Search

Shifts in the timing of key life history events are some of the most documented biological responses to climate warming. Synchronous timing between a consumer and ephemeral resource can be important for consumer fitness, with asynchrony between the two parties often having a negative effect, as outlined by the match-mismatch hypothesis (MMH). Differences in the thermal sensitivity of phenology among species occupying different trophic levels leave consumers at risk of asynchrony within time-sensitive trophic interactions under climate warming. There is abundant evidence that secondary consumers shift their phenology by less than their resources, resulting in them lagging behind the resource as temperatures increase. As such, phenology and the MMH have become prominent in research regarding the impacts of climate warming on biological systems. The phenology of individuals within a population is often summarised using the mean. However, interactions are not just sensitive to the mean of each trophic level, but instead the synchrony among individuals. The range of timings among individuals forms the phenological distribution, described by the maximum number of individuals (height), the range of dates over which the event will occur (duration), and the relative shape (width). Differences in the height, duration and width of the phenological distribution of an ephemeral resource are likely to have implications for the MMH, affecting the prevalence of food available to the consumer at any point in time, and also at different levels of asynchrony to the mean timing of the resource. For example, with a wider resource distribution, the difference in food available to synchronous consumers versus asynchronous consumers will be less than when there is a narrower resource distribution. As such, if the fitness of consumers is affected by variation in the phenological distribution of food availability, differences in the distribution of a resource could affect consumer mean population fitness and the strength of selection on timing. Given the possible implications that shifts in these aspects of the full phenological distribution have for trophic interactions, it is surprising that it has been the focus of minimal research to date. In this thesis, I utilise data on the spring peak in arboreal caterpillars within deciduous woodland systems and the breeding phenology and breeding success of blue tits (Cyanistes caeruleus). I explore variations in the phenological distribution of caterpillars and the implications for the phenological fitness function of blue tits, an insectivorous bird that relies on caterpillars to feed young. Data have been collected across 44 sites spanning an elevation range of 10-433m and 2o latitude through Scotland over nine years. The sites vary both in woodland tree composition and in temperature. In Chapter 1, I introduce phenology, the MMH, the full phenological distribution and the deciduous woodland study system. Chapters 2 and 3 then focus on habitat and temperature as drivers of the phenological distribution of caterpillars. In Chapter 4, I examine the consequences of the full phenological distribution of caterpillars for blue tit fitness relative to breeding phenology, and in Chapter 5 I discuss the broader implications of my work, limitations and future directions and opportunities. In Chapter 2 I examine variation in the phenological distribution of caterpillars in relation to woodland tree composition using data on the abundance and mass of caterpillars on ten different tree taxa throughout spring, and information about the local tree taxa composition more specifically. I find that there is substantial variation in the abundance of caterpillars supported by different deciduous tree taxa and that the prevalence of oak trees in the local woodland has an additional effect, increasing the abundance of caterpillars found throughout the site. I also found that there is substantial variation in the height of the caterpillar peak hosted by different tree taxa, and some minimal variation in the timing of the peak and the duration. There was no significant difference in the ultimate mass of caterpillars among taxa meaning that the differences in the peak in biomass among tree taxa largely mirrored patterns seen for the peak in abundance. These results show that woodland tree composition contributes to variation in the phenological distribution of caterpillars between locations, suggesting that studying the MMH in a range of woodland types may be of benefit to accurately predict the consequences of climate warming at the meta-population level. In Chapter 3, I identified the time periods within spring during which average temperature best predicted the timing, height and width of the phenological distribution of caterpillar abundance. I found mean timing to be better predicted by temperatures earlier in spring with height predicted by temperatures that were slightly later, whilst there was substantial uncertainty in the period over which average temperature best predicted width. In this chapter I develop a novel method to phenology research, testing for thermal sensitivity in the mean timing, height and width of the phenological distribution of caterpillars from raw data of abundance over time. I found that the height of the phenological distribution increases substantially with an increase in temperature and that the shape becomes more steeply peaked, falling over a similar duration of time. These results suggest climate warming is having a substantial impact on the phenological distribution of a key resource in temperate woodlands and provides additional insights into the impacts of climate change past shifts in mean timing alone. In Chapter 4 I present an extension to the MMH in which the full phenological distribution of caterpillars affects the full phenological fitness function of blue tits. I achieve this by extending the statistical method from Chapter 3 and use site and year specific estimates of the mean timing, height and width of the caterpillar peak to predict the equivalent metrics in the blue tit fledging success fitness function. This framework has the potential to allow tests of how the timing of the caterpillar peak affects the optimal breeding timing for blue tits, whether the maximum height of the caterpillar peak affects the maximum fitness of blue tits, and whether the width of the caterpillar peak affects the width of the phenological fitness function which may have implications for the strength of selection. However, having identified no optimum hatch date for blue tits within the fledging success data, the data do not conform with the structure of the framework described. As an alternative, I examined the effect of the mean timing, height and width of the caterpillar phenological distribution among site by year combinations on a straight-line fitness function for the birds, allowing the intercept and slope of fledging success over time to vary with each caterpillar parameter. I found no effect of the mean timing, height or width of the caterpillar peak on the blue tit phenological fitness function which shows no evidence of the MMH affecting these populations. Alternatively, I suggest parental quality may have a stronger effect than environmental factors on fledging success. As the effect of the MMH has varied among fitness traits in other passerine species this does not rule out any effect of the phenological distribution extension of the MMH, but presents an opportunity for further work on this topic. Nevertheless, the conceptual framework put forwards holds an interesting progression for how we perceive the dynamics within these ephemeral trophic interactions. In this thesis, I have introduced approaches that allow more nuanced insights into the impacts of climate change on phenology and the consequences of phenology for fitness. The phenological distribution of caterpillars varies with woodland tree composition, contributing to spatial heterogeneity in a key resource and guild of herbivores. The increase in height and narrowing shape of the phenological distribution with an increase in temperature shows that reducing the caterpillar peak to mean timing misses important information about how the resource is responding to a warming climate. By extending the conceptual framework of the MMH we gain insight into how the resource distribution can influence the strength of selection and population persistence, highlighting the value of considering the full phenological distribution in phenology and MMH research. The theory and statistical methods employed in this thesis are transferable to a vast range of species and interactions. To accurately interpret and predict the impacts of climate warming in biomes driven by the strong seasonality of events, it may be essential to move on from focusing on mean timing and begin to incorporate the full phenological distribution.

Across temperate environments, climate warming is leading to a general advancement of spring phenology in a wide range of ecologically and taxonomically diverse species. For taxa that depend on interactions with other species-predators and prey, pollinators, parasites and hosts-widespread phenological changes may cause severe problems. Divergent phenological responses to spring temperature changes among taxa could result in these crucial biotic interactions becoming mistimed. This may cause significant negative fitness effects that could ripple through a population, across trophic levels, and perhaps entire ecosystems. This concept, formalised as the match-mismatch hypothesis (MMH) has become the subject of intense speculation and debate in recent decades. Much of our understanding of the occurrence and significance of 'phenological mismatch' (negative fitness consequences brought about by mistiming between interacting species) due to climate change comes from the trophic interactions in the classic temperate woodland tree/caterpillar/bird food-chain. This work, however, suffers from many limitations. Spring-feeding caterpillars, forming the central link in this food-chain, are particularly important in that fluctuations in their populations can affect both higher and lower trophic levels. In the tree/caterpillar link, previous literature focuses largely on a single host and caterpillar species pairing: oak (Quercus robur) and the winter moth (Operophtera brumata). It has been argued that these caterpillars could respond more strongly than their host-plants to climate warming in terms of shifting their phenology, and that even slight mistiming between the two trophic levels has significant negative fitness effects for them. Caterpillars that hatch before bud-burst on their host tree will likely starve, and those that hatch too late are forced to feed on less palatable mature foliage. This rather narrow view, however, overlooks the fact that these caterpillars may be resilient to mistiming in many instances, and that the oak/winter moth trophic interaction may not necessarily be representative of the many other caterpillar species, or alternative host-plant species. In this thesis, I attempt to expand our knowledge and understanding of the operation of the MMH in this system by specifically addressing some of these key caveats. First, in Chapter 2, in order to address the role of different plant species in the diet of the winter moth and the relative importance of oak as a host-plant species, I consider the effects of host-plant species on survival, growth, and development of the caterpillars, across four British populations. I find that winter moth caterpillar fitness varies substantially across host-plant species, but that there are also strong population-specific responses consistent with genetic divergence. In contrast to the assumptions typically made in the literature that oak is the "primary", "principal", or most significant host-plant species in the field, I find that caterpillar performance on this species is consistently poor relative to other abundant and widespread host-plant species. Reconciling this apparent inconsistency represents an obvious avenue for future research. A taxonomically broad diet may serve to buffer winter moth caterpillars against the effects of mismatch on any one host-plant species-phenology varies across hosts and, averaged across a population, this might ensure there are always some food resources available for individuals to exploit. Next, in Chapters 3 and 4, to determine whether the impacts of mismatch generalise across caterpillar and host-plant species, I directly test the effects of mistiming across a range of British spring-feeding caterpillar species, including the winter moth. In Chapter 3, I consider the effects of late-hatching asynchrony on performance (and fitness in the winter moth) of up to 65 days. I find that the effects of asynchrony on performance are contingent on the particular caterpillar/host-plant species pairing in question. Depending on the host-plant species, some caterpillar species show little to no decline in performance across a period of several months (e.g. vapourer Orgyia antiqua on birch Betula pendula or sycamore Acer pseudoplatanus), while others show precipitous declines in a matter of days (e.g. fitness in winter moth on sycamore or sallow Salix caprea). This highlights the danger of extrapolating from a single caterpillar/host-plant species pairing. Indeed, in both cases, the winter moth and oak appear to be exceptional-performance of the former typically showing a steeper than average decline with increasing asynchrony, and the latter being a generally poor host for most spring-feeding caterpillar species, on which performance declines at a greater rate than other host-plants with asynchrony. Overall, I find that, in contrast to the prevailing view in the literature, synchrony is important for caterpillar fitness, but within fairly broad bounds (at a scale of weeks and months, rather than days), though this varies across hosts and species. In Chapter 4, I consider asynchrony in the opposite direction, and investigate the ability of spring-feeding caterpillars to cope with hatching too early, before bud-burst on their natal tree. Early hatching caterpillars can simply tolerate a lack of food and wait until it becomes available, or they may be able to exploit the unopened buds of their host-plants as a food source in the intervening period. I found that across five spring-feeding caterpillar species, there is often a considerable ability to tolerate starvation, ranging from several days in the winter moth and mottled umber Erannis defoliaria, to over thirty in the black arches Lymantria monacha. Increased temperatures, however, significantly reduced the time which caterpillars could survive without food, often by a substantial margin (e.g. by twenty days in the black arches moth at temperatures of 21°C versus 5°C). In the winter moth, I show experimentally for the first time that caterpillars are indeed able to feed on the unopened and opening buds of a range of their host-plant species. However, the likelihood of establishment on buds is initially low and increases steeply as buds mature and softer tissue becomes more exposed. Nonetheless, this clearly demonstrates that many spring-feeding caterpillar species have at least some ability to tolerate early hatching on their host tree. In Chapter 5, I consider in more detail the widely-held assumption that foliage becomes unsuitable for caterpillar consumption very soon after bud-burst. In contrast to Chapter 3, I reared caterpillars on frozen foliage collected from a sample of trees across a two week period after bud-burst, to determine the effects of any changes in their structure and secondary chemistry across this period on palatability. Specifically, I focussed here on the effects of caterpillar asynchrony on growth rate and rate of survival across time, both of which have distinct fitness implications versus overall mass attained or survival probability (cf. Chapter 3). I find no consistent effects of leaf age on rates of mortality across time, suggesting that leaf maturation occurring within the first two weeks after bud-burst generally has little effect on caterpillar performance. There are, however, significant effects of host-plant species and age on growth rates-on older oak foliage, growth rates are higher, the implications of which are unclear. Additionally, I find that there is substantial variation in caterpillar performance between individual trees and broods. Taken together, these findings may indicate that phenological variation between individual trees could serve to ameliorate mismatch, buffering against it at the population level. Finally, in Chapter 6, I discuss the concept of 'buffering' in detail-a phrase widely used but little considered. I argue that buffering is related to concepts of stability in living systems, and that it represents the means by which stability is maintained, via a range of 'buffering mechanisms'. I define buffering as "the amelioration of any fitness effects resulting from an environmental change". I explore the concept specifically within the spring-feeding caterpillar system, and argue that the very unpredictability and uncertainty that is an inherent part of their niche has driven the evolution of many of the buffering mechanisms by which that variation can be tolerated.