An integrative phenology and climatic suitability model for emerald ash borer

Abstract

IntroductionDecision support models that predict both when and where to expect emerald ash borer (EAB), Agrilus planipennis Fairmaire (Coleoptera: Buprestidae), are needed for the development and implementation of effective management strategies against this major invasive pest of ash (Fraxinus species) in North America and other regions such as Europe. We present a spatialized model of phenology and climatic suitability for EAB for use in the Degree-Days, Risk, and Phenological event mapping (DDRP) platform, which is an open-source decision support tool to help detect, monitor, and manage invasive threats.MethodsWe evaluated the model using presence records from three geographic regions (China, North America, and Europe) and a phenological dataset consisting primarily of observations from the northeastern and midwestern United States. To demonstrate the model, we produced phenological event maps for a recent year and tested for trends in EAB’s phenology and potential distribution over a recent 20-year period.ResultsOverall, the model exhibited strong performance. Presence was correctly estimated for over 99% of presence records and predicted dates of adult phenological events corresponded closely with observed dates, with a mean absolute error of ca. 7 days and low estimates of bias. Climate stresses were insufficient to exclude EAB from areas with native Fraxinus species in North America and Europe; however, extreme weather events, climate warming, and an inability for EAB to complete its life cycle may reduce suitability for some areas. Significant trends toward earlier adult emergence over 20 years occurred in only some areas.DiscussionNear real-time model forecasts for the conterminous United States are available at two websites to provide end-users with decision-support for surveillance and management of this invasive pest. Forecasts of adult emergence and egg hatch are particularly relevant for surveillance and for managing existing populations with pesticide treatments and parasitoid introductions.

Phenological Mapping of Invasive Insects: Decision Support for Surveillance and Management

“…Our current assessment is that using simple linear degree-day accumulations to display either developmental stages (via lookup table) or dates of phenological events are often adequate in their predictive accuracy, while still simple enough to be readily adapted for dozens, or even hundreds, of species. For example, validation analyses of certain models used by the DDRP, Pheno Forecast, and SAFARIS platforms have revealed evidence of overall good predictive performances [15,16,46,47]. To our knowledge, degree-day maps based upon nonlinear temperature response rates have not reached implementation for tactical decision support, despite the trend of using nonlinear equations to model the temperature-development response of insects [24,49,52,67,97].…”

“…Phenological maps may be used to assess the potential impacts of climate change on invasive insect phenology. For example, Barker et al [46] tested for trends in the date of the first adult emergence of the emerald ash borer for North America and Europe after combining model outputs for each year over a recent 20-year period. Other studies have modeled the phenology of invasive insects under future climate change scenarios to estimate the impacts of global warming and altered precipitation patterns on important phenological events, potential voltinism, and population growth [42,[68][69][70][71][72].…”

“…However, historical datasets may be used for model development and validation, such as those for Europe [83], continental North America and Hawaii [84,85], Brazil [79,86], China [80,81], India [82], and Bangladesh, Nepal, and Pakistan [78]. For example, the validation of DDRP models for the emerald ash borer and the small tomato borer [Neoleucinodes elegantalis (Guenée)] used E-OBS and BR-DWGD data for Brazil and Europe, respectively [16,46]. Some phenological mapping studies overcame an absence of readily available gridded daily climate data by interpolating weather station data over a landscape of interest using custom software [31,40,41,[87][88][89][90].…”

“…Phenology model validation requires a set of observations not used in model development. Observations could be resampled using jackknifing or cross-validation methods, split into separate sets (e.g., 75% for development, 25% for validation), or originate from different years or areas [31,87,90,120,121] [15,16,46,47]. Caution should be taken when using citizen science collected data (e.g., from iNaturalist) because they may contain errors, such as species misidentifications or incorrect dates and locations [93,94].…”

“…Climate data at fine-scale spatial resolution increased the prediction accuracy because local effects (e.g., of mountains, valleys, or large bodies of water) that might influence phenological events could be modeled [43][44][45]. Additionally, advancements in satellite, cellular, and internet communications allowed meteorological observations to be released within hours of being collected, enabling researchers to produce phenology maps in near real-time to gain insight into the current development and activities of a species [15,16,46,47]. For example, the advent of free, near real-time (henceforth real-time for brevity) daily climate data from the Parameter-elevation Regressions on Independent Slopes Model (PRISM) database [http://www.prism.oregonstate.edu (accessed on 19 December 2023)] in 2011 was a major breakthrough for phenological mapping for the contiguous United States (CONUS), as PRISM data prior to this time were only available as monthly grids that were released months after data collection [43,44].…”