Dynamic selection of environmental variables to improve the prediction of aphid phenology: A machine learning approach

Abstract

Insect pests now pose a greater threat to crop production given the recent emergence of insecticide resistance, the removal of effective compounds from the market (e.g. neonicotinoids) and the changing climate that promotes successful overwintering and earlier migration of pests. As surveillance tools, predictive models are important to mitigate against pest outbreaks. Currently they provide decision support on species emergence, distribution, and migration patterns and their use effectively gives growers more time to take strategic crop interventions such as delayed sowing or targeted insecticide use. Existing techniques may have met their optimal usefulness, particularly in complex systems and changing climates. Machine learning (ML) arguably is an advance over current capabilities because it has the potential to efficiently identify the most informative timewindows whilst simultaneously improving species predictions. In doing so, ML is likely to advance the length of any integrated pest management opportunity when growers can intervene. As an example, we studied the migration of 51 species of aphids, which include some of the most economically important pests worldwide. We used a combination of entropy and C5.0 boosted decision trees to identify the most informative time windows to link meteorological variables to aphid migration patterns across the UK. Decision trees significantly improved the accuracy of first flight prediction by 20% compared to general additive models; further, meteorological variables that were selected by entropy significantly improved the accuracy by a further 3-5% compared to expert derived variables. Coarser (e.g. monthly) weather variables resulted in similar accuracies to finer (e.g. daily) variables but the most accurate model included multiple temporal resolutions with different period lengths. This combined resolution model alone highlights the ability of machine learning to accurately predict complex relationships between species and their meteorological drivers, largely beyond the experience of experts in the field. Finally, we identified the potential of these models to predict long-term first flight patterns in which machine learning attained equally high predictive ability as shorter-term forecasts. Whilst machine learning is a statistical advance, it is not necessarily a panacea: experts will be needed to underpin results with a mechanistic understanding, thus avoiding spurious relationships. The results of this study should provide researchers with an automated methodology to derive and select the most appropriate environmental variables when predicting ecological phenomena, while simultaneously improving the accuracy of such models.

Effects of Citrus Overwintering Predators, Host Plant Phenology and Environmental Variables on Aphid Infestation Dynamics in Clementine Citrus

“…Temperature and precipitation do not seem to directly limit the development of A. gossypii and A. spiraecola in western Mediterranean citrus agroecosystems (Ro et al 1998;Alyokhin et al 2011;Lu et al 2015;Holloway et al 2018). Both the lower and higher thermal developmental thresholds for the two aphid species (A. spiraecola: 2.3 ºC and 35 ºC; A. gossypii: 6.2 ºC and 35 ºC) (Wang and Tsai 2000;Kersting et al 1999) were never reached throughout our study period.…”

A review of the methods for studying biotic interactions in phenological analyses

“…Selection of abiotic parameters is well established in phenology, but it remains important to consider the representation of variables (e.g. mean, max, min) and the appropriate spatial and temporal resolution (Holloway et al., 2018; Simmonds et al., 2019; Van de Pol et al., 2016). From these variables, the modelling approaches identified through Sections 2–5 should then be selected according to the most appropriate data, interaction, spatiotemporal dimension and biotic representation.…”

“…Therefore, in the case of (local or global) extinction of one of the interactors or asynchronies between partners, the network could buffer the effect of such events, but there would be uncertainties related to how resilient species might be to such changes (Figure 1b,e). Moreover, many species initiate phenological events based on species interactions, with many aphid species initiating migration based on the senescing of host plants (Dixon & Glen, 1971; Watt & Dixon, 1981), which may or may not be captured by changes in the abiotic conditions alone (Holloway et al., 2018). In addition, the effect of climate change can disrupt biotic interactions in many ways, for example, turning from a favourable scenario of facilitation to competitive exclusion or competence (Blois et al., 2013) or by advancing or delaying phenology of species that exhibit changes in their interaction types through ontogeny, which is where species shift their relationship from competition to predation, facilitation to competition or herbivory to mutualism during their life cycle (Yang & Rudolf, 2010), illustrated in Figure 1c,f.…”

Artificial neural networks for monitoring network optimisation—a practical example using a national insect survey

“…This also affects our ability to disentangle complex relationships between measurements and environmental covariates. Machine learning approaches, such as random forests, are usually less hindered by aforementioned conditions (Holloway et al, 2018;Fouedjio and Klump, 2019). A key limitation, however, is that machine learning methods are often unable to associate their predictions with uncertainty estimates.…”