Within-plant distribution of a galling adelgid (Homoptera: Adelgidae): the consequences of conflicting survivorship, growth, and reproduction

Abstract. 1. Adelges cooleyi (Gil.) was rarely found in the top-most whorl of shoots in its host plant Engelmann spruce.
2. The survivorship of the few Adelges fundatrices which did colonize the top whorl was at or near zero for the 3 years in which survivorship was assessed, while survivorship in the bottom whorls was at least 60%.
3. In contrast, when an Adelges fundatrix did survive in the top whorl, her growth and reproductive output, and the growth of her progeny, the gallicolae, were at least double that of bottom whorl fundatrices.
4. The distribution of fundatrices within spruce trees was strongly correlated ( R ²=0.934) with fundatrix fitness (=survivorship × fecundity × growth).
5. We hypothesize that the vigorous growth of top whorl shoots promotes Adelges growth and fecundity, but the earlier growth of these shoots results in low temperature mortality of fundatrices.     

Post‐gall induction performance of Adelges Abietis (L.) (Homoptera: Adelgidae) is influenced by clone,shoot length,and density of colonising gallicolae

1. We evaluated the effect of clone (one susceptible and one resistant clone), shoot length, crown level, and gallicola density on post‐gall induction performance of Adelges abietis. Galls that had been successfully induced by one fundatrix on a range of shoot sizes were selected, and the number of gallicolae that could colonise the gall was manipulated. 2. Post‐induction gall development success was inversely related to shoot length and was higher on the susceptible clone than on the resistant clone. As gallicola density did not influence the proportion of galls that successfully completed development, reduced post‐induction gall development on large shoots was not likely to be result of an insufficient stimulus from gallicolae. 3. Clone was the only factor that significantly influenced gall volume and galls were larger on the susceptible clone than on the resistant clone. As gall volume did not increase when more gallicolae attempted to colonise a gall, competition within a gall increased. Gallicola survival was inversely related to the number of colonising gallicolae. Our results suggest that gall size may be limiting at natural densities. 4. Previous studies report positive relationships between gall induction success and fundatrix density, and between gall size and fundatrix density. As each fundatrix produces one egg mass of gallicolae, this study suggests that there may be a trade‐off between the successful induction of a large gall and subsequent survival of gallicolae. 5. In the present study, clone influenced all measures of post‐gall induction performance. Performance was always higher on the susceptible than on the resistant clone.     

Fitness effects of two facultative endosymbiotic bacteria on the pea aphid,Acyrthosiphon pisum,and the blue alfalfa aphid,A. kondoi

The effects of two bacterial endosymbionts, designated PASS and PAR, were evaluated on the pea aphid, Acyrthosiphon pisum (Harris) (Hemiptera:Aphididae), in which they occur facultatively, and on the blue alfalfa aphid, A. kondoi Shinji, in which these bacteria have not been found in natural populations. Subclones of pea aphids and blue alfalfa aphids, derived from parent aphid clones that did not contain PASS or PAR, were infected with one or both bacteria, generating PASS- and/or PAR-positive subclones with minimal genetic differences from the parent clones. Under laboratory conditions at 20 °C, PAR consistently reduced the fecundity (by between 19 and 60%) of subclones derived from three different parent pea aphid clones on bur clover, Medicago hispida Gaertn. PAR had intermediate effects on pea aphids reared on sweet pea, Lathyrus odoratus L., and had no significant effect on pea aphids on alfalfa, Medicago sativa L. The effect of PASS was either neutral or negative, depending on parent clone as well as host plant. Also at 20 °C, PASS reduced fecundity (70–77%) and longevity (40–48%), and increased the age of first reproduction (by up to 1.5 days) of blue alfalfa aphid reared on alfalfa and clover. PAR had a less dramatic effect (e.g., 30–39% reduction in fecundity) on these traits of blue alfalfa aphid. In contrast, PAR and PASS increased the fitness of pea aphid subclones of one parent clone reared for three generations at 25 °C on each of the three test plants. Without facultative bacteria, fecundity of the parent clone was reduced to a mean total of < 6 offspring per adult at this elevated temperature, but with PASS or PAR, mean total fecundity of its subclones was > 35. However, this ameliorative effect of facultative bacteria at 25 °C was not found for two other sets of parent clones and their derived subclones. Alate production in pea aphids was significantly increased in large populations of two PASS- and PAR-positive subclones relative to their parent clones. Attempts to transmit PASS or PAR horizontally, i.e., from aphid to aphid via feeding on host plants (bur clover), were unsuccessful.     

A galling aphid with extra life-cycle complexity: Population ecology and evolutionary considerations

In most gall-forming aphids, only the fundatrix is able to induce a gall on the host plant. In Smynthurodes betae Westw. (and a few other species), F₂ descendants emerge from the mother gall and induce their own, morphologically different galls. This constitutes an added complexity to the already very complex life cycle of gall-forming aphids. We investigated the ecology of S. betae on marked trees and shoots at four sites in Israel. Gall initiation, gall distribution and density, and temporal changes in clone size within the galls were investigated during two consecutive years. We discuss the possibility that the two-gall life cycle evolved from the typical one-gall system of most gall aphids, and the possible selective advantage of this added complexity in the life-history strategy of gall aphids. Although the total reproductive output of S. betae is not higher than in related species with a single gall per life cycle, there seems to be an advantage in the subdivision of each aphid clone into several galls, thus reducing the risk of the accidental extinction of the clone (genotype) by environmental factors, including parasites and predators.     

Hormaphis hamamelidis and gall size: a test of the plant vigor hypothesis

We tested the Plant Vigor Hypothesis by determining the distribution of galls formed on leaves of witch hazel, Hamamelis virginiana , by the aphid Hormaphis hamamelidis , and by determining various factors that affect the fecundity of the gall-forming fundatrices. We also studied the role of the fundatrix in host plant manipulation. While the mean number of galls per leaf was low, galls had an aggregated distribution among leaves. Among trees, the average number of galls per leaf was not related to the mean leaf size, contrary to the preference prediction of the Plant Vigor Hypothesis. While fundatrices preferred the distal leaves of buds, which grew more than the proximal leaves, being on distal leaves conferred no increase in fecundity for fundatrices, contrary to predictions of the Plant Vigor Hypothesis. Gall size was the factor that explained the largest proportion of variation in fundatrix fecundity; fundatrix size explained somewhat less of the variation. Also, gall position on the leaf, number of aphid galls on the leaf, and on which leaf of the bud the gall was located all played small, statistically significant roles in explaining fundatrix fecundity, but their effects were variable between experiments. Removal of fundatrices shortly after galls had enclosed them limited the growth of galls, indicating the role of the fundatrices in gall growth. We compare and contrast this system versus other gall-forming insects, as well as discuss the adaptive significance of the aphid manipulation of the host plant. Much of the data contradict predictions of the Plant Vigor Hypothesis, and we discuss how gall size, as a measure of plant growth caused by insect manipulation, explains the observed patterns.     

Performance of clones and morphs of two cereal aphids on wheat plants with high and low nitrogen content

The great variability of the aphid life cycle and their tendency for host alternation gives rise to aphid clones and morphs. Inter‐ and intraclonal variability may be observed in the responses of aphids to various environmental factors. In this study we aimed to evaluate the influence of intrinsic factors (clone and morph type) on the intrinsic rate of increase (r_m) of the English grain aphid, Sitobion avenae (Fabricius), and the bird cherry‐oat aphid, Rhopalosiphum padi (Linnaeus). For each species four apterous clones were collected from established laboratory colonies and compared to assess their relative fitness on high‐ and low‐nitrogen wheat plants under laboratory conditions. The clones had significantly different intrinsic rates of increase on high‐ and low‐nitrogen plants. All R. padi clones had a higher intrinsic rate of increase and mean relative growth rate than S. avenae. Experiments were also conducted to compare the mean fecundity of apterous and alate morphs of S. avenae and R. padi clones on high‐ and low‐nitrogen wheat plants. On high‐nitrogen plants the apterous morphs of S. avenae clones had significantly higher mean fecundity than alate morphs. There were no significant differences between the mean fecundity of alate morphs of the same species on high‐ and low‐nitrogen plants. The results support the idea of better fitness of specific clones/morphs on certain host plants due to higher and lower intrinsic rates of increase.     

The effects of host-plant properties on gall density,gall weight and clone size in the aphidAploneura lentisci (Pass.) (Aphididae: Fordinae) in Israel

Summary Samples of shoots ofPistacia lentiscus carrying galls of the aphid,Aploneura lentisci, were collected at three localities in Israel. Shoots growing near pruning scars carried more galls than elsewhere on the plant, but these galls weighed less and contained fewer aphids (smaller clones). The proportion of empty galls increased with gall density. Crowding of galls at such sites may be due to the early burst of buds at the time of aphid emergence from the overwintering eggs, and not to active search for preferred sites. Shoots bearing larger numbers of leaves carried heavier galls, which contained larger aphid clones. The position of the galled leaf on the shoot had no effect on gall weight nor on clone size. The physiological condition of the plant may be an important environmental (ecological) factor affecting the variation in clone-size and in aphid morphology among galls.     

Maternal death relaxes developmental inhibition in nymphal aphid defenders

In certain aphids, first-instar nymphs defend their gall by attacking intruding arthropod predators. One correlate of such defensive behaviour is a lengthened duration of the first nymphal stadium during the galling phase of the life cycle. A prolonged first stadium allows a large army of first-instar defenders to accumulate, which may be advantageous for gall defence. The factors determining developmental delay have been unclear, however. Our field experiment with Pemphigus obesinymphae, a North American gall-forming aphid with defensive first-instar nymphs, tests whether first-stadium duration is influenced by the death of the colony''s fundatrix (mother). We killed fundatrices in certain galls, left those in control galls alive, and counted aphids in each stadium in each gall. Galls in which fundatrices were killed contained a lower proportion of first-instar defenders and more late-instar nymphs than did galls with living fundatrices, indicating that maternal death dramatically increased developmental rate of nymphs. Possibly nymphal aphids respond adaptively to environmental cues that signal a threat to the colony''s welfare. Alternatively, the fundatrix actively suppresses offspring development in order to maintain a large army of soldiers to protect her gall. The results add a new layer of complexity to our understanding of social aphid systems.     

Parasitoid-host relationship betweenTrioxys (Binodoxys) Indicus [Hymenoptera: Aphidiidae] andAphis Craccivora [Hemiptera: Aphididae]. VI. Impact of males on the number of progeny of the parasitoid reared on certain host plants

The present paper elucidates the impact of males on the reproductive rate ofTrioxys (Binodoxys) indicus reared on the aphids ofCajanus cajan, Dolichos lablab andSolanum melongena host plants. The F₁ offspring increases with the increase of parasitoid number. This increase is maximum inC. cajan followed byD. lablab andS. melongena reared aphids. The presence of male parasitoids caused a significant decrease in the emergence of F₁ generation.      

Plant module size and dose of gall induction stimulus influence gall induction and galler performance

Several different hypotheses attribute large variations in gall abundance and galler performance to variations in plant growth rate and/or module size. The plant vigor hypothesis (PVH) predicts that galler performance will be greatest on large, fast‐growing plant modules due to their large, rapidly dividing cells that allow them to make large galls, where survival and size of survivors are usually greater. The optimal module size hypothesis (OMSH) predicts that galler performance will be greatest on intermediate‐sized modules, based on a tradeoff between ease of gall induction on small modules and increased resource availability in large modules. Here, we evaluate predictions of the PVH and OMSH during a two‐year field study by examining gall induction and full gall development success, as well as subsequent galler performance in successfully developed galls, using a shoot galling adelgid, Adelges abietis, on white spruce Picea glauca. In addition, we tested whether increased dose of gall induction stimulus on different sized modules, achieved by observing differing A. abietis densities per bud, could increase the rate of successful gall induction, as predicted by the OMSH. Galls of A. abietis appeared to be induced by a dose‐dependent stimulus from fundatrices in spring. Furthermore, the critical stimulus dose required to induce a gall appeared to increase with shoot length. These findings support the OMSH and not the PVH. Galler performance (gall volume and the number of gallicolae that emerged from each gall) in successfully developed galls was positively related to shoot length, supporting both the PVH and OMSH. We conclude that the PVH represents one component of the more inclusive OMSH, which considers both ease of gall induction on different‐sized modules and the effect of plant module size on gall size.     

Host selection and performance of the giant willow aphid, Tuberolachnus salignus Gmelin – implications for pest management

1 The recent increase in planting of selected willow clones as energy crops for biomass production has resulted in a need to understand the relationship between commonly grown, clonally propagated genotypes and their pests. 2 For the first time, we present a study of the interactions of six willow clones and a previously unconsidered pest, the giant willow aphid Tuberolachnus salignus. 3 Tuberolachnus salignus alatae displayed no preference between the clones, but there was genetic variation in resistance between the clones; Q83 was the most resistant and led to the lowest reproductive performance in the aphid 4 Maternal effects buffered changes in aphid performance. On four tested willow clones fecundity of first generation aphids on the new host clone was intermediate to that of the second generation and that of the clone used to maintain the aphids in culture. 5 In the field, patterns of aphid infestation were highly variable between years, with the duration of attack being up to four times longer in 1999. In both years there was a significant effect of willow clone on the intensity of infestation. However, whereas Orm had the lowest intensity of infestation in the first year, Dasyclados supported a lower population level than other monitored clones in the second year.     

The cost of conflict in aphid societies

Most social aphids are found within plant galls, inside of which clonally‐derived family groups feed, and specialized larval castes forego reproduction and perform various cooperative tasks, including group defence. When unrelated aphids move between clones, conditions are ripe for conflict because galls and cooperative defence are shared resources that are vulnerable to exploitation. A key unknown is whether conflict is costly in aphid social groups. We show that diversity within groups is negatively correlated with performance in the North American social aphid, Pemphigus obesinymphae. A substantial fraction of productivity is invested into drifting. However, drifting aphids tend to mature and depart non‐natal galls prior to the seasonal peak in fecundity. These results suggest that when unrelated individuals move between groups, social aphids may experience conditions consistent with a tragedy of the commons. These results also emphasize the strongly convergent properties associated with conflict across the spectrum of animal and microbial sociality.     

Effects of augmentative releases of the coccinellid, Adalia bipunctata, and of insecticide treatments in autumn on the spring population of aphids of the genus Dysaphis in apple orchards

The impact of augmentative releases of indigenous predators and insecticide applications to control the autumn aphid forms of the genus Dysaphis (Homoptera: Aphididae), major pest insects on apple trees, was assessed in one-year field experiments in Switzerland. Eggs and larvae of the two-spot ladybird beetle Adalia bipunctata (L.) (Coleoptera: Coccinellidae) were released on 4-year old apple trees in various numbers at five different dates in autumn 1998 when sexuales of the aphids were present. Additionally, Pyrethrum HP was sprayed at the same five dates to compare the effectiveness of these augmentative releases to a commonly applied insecticide. Augmentative releases of larvae before mid-October significantly prevented the deposition of overwintering eggs by aphids of the genus Dysaphis and consequently reduced the number of hatched fundatrices in spring, 1999. There was a significant negative functional response among the number of released coccinellid larvae in autumn 1998 and the number of observed fundatrices on apple trees in spring 1999. Applications of Pyrethrum HP before mid-October were more effective than augmentative releases of larvae of A. bipunctata. The release of coccinellid eggs had no impact on the number of fundatrices of the genus Dysaphis in the next spring because they did not hatch due to bad weather conditions. The weather conditions in autumn seemed to have an impact on the autumn migration of the winged aphids back to their primary host. The prevention of egg deposition of aphids in autumn is a promising control strategy and deserves further exploration for practical use.     

Relationship between cell size and specific thrombopoietin productivity in chinese hamster ovary cells during dihydrofolate reductase-mediated gene amplification

When parental Chinese hamster ovary (CHO) cell clones that are capable of producing thrombopoietin (TPO) were subjected to high methotrexate (MTX) concentrations, clonal variations in cell growth were apparent. In the clones that had no significant enhancement in specific TPO productivity (q _Tpo) when a higher level of MTX was administered, their growth was not depressed significantly nor their cell size changed significantly. On the other hand, those clones that showed a significant enhancement inq _Tpo at higher a MTX dosage, cell growth was depressed initially but recovered during successive sub-cultures. Furthermore, their cell size increased, which suggested that changes in cell size may be indicative of an enhancedq _Tpo. When the enhancement of theq _Tpo of 9 clones after a high MTX dosage was plotted against the extent of the increase of their size, there was a linear correlation (r ²=0.80,P<0.001, ANOVA), which suggested that an enhancement ofq _Tpo after high MTX administration can be measured by the increase in their cell size. Taken together, our data demonstrate that the selection of amplified CHO cell clones with enhancedq _Tpo can be done based upon their increased size and growth pattern. This facilitates the development of highly productive recombinant CHO cell lines.     

Genetic variation in the effect of a facultative symbiont on host-plant use by pea aphids

Ecological specialisation on different host plants occurs frequently among phytophagous insects and is normally assumed to have a genetic basis. However, insects often carry microbial symbionts, which may play a role in the evolution of specialisation. The bacterium Regiella insecticola is a facultative symbiont of pea aphids (Acyrthosiphon pisum) where it is found most frequently in aphid clones feeding on Trifolium giving rise to the hypothesis that it may improve aphid performance on this plant. A study in which R. insecticola was eliminated from a single naturally infected aphid clone supported the hypothesis, but a second involving two aphid clones did not find the same effect. We created a series of new pea aphid–R. insecticola associations by injecting different strains of bacteria into five aphid clones uninfected by symbionts. For all aphid clones, the bacteria decreased the rate at which aphids accepted Vicia faba as a food plant and reduced performance on this plant. Their effect on aphids given Trifolium pratense was more complex: R. insecticola negatively affected acceptance by all aphid clones, had no effect on the performance of four aphid clones, but increased performance of a fifth, thus demonstrating genetic variation in the effect of R. insecticola on pea aphid host use. We discuss how these results may explain the distribution and frequency of this symbiont across different aphid populations. Julia Ferrari and Claire L. Scarborough contributed equally to the work.     

Presence of two host races of Aphis gossypii Glover (Hemiptera: Aphididae) collected in Turkey

The aphid, Aphis gossypii, is a primary pest of citrus, cotton, cucurbits and greenhouse‐grown vegetables in Turkey and throughout Europe. There is some previous empirical data suggesting that host‐adapted genotypes of this aphid exist which may in fact be host‐races. To determine if host races of A. gossypii are indeed present in the eastern Mediterranean region of Turkey, reciprocal host transfer experiments and life table analyses were performed with multiple asexual lineages (= clones) of the aphid collected from different hosts. The collection hosts included citrus, cucumber, eggplant, okra, sweet pepper and cotton. Aphid developmental times on the host from which the aphid was originally collected (= collection or natal host) were shorter (5.2–6.0 days) and had a higher intrinsic rate of population growth (r_m = 0.25–0.44) than the 6.6–7.3 days required when the aphid was reared on a non‐original collection host (= non‐collection host or non‐natal host) and had r_m = 0.03–0.30. Total immature mortality of the cotton clone, especially in the first nymphal stage, was high (51–100%) with low r_m (0–0.03) on cucumber, citrus and sweet pepper. Aphid populations transferred from citrus, eggplant and okra to cotton (r_m = 0.29–0.30) did not differ significantly in their performance from that of the cotton population on cotton (r_m = 0.34), whereas that from sweet pepper and cucumber populations (r_m = 0.22–0.24) were significantly lower. These data have allowed us to separate A. gossypii into two distinct biological groups: (a) a ‘generalist’ population obtained from cucumber, sweet pepper, citrus, eggplant and okra which exhibited statistically better development on cotton; versus (b) a population from cotton which, by comparison on reciprocal hosts, developed poorly on non‐natal hosts except on eggplant. Development of the cotton clone on cucumber and okra was not improved after four successive generations on the non‐natal host. The good development of A. gossypii from eggplant and cotton on these reciprocal hosts suggests that these particular clones were similar, if not identical, host races.     

Drought stress and cereal aphid performance

The performance of clones of Rhopalosiphum padi and Sitobion avenae from England and Spain was examined on drought-stressed tillering winter wheat in an environment chamber at 14 ± 1°C. Two different levels of drought stress and an unstressed control were established by different watering regimes which resulted in drought-stressed plants being smaller at the end of the experiment. The effect of drought stress to plants on aphid performance was not significantly different between the clones tested. Drought stress had no effect on aphid development time, nymphal mortality, the weight of teneral adults and the number of embryos in teneral adults up to the onset of reproduction in the first F₁ generation. The subsequent reproductive capacity, as measured by the effective and potential fecundity, and the reproductive rate, were much reduced on drought-stressed plants. However, there was only a small decrease in the intrinsic rate of increase (r_m). Overall the clone of R. padi from Spain performed better than that from England, the development and prereproductive times being shorter and the fecundity higher in the Spanish clone, giving a higher r_m. There were no differences in the fecundity and the r_m between the Spanish and the English clones of S. avenae. The proportion of the F₂ generation that was alate differed greatly between clones, and only the English S. avenae produced significantly more alatae on drought-stressed than on unstressed plants.     

Aphid galls accumulate high concentrations of amino acids: a support for the nutrition hypothesis for gall formation

The nutrition hypothesis for the adaptive significance of insect gall formation postulates that galls accumulate higher concentrations of nutritive compounds than uninfested plant tissue, resulting in a high performance of the gall former. This hypothesis has been supported by some taxa of gall insects, but not by taxa such as cynipid wasps. Aphid galls are expected to require higher levels of nitrogen than other insects’ galls with a single inhabitant, because aphid galls are required to sustain a number of aphids reproducing parthenogenetically over two generations. The present study tested this hypothesis by evaluating aphid performance and amino acid concentration in phloem sap, using the aphid Rhopalosiphum insertum (Walker) (Homoptera: Aphididae), which establishes colonies on leaves of Sorbus commixta Hedlund or in galls of the aphid Sorbaphis chaetosiphon Shaposhnikov (Homoptera: Aphididae). We prepared the gall and non‐gall treatments on trees of S. commixta, in which R. insertum fundatrices were reared and allowed to reproduce. In S. chaetosiphon galls, R. insertum colonies propagated more rapidly, and the second generation grew larger and more fecund than on ungalled leaves. The amount of amino acids exuding from cut galled leaves was fivefold that in ungalled leaves; however, there was no significant difference in the amino acid composition between galled and ungalled leaves. In the intact leaves, total amino acid concentration in the phloem sap declined rapidly from late April to late May; however, the galls retained this high amino acid concentration in developing leaves for 1 month. These results indicate that the improved performance in R. insertum is ascribed to the increased concentration of amino acids in galled leaves. We suggest that S. chaetosiphon galls function to promote the breakdown of leaf protein, leading to an increased performance of gall‐inhabiting aphids.     

The evolution of gall parasitism accompanied by a host shift in the gall aphid, Eriosoma yangi (Homoptera: Aphidoidea)

Eriosoma aphids form leaf-roll galls on various elm species. Eriosoma yangi occurring on the Japanese elm is an obligatory "cuckoo" parasite, invading and usurping galls of other Eriosoma species. But another host race of E. yangi induces its own galls on the Chinese elm. Multivariate morphometries show that the two host races differ significantly in quantitative characters. Overwintering eggs of the parasitic form were experimentally transferred to the Chinese elm. The resultant larvae still exhibited a parasitic habit and developed into adults, which multivariate morphometries classified as the parasitic form. This form is considered different enough from the gall maker to be given a species rank. Eriosoma aphids facultatively invade other Eriosoma galls, where they sometimes deposit offspring. It is hypothesized that the parasitic form on the Japanese elm originated from a founder population which had migrated from the Chinese elm. On this new host, only those fundatrices that invade other species' galls will be able to leave offspring, because aphid fundatrices cannot gall plants other than their primary hosts. It is assumed that the parasitic form was established under intense selective pressure following a population bottleneck. This hypothesis was corroborated by field investigations at sites where the two forms occur sympatrically.     

Preference–performance relationship in the gall midge Rabdophaga rosaria: insights from a common‐garden experiment with nine willow clones

1. Oviposition preferences of herbivorous insects are predicted to match offspring performance on different host taxa or on conspecific host genotypes. In gall‐inducing insects, host‐plant properties such as growth rate and gall size, which are determined by plant genotype and growing conditions, may have a significant impact on offspring performance and, hence, should influence oviposition site selection. 2. The present study investigated host preference of the European rosette willow gall midge Rabdophaga rosaria (Loew) in relation to offspring success on seven clones of Salix myrsinifolia Salisb. and two naturally hybridised S. myrsinifolia × phylicifolia L. clones growing in a replicated design in an experimental field under two fertilisation regimes. For each clone, the average growth rate, number of shoot tips, and leaf and gall size were determined, and their effects on midge preference and larval survival were examined. 3. Main shoot height, number of shoot tips, and gall size were significantly affected by clone. The midges clearly preferred certain clones over the others, but preferences were not related to willow growth traits or to gall size. Survival probability was higher in large than in small galls, but females did not prefer large‐leaved clones that produced the biggest rosette galls. Midge oviposition was also uncorrelated with prior rates of leaf‐rust infection and with feeding preferences of voles and folivorous insects. 4. The weak preference–performance relationship of R. rosaria within S. myrsinifolia is probably explained by evolutionary constraints that prevent generalist insects from achieving an ability to discriminate among conspecific hosts of variable quality.