Effects of elevated temperature on multi-species interactions: the case of Pedunculate Oak, Winter Moth and Tits



1. The effects of temperature on the Oak–Winter Moth–Tit food chain were studied at Wytham Wood, Oxford, and experimentally in the controlled environment solardomes at the Institute of Terrestrial Ecology, Bangor.

2. Tree cores from Wytham indicated that mature Oaks grew best at high temperatures and rainfall, but with low caterpillar populations. Young trees grew less well at elevated temperature, probably because they lost more water than they gained. Elevated temperatures advanced budburst, reduced foliar nitrogen and increased leaf toughness.

3. Moth eggs laid later or maintained at cooler temperatures than average required fewer heat units to hatch. Caterpillars took up to 50 days to complete growth at field temperatures but did so in only 20 days at a constant 15 °C.

4. The mass of Tit chicks at day 15 (day 1 = egg hatch) was positively correlated with temperature and negatively correlated with rainfall during the growing period.

5. At elevated temperature, budburst and moth egg hatch were synchronized, but earlier. Late feeding larvae and larvae fed on leaves from trees grown at elevated temperature produced smaller pupae. Pupal mass was unaffected when caterpillars and trees were maintained together under the same conditions.

6. Delaying egg hatch in Tits, to simulate conditions at elevated spring temperatures, resulted in reduced chick mass, body size and fledging success. This occurred because the chicks were fed later and prey quality was poorer, because the peak of caterpillar biomass was missed.

7. We predict that moth reproductive output will be retained at elevated temperatures because both leaves and caterpillars develop faster. Brood size in birds may be reduced because they cannot lay early enough to coincide with the narrower peak of food abundance.


The relationship between Pedunculate Oak (Quercus robur L.), Winter Moth (Operophtera brumata L.), and Great and Blue Tits (Parus major L. and Parus caeruleus L.) has been studied for many years in Wytham Wood, near Oxford, since the classic population studies of Varley and Gradwell (Varley 1970). The interaction between these organisms is complex. The suitability of one species (e.g. Oak) as a food resource for another (e.g. Winter Moth) will depend on its quantity and nutritional quality, which are, in turn, affected by the current environment and the performance of both food and consumer species in previous seasons. This three-step food chain, however, forms only a small part of a complex web of interactions (Varley, Gradwell & Hassell 1973). For example, although Oak is the preferred food plant, Winter Moth is polyphagous and can feed on a wide range of plant species (Wint 1983); birds will feed their young on species other than Winter Moth caterpillars (Betts 1955); species other than birds, such as parasitoids, will use the caterpillars as a food source (Kerslake et al. 1996). The relationship between the three species is dynamic; the nutritional quality of leaves and caterpillars and the nutritional requirements of caterpillars and birds are continuously changing as leaves, caterpillars and birds develop simultaneously. Consequently, the synchronization between budburst of Oak and egg hatch of Winter Moth, and the parallel timing of leaf development and caterpillar growth, or caterpillar growth and fledgling development, are critical to the final reproductive output of all the species.

Current predictions suggest that global levels of atmospheric carbon dioxide will increase from the present concentration of about 360 ppmv to about 500 ppmv by the year 2100, with a consequent global temperature increase of about 2 °C (IPCC 1996). Increased temperature would be expected to reduce the nitrogen concentration and increase the condensed tannin content of Oak leaves (Dury et al. 1998), thereby reducing the performance of the caterpillars which feed on them (Scriber & Slansky 1981). Elevated temperature would also accelerate the developmental rate of caterpillars, which, consequently, would affect the breeding success of Tits (Perrins 1991; van Noordwijk, McCleery & Perrins 1995) whose laying date is negatively correlated with temperature (Blondel 1985; Perrins & McCleery 1989). Although there is undoubtedly an interaction between plants, insect herbivores and predators, Stamp (1993) emphasizes that it is unclear what the relative effects of temperature, food quality and consumers on such a system are, and to what degree the three interact. Here we examine some of the effects of environmental change on the plant/herbivore/predator interaction, using the Oak–Winter Moth–Tit food chain as an example.

Our null hypotheses are, first, that elevated temperatures will have no effect on the interaction between Oak and caterpillars and, second, that it will similarly have no effect on the interaction between caterpillars and birds. The overall effect can best be measured as an effect on the performance and eventual reproductive output of each species, measured as their fecundity. The effect of elevated temperature on each species is considered individually before studying its effect on the interaction between them.

Several individuals and organizations have been involved in this collaborative study. The results presented therefore are taken from a variety of experiments and observations. All have emerged during the TIGER (Terrestrial Initiative in Global Environmental Research) programme during the period 1993–1997. Details are already available of some of the methods and results emanating from studies of paired interactions, i.e. between Oak and Winter Moth (Dury et al. 1998; Buse & Good 1996; Buse et al. 1998) and between Winter Moth and birds (Woodburn 1997). The main objective of this paper is to present an up-to-date, integrated picture of how these three species interact at ambient and elevated temperature, and to predict what effect climate change might have on this food chain.

Materials and methods


The field site was Wytham Wood (National Grid Reference SP 4708; longitude 1 ° 20 ′W, latitude 51 ° 46 ′N) near Oxford, UK. This is an area of about 320 ha consisting of former Ash–Maple–Hazel coppice with Oak standards, partly invaded by sycamore (NVC type W8a; Rodwell 1991). It is typical of the woodland habitat through much of lowland Britain. Because Wytham is one of the most intensively studied woodland sites in Britain, the selection of this site for our study enabled us to include data, collected over about 30 years, on the effects of year-to-year temperature variations on the three species. For our study, elevated walkways were erected to enable us to gain access to the crowns of several mature (around 100-year-old) Oak trees (Walkway Site). These trees are located at an altitude of c. 160 m above sea level. The birds were studied intensively in this area and more generally in other parts of the wood.

The effects of simulated elevated temperature environments on entire, 4-year-old Oak trees and their interactions with Winter Moth were examined at the Climate Change Facility (solardomes) at the Institute of Terrestrial Ecology, Bangor. Each tree stood in 2 cm water to prevent drought and was fed twice during the year with 15:30:15 nitrogen:phosphorus:potassium liquid fertilizer to provide an adequate nutrient supply. The eight hemispherical glasshouses, each with a floor area of 10 m2, provide two replicates of a 22 factorial combination of ambient and elevated temperature (+ 3 °C) and two carbon dioxide concentrations (ambient and ambient + 340 ppmv) (Rafarel, Ashenden & Roberts 1995). There were thus two replicates of each of the combinations of ambient temperature/ambient CO2, ambient temperature/elevated CO2, elevated temperature/ambient CO2, and elevated temperature/elevated CO2. There was also a replicated treatment outside the domes. Only the effects of the contrasting temperature treatments are considered in this paper.


In 1994, the date of budburst on the Oaks in the solardomes was measured as well as the duration of leaf development [defined as the period when the bud is beginning to open and a newly hatched caterpillar can enter (Varley et al. 1973), to full-size leaf].

Core samples were taken from the same trees as Varley and colleagues had monitored, and the tree ring widths measured for the period 1951–1995. The annual growth rates were compared with mean annual temperatures for this period, recorded at the Radcliffe Observatory, 5 km east of Wytham. The effect of elevated temperature on tree growth in the solardomes was investigated by recording the change in total shoot length, total number of shoots and trunk diameter on nine Oak trees in each of the replicated temperature treatments (36 trees in all) between the winter of 1993 and the winter of 1994. The dry mass of leaves falling from the trees in each treatment was also recorded.

Samples of developing Oak leaves in Wytham and in the contrasting temperature treatments in the solardomes were collected at intervals and immediately freeze-dried. Total nitrogen, total phenolics, condensed tannin levels, total carbon and the carbon:nitrogen ratio were determined for these leaves, using the methods described by Dury et al. (1998). Leaf toughness was also considered to be an important factor in palatability for Winter Moth larvae and this was assessed as the force required to pierce mature leaves with a penetrometer, similar to that designed by Feeny (1970).


In the solardomes, the dates of emergence of caterpillars from eggs experimentally attached to trees were measured, as were those for eggs inserted in crevices on the north and south side of mature trees in Wytham. In a laboratory experiment, groups of eggs from the same females were subjected to differing, but constant, temperatures by introducing them simultaneously at 4 °C intervals along a temperature gradient from 2 to 34 °C. The date of hatching was recorded.

The parental effect, i.e. the influence of the date at which the eggs were laid in November or December on the time taken before hatching, was investigated by maintaining some eggs from batches laid on different dates at a mean of 9 °C until they hatched.

During the feeding trials (described below), caterpillars in the ambient and elevated temperature domes and at a constant temperature of 15 °C were fed on leaves from one source. The effect of temperature on the duration of the caterpillar stage was measured.


Data on breeding success of Great Tit nests in Wytham Wood was available for 2545 broods during the period 1983–1995. An analysis of reproductive success from hatching up to day 15 was compared with temperature data for this period, recorded at the Radcliffe Observatory.


In the solardomes, eggs, either attached to trees or in ventilated glass tubes, were introduced into the contrasting temperature treatments in early winter in 1994 and the date of egg-hatch and budburst monitored.

Groups of caterpillars were fed either with leaves from trees grown at ambient temperature or with leaves from the elevated temperature treatment. The caterpillars were reared either in the laboratory at a constant temperature of 15 °C, where they were kept in individual, closed Petri dishes, or in the solardomes at ambient or elevated temperature, where they were similarly kept in dishes, or sleeved on the trees. The resulting pupae were weighed. As an indication of the interaction with foliar quality in Wytham Wood, fully fed caterpillars falling to pupate from early leafing and late-leafing trees were collected daily in funnels. The resulting pupae were maintained until the adults emerged, when the females were weighed.

The annual growth rates of Oaks estimated from tree core measurements were compared with the annual caterpillar population density in the wood. As any effect observed could be owing to herbivory by all caterpillar species present, Oak growth was compared with the Winter Moth equivalent (Varley et al. 1973), i.e. the equivalent density of all caterpillar species in terms of Winter Moth, calculated by comparing their rates of feeding.


One effect of elevated temperature may be that the caterpillars will develop faster and so the breeding birds will not be able to coincide their broods with the caterpillar peak. In order to mimic this, the hatching of Blue Tit broods was delayed by removing (under licence from English Nature) clutches for a period of 7 days after laying (or replacing them with dummy eggs). The eggs were then returned to the nests and the parents allowed to incubate them normally. The nestlings were weighed at 2 day intervals, tarsus measurements taken on day 13 and visit rates by the adults to feed the young recorded by a counter, triggered by a microswitch at the entrance to the nest boxes.


ANOVA was used for the data from the replicated solardome treatments (using means to preclude pseudoreplication). ANOVA and t-tests were used for the bird data. The Winter Moth equivalents and tree ring data were both ranked before regression analysis to prevent outlying data from exerting excessive influence on the result. The Mann–Whitney test was used to compare the mass of female moths from different sources.



In Wytham Wood, trees at the top of the hill, where the microclimate was cooler, tended to flush 5–7 days later than those lower down. In the solardomes, a 3 °C elevation above ambient temperature advanced the mean date of 50% budburst by about 12 days (F = 48·1, df = 1,4, P < 0·001). The extent of this advance is likely to vary from year to year. The mean time from budburst to full-size leaf responded similarly, taking 30 days at elevated temperature and 51 days at ambient temperature (F = 40·0, df = 1,4, P < 0·001).

The annual growth of mature Oaks was significantly greater when summer temperatures and rainfall were higher. The young trees in the solardomes grew significantly more and longer shoots (F = 16·3, df = 1,4, P < 0·05 and F = 8·1, df = 1,4, P < 0·05, respectively) in ambient than in elevated temperature treatments, but there was no effect on trunk diameter (F = 54, df = 1,4, P = 0·08). There was, however, a significantly greater mean dry mass of leaves falling in the ambient temperature treatments (148 g) than in the elevated temperature treatments (130 g) (F = 49·5, df = 1,4, P < 0·01). The increased growth in ambient temperature conditions is as a result of increased secondary growth; although the trees were standing in water and were supplied with luxury levels of essential nutrients, the additional 3 °C appears to have produced ‘physiological drought’ conditions during the peak summer temperatures when the secondary shoots were developing.

Table 1, summarized from Dury et al. (1998), identifies how the normal changes in leaf chemistry which occur during leaf development are affected by elevated temperature; there was no marked difference in the normal changes between leaves from the mature trees in Wytham and the saplings in the solardomes. Elevated temperature significantly decreased organic nitrogen levels, although this effect was more pronounced and prolonged in 1995 than in 1994. Condensed tannins accumulated slowly for the first 30–35 days in Wytham and both temperature treatments. By 42 days after budburst, however, leaves from trees grown at elevated temperature showed increased levels of condensed tannin. As with nitrogen levels, condensed tannins were consistently higher in some trees in Wytham than others. Under ‘normal’ conditions, total phenols increased for the first 20 days after budburst, then steadily decreased, reflecting the increase in condensed tannins. The carbon:nitrogen ratio increased as leaves matured because foliar nitrogen was decreasing, but there was no significant trend in organic carbon content either over time or with treatment.

Table 1.  . The significant effects of elevated temperature treatment on the nutrient quality of Oak leaves during leaf growth, the time at which they are eaten by Winter Moth larvae. Leaf toughness was measured outside this period. From Dury et al. (1998) Thumbnail image of

Mature leaves grown under elevated temperature were tougher than those grown at ambient temperature (F = 32·5, df = 1,4, P < 0·01). An effect of maturation was not excluded in this experiment because the former had reached full-size 1 week before the latter.


Eggs hatched on average about 10 days earlier (F = 219·3, df = 1,4, P < 0·001) under elevated temperature. Similarly, eggs on the north side of tree trunks hatched later than those on the south. The date the eggs hatched was positively correlated with the date they were laid (r2 = 0·81, n = 64, P < 0·001), although fewer accumulated day-degrees were required when they hatched later (r2 = 0·60, n = 64, P = 0·001). Between 10 and 25 °C in the laboratory temperature gradient, hatching date was negatively correlated with temperature. Below 10 °C, however, fewer day-degrees (> 4 °C) were required for hatching at lower temperatures until, below about 5 °C, no eggs hatched.

In the feeding trials, caterpillars took a mean of 49 days to develop at ambient temperature and of 36 days at 3 °C above ambient (F = 24·8, df = 1,4, P < 0·01). Development required only about 20 days at a constant 15 °C.


Analysis of total covariance suggested that total brood mass and mean chick mass up to day 15 were positively related (F = 12·8, df = 1,2527, P < 0·001 and F = 11·4, df = 1,2527, P < 0·001, respectively) to daily maximum temperatures.


In the solardomes, the Winter Moth eggs, whether on the trees or caged, generally hatched in synchrony with budburst; calculating from 1 March, mean date of egg darkening, which preceded egg hatch by 5–10 days, was at 18·7 and 29·4 (SED 0·8) days and mean date of budburst at 26·3 and 38·7 (SED 1·6) days at elevated and ambient temperatures, respectively.

The effect of being fed on leaves from trees grown under ambient or elevated temperatures on caterpillar development differed according to whether they were being fed at constant temperature or at the same temperature as the leaf source. When reared at a constant temperature of 15 °C, mean pupal mass was 38·6 and 42·9 mg, (F = 4·6, df = 42, P < 0·05) on leaves grown at elevated or ambient temperature, respectively. In contrast, when reared under the same conditions as the trees from which they were fed, there was no effect of temperature on pupal mass (means of 28·0 and 27·5 mg, F = 0·0, df = 1,4, P = 0·97).

In Wytham, adult females, originating from caterpillars collected as they fell from trees to pupate, were heavier from late-leafing than earlier-leafing trees, with means of 37 and 43·5 mg, respectively (F = 9·82, df = 50, P < 0·01).

There was a significant reduction in tree ring width when caterpillar populations were high (Fig. 1).

Figure 1.

. The relationship between the density of caterpillars in Wytham Wood in various years, measured as Winter Moth equivalents (calculated by comparing the rate of feeding of each species with Winter Moth), and Oak tree growth, measured as tree ring width. To reduce the influence of outlying data points, the data were ranked before analysis. Spearman rank correlation coefficient rs = 0·46, n = 20, P < 0·05.


The 7 day experimental delay in chick hatch resulted in mean chick mass and mean chick body size being reduced. At day 13, in the control and treatment, respectively, the mean mass was 11·17 ± 0·08 and 9·97 ± 0·49 g (t = 3·03, df = 37, P < 0·01) and mean tarsus length was 16·28 ± 0·06 and 15·66 ± 0·23 mm (t = 3·2, df = 37, P < 0·01). Fledging success was also significantly reduced, the proportion of hatched eggs that fledged being 0·97 ± 0·01 in the controls and 0·87 ± 0·06 in the experimental broods (t = 2·1, df = 35, P < 0·05). Visits by the adults to feed the young were significantly more frequent in the experimental broods than in the controls (F = 56·1, df = 332, P < 0·001), suggesting that prey brought to the former were smaller or of lower food quality.


The results of our study suggest that there would be little overall effect of predicted levels of climatic warming on the Oak/Winter Moth interaction or, consequently, on reproductive output of moths. At elevated temperature, buds burst earlier, leaves develop faster and lay down tannins sooner, but caterpillars also appear earlier and develop faster, cancelling any possible advantage to the trees (Fig. 2). Other effects, such as increased leaf toughness, occur too late to influence Winter Moth development, but might affect later-emerging caterpillar species. The timing of the commencement of breeding in Tits is closely correlated with the time of caterpillar emergence. At elevated temperatures, however, the period from hatching to fledging is unaffected, whereas the developmental period of caterpillars is shortened (Fig. 2). The figure illustrates how this results in the breeding season of the Tits becoming late relative to the peak in caterpillar biomass, with a consequent shortage, and poorer quality, food for the chicks and hence poorer performance of offspring.

Figure 2.

. A schematic representation of the effect of a 3 °C elevation in spring temperature on the normal timing and duration of the developmental period of leaves, caterpillars and birds from minimum to maximum biomass. (Based on the results of this study).

The first link in the relationship between the moth and its Oak host, i.e. the synchrony between egg hatch and budburst, is believed to be an important determinant of defoliation levels in Oak woodland (Hunter 1992) and in the eventual mass of the adult Winter Moth (van Dongen et al. 1994). The phenology of the larval stage has evolved to be as early as possible because of the marked seasonal decline in foliar quality, particularly an increase in intensity of host-plant ‘defences’ and a reduction in the available nitrogen and water (Wint 1983). The maintenance of synchronization is ensured not only by the tendency for the wingless females to return to the same tree (Embree 1965; Graf et al. 1995), the order of bud opening between trees tending to be the same each year (Varley & Gradwell 1956), but also by the winged males having very limited dispersion (van Dongen, Matthysen & Dhondt 1996). Both tend to minimize genetic variability in egg hatch date and thus increase specificity to a particular tree. Also, the reduced heat units required for hatching by later laid eggs tends to improve synchronization. In our experiments, both budburst and egg hatch advanced at elevated temperature and the degree of synchronization between Winter Moth and Oak was generally unaffected by elevated temperature. Similarly in Wytham, eggs introduced on to the trees in autumn hatched during the period of budburst of the Oaks (L. R. Cole, personal communication). The situation is different in other Winter Moth/host interactions. Dewar & Watt (1992) predicted that climatic warming would result in decreased synchrony with budburst in Sitka Spruce (Picea sitchensis) and Kerslake & Hartley (1997) suggested that synchronization with Heather (Calluna vulgaris) is unimportant. If egg hatch occurs before or after Oak budburst, the caterpillars can disperse to other hosts (Smith 1972; Kirsten & Topp 1991), but there is little understorey in Wytham, or other mature woodlands, for them to feed on. In any case, Oak is considered to be the optimum host for Winter Moth (Wint 1983) and, because dispersal must be dangerous, it is probably a last resort.

In our experiments, pupal mass was unaffected by whether caterpillars and their host were at ambient or at elevated temperature, despite the accelerated reduction in food quality at elevated temperature. Thus, the increased rate of mean caterpillar development at higher temperatures, about 20 days at a constant 15 °C but extending to 40 days at ‘normal’ temperatures, matched the increased rate of development of the leaves. There was, however, a decline in pupal mass when caterpillars were fed at constant temperature with leaves taken from elevated temperature trees and when caterpillars were developing later than normal in the field, or when fed on leaves from early leafing trees than from later-leafing trees. This decline with time reflects declining leaf quality. It is well-known that caterpillars forced to feed on old leaves metamorphose into small adults or fail to pupate (Feeny 1968). It is those caterpillars which, because they are comparatively early or late hatching, or feed on early or late-leafing trees, have a mismatch in timing with optimum quality of leaf development that will have reduced reproductive output. In Wytham, later hatching caterpillars produced significantly lighter pupae than earlier-hatching caterpillars on the same tree (L. R. Cole, personal communication). Because pupal mass is positively correlated with fecundity (Buse et al. 1998), this signifies that fewer eggs will subsequently be laid. These ‘bottom-up’ forces (i.e. related to host-plant quality) explain 17% of the spatial and temporal variation in the Winter Moth population in Wytham Wood (Hunter, Varley & Gradwell 1997). If food quality becomes poor, caterpillars could disperse to other hosts, but considerable losses will still occur (Feeny 1970), reducing eventual reproductive output of the moth.

The interaction between caterpillars and Oak is a two-way process; reproductive success at one trophic level can mean reduced performance at a lower level. The effect of the size of caterpillar populations on tree growth, measured by tree ring widths, was greater than expected (Fig. 1). Repetition of attack can lead to severe reductions in growth and even dieback (Rubstov 1996). Defoliation, along with drought and frost damage, is a frequent initiator of stress-induced diseases (Wargo 1996). It is believed that climatic variation results in the wide fluctuation of insect populations on trees of deciduous woodlands (Strong, Lawton & Southwood 1984). Both our results and those of Pilcher & Gray (1982) suggest that elevated temperature increases Oak growth, provided that sufficient water is available. This might be aided by a reduction in feeding by caterpillar ‘stragglers’ owing to the more rapid maturation of leaves reducing their ability to survive. A greater food resource might thus be available for caterpillars in the following year. The envisaged increase in temperature as a result of global warming will, however, occur gradually over many years, allowing the caterpillars (yearly breeders), but not the Oaks, to adapt by selection.

There is evidence that the inter-relationship between caterpillar and Tit chick development is affected by elevated temperature. The Wytham study has shown that recruitment into the Tit breeding population is heavily biased towards early broods. It is suggested that this is because, in order to have their broods in the nest when caterpillars are abundant, the birds have to start breeding at a time when food supply is still poor (Perrins 1991). Our experiment mimicked the effect of climate change by delaying hatching so that peak caterpillar availability was relatively early. This resulted in chick performance being reduced. Similarly, in ‘real’ warmer-than-average springs, despite evidence from long-term studies that brood success was enhanced up to day 15 after hatching (Woodburn 1997), the young showed poorer performance after fledging (Perrins 1991; van Noordwijk et al. 1995), the resultant fledglings being lighter and smaller, and having poorer survival. This reduced performance appears to be related to a change in the duration of caterpillar feeding at elevated temperature.

The availability of caterpillars as food for birds at any specific time will be the product of decreasing caterpillar numbers (depending on hatching, deaths, dispersal and leaving the tree to pupate) and increasing caterpillar size (depending on the age of the caterpillars and the availability and quality of Oak leaves as food). This is diagramatically represented in Fig. 3, which is based on the following parameters: the number of young caterpillars entering the population and the maximum mass of individual caterpillars is the same at both normal and elevated temperature; most larval loss occurs during the first instar (Varley & Gradwell 1960; Holliday 1985) and when the larvae pupate or are eaten by birds; larval biomass increases rapidly only in the final instars (four and five). The result is a narrow peak of food availability for birds, even at normal spring temperatures, when, for optimum chick growth, the chicks need to be in the nest. The critical synchronization between caterpillar and chick development can become out of phase at elevated temperature, peak food availability declining before the chicks fledge. This effect would be particularly marked if temperature increases after the chicks have hatched; caterpillar development will accelerate, whereas chick development cannot change (Fig. 2).

Figure 3.

. A diagram of the relationship between relative caterpillar number, caterpillar mass and food availability for birds (the product of caterpillar number and mass) during caterpillar development at normal and elevated (+ 3 °C) spring temperatures. This is based on our observations that the size of the caterpillar population and the mass of individual caterpillars is the same at both temperatures, but feeding duration is reduced at elevated temperature. The units on the y-axis are arbitrary.

The shortfall in suitable food can, to some extent, be compensated for by more frequent visits to the nest with smaller caterpillars or with pupae. The latter, however, which are in cocoons in the litter layer or soil (having dropped from the trees as fully fed larvae), are more difficult for Tits to find than the caterpillars on the trees. Similarly, the alternative food source of Tortrix pupae is difficult to find on the leaves of the trees. Elevated temperature throughout the nesting season, however, will have an even more marked effect on the birds. The compressed duration of leaf and caterpillar development at elevated temperature leads to a narrowing, and raising, of the peak of food availability, assuming that caterpillar mass and total numbers remain the same (Fig. 3). This means that, at elevated temperature, the present synchronization between chick age and food availability will change. This earlier, but reduced duration, of the peak of food availability might explain why, in warmer springs, brood success is improved during chick development but reduced by the time of fledging. It also explains why the smaller clutch sizes reported in the warmer environment of southern France (Perrins 1965; Blondel 1985) would be of advantage by (1) shortening the time between the start of laying and the date of hatching and (2) concentrating the available food resources on fewer offspring. The reduced clutch size of late-laying birds in Wytham (Perrins & McCleery 1989), which will miss the peak in food availability, may be a similar response.

Our first hypothesis, that climatic warming will have no effect on the interaction between Oak and Winter Moth caterpillars is supported by the present study, whereas the second, that there will be no effect on the interaction between caterpillars and birds, is not. Despite the potential adverse effect of elevated temperature on Tit populations, the conclusion of van Noordwijk et al. (1981) that the mean laying date of a population of Great Tits can change by 0·5 days per generation owing to selection and the consistently earlier laying dates and smaller clutch sizes in southern France than Wytham (Perrins 1965; Blondel 1985) suggest that Tits would adapt to the new conditions. The predicted gradual increase in mean annual temperature with climatic warming would allow such adaptations to occur in the life cycles of the moth and bird species in the food chain, but not in Oak, because recruitment is only occasional.

Our approach to examining the effect of one aspect of climate change, namely temperature, on a three-step food chain has been to mimic natural conditions as closely as possible. This was possible because the solardomes track ambient environmental conditions (while allowing study of a 3 °C elevation above ambient temperature) and necessary because we included field studies of trees and birds in a mature Oak woodland in our study. Thus, fluctuations in temperature and light were natural in both solardome and field studies, but out of our control. The solardomes have also allowed the effects of enhanced CO2, on its own and combined with temperature, to be examined (Buse & Good 1996; Buse et al. 1998), which would not be possible in the field. In contrast, environmental chambers, such as the Ecotron, allow complete, but miniature, ecosystems to be run under controlled conditions for long periods. These enable the effects of enhanced CO2 and enhanced temperature on population, community and ecosystem dynamics to be studied over several generations (Lawton 1996). Examples are the effects of elevated CO2 on soil microbial biomass and four annual plant species (Kampichler et al. 1998) and on fungi and fungal-feeders (Collembola) (Jones et al. 1998). The approaches are complementary; the Ecotron allows complete control of many environmental factors, whereas the solardomes allow control of some factors, and monitoring of others, under more natural ambient conditions.


We wish to thank John Castle for technical support and advice concerning the chemical analysis of foliar samples, Irene Mueller-Harvey for advice on tannin assays and Ray Rafarel for technical support at the solardome facility. We also thank Lionel Cole for allowing us to include unpublished data on the Winter Moth in Wytham Wood. This work was supported by the Natural Environment Research Council through its TIGER (Terrestrial Initiative in Global Environmental Research) programme, award number GST/4(T91/173): species interactions.