Predicting future evolutionary change is a critical challenge in the Anthropocene as geographic range shifts and local extinction emerge as hallmarks of planetary change. Hence, spatial sorting – a driver of rapid evolution in which dispersal-associated traits accumulate along expanding range edges and within recolonized habitats – might be of growing importance in ecology and conservation. We report on the results of a natural experiment that monitored re-colonization of host plants by the seed-feeding, red-shouldered soapberry bug, Jadera haematoloma, after local extinctions from catastrophic flooding in an extreme hurricane. We tested the contribution of spatial sorting to generate rapid and persistent evolution in dispersal traits, as well as in feeding traits unrelated to dispersal. Long-winged dispersal forms accumulated in re-colonized habitats, and due to genetic correlation, mouthparts also became longer, and this shift persisted across generations. Those longer mouthparts were likely adaptive on one host plant species but maladaptive on two others based on matching the optimum depth of seeds within their host fruits. Moreover, spatial sorting eroded recently evolved adaptive divergence in mouthpart length among all host-associated biotypes, an outcome pointing to profound practical consequences of the extreme weather event for local adaptation, population resilience, and evolutionary futures.

Field sampling soapberry bug populations

Our study was conducted as a long-term natural experiment in a 1,200 km² region in the Houston area that spanned the year prior to and the three years after the 2017 flooding event caused by Hurricane Harvey. In doing so, we monitored soapberry bug populations that inhabit three locally available host plant species, balloon vine (Cardiospermum halicacabum), western soapberry (Sapindus v. drummondii), and goldenrain tree (Koelreuteria elegans), (Fig. 2d-f). Our study included 15 sites (three C. halicacabum, three S. drummondii, and nine K. elegans; Extended Figure 1) sampled 72 times over the three-year period after the hurricane (September 2017 to September 2020). Sampling was performed on days without precipitation and was conducted between the hours of 10:00 and 15:00. Each site was sampled once every 14 days. Of the 15 monitored sites, nine sites flooded (three S. drummondii, six K. elegans, two C. halicacabum), while four sites did not flood and serve as our controls (three K. elegans and one C. halicacabum; Supplemental Table 1). Flooded sites in this study were inundated with standing water from 7 days to 90+ days. No S. drummondii sites were included in our control as this natural experiment relied on the serendipitous overlap of flood patterns with our previous history of insect sampling. Taking advantage of haphazard flooding also created discrepancies in host plant abundance, which varied between sampling site with Koelreuteria- and Sapindus-associated sites consisting of anywhere from one to seven trees and Cardiospermum-associated sites consisting of five to twenty herbaceous plants.

For each sampling effort, insects were collected by hand over a period of 30 minutes from a 50 m² region around the base of the host plant. Captured insects were temporarily stored in a plastic container lined with Insect-a-Slip (polytetrafluoroethylene coating, Bio Quip Products, Inc., Rancho Dominguez, CA). For each individual soapberry bug, insect beak-length was measured in the field using a digital caliper (Stainless Steel iGaging ABSOLUTE ORIGIN 0–6ʺ Digital LCD Electronic Caliper-IP54; iGaging, Los Angeles, Calif.), and wing morphology (macropterous = full flight wings; brachypterous = reduced wings) was visually assessed by the presence of developed corium of the distal membrane of the insect’s forewing (Fig. 2b). Captured insects were immediately returned to their habitat after morphological measurements to avoid artificial selection on any population. Given that some flooded sites were colonized earlier than others, we limited our data analysis to individuals collected in the first year (26 sampling periods) after colonization at flooded sites and after the hurricane at control sites to avoid overrepresentation of any one individual site. 

Field measurements of soapberry bug morphology

The morphological measurements of the focal insect traits included beak length, wing-form, and insect forewing length. We measured beak length as the distance from the clypeus margin of the insect's lower sclerites to the most distal point of the labium. The categorical variable wing-form was determined to be either “brachypterous” or “macropterous” based on visual inspection. An individual was classified as brachypterous when the medio-cubital and apical folding fields of the hind wing were underdeveloped and a developed distal membrane on forewing was absent; an individual was classified to be “macropterous” if the medio-cubital and apical folding fields of the hindwing were fully developed and the distal membrane of the forewing was present (Fig. 2b,c). Forewing length was measured as the distance from the proximal margin of the scutellum to the distal edge of the forewing membrane. Individuals where trait measurements were obscured by damage were excluded. 

This data was entered in Excel, and analysis was conducted in R.