Search

The movements of critically endangered vultures, equipped with satellite-based tracking devices in Namibia, were inspected using Generalized Additive Models. Models incorporated spatially adaptive (1D and 2D) smooths via the Spatially Adaptive Local Smoothing Algorithm (SALSA) and Complex REgion Spatial Smoother (CReSS) method. The correlated nature of geo-location data was address via robust standard errors. The results of this thorough and integrative study of movement ecology have an unprecedented level of detail, far exceeding what is available in the literature. Namely, vultures were seen throughout Namibia and its five neighbouring countries with three individuals visiting locations farther than 1,000 km from where they were initially seen. Large variability was found both within and between birds. Differences were perceived in four daily movement properties, even though temporal differences were only captured for daily distance travelled (monthly) and daily maximum displacement (seasonally). There was noticeable variation in the size of the areas each bird used from month to month, often showing very little spatial overlap. Home ranges varied greatly; one bird expanded its monthly home range as much as nineteen times its smaller size. Contrastingly, core areas remained sometimes constant. Home ranges were three to five times larger than the respective core areas, clearly indicating a non-uniform use of the environment. The extensive study area (2.3 million sq.km) was characterised using habitat features, climate conditions and indices of human presence. Vegetation index, minimum distance to river and minimum distance to road were consistently important in explaining the probability of bird presence. Nonetheless, each vulture used its environment in its own way. These novel findings support trans-frontier conservation measures, represent crucial support to revise the geographic extent of existing conservation actions and constitute the basis to predict the risk of exposure of vultures to lethal threats or to assess changes under Climate Change scenarios.

In the management of agricultural insect pests, short-term costs must be balanced against long-term benefits. Controls should be selected to account for both their immediate and downstream effects upon the demography and genetics of the pest, enabling suppression today without threatening suppression tomorrow. The iterative, algorithmic method of dynamic programming can provide optimal solutions to problems of this type, in which actions are taken sequentially and each action may influence those which follow it. However, this approach is fundamentally constrained with regards to the magnitude of the problems it may solve. As questions of insect pest management can be subject to ecological and evolutionary complexities, this may place them beyond the scope of dynamic programming. When it is the intricacies of a problem that are of interest, it may be more productive to utilise approximate dynamic programming (ADP) methods which can attempt problems of arbitrary complexity, although at the expense of no longer guaranteeing optimality. In this thesis I first challenge a dynamic programming algorithm with the management of a hypothetical insect pest feeding upon a transgenic insecticidal crop. The model explores how different realisations of fitness costs to resistance influence the algorithms suggested actions. I then apply a brute-force variant of ADP, a lookahead policy, to the management of a stage-structured, continuously reproducing pest population. This was to explore the extent to which an algorithm with a limited temporal perspective is able to balance the timetable of pest demography against the timescale over which insecticidal sprays and bisex-lethal sterile insect releases unfold. This same decision framework is then applied to a modified problem in which resistance to insecticidal toxins may evolve and releases are now male-selecting. This was used to assess the efficacy with which simple lookahead policies utilise a control with delayed benefits (the male-selecting releases) and possible constraints on their capacity to respond to resistance evolution. Dynamic programming and ADP methods offer a versatile toolbox for accounting for the potential impacts of the evolutionary and ecological peculiarities of particular pests upon control decisions.

Picophytoplanktonic cells (0.2-2 μm) are the dominant phytoplankters in the largest marine biomes on Earth: the subtropical gyres. The overaching aim of this thesis is to develop algorithms that use remote-sensing observables to map the distribution of the smallest and most abundant member of picophytoplankton, Prochlorococcus, and assess its contribution to the marine carbon cycle. To understand how the photoacclimatory status and growth of Prochlorococcus and its sister genera Synechococcus are influenced by light and nutrients, experiments were conducted in the South Atlantic Gyre (SAG). Results from the manipulation experiments show that, in the central region of the SAG, nutrient addition can induce marked changes in the optical properties of Prochlorococcus cells when subjected to saturating light levels, leading to a decrease in cell abundance, whereas at the gyre periphery no substantive changes in cell growth or optical characteristics were observed. Since light plays a central role in shaping the distribution of cyanobacteria, an empirical algorithm based on relationships between Prochlorococcus abundance and remotely-sensed observables was developed. The outputs were then used in a modified primary production model to predict the vertical distribution of carbon fixation by Prochlorococcus. The models estimate that ∼ 3.4 x 1027 Prochlorococcus cells in the global ocean fix 4.7 Gt C year-1. Most of the cell biomass and primary productivity is concentrated in the subtropical gyres and areas near the Equatorial Convergence, and 61% of the carbon fixation occurs in the upper water column (0-45 metres), where only 43% of the cells reside. However, in the gyres, carbon fixation is highest (62%) in deeper layers (45-200m), and both cell abundance and carbon fixation show marked seasonal patterns. The models developed in this study provide an unprecedented view of the vertical distribution of Prochlorococcus cells and their corresponding rates of carbon fixation in the global ocean.

The Western Honeybee (Apis mellifera) is an important species, not only ecologically and economically, but as a source of recreation to many. The pollination services the species provides benefit a number of crops worldwide, and, as the honeybee is domesticated and kept in hives, can be directed commercially. Recently, although overall global stocks are growing, there have been reports of high colony losses worldwide. Due to the value of this species, this is a worrying trend. There are many stressors facing the honeybee, both natural and anthropogenic in origin. Two of the most prevalent, both in the popular media and in monitoring studies of colonies are insecticidal pesticides and the parasitic mite Varroa destructor. Due to the difficulties and expense of carrying out large-scale field studies required to properly investigate the multiple stressors and their interaction, the use of modelling to explore the problem and direct field work is a vital resource. In this thesis, I present research using the BEEHAVE model and a novel model to explore the exposure and potential impacts of pesticides and the varroa mite. The results show that the timing of a pesticide exposure in the year greatly changes the resultant impact on the colony. Pesticides can have many impacts on different stages of the honeybee, and I show that increased mortalities of different life stages of the honeybee (larvae, in-hive adults, foragers) and decreasing egg-laying rate, affect the development of the colony to different extents at different times of the year, with the colony being highly sensitive to losses of in-hive bees during the summer, and the over-wintering bees at the beginning and end of the year. A novel model is presented exploring the in-hive distribution of pesticide-containing nectar and the effect it has on the exposure of in-hive receiving bees and larvae. The results from this model show that, in-hive distribution is not important to consider for the adults, but may be for the larvae. The landscape, specifically the distance to pesticide-treated forage in relation to untreated forage also has an impact on the result of a pesticide exposure, and this is a potential avenue for the mitigation of pesticide impacts. I also present work towards the validation of BEEHAVE with regards to varroa mite infestation, finding that the model results are close to empirical data, both for datasets from the UK and USA, but the impact of varroa is underestimated. The results are discussed in the context of pesticide risk assessment, the mitigation of potential stressors and the modelling of the varroa mite. The BEEHAVE model is a vital tool for many applications, one being the risk assessment of pesticides. A review of the model by the European Food Security Agency (EFSA) highlighted extensions to the model required before it can be incorporated. This research begins to answer some questions asked in that review.

The use of spatial statistics to investigate ecological processes in plant communities is becoming increasingly widespread. In diverse communities such as tropical rainforests, analysis of spatial structure may help to unravel the various processes that act and interact to maintain high levels of diversity. In particular, a number of contrasting mechanisms have been suggested to explain species coexistence, and these differ greatly in their practical implications for the ecology and conservation of tropical forests. Traditional first-order measures of community structure have proved unable to distinguish these mechanisms in practice, but statistics that describe spatial structure may be able to do so. This is of great interest and relevance as spatially explicit data become available for a range of ecological communities and analysis methods for these data become more accessible. This thesis investigates the potential for inference about underlying ecological processes in plant communities using spatial statistics. Current methodologies for spatial analysis are reviewed and extended, and are used to characterise the spatial signals of the principal theorised mechanisms of coexistence. The sensitivity of a range of spatial statistics to these signals is assessed, and the strength of such signals in natural communities is investigated. The spatial signals of the processes considered here are found to be strong and robust to modelled stochastic variation. Several new and existing spatial statistics are found to be sensitive to these signals, and offer great promise for inference about underlying processes from empirical data. The relative strengths of particular processes are found to vary between natural communities, with any one theory being insufficient to explain observed patterns. This thesis extends both understanding of species coexistence in diverse plant communities and the methodology for assessing underlying process in particular cases. It demonstrates that the potential of spatial statistics in ecology is great and largely unexplored.

The central question addressed in this thesis is whether descriptions of the dynamics of ecological systems at one scale may be effectively used as descriptions of the dynamics of ecological systems at larger scales. This question is addressed in the context of the dynamics of fungal communities. A simple experimental system and complementary theoretical approach, in the form of an individual-based (cellular automaton) model, is presented. Experimental results derived from small-scale systems are used to quantify parameters of the model; results from large-scale experimental systems serve to test the model. The theoretical analyses clearly demonstrate that the dynamics observed are a result of both local and non-local features of the experimental system. In cases such as this the immediate extrapolation of results derived from xperiments conducted out of the context of the community to represent system scale behaviour is not possible. In response to this observation a generic framework is developed to allow the consideration of effects at a range of scales through contextual parameterisation of localised dynamics. The framework is directed toward plant systems where a large body of experimental data exists, and may be parameterised by that experimental data. It represents the essential features of individual interactions in terms of competition for space and resource, and the behaviour of a given plant is described in terms of functional traits. Model runs demonstrate complex community patterns suggestive of a known biological phenomena, succession, that arises as a consequence of the coupling between the community and environment. This coupling may allow the long-term coexistence of species through some particular balance in individual function (traits) across the community. A search mechanism is determined to allow combinations of trait values at the scale of the individual to be assessed for a particular community-scale phenomenon. Initial results demonstrate that this mechanism may identify and converge on combinations of trait values that give rise to, in this case, a simple measure of diversity. The manner in which the generic framework developed may be applied to further the investigation into fungal community dynamics is addressed.

Neural network models were developed to predict the number of R. padi caught during the autumn flight period, at Lincoln, Canterbury. The models were based on weather data and aphids caught in a suction trap over the period 1982-2000. The first neural network model was trained using weekly data over a time-series of 15 years. The network predicted the yearly bimodal flight accurately, but could not generalise well to new data because of over-fitting. To eliminate stochastic noise from the data, four main modifications were carried out, 1) the autumn flight period only was used for modelling; 2) the total number of aphids caught during that period was used as the dependent variable to be predicted; 3) years were arranged randomly to prevent the network models from learning major trends; and 4) the data was pre-processed, by identifying and selecting periods in which each weather variable showed a strong correlation with the total number of aphids caught during the autumn flight. The capability of the new models to generalise improved significantly when validated with 13 years of jack-knifed data. A very high r value between the observed and predicted number of aphids (0.9575) was obtained. The neural networks were compared with a multiple regression model developed using the same data. Despite the r value between observed and predicted number of aphids for the multiple regression model was high (0.8591), the neural network model performed better. The absolute mean error (ABSME) for the multiple regression model was 313.58 aphids, whereas for neural network model it was 121.51 aphids. A further development was the design of a calibration model. The calibration model was validated by jack-knifing each year and predicting the number of aphids in the autumn flight. Predicted and observed numbers of aphids were compared, giving a very high r value (0.91) and an ABSME of 94 aphids over the 19 years in the time series. Further development of the neural network models is discussed.