Analysis of the invasiveness of spotted wing Drosophila (Drosophila suzukii) in North America, Europe, and the Mediterranean Basin

Abstract

The polyphagous Asian vinegar fly Drosophila suzukii (spotted wing Drosophila) is a native of Eastern and Southeastern Asia. It emerged as an important invasive insect pest of berries and stone fruits in the Americas and Europe beginning in 2008. Species distribution models are commonly used for analyzing the extant and potential range expansion of invasive species. Previous modeling efforts for D. suzukii include a degree-day model, a MaxEnt ecological niche model, a demographic model incorporating the effects of temperature, and a preliminary mechanistic physiologically-based demographic model (PBDM). In the present analysis, we refine the PBDM for D. suzukii based on biological data reported in the literature. The PBDM is used to assess the effects of temperature and relative humidity from a recently published global climate dataset (AgMERRA) on the prospective geographic distribution and relative abundance of the pest in the USA and Mexico, and in Europe and the Mediterranean Basin. Our focus is on areas of recent invasion and of predicted higher invasiveness in these areas. Although the species is native to Asia and is of putative temperate origins, it has established in subtropical to north temperate zones worldwide where it infests a wide range of wild and domesticated berries and stone fruits. The model captures the observed phenology of D. suzukii at specific locations, as well as the potential geographic distribution and relative favorability across larger regions. The main limiting factor is cold winter temperature in northern areas, though high temperatures and low relative humidity may be limiting in arid areas. The effect of greater cold tolerance in winter morph adults is explored.

Appendix

A review of physiologically based demographic model (PBDM)

Physiologically-based demographic system models (PBDMs) explicitly capture the mechanistic weather-driven biology and dynamics of species at all trophic levels to predict the weather driven phenology, dynamics and distribution of single and multiple species across wide geographic areas on a daily basis—a time step rarely used in macro-ecological modeling (see Gutierrez and Baumgärtner 1984; Gutierrez 1996; Gutierrez and Ponti 2013). For some applications, the time step could be smaller or larger (Gilioli et al. 2016). The model captures via sub-models the processes of resource acquisition and allocation, and the birth–death rates in great detail or using simple functions that capture the relevant biology (see Gutierrez 1996; Gutierrez and Ponti 2013; Gilioli et al. 2016). PBDMs are sufficiently detailed to be realistic, and yet complexity is kept to a minimum by applying the same dynamics model and process sub models to all trophic levels. The complexity enters the model at the conceptual level and running the model requires minimal computational capacity. These models have contributed to basic theory and helped solve many applied field problems because they bridge the gap between purely theoretical analytic models and overly complicated simulation models. Physiological analogy across trophic levels is a powerful conceptual tool and is used as a way to tackle the huge challenges facing global ecosystem modeling.

A demographic dynamics model with distributed maturation times

The biology of resource acquisition and allocation is embedded in a demographic model with distributed maturation times (Manetsch 1976; Vansickle 1977) and is used to simulate the dynamics of age (and mass) structured populations where time (t) is chronological time and age (a) is in physiological time units (e.g., proportional development, degree days). But other dynamics models could also be used (e.g., Gutierrez 1996; DiCola et al. 1999; Gilioli et al. 2016).

The general distributed maturation time model for the ith age class of a population as a rate (r _i ) is

$$\frac{{dr_{i}}}{dt} = \frac{k\,\Delta x}{del}[r_{i - 1} (t) - r_{i} (t)] - \mu_{i} (t)r_{i} (t).$$

(11)

The density of the ith cohort is $N_{i} (t) = \frac{del}{k}r_{i}$, where Erlang parameter k is the number of different age cohorts (stages), del is the expected mean developmental time, Δx is an increment in time varying physiological age, $\frac{k\,\Delta x}{del}$ scales the time varying flow rate of individuals between age classes, and $- \infty < \mu_{i} (t) < \infty$ is the proportional net loss rate that could include age-species death and net immigration and in mass dynamics model the age specific growth rate. Note that births enter the first age class with age zero. The density of all k cohorts is $N(t) = \sum\nolimits_{i = 1}^{k} {N_{i}} (t)$, but the density can also be summed by life stage. The maturation across the k age classes is depicted in Fig. 8a, while the patterns of emergence times for different values of k are depicted in Fig. 8b (see Manetsch 1976; Vansickle 1977). Assuming k = 25 age classes, the standard deviation of developmental times for the immature and adult life stages (i.e., $std = \sqrt {del^{2}/k}$) is ~20 % of the mean. Similar schemes could be used for each life stages (Fig. 8c).

Fig. 8

A distributed maturation time model: a the general model across all life stages including flow rates between stages (r), net mortality rates [i.e. μ _i (t), i = 1, …, k] due to biotic and abiotic factors where x ₀(t) are births and y(t) is death at maximum age, b the stylized distribution of cohort maturation times given different values of the Erlang parameter k, and c the general model with the dynamics of a specified for each life stage (S) including stage specific flow rates between stages (y _stage ), net mortality (i.e. μ_stage) due to biotic and abiotic factors (see text)