C. H. Jarvis

The role of climatic mapping in predicting the potential geographical distribution of non-indigenous pests under current and future climates

Abstract Climatic mapping, which predicts the potential distribution of organisms in new areas and under future climates based on their responses to climate in their home range, has recently been criticised for ignoring dispersal and interactions between species, such as competition, predation and parasitism. In order to determine whether these criticisms are justified, the different procedures employed in climatic mapping were reviewed, with examples taken from studies of the Mediterranean fruit fly ( Ceratitis capitata ), Karnal bunt of wheat ( Tilletia indica ) and the Colorado potato beetle ( Leptinotarsa decemlineata ). All these studies stressed the key role played by non-climatic factors in determining distribution but it was shown that these factors, e.g., the availability of food and synchrony with the host plant, together with the difficulties of downscaling and upscaling data, were different to those highlighted in the criticisms. The extent to which laboratory studies on Drosophila populations, on which the criticisms are based, can be extrapolated to general predictions of species distributions was also explored. The Drosophila experiments were found to illustrate the importance of climate but could not accurately determine potential species distributions because only adult and not breeding population densities were estimated. The experimental design overestimated species interactions and ignored other factors, such as the availability of food. It was concluded that while there are limitations, climatic mapping procedures continue to play a vital role in determining what G.E. Hutchinson defined as the “fundamental niche” in studies of potential distribution. This applies especially for pest species, where natural dispersal is generally less important than transport by man, and species interactions are limited by the impoverished species diversity in agroecosystems. Due to the lack of data, climatic mapping is often the only approach which can be adopted. Nevertheless, to ensure that non-climatic factors are not neglected in such studies, a standard framework should be employed. Such frameworks have already been developed for pest risk analyses and are suitable for general use in studies of potential distribution because, in order to justify the phytosanitary regulation of international trade, they must also consider the potential for pests to invade new areas and the impacts of such invasions.

The sensitivity of a neural network for classifying remotely sensed imagery

Abstract A series of experiments are conducted on a feed-forward backpropagation neural network which is used to classify land cover from Landsat TM data. By investigating the effects of changing the numbers of network nodes in the input and hidden layers, potentially surplus nodes can be identified and removed to create a more compact network, without loss of classification accuracy. By exploring how momentum can be used with different rates of network learning, an optimal pairing is found which leads to a more rapid convergence and better classification of urban land cover than obtained in previous studies where momentum rarely was used. These optimal network parameters are used to classify an extract of a Landsat TM image of a dockland area with accuracy equal to that obtained using the maximum likelihood method. Given that in this case, the nature of the image data is ideal for a parametric method, this result is not unexpected. The competence of the neural technique is however demonstrated and criteria are given to help determine in advance when neural techniques may be preferable to parametric classifiers. Taken together, the findings show that careful balancing and adjustment of network parameters may be required to obtain a satisfactory result. The method can guide new users in configuring a popular neural network to suit their image data. Given the specific nature of our results, further research on neural networks in remote sensing could benefit from more systematic reporting of network parameters, training times and accuracies obtained.

The impact of interpolated daily temperature data on landscape-wide predictions of invertebrate pest phenology

Insect phenology depends upon temperature, and data from scattered synoptic weather stations are the principle inputs for phenology models used in decision support systems. The paper assesses the spatial dynamics of the penalty, as measured through errors in the timing of predicted insect development stages, that results when entomologists use daily maximum and minimum temperature data from the nearest station to a location, in comparison with an interpolated temperature equivalent, to drive their models. Jack-knife cross-validated estimates of temperature were propagated through an example phenology model, in this case for codling moth (Cydia pomonella). The intention was to contrast the effect of two interpolation methods on phenological results through time at different geographical locations. Use of weather data from the nearest UK meteorological data station (174 points) for phenological modelling doubled the error in predicted development dates for first generation development when compared with the use of landscape-wide interpolated daily temperature data. The results are based on a partial thin plate spline interpolation methodology: the figures are spatial and temporal averages for mainland England and Wales. Overall, spline interpolations provide phenology results that either exceed or are as good as nearest neighbour techniques for 75% of locations over England and Wales, taking first and second generation developmental stages into account. In a minority (21%) of cases nearest neighbour strategies (Voronoi methods) performed better, with an average 18-day improvement in the predictions of development date over the spline method on those occasions. Where splines performed best, their performance exceeded that of the Voronoi method by an average of 25 days. Nearest neighbour techniques did not necessarily perform well in lowland areas, indicating findings of potential significance to those considering input data requirements when modelling insect ecology.

GEO_BUG: a geographical modelling environment for assessing the likelihood of pest development

Abstract This paper describes software designed to explore pest phenology (development) over space and time. The framework presented links sequences of interpolated daily maximum and minimum temperatures with a variety of process-based phenology and accumulated temperature models. The flexibility offered by this approach is demonstrated using examples of gridded maps of pest phenology on target dates, graphs of the sequences of pest development at individual locations and assessments of error in the predicted dates over the course of a model run. Finally, the potential application of the software in support of agricultural management systems, policy development and integrated research is discussed.

Towards a British framework for enhancing the availability and value of agro-meteorological data

Abstract Geographical modelling methods offer opportunities for enhancing the resolution in time and space of agro-meteorological data. Given the number of emerging agricultural decision support initiatives in Britain, together with the growing use of weather-driven environmental models, exploring this potential is a matter of current importance. The paper discusses a geographical framework that focuses particularly on the spatial and temporal resolution of the information required, suggesting methods for enhancing the underlying real-time network of meteorological data observations to meet these challenges. Finally, we advance internet delivery as a means by which seamless data collation and distribution may be achieved through multi-agency collaboration.

Accounting for error when modelling with time series data: estimating the development of crop pests throughout the year

In order to attach some statement of reliability to mesoscale maps of how pest risk may develop over time, methods were developed to enable the detection and evaluation of errors in predictions that arise from the use of input data series from remote point sources. Firstly, we investigated how predicted model results may differ as a result of the ordering of the spatial interpolation and the model procedures. Principles of logic were used to detect errors occurring in the daily sequences of predicted pest development. Analyses of spatial autocorrelation within the gridded results showed that areas where a pest was predicted to reach a certain stage of development become more fragmented as a model run progressed over time. We identified that the less intensive approach of running a model only at data points and subsequently interpolating these to a grid can, in some cases, result in errors of logic and unrealistic degrees of autocorrelation. These errors occurred particularly when mapping a non-indigenous, marginal, pest at the later stages of its development. As a strategy for error evaluation, deterministic process models were run using point-based estimates of interpolated daily temperature to give RMS data errors at the sample points. This enabled us to investigate how the component of error related to sparsely distributed point data contributed to errors in the gridded estimates of pest development over time. The error detection and evaluation methods outlined are tractable and applicable to a wide variety of cases where point based models running over multiple time steps are extended to provide spatially continuous, landscape-wide, mappable results.