Ebd06.ebd.csic.es

Variation in herbivory by Yponomeuta mahalebella on its only host plant Prunus mahaleb along an C O N C H I T A A L O N S O EstacioÂn BioloÂgica DonÄana, C.S.I.C., Sevilla, Spain Abstract. 1. The effect of natural variation in abiotic conditions on the herbivory interaction between Prunus mahaleb (Rosaceae) and its monophagous folivore, larvae of Yponomeuta mahalebella (Lepidoptera, Yponomeutidae), was analysed for 2 consecutive years along an elevational gradient in Sierra de Cazorla, south-east 2. There was a negative correlation between site elevation and mean population herbivory level measured at the end of the growing period of Y. mahalebella.
Mortality during larval development was higher at higher elevation sites, and mean adult body mass was higher in lower elevation populations.
3. Variation in temperatures recorded during the larval growth period at different altitudes was the only study factor related to abundance of Y. mahalebella larvae; neither differences in parasitisation rates nor plant features covaried signi®cantly 4. These results support the existence of geographical variation in plant±animal interactions in relation to environmental heterogeneity.
Key words. Elevational gradient, herbivory, Lepidoptera, natural enemies, plant± insect interaction, temperature, Yponomeutidae.
dinal range distribution of P. mahaleb at Sierra de Cazorla, JaeÂn province, south-east Spain (37°59¢N, 2°54¢W). Prunus Thompson's (1994) review of plant±animal interactions mahaleb is a small deciduous tree (2±8 m tall) of central and pointed out that these interspeci®c relationships can be southern Europe that, within the Sierra de Cazorla, normally modi®ed by environmental conditions (e.g. Janzen, 1985; grows in small groups of less than 10 individuals, with a few Dudt & Shure, 1994; Herrera, 1995a,b and references therein; scattered larger populations. Leaf ¯ush begins in mid-April Louda & Rodman, 1996). Studies of plant±animal interactions and development continues to the end of May. There is an along natural abiotic gradients, e.g. latitudinal and altitudinal approximately 10-day delay in ¯ushing between populations (McCoy, 1990; Stevens, 1992; Begon et al., 1996; Gaston at different elevations within the range considered in this et al., 1998), can help to evaluate the role of environmental study (C. Alonso, unpublished). The leaves of P. mahaleb factors in such relationships. Such studies are, however, scarce contain coumarins and hydroxicoumarins (Fung & Herreb- (e.g. Koptur, 1985; Hill & Hodkinson, 1992), and conducted out, 1987) that may play a feeding-deterrent role for many mainly in extreme climates, such as the Arctic, where abiotic factors are limiting for most species (Strathdee & Bale, 1998).
In late summer, females of Y. mahalebella deposit egg More general conclusions may be obtained by studying masses around the stems of P. mahaleb (Kooi, 1990). Eggs altitudinal gradients in less extreme conditions.
hatch before hibernation and ®rst-instar larvae remain inside This study focuses on the highly speci®c relationship the batch shelter until the following spring. Caterpillars ®rst between the Saint Lucie's Cherry (Prunus mahaleb L., become active in early April at lower sites and development Rosaceae) and the small ermine moth (Yponomeuta mahale- continues until late June. From the second instar on, larvae, bella Latr., Lepidoptera: Yponomeutidae), within the altitu- which are monophagous on P. mahaleb leaves, spin a conspicuous silk tent and feed gregariously on leaves therein.
Correspondence: Conchita Alonso, Section of Ecology, Department When food within the tent becomes limiting, caterpillars of Biology, University of Turku, FIN-20014, Turku, Finland. E-mail: extend the tent along the same branch. The cylindrical tents can be measured easily in the ®eld.
The work reported here aimed to investigate which factors the largest P. mahaleb populations located mainly between affect the study plant±herbivore system and particularly whether abiotic factors may be relevant for herbivory Observations were carried out during the period in which Y.
interactions in a temperate area. Patterns of covariation mahalebella larvae develop, from April to June, in 1994 and between air temperature, herbivory level, and herbivore 1995. All P. mahaleb trees at each study population were performance were studied over 2 consecutive years in six P.
marked initially in 1994 (see Table 1), except at the Torcal del mahaleb populations located at different elevations in the study Cerecino and Nava las Correhuelas sites, where only a area. Variations in the impact of natural enemies and plant subsample was marked owing to the large number of features were also considered as alternative explanations to individuals present. At the same time, a minimum±maximum differences in herbivory along the study abiotic gradient. The thermometer was attached to the shaded trunk of one tree at speci®c questions addressed were: (1) How are Y. mahalebella each population, and minimum and maximum weekly larvae distributed among P. mahaleb populations? (2) Is their temperatures were recorded over the study period (Table 1).
abundance related to site elevation? (3) Can natural variation Mean weekly temperature was computed as (maximum + in temperature, plant features, or natural-enemy populations Marked trees were examined weekly and all Y. mahalebella tents found were marked with numbered, coloured tape. In 1994, numbers of Y. mahalebella larvae within tents were At the end of the season, total lengths of silk tents spun by Y.
Six P. mahaleb populations with altitudes ranging from 1300 mahalebella larvae were estimated for each tree. The length of to 1700 m (Table 1), and distances between sites varying from all individual tents in the tree, both with and without larvae 2 to 8 km, were selected in the southern part of Sierra de inside, was summed. The incidence of Y. mahalebella at each Cazorla Natural Park, a 214 000 ha protected area in south-east site was calculated as the mean number of tents and mean tent Spain (Fig. 1). The elevation of study populations was determined directly by using an altimeter; however, because Finally, to evaluate differences in herbivore ®tness among the exact elevation for all P. mahaleb populations in the area populations, all tents holding live larvae were collected was not available, altitudinal distribution of P. mahaleb between 21 and 24 June 1994, and 18 and 20 June 1995, by (Fig. 2) was obtained by cartographic analysis, using a clipping the branch portion they occupied. Branches were 1 : 50 000 scale map and detailed distribution information placed in polythene bags and kept at room temperature until (Fig. 1; C. M. Herrera and P. Jordano, unpublished). A map of emergence of moths or parasitoids. Moths were killed, sexed, the area was reticulated and the average elevation for every dried at room temperature, and weighed. Parasitoid incidence 1-km2 area where P. mahaleb was present (n = 127, Fig. 1) was was evaluated in collected samples as the proportion of calculated as the mean of their minimum and maximum parasitised individuals per tent. Although there are some elevations. An altitudinal pro®le of the southern half of the methodological problems with estimation of parasitoid in- park where P. mahaleb is mainly located (Fig. 1) was drawn cidence from ®eld collections (Van Driesche et al., 1991), it is using the same method (i.e. calculating the mean of minimum unlikely that parasitised larvae changed behaviour, because and maximum elevation per 1 km2), and choosing a similar this has not been reported for the same parasitoid species in number of squares randomly (n = 124). This random sample other Yponomeuta sp. (Dijkerman et al., 1986; Kuhlman, represented 10% of the study area. This procedure indicated 1995). Parasitism rate data should be therefore treated with that P. mahaleb distribution in the area ranges from 900 to caution, but they were suitable for comparisons among 1900 m, with 50% of locations between 1400 and 1650 m, and Table 1. Abiotic conditions of studied populations: elevation, and maximum, minimum, and mean weekly temperatures recorded during the 10- week study period. Note that the differences among populations in elevation are not of the same magnitude, with two low, two intermediate, and two high elevation sites. n = number of marked trees. Mean temperature was calculated as (maximum + minimum)/2.
# 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379 Herbivory along an elevational gradient 373 for 100 tents collected in late June 1994. The relationship between number of larvae and length of the tent was also Length of Y. mahalebella tents was used as an estimate studied experimentally. On 2 June 1995, ®ve tents were of herbivore consumption and abundance. To test the collected from an unmarked tree. Caterpillars from each accuracy of this measurement, length of the tent, the tent were taken out of the tent and divided in three groups, number of caterpillars (herbivore load), and the number of each with double the number of larvae of the previous leaves consumed (herbivory level) within it, were recorded group (e.g. 8±16±32). Fifteen branches from another Fig. 1. Map of the study area showing the location of Sierra de Cazorla Natural Park, the distribution of P. mahaleb populations within it (dots), and the study sites (dashed squares).
# 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379 All analyses, except those stated above, were conducted using the SAS package (SAS Institute, 1996). Means are Length of the tent in centimetres (X) was related directly to the number of eaten leaves within the tent (Y = 1.8 X); the relationship was highly signi®cant (F = 242.7, d.f. = 1,98, P < 0.001) and the regression explained a high percentage of variance (R2 = 0.92). Tent length (X) was also related signi®cantly to the number of caterpillars present at the end of the growing season (Y = 1.25 + 0.21 X; F = 40.8, d.f. = 1,97, Fig. 2. Altitudinal distribution of P. mahaleb in the Sierra de P < 0.001), however this model explained a lower percentage of Cazorla Natural Park. Black bars represent the percentage of P.
variance (R2 = 0.26). For this reason, the relationship was also mahaleb populations (n = 127) at different altitudes, with grey areas studied using a manipulative experiment where the number of showing populations with > 10 individuals. Dashed ®gure represents caterpillars present was the only factor related signi®cantly to the percentage of land area at different altitudes, estimated from a the length of the tent after 5 days (F = 41.6, d.f. = 1,5, P = 0.001).
random sample of 1-km2 squares (n = 124) within the area (see Number of larvae explained a high percentage of variance in tent length (R2 = 0.83) in a model including also origin of the larvae and origin 3 number interaction.
At tree level, the total tent length was correlated signi®- cantly with the number of tents [rs (1994) = 0.88, n = 122, unmarked tree, each » 25 cm long, were cut and put into P < 0.001; rs (1995) = 0.98, n = 122, P < 0.001].
clip vials. Each group of caterpillars was put onto a branch portion and covered with transparent plastic bags to prevent escape. Branches were kept at room temperature, and the Among-populations variation in air temperature length of tents was measured after 5 days.
Size of the trees did not affect the estimates at tree level Maximum, minimum, and mean weekly temperatures because larvae were never constrained by scarcity of free space recorded between mid-April and late-June did not differ signi®cantly from 1994 to 1995 (d.f. = 1,109, P = NS, for three separate ANOVAs; Table 1). Both maximum and minimum temperatures differed signi®cantly among sites [F (maxi- mum) = 3.1, d.f. = 5,105, P < 0.05; F (minimum) = 3.2, d.f. = 5,105, P < 0.05]. Maximum temperatures at the CanÄada Differences in herbivory level among populations were la Medianega site were signi®cantly lower than at the other evaluated using the nonparametric Kruskal±Wallis test, study sites, and minimum temperatures were signi®cantly because neither the number of tents nor the total tent length lower at Nava las Correhuelas than at the other sites. Mean temperatures were higher at the two low elevation sites than at The relationships between tent length and both herbivory the highest elevations (Table 1), but differences among sites in level and herbivore load were analysed by robust regression mean temperature were not statistically signi®cant (F = 2.05, using the least median squared method within the PROGRESS program (Rousseeuw & Leroy, 1987).
When there was an ordered prediction about differences among populations, isotonic regression was used (Gaines & Differences in herbivory among populations Finally, tents were used as sample units to evaluate Study populations differed signi®cantly in herbivory level [c2 differences in herbivore performance among populations.
(1994) = 40.85, d.f. = 5, P < 0.001; c2 (1995) = 46.11, d.f. = 5, Caterpillars growing within the same tent were not indepen- P < 0.001, Kruskal±Wallis tests]. Although overall herbivory dent, probably belonging to the same brood, hatching at the was higher in 1994, differences among populations were same place, and growing in the same micro-environment, and consistent in both study seasons (rs = 0.94, n = 6, P < 0.01; thus were not statistically independent. The percentage of collected larvae reaching the adult stage was used to estimate In 1994, herbivory at the population level was correlated survival, and adult body mass was used as an estimate of negatively with site altitude (rs = ± 0.83, n = 6, P < 0.05) and positively with mean temperature recorded during the study # 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379 Herbivory along an elevational gradient 375 Fig. 3. Mean population herbivory level observed in the six P.
mahaleb study populations in 1994 and 1995. Symbols represent mean values, bars = ±SEM, and the dotted line represents equality between years. Populations are ordered by altitude from lowest Fig. 5. Phenology of Y. mahalebella tent appearance at different (Roblehondo) to highest (Cabeza del Tejo) elevation.
locations, showing the percentage of trees at every location where there was at least one Y. mahalebella tent during the 10-week study period. Note that all trees presented Y. mahalebella tents at the end of the season in only two populations. Populations are ordered by altitude from lowest (Roblehondo) to highest (Cabeza del Tejo) lower altitude populations, but they reached the last instar simultaneously at all sites, indicating longer growth periods at Adult body mass. Adult body mass was used as an estimate of potential fecundity. As noted above, larvae within the same tent could not be treated as independent samples, so analyses were conducted after computing means for individual tents and using numbers of individuals per tent as a weighting factor (SAS Institute, 1996). Females were signi®cantly heavier (7.72 6 1.98 mg, n = 289)(F = 233.56, d.f. = 1,532, P < 0.001).
Mean adult body mass per tent was therefore calculated Fig. 4. Relationship between mean population herbivory level and Effects of sex, year, population, and year 3 population mean temperature recorded during the 10-week study period (April± factors on adult body mass were evaluated. All factors June) of Y. mahalebella larval development. Symbols represent mean except sex were considered as random effects in a general values, and bars = ±SEM. Different symbols identify study linear model (Proc GLM; SAS Institute, 1996). The populations, with black symbols representing 1994 data and white symbols representing 1995 data. Populations are ordered by altitude d.f. = 11,145, P < 0.001) and explained 54% of variance in from lowest (Roblehondo) to highest (Cabeza del Tejo) elevation.
adult body mass. Neither year (F = 0.11, d.f. = 1,8.58, P = NS) nor population (F = 0.76, d.f. = 5,4.12, P = NS) showed a signi®cant effect on adult body mass (Table 2), season (rs = 1, n = 6, P < 0.01; Fig. 4). This pattern was but there was a signi®cant year 3 population interaction supported by the 1995 results (isotonic regression: (F = 4.03, d.f. = 4,145, P < 0.01) in addition to the sex effect.
E2 = 0.273, n = 6, P < 0.001), where mean population herbivory Separating the data by year, population had a signi®cant increased steadily with mean population temperature.
effect in both data sets [F (1994) = 3.08, d.f. = 5,72, P < 0.05; F (1995) = 6.47, d.f. = 4,72, P < 0.001].
The relationships between adult body mass and both Consequences for herbivores growing within different tree population temperature and elevation were evaluated sepa- rately for each study season owing to the signi®cant year 3 population interaction in the general linear model (see Phenology. Date of Y. mahalebella tent appearance was above). In 1994, there was a positive relationship between correlated with site altitude (Fig. 5). Tents appeared earlier in adult body mass and mean temperature (Fig. 6), with a # 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379 Table 2. Adult body mass of Y. mahalebella individuals reared from ®nal-instar larvae collected from different populations. Mean population dry body mass 6 SD (mg) is presented separately by sex and year. Population means were calculated based on means per tent and weighted for number of individuals within each tent. Number of tents (nt) and number of individuals (ni) are shown. Missing values correspond to absence of (F = 1.17, d.f. = 3,43, P = NS), however the ®nal numbers differed signi®cantly among populations (F = 9.59, d.f. = 3,43, P < 0.001; Table 3). The percentage of lost larvae during the growing period was also signi®cantly different among populations (F = 8.77, d.f. = 3,43, P < 0.001), with greater losses recorded in higher altitude populations (Table 3), however it was not possible to evaluate the effect of dispersal Differences in parasitism rates among populations and consistency between years were measured using data for the three populations with adequate sample sizes in both study years (Poyo Manquillo, Torcal del Cerecino, and CanÄada la Medianega; Table 4). Differences among populations were statistically signi®cant (generalised logits model: c2 = 11.06, d.f. = 4, P < 0.05), with the parasitism rate lowest at Poyo Manquillo site. The population 3 year interaction was also signi®cant (c2 = 33.86, d.f. = 4, P < 0.001), indicating that there were differences among populations in larval performance Fig. 6. Relationship between mean adult Y. mahalebella body mass and mean temperature recorded at different populations during the larval development period. Symbols represent mean values, and bars = ±SEM. Populations are ordered by altitude from lowest (Poyo Manquillo) to highest (Cabeza del Tejo) elevation.
The incidence of herbivores on plants is usually estimated in the literature as either plant defoliation or herbivore abun- signi®cant steady increase of adult body mass at lower dance, but only rarely have both kinds of measurements been altitudes (isotonic regression E2 = 0.12, n = 5, P < 0.001 for considered simultaneously (see Alonso & Herrera, 1996). The females; E2 = 0.08, n = 5, P < 0.001 for males). A similar result tight relationship found here between tent length and the was found for 1995 data where only four populations could be number of consumed leaves within it, along with the used due to reduced sample sizes (E2 = 0.08, n = 4, P < 0.05 for experimental evidence that tent length was highly dependent females; E2 = 0.19, n = 4, P < 0.001 for males).
of the number of larvae spinning the tent, served to demonstrate that the length of Y. mahalebella tents may be Herbivore mortality. Two different aspects were considered: used as a reliable indirect estimate of both defoliation and larval disappearance during the growing season and parasitism.
The mean number of caterpillars per tent was signi®cantly Estimated abundance of Y. mahalebella larvae was corre- lower at the end of the growing season than at the beginning lated inversely with site altitude over 2 successive years, a (t = 7.9, n = 47, P < 0.001; Table 3). Initial mean number in pattern supported by casual observations over many years different populations varied between 19 and 24 caterpillars per (C. M. Herrera, pers. comm.), indicating that P. mahaleb tent, and did not differ signi®cantly among populations populations above 1600 m never receive massive defoliations, # 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379 Herbivory along an elevational gradient 377 Table 3. Caterpillar disappearance during the 1994 growing season. ni is the initial estimated number of caterpillars within a tent, nf is the number of caterpillars collected at the end of the growing season. Per cent loss was calculated from the difference between ni and nf in collected tents. Different letters in the same column indicate statistically-signi®cant differences (Student±Newman±Keuls test, P < 0.05).
Table 4. Growth result of rearing ®nal-instar larvae collected from different populations. nt is the number of collected tents and nc the number of collected caterpillars. Mean percentages of adults (A), parasitised (P), and dead (R) caterpillars were calculated based on individual tent data.
but that these are more frequent below 1400 m. Yponomeuta warmer sites also have longer growing periods. Finally, adult mahalebella was more abundant at lower elevations, where P.
moths at lower elevations were heavier (Fig. 6), in accord with mahaleb populations were less well represented (Fig. 2), and the model for spatial heterogeneity proposed by Sibly and was uncommon at the middle and high elevations, where the Atkinson (1994) in which `the optimal plastic strategy is larger largest P. mahaleb populations were present (Fig. 2). This adult body size¼in warmer habitats'. As body size is usually suggests that there was some factor(s) affecting the species linked to fecundity in Yponomeuta (Kooi et al., 1989; Leather distributions differentially. A species such as P. mahaleb, at & Mackenzie, 1994), individuals at lower elevations probably the southern limit of its distribution, may be less stressed and have higher fecundity, which contributes to their higher therefore more abundant at higher elevations, where abiotic conditions are more similar to those recorded at more northerly Losses of Y. mahalebella larvae during development were also fewer at the lower sites (Table 3). Climate unpredict- Factors determining the elevational pattern of Y. mahale- ability, plant phenology, and natural enemies could not bella abundance were less evident. Many factors can affect apparently predict the observed pattern of mortality. Myers insect herbivore performance along an elevational gradient, (1981) suggested that larvae commencing growth late in spring including plant phenology (Hodkinson, 1997), plant secondary will tend to have lower mortality because climate variability is compounds (Louda & Rodman, 1983; Koptur, 1985), leaf traits less than in early spring, however Y. mahalebella larval (Kudo, 1996), and abundance of natural enemies (Koptur, disappearance was highest in the upper populations, which 1985). In addition, variation in abiotic conditions related to began to grow later (Fig. 5). More detailed studies are needed elevation can similarly affect larval development and abun- to evaluate whether small differences in synchrony between dance. Finally, both biotic and abiotic factors may also vary leaf ¯ush and the start of larval activity (Aide & LondonÄo, among sites at similar elevations (Sork et al., 1993). The highly 1989) explain differences in disappearance rates, although on signi®cant correlation found between herbivore abundance and the coarse scale studied, phenology does not appear to explain site temperature (Fig. 4) suggests that temperature in¯uences differences in Y. mahalebella abundance. Tent appearance was herbivore abundance at different altitudes, probably through delayed by approximately 2 weeks at higher elevation sites differences in degree-day accumulation (Whittaker & Tribe, (Fig. 5), but because the host trees also ¯ushed leaves later, 1996; Hodkinson, 1997; RuohomaÈki et al., 1997 and references interspeci®c synchronisation should not be affected unduly, therein). Furthermore, differences in temperature among sites especially because external folivores such as Y. mahalebella may be ampli®ed by tent structure, because the temperature larvae are unlikely to be constrained by a lack of suitable food within tents is related to sun exposure (Alonso, 1997a), and the # 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379 Both parasitisation and predation rates also fail to explain Alonso, C. (1997b) Variaciones en las relaciones planta-insectos larval disappearance during development. Toxic compounds ®toÂfagos: efectos de factores bioÂticos y abioÂticos. PhD thesis, contained in Y. mahalebella larvae (Fung & Herrebout, 1987; Fung, 1988), combined with gregariousness (SilleÂn- Alonso, C. & Herrera, C.M. (1996) Variation in herbivory within and Tullberg & Leimar, 1988; Gamberale & Tullberg, 1996) among plants of Daphne laureola (Thymelaeaceae): correlation and protection within tents all serve as predator deterrents.
with plant size and architecture. Journal of Ecology, 84, 495±502.
Begon, M., Harper, J.L. & Townsend, C.R. (1996) Ecology: In addition, it might be expected that mortality would be Individuals Populations and Communities, 3rd edn. Blackwell, higher where larval growing periods were longer, giving greater opportunity for predator attack (Evans, 1982; but Benrey, B. & Denno, R.F. (1997) The slow-growth-high-mortality see Benrey & Denno, 1997), but the observed pattern was hypothesis: a test using the cabbage butter¯y. Ecology, 78, 987± just the opposite. Finally, parasitoids were not expected to affect mortality during the early developmental stages, Dijkerman, H.J., Groot, J.M.B. & Herrebout, W.M. (1986) The because the main parasitoid species, such as Ageniaspis parasitoids of the genus Yponomeuta Latreille (Lepidoptera, fuscicollis Dalman and Diadegma sp., emerge from ®nal- Yponomeutidae) in the Netherlands. Proccedings of Kononklijke instar larvae (see Kuhlman, 1995 and references therein).
Netherlandse Akademie Van Wetenschappen C, 84, 379±398.
Moreover, parasitism rates recorded from ®nal-instar larvae Dudt, J.F. & Shure, D.J. (1994) The in¯uence of light and nutrients on were higher in the Poyo Manquillo population where the foliar phenolics and insect herbivory. Ecology, 75, 86±98.
highest herbivore abundances were recorded. Otherwise, Evans, E.W. (1982) In¯uence of weather on predator/prey relations: stinkbugs and tent caterpillars. New York Entomological Society, 90, neither differences in plant size nor the mineral or physical properties of P. mahaleb leaves were correlated signi®- Fung, S.Y. (1988) Butenolides in parasitoids of small ermine moths, cantly with recorded herbivore abundance in the study Yponomeuta spp. (Lepidoptera: Yponomeutidae). Proceedings of Koninklijke Netherlandse Akademie Van Wetenschappen C, 91, The observed elevational pattern of Y. mahalebella abun- dance therefore appears to be related mainly to variation in Fung, S.Y. & Herrebout, W.M. (1987) Coumarins in Prunus mahaleb temperature affecting fecundity. The relationship between and its herbivore the small ermine moth Yponomeuta mahalebellus.
temperature and larval mortality remains unknown, but biotic Journal of Chemical Ecology, 13, 2041±2047.
factors did not explain the lower mortality found at the lower Gaines, S.D. & Rice, W.R. (1990) Analysis of biological data when elevation sites. A detailed study of thermal requirements of Y.
there are ordered expectations. The American Naturalist, 135, 310± mahalebella larvae will contribute to elucidating whether temperature can also be related to larval mortality. Never- Gamberale, G. & Tullberg, B.S. (1996) Evidence for a more effective signal in aggregated aposematic prey. Animal Behaviour, 52, 597± theless, these results suggest that the magnitude of herbivore pressure can be modi®ed by environmental conditions within a Gaston, K.J., Blackburn, T.M. & Spicer, J.I. (1998) Rapoport's rule: geographical range (Thompson, 1994).
time for an epitaph? Trends in Ecology and Evolution, 13, 70±74.
Herrera, C.M. (1995a) Floral biology, microclimate and pollination by ectothermic bees in an early-blooming herb. Ecology, 76, 218±228.
Herrera, C.M. (1995b) Microclimate and individual variation in pollinators: ¯owering plants are more than their ¯owers. Ecology, I am particularly indebted to Carlos M. Herrera for teaching, encouragement and valuable advice. He, Jose M. GoÂmez, Hill, J.K. & Hodkinson, I.D. (1992) Effects of temperature on phenological synchrony and altitudinal distribution of jumping Erkki Haukioja, Ian D. Hodkinson, Simon R. Leather, Lauri plant lice (Hemiptera: Psylloidea) on dwarf willow (Salix Oksanen and two anonymous reviewers made useful com- lapponum) in Norway. Ecological Entomology, 20, 237±244.
ments on earlier versions of this manuscript. Manolo CarrioÂn, Hodkinson, I.D. (1997) Progressive restriction of host plant exploita- Alicia Prieto, and RocõÂo Requerey assisted with sampling and tion along a climatic gradient: the willow psyllid Cacopsylla processing data. The Agencia de Medio Ambiente supported groenlandica in Greenland. Ecological Entomology, 22, 47±54.
this work in Cazorla, and Spanish DGICYT and Ministerio de Hodkinson, I.D. & Bird, J. (1998) Host-speci®c insect herbivores as EducacioÂn y Ciencia funded it through grant PB91±0114, and a sensors of climate change in Arctic and Alpine environments. Arctic predoctoral fellowship. While writing the paper I was supported by a European Commission-TMR fellowship.
Janzen, D.H. (1985) A host plant is more than its chemistry. Illinois Natural History Survey Bulletin, 33, 141±174.
Kooi, R.E. (1990) Host-plant selection and larval food-acceptance by small ermine moths. PhD thesis, University of Leiden. Studies in Yponomeuta, 9. ICG Printing, Dordrecht, The Netherlands.
Kooi, R.E., Herrebout, W.M. & van de Water, T.P.M. (1989) Aide, T.M. & LondonÄo, E.C. (1989) The effects of rapid leaf expansion Observations on pupal and imaginal weights of small ermine moths on the growth and survivorship of a lepidopteran herbivore. Oikos, (Lepidoptera: Yponomeuta). Proceedings of the Koninklijke Neder- landse Akademie Van Wetenschappen C, 92, 81±88.
Alonso, C. (1997a) Choosing a place to grow. Importance of within- Koptur, S. (1985) Alternative defenses against herbivores in Inga plant abiotic microenvironment for Yponomeuta mahalebella.
(Fabaceae: Mimosidae) over an elevational gradient. Ecology, 66, Entomologia experimentalis et applicata, 83, 171±180.
# 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379 Herbivory along an elevational gradient 379 Kudo, G. (1996) Intraspeci®c variation of leaf traits in several SAS Institute (1996) SAS/STAT Software: Changes and Enhancements deciduous species in relation to length of growing season.
through Release 6.11. SAS Institute, Cary, North Carolina.
Sibly, R.M. & Atkinson, D. (1994) How rearing temperature affects Kuhlman, U. (1995) Impact of insect predators and parasitoids on optimal adult size in ectotherms. Functional Ecology, 8, 486±493.
populations of the apple ermine moth, Yponomeuta malinellus, SilleÂn-Tullberg, B. & Leimar, O. (1988) The evolution of gregarious- (Lepidoptera: Yponomeutidae). PhD thesis, International Institute of ness in distasteful insects as a defense against predators. The Biological Control, DeleÂmont, Switzerland.
Leather, S.R. & Mackenzie, G.A. (1994) Factors affecting the Sork, V., Stowe, K.A. & Hochwender, C. (1993) Evidence for local population development of the bird cherry ermine moth, Ypono- adaptation in closely adjacent populations of northern red oak meuta evonymella (L.). The Entomologist, 113, 86±105.
(Quercus rubra L.) expressed as resistance to leaf herbivores. The Louda, S.M. & Rodman, J.E. (1983) Ecological patterns in the glucosinolate content of a native mustard, Cardamine cordifolia, in Stevens, G.C. (1992) The elevational gradient in altitude range: an the Rocky Mountains. Journal of Chemical Ecology, 9, 397±422.
extension of the Rapoport's latitudinal rule of altitude. The Louda, S.M. & Rodman, J.E. (1996) Insect herbivory as a major factor in the shade distribution of a native crucifer (Cardamine cordifolia Strathdee, A.T. & Bale, J.S. (1998) Life on the edge: insect ecology in A. Gray, bittercress). Journal of Ecology, 84, 229±237.
arctic environments. Annual Review of Entomology, 43, 85±106.
McCoy, E.D. (1990) The distribution of insects along elevational Thompson, J.N. (1994) The Co-evolutionary Process. University of Myers, J.H. (1981) Interactions between western tent caterpillars and Van Driesche, R.G., Bellows, T.S., Elkinton, J.S., Gould, J.R. & Ferro, wild rose: a test of some general plant herbivore hypotheses.
D.N. (1991) The meaning of percentage parasitism revisited: Journal of Animal Ecology, 50, 11±25.
solutions to the problem of accurately estimating total losses from Rousseeuw, P.J. & Leroy, A.M. (1987) Robust Regression and Outlier parasitism. Environmental Entomology, 20, 1±7.
Detection. Wiley-Interscience, New York.
Whittaker, J.B. & Tribe, N.P. (1996) An altitudinal transect as an RuohomaÈki, K., Virtanen, T., Kaitaniemi, P. & Tammaru, T. (1997) indicator of responses of spittlebug (Auchenorrhyncha: Cercopidae) Old mountain birches at high altitudes are prone to outbreaks of to climate change. European Journal of Entomology, 93, 319±324.
Epirrita autumnata (Lepidoptera: Geometridae). Environmental # 1999 Blackwell Science Ltd, Ecological Entomology, 24, 371±379

Source: http://ebd06.ebd.csic.es/personal/cam/Alonso.1999.Ecol.Entomol.pdf