The impacts of climate change on the abundance and distribution of the Spotted Wing Drosophila (Drosophila suzukii) in the United States and Canada

Abstract

D. suzukii is a relatively recent and destructive pest species to the North American soft-skinned fruit industry. Understanding this species' potential to shift in abundance and range due to changing climate is an important part of an effective mitigation and management strategy. We parameterized a temperature-driven D. suzukii population dynamics model using temperature data derived from several Global Circulation Models (CMIP5) with a range of relative concentration pathway (RCP) predictions. Mean consensus between the models suggest that without adaptation to both higher prolonged temperatures and higher short-term temperature events D. suzukii population levels are likely to drop in currently higher-risk regions. The potential drop in population is evident both as time progresses and as the severity of the RCP scenario increases. Some regions, particularly in northern latitudes, may experience increased populations due to milder winter and more developmentally-ideal summer conditions, but many of these regions are not currently known for soft-skinned fruit production and so the effects of this population increase may not have a significant impact.

Keywords: CMIP5; Drosophila suzukii; Elevated temperatures; Global circulation model; Invasive species; Soft-skinned fruit industry.

Figure 1. Schematic of the D. suzukii population dynamics model presented in Langille et al. (2016).

This schematic highlights population stage structure and relationships between state variable equations including principal input and output parameters. Fecundity (f), development (d) and mortality (μ) processes are temperature (T) dependent while development and mortality also rely on the fruit quality sub model (fr). In addition to eggs, pupa and adult males, the model contains three juvenile instar stages and seven adult female stages in order to account for variation in egg viability (v). θ represents the ratio of males to females and is typically set at 0.5. Solid arrows indicate developmental stage transition, dashed arrows indicate fecundity and dot-dashed arrows indicate mortality.

Figure 2. (A) Sample temperature profiles for Burlington, New Jersey from the CCSM4 model with different RCP-date combinations. Model-RCP combinations were selected to illustrate a range of output possibilities. (B) and (C) illustrate fruit quality submodel response and model population based on temperature profiles in (A).

Figure 3. Aggregated consensus maps of the modelled populations.

Data are the computed mean normalized total population (all life stages) from all four climate models. Normalization is based on the entire pre-mean data set. White cells located in the interior parts of the maps reflect missing values in the downscaled climate model outputs and a 0.25 degree bi-linear interpolation was applied to all data cells.

Figure 4. (A) Mean normalized population distribution across North America per-timeframe. Normalization occurs across the entire data set and mean is across all models and RCPs. (B) Corresponding per-timeframe coefficient of variation across all models and RCPs.

Coefficient of variation is based on the normalized population data. White cells located in the interior parts of the maps reflect missing values in the downscaled climate model outputs and a 0.25 degree bi-linear interpolation was applied to all data cells.

Figure 5. Grid cell coefficient of variation in population sizes versus the grid cell mean population size for the 2020s, 2050s and 2080s (A, B, C respectively).

The uncertainty in the population sizes that result from the differences between the climate model–RCP combinations tends to be larger as the mean of those population sizes gets smaller. Not surprisingly, the uncertainty is also larger later in the century than  earlier.

Figure 6. Normalized populations for historic temperatures and GCM projections.

GCM projections are averaged over all climate model-RCP combinations where × denotes mean population, whiskers denote maximum and minimum values, interior bar marks quartiles and small circles denote outliers. For averaged historic temperatures, large circles represent population normalized to GCM results.

Figure 7. (A) and (B) Normalized populations over 20 years of historic data (1993 through 2013) for Hillsborough, Florida and Fraser Valley, British Columbia respectively.

The dashed line indicates the mean annual population while the bold-solid line indicates the population for the mean temperatures across all years. (C) and (D) illustrate the daily temperature data for all years and the mean daily temperatures across all years for Hillsborough, Florida and Fraser Valley, British Columbia respectively.