From identification to forecasting: the potential of image recognition and artificial intelligence for aphid pest monitoring

Insect monitoring has gained global public attention in recent years in the context of insect decline and biodiversity loss. Monitoring methods that can collect samples over a long period of time and independently of human influences are of particular importance. While these passive collection methods, e.g. suction traps, provide standardized and comparable data sets, the time required to analyze the large number of samples and trapped specimens is high. Another challenge is the necessary high level of taxonomic expertise required for accurate specimen processing. These factors create a bottleneck in specimen processing. In this context, machine learning, image recognition and artificial intelligence have emerged as promising tools to address the shortcomings of manual identification and quantification in the analysis of such trap catches. Aphids are important agricultural pests that pose a significant risk to several important crops and cause high economic losses through feeding damage and transmission of plant viruses. It has been shown that long-term monitoring of migrating aphids using suction traps can be used to make, adjust and improve predictions of their abundance so that the risk of plant viruses spreading through aphids can be more accurately predicted. With the increasing demand for alternatives to conventional pesticide use in crop protection, the need for predictive models is growing, e.g. as a basis for resistance development and as a measure for resistance management. In this context, advancing climate change has a strong influence on the total abundance of migrating aphids as well as on the peak occurrences of aphids within a year. Using aphids as a model organism, we demonstrate the possibilities of systematic monitoring of insect pests and the potential of future technical developments in the subsequent automated identification of individuals through to the use of case data for intelligent forecasting models. Using aphids as an example, we show the potential for systematic monitoring of insect pests through technical developments in the automated identification of individuals from static images (i.e. advances in image recognition software). We discuss the potential applications with regard to the automatic processing of insect case data and the development of intelligent prediction models.


Ethiopia
Brevicoryne brassicae aphid population dynamics 2020 "This study provided information on time of initial infestation and also the peak activity of the particular insect in relation to temperature." Shonga and Getu (2021) Mustard India Lipaphis erysimi aphid population dynamics, first aphid emergence 2005 "Hence, the present study was undertaken to develop forecasts for crop age at time of attack by L. erysimi , peak number of aphids on the crop in the season and crop age at peak population of the aphid." Chattopadhyay et al. (2005) Mustard India Lipaphis erysimi Revision previous model, aphid population dynamics 2008 "This paper extends the previous cumulative-size dependent models to include immigration." J. H.  Mustard India Lipaphis erysimi aphid population dynamics 2014 "Experimental data from six north Indian locations were used to study the role of weather on the incidence and development of mustard aphid." Rao et al. (2014) Mustard India Lipaphis erysimi aphid population dynamics 2020 "In this study, an effort is made to develop forewarning model using the field data on aphid for 12 consecutive rabi seasons from 2003-2004 to 2014-2015 under different agro-climatic locations in India." Tharranum et al. (2020) Oilseed rape Australia Myzus persicae Beet western yellows virus (BWYV) incidence, aphid population dynamics 2010 "The aims of this study were to: (i) adapt the hybrid mechanistic/statistical model of Maling et al. (2008) to predict aphid vector activity and epidemics of BWYV spreading from a nearby external source to a B. napus crop […]." Maling et al. (2010) Oilseed rape Iran Brevicoryne brassicae 50% aphid emergence 2016 "The aim of the study was to develop a model based on degreedays for predicting 50% emergence of the cabbage aphid population in canola fields." Nematollahi et al. (2016) Oilseed rape Australia Myzus persicae aphid population dynamics 2021 "Here, we develop a mechanistic model to explore the population dynamics of M. persicae and D. rapae during the canola growing season in southern Australia." Barton et al. (2021) Rapeseed mustard

India
Lipaphis erysimi aphid population dynamics 2012 "Thus, an attempt has been made here to predict the appearance and development of aphids on mustard crop for the eastern plains of Rajasthan agro-climatic conditions which ultimately may culminate into a Decision Support System (DSS) for the management of aphids." Rao et al. (2012) Rapeseed mustard GTLAUS tri-trophic model for simulation wheat-aphid-antagonist interaction developed by Freier et al. (1996) has been revised with the aim of improving the model details for better modelling of antagonist effects, etc. This paper describes the latest version, GET-LAUS01, and documents the results of model validation and predator effect simulations using field data collected over 8 years." Gosselke et al. fall and spring migration 2009 "Accordingly, the aim of our investigations was to determine relationships between mostly meteorological parameters and cereal aphid flight activity in order to identify reliable predictors of aphid immigration into winter wheat and winter barley crops in autumn and spring for modelling purposes." Klueken et al. (2009b) Barley USA Diuraphis noxia aphid population growth 2009 "We would like to emphasize that it is not the objective of this study to build predictive models for RWA population dynamics; instead, we try to explore the intrinsic properties of RWA population growth by building a catastrophe theory model with the population intrinsic rate ofgrowth (rm ) various environmental conditions."

Rhopalosiphum padi
Barley yellow dwarf virus (BYDW) incidence 2019 "Here, our aim was to explain aphid occurrences using Czech weather data and to develop predictive models that could enable farmers in Central Europe to predict the aphid migration time, duration, and total aphid numbers in the weeks ahead of their arrival." Jarošová et al.
Corn Egypt Rhopalosiphum maidis yield loss 2004 "The aim of the present study is to assess the corn yield losses and to build forecasting models of yield related with aphid infestation." Al-Eryan and El-Tabbakh (2004) Corn, sorghum, wheat, wheat volunteer France Rhopalosiphum padi aphid population dynamics 2021 "In this study, we used a spatially explicit simulation model to (1) determine how climate, pest immigration, and spatial-temporal variation in habitat availability influence seasonal variation of pest populations in both individual crops and at the landscape level, and (2) explore agronomic scenarios to identify potential effects of changes in practices on aphid populations and help inform future decisionmaking." Thierry et al.
Sorghum USA Melanaphis sacchari spatio-temporal infestation dynamics / invasion 2020 "We describe a prototype of a computational framework that could be used to forecast sugarcane aphid invasions of sorghum in near-real-time, as supported by timely field reports on current aphid infestation status." Koralewski et al.
Sorghum USA Melanaphis sacchari spatio-temporal infestation dynamics / invasion 2020 "We use an integrated ecological model to simulate local and regional infestation dynamics of sugarcane aphids, Melanaphis sacchari (Zehntner) (Hemiptera: Aphididae), on sorghum, Sorghum bicolor (L.) Moench (family Poaceae), in the southern to central Great Plains of the United States. Local dynamics of aphid populations on sorghum are simulated by a spatially explicit, individual-based model, whereas regional aphid migration is simulated by an atmospheric model that computes inert air particle (aphid) transport, dispersion, and deposition." Wheat Denmark Cereal aphids spatial aphid population growth 2020 "In this study, we have sampled aphid populations in a spatial set-up during both the initial and the epidemic phase to fit a population growth model. More specifically, we have developed a spatio-temporal stochastic aphid population growth model and fitted the model to empirical spatial time series aphid population data using a Bayesian hierarchical fitting procedure." Damgaard et al. (2020) Winter barley UK Rhopalosiphum padi aphid population dynamics 2000 "The aims of this paper are to utilize simulation analysis techniques to develop a model describing the population development of R. padi , to use the model to examine the relative importance ofvarious population processes in pest dynamics and to determine whether the model could be used as a basis for forecasting aphid outbreaks." Morgan (2000) Winter cereals England Rhopalosiphum spp., Sitobion spp.
Barley yellow dwarf virus (BYDV) incidence 1992 "A computer model is described which simulates the spread of barley yellow dwarf virus (BYDV) by aphids in winter cereals in SW England." Kendall et al. (1992) Winter wheat EU Sitobion avenae aphid population dynamics 1982 "Out aim in this monograph are to to explain the population development if S. avenae on cereals and to indicate how cereal aphid outbreaks might be predicted." Carter et al. (1982) Winter wheat UK Sitobion avenae aphid population dynamics 1985 "The simulation model described below was developed to gain understanding of the popula-tion dynamics of the grain aphid, Sitobion avenae (F.), in order to develop a reliable forecasting scheme." Carter (1985) Winter wheat UK, Netherlands Sitobion avenae peak population density 1986 "This paper sets out to test the hypothesis that both the population density of S. avenae on wheat and its observed rate of increase are important predictors of peak population size. It also aims to test whether they can be used together to forecast peak S. avenae density, and whether the anomalous suction trap forecasts cited above can be accounted for." Entwistle and Dixon (1986) Winter wheat Netherlands Sitobion avenae aphid damage / yield loss 1986 "Therefore, to quantify the effect of various dynamic reduction processes on growth and development of winter wheat, a simulation approach was adopted, using quantitative data from laboratory studies." Winter wheat Netherlands Sitobion avenae aphid damage / yield loss 1991 "The model is used to assess the contribution of various injury components of S. avenae to damage and to evaluate the consequences of the lack of detailed information on some processes in the aphid -winter wheat system for predicted yield." Rossing (1991a) Winter wheat Netherlands Sitobion avenae aphid damage / yield loss 1991 "Regression models are constructed that relate simulated aphid damage both during various periods of crop development and averaged over the entire post-anthesis phase, to the simulated attainable yield level." Rossing (1991b) Winter wheat Germany Sitobion avenae summer population size 1994 "The implemented model 'LAUS' for the grain aphid in early summer is presented with its weather-driven run through submodels for wheat microclimate and phenology and for the population dynamics of aphid antagonists." Friesland (1994) Winter wheat Germany Sitobion avenae, Rhopalosiphum padi, Metopolophium dirhodum aphid population dynamics 1996 "In this report, GTLAUS (version 3.7), a discrete simulation model of wheat -cereal aphid -predator interaction, which can be used for complex ecological studies in a representative tritrophic system of arable farming, is described and presented with selected scenario runs." Freier et al. (1996) Winter wheat UK Sitobion avenae aphid population dynamics 1997 "This paper describes the derivation of a mechanistic computer model of the interaction between the cereal aphid Sitobion avenae and the coccinellid Coccinella septempunctata, and its validation using all the limited data available." Skirvin et al. (1997) Way et al. (1981) Broad beans UK Aphis fabae aphid infestation 1994 "This paper describes the decision problem of controlling A. fabae , the knowledge acquisition and engineering, and the encoding of the information into a computer program." Knight and Cammell (1994) Luoin  Barlow and Dixon (1980) Lime tree New Zealand, UK Eucallipterus tiliae aphid population dynamics 1981 "This paper considers two approaches to the modelling of aphid populations. Using the lime aphid (Eucallipterus tiliae L.) as an example, it discusses the well-established role of detailed simulation models in the study of aphid population dynamics then considers the possible application of a simple, general herbivore/plant model to aphids." Barlow (1981) Orchards (apple)

Switzerland
Dysaphis plantaginea time of egg hatching 2006 "The objective of the present study was to develop a basis for a reliable, temperature-driven tool to forecast the phenology of D. plantaginea in order to facilitate decision making of apple growers with respect to aphid control before bloom." Graf et al. (2006) Orchards ( spring population size 2001 "The primary objective of this study was to quantify the e€ ffects of autumn and winter meteorological variables on the aphid species populations the following spring." Rongai et al. -UK Various initial flight activity 1991 "Studies have related suction trapping data to weather data […]. Results presented in this paper extend these studies to more sites, species and years and use more weather variables in an improved multiple regression approach." Harrington et al. (1991) -UK Myzus persicae initial flight activity 2009 "This paper describes the multiple regression technique in greater detail and assesses the abilities of the models using this technique, and using simple regression with mean temperature, to predict the date of the first record of Myzus persicae in the Rothamsted suction trap in the years from 1989 to 1992." Howling et al. (1993)