Materials and Methods:
The climate suitability analyses presented here follow the methods of Andersen and Elkinton (2022) using locality records obtained from the invaded distributions of knotweed species to predict regions in the native range of these species where candidate biological control agents might be most successfully established. As per Andersen and Elkinton (2022), we used host-records (i.e., knotweed records) as a proxy for their specialist-parasites (i.e., A. itadori ), as the records for hosts are often more readily available in public databases (Andersen & Elkinton, 2022; Johnson et al., 2019; Schneider et al., 2022).
Climate suitability analyses were based on the use of published records for all species of Reynoutria (knotweeds) obtained from the GBIF database (accessed on September 28th, 2022: GBIF Occurrence Download https://doi.org/10.15468/dl.pdjdh8 ). This dataset was then filtered to remove all records that lacked geographic locality information, and then subdivided by focal species, resulting in one dataset each for R. japonica , R. × bohemica , and R. sachalinensis . The species datasets were then further subdivided into geographical bins, with separate bins for samples from North America (those samples located between 0°N, 180°W and 90°N, 20°W) and from Europe and western Asia (those samples located between 0°N, 20°W and 90°N, 60°E).
To reduce the effects of sampling biases in our analyses we followed the recommendations of Hijmans and Elith (2021). In the R statistical language environment (R Core Team, 2022), we used the packages ‘raster’ (Hijmans & van Etten, 2012) and ‘dismo’ (Hijmans et al., 2015) to randomly select one observation per 1 minute x 1 minute grid cell within each dataset. The final datasets were then used to independently estimate climate suitability envelopes in MaxEnt v 3.3.3e (Phillips et al., 2006; Phillips & Dudik, 2008) based upon the 5-minute resolution WorldClim v 2.1 dataset (available at https://www.worldclim.org ). Jack-knife analyses were performed on each dataset to measure the relative importance of each climate variable, and results were mapped in ArcGIS v 10.8 (Esri®, Inc., Redlands, CA)