 g dry weight (95% CI ± 0.06) for S. caprea, 0.17 g dry weight (95% CI ± 0.04) for Fl hybrids, and 0.13 g dry weight for S. repens (95% CI ± 0.03).

Insects
First instar P. vitellinae larvae were collected on July 1 in the field from S. myrsinifolia (L.). On each experimental plant a larva was enclosed on a randomly chosen shoot, with sufficient leaves to support its growth. Larvae for each block were taken from the same wild plant, and thus probably originated from the same clutch. The larvae were left to forage and develop on the plant for 46 days (until August 16), when the first beetles hatched from their pupae. At this point I measured survival, the developmental stage the larvae had attained and the area of consumed leaf. When the experiment was terminated no larvae had eaten all of the enclosed leaves.

Statistical methods
I analysed whether there were any differences between plant categories and treatments in the probability of a larva surviving until the termination of the experiment. Among the surviving individuals I analysed the influence of plant category and simulated herbivory treatment on the probability of a P. vitellinae individual reaching pupal or adult stage at the end of the experiment. For both analyses I used logistic regression with plant category, treatment and the interaction between them as factors in the model [39,40]. To evaluate differences in the consumed leaf area I used two-factorial analyses of variance with plant categories and treatments as independent factors. For all statistical tests I used R version 1.2 [41].