Predicting species invasiveness with genomic data: Is genomic offset related to establishment probability?

Abstract

Predicting the risk of establishment and spread of populations outside their native range represents a major challenge in evolutionary biology. Various methods have recently been developed to estimate population (mal)adaptation to a new environment with genomic data via so-called Genomic Offset (GO) statistics. These approaches are particularly promising for studying invasive species but have still rarely been used in this context. Here, we evaluated the relationship between GO and the establishment probability of a population in a new environment using both in silico and empirical data. First, we designed invasion simulations to evaluate the ability to predict establishment probability of two GO computation methods (Geometric GO and Gradient Forest) under several conditions. Additionally, we aimed to evaluate the interpretability of absolute Geometric GO values, which theoretically represent the adaptive genetic distance between populations from distinct environments. Second, utilizing public empirical data from the crop pest species Bactrocera tryoni, a fruit fly native from Northern Australia, we computed GO between "source" populations and a diverse range of locations within invaded areas. This practical application of GO within the context of a biological invasion underscores its potential in providing insights and guiding recommendations for future invasion risk assessment. Overall, our results suggest that GO statistics represent good predictors of the establishment probability and may thus inform invasion risk, although the influence of several factors on prediction performance (e.g., propagule pressure or admixture) will need further investigation.

Keywords: Bactrocera tryoni; GEA; biological invasions; genomic offset; local adaptation.

FIGURE 1

Schematic representation of the three simulated native environment types. For each environment type (columns), the two grids represent the optimal values of environmental variables $e_1$ (top) and $e_2$ (bottom) for each of the 25 populations. Population indices are indicated on the top left panel and specify the source populations used for invasion (see Simulations design: invaded area).

FIGURE 2

Mean R ² values between GO and $\log \left(p_e\right)$ (for the low migration rate and 10 invading individuals) for the different native environment types (L on the left; M in the middle; and R on the left) as a function of the covariables included in the computation of GO (two causal variables on top; eight covariables including two causal and six confounding covariables on center; and five PCs at bottom). Each panel represents the mean R ² value over 90 observations for each of the three possible source population for invasion (−1/−1, 0/0, and 1/1); specified on the x‐axis; and over 10 replicated simulations for the different GO estimators (${\mathrm{GO}}_{\mathrm{gf}}$, ${\mathrm{gGO}}_{\text{lfmm}}$, ${\mathrm{gGO}}_{\text{is}}$, and ${\mathrm{gGO}}_{\mathrm{mc}}$, see the main text for details) alongside with Euclidean environmental distance. All SNPs were used for GO computation.

FIGURE 3

Mean R ² values between GO and $\log \left(p_e\right)$, for the low migration rate and 10 invading individuals, depending on the type of native environment (L, M, or R). Each panel represents the mean R ² value (for each of the three possible source population for invasion, −1/−1, 0/0, and 1/1) between $\log \left(p_e\right)$ and GO obtained through five gGO computation methods, including ${\mathrm{gGO}}_{\text{lfmm}}$ and ${\mathrm{gGO}}_{\mathrm{mc}}$ modified in order to treat variables independently (noted as “univariate”), alongside Euclidean distance. All SNPs and covariables were used for GO computation.

FIGURE 4

Comparison between gGO and $f_2$ among QTNs (i.e., “ground truth” GO value) or among QTNs and neutral SNPs. The estimated values with the two gGO estimators (${\mathrm{gGO}}_{\text{lfmm}}$ and ${\mathrm{gGO}}_{\mathrm{mc}}$) were obtained using QTNs and neutral markers (with MAF $>0.01$) for the scenarios with low (left panel) or high (right panel) migration within the native area under the L (linear) environment. The inset in each panel gives the two corresponding MAPEs separated by a slash.

FIGURE 5

Application of GO to B. tryoni populations. GO was estimated between population 1 (“source” population, identified with a blue triangle) and a large area in Oceania, with ${\mathrm{gGO}}_{\mathrm{mc}}$, ${\mathrm{gGO}}_{\text{lfmm}}$, and ${\mathrm{GO}}_{\mathrm{gf}}$. Squared Euclidean distance to the source population is also displayed. Shades of yellow indicate lower GO values, while red shades higher GO values. Grey pixels represent outliers values, and black dots the studied populations.