Discussion

 
We employed a geometric morphometrics framework to demonstrate strong differentiation in wing venation geometries across three sympatric species of Halictus bees, as well as significant (though less pronounced) differentiation across island and mainland populations of H. tripartitus. Species-level variation in wing venation is well established for many bee taxa, serving as useful characters for identification to family, genus, or species levels (Michener, 1994). Several studies have even successfully discriminated between distinct insect subspecies or genetic lineages within species using wing morphometrics (Akahira & Sakagami, 1959; Francoy et al., 2008, 2011; Carneiro et al. 2019)[MOU3] [CT4] . Fewer studies, however, have investigated population-level variation in wing venation (Francoy et al., 2011; Rossa et al., 2016). This variation in wing morphology can serve as a useful proxy for assessing the extent of divergence between closely related populations[MOU5] [CT6]  (Oleksa & Tofilski, 2015; Peil & Aranda, 2021). More broadly, these results provide robust evidence for microevolutionary change in wing morphology across reproductively isolated bee populations.
            As for many island bee populations, it is unknown how and when this population of H. tripartitus colonized Santa Cruz Island. Island dispersal by bees is generally poorly understood, though phylogenetic analyses and behavioral studies can offer clues to potential avenues for colonization events. On Santa Cruz Island, colonization by many bee species could have occurred during the Last Glacial Maximum (17,000-18,000 years ago), when lowered sea levels reduced the distance from the mainland to about 6 km (Miller, 1985). Gene flow across the channel may have continued for an unknown period, depending primarily on the distance of the water barrier and the dispersal capabilities of H. tripartitus. Dispersal timing aside, it is evident from our results that the Santa Cruz Island population has diverged morphologically from the mainland population. The unique selective environment of the island (i.e., including climatic and ecological differences from the mainland) may contribute to this population divergence, in addition to founder effect and genetic drift. Future sampling across the entire Channel Island archipelago could shed light on historical patterns of dispersal and population divergence, inferred from patterns of differentiation in wing landmark geometries. Finally, comparisons to other populations across the considerable geographic range of H. tripartitus would provide interesting context for assessing the relative magnitude of phenotypic divergence in the Santa Cruz Island population.
These results highlight the utility of geometric morphometrics for quantifying complex patterns of phenotypic variation that elude observation via simple measurement techniques. The application of geometric morphometrics to insect wing venation patterns is still a relatively recent development, but already has shown promise for species identifications (Aytekin et al., 2007; T. Francoy et al., 2009; Rattanawannee et al., 2010). Our accurate discrimination between three Halictus species likewise supports a role for geometric morphometrics in taxonomic identification to the species level. Further, geometric wing morphometrics may be useful for distinguishing among populations (Rossa et al., 2016; Henriques et al., 2020)[MOU7] [CT8]  and between species within complexes (Francoy et al., 2011). Identifying features of wing venation have even been successfully integrated into computer-aided identification systems, which can accurately identify bee specimens to species and even subspecies from images of wings (Buschbacher et al., 2020; Rattanawannee et al. 2012).[MOU9] [CT10]  Our results indicate that population variation in wing venation can be successfully discriminated using geometric morphometrics, and suggest that these patterns could be usefully extended toward automated identification systems with the aim of further classifying specimens to the population level.
Beyond its use in population identification, wing morphometry holds valuable potential for large-scale population studies, by providing a tractable alternative to more costly and time-consuming molecular methods for analyzing population structure. Unlike some morphological traits that can degrade over time, wing venation is strongly preserved in museum specimens, presenting opportunities for sampling of existing specimens in place of conducting new surveys. Wings represent powerful candidates for geometric morphometric analysis because their two-dimensional surfaces lend themselves to straightforward imaging, in contrast to three-dimensional traits that require additional protocols to standardize orientation within images. Future studies seeking to identify bees to species or population level may find this methodology viable and potentially more adaptable than traditional taxonomic identifications using dichotomous keys. In particular, we envision that wing morphometrics could increase the feasibility of large-scale monitoring projects by reducing taxonomic labor (Engel et al., 2021).
            In conclusion, we demonstrated species- and population-level variation in Halictus wing venation. Our results provide evidence for the divergence of wing venation patterns in isolated island and mainland populations of H. tripartitus. Our study emphasizes that wing venation patterns can act as quantifiable indicators of phenotypic differentiation within species, and may be useful for inferring the extent of variation among reproductively isolated populations. Morphological population signatures such as these hold enormous potential for enabling broader assessments of evolutionary change across insect populations over time, over geographic space, or with climatic variables.

 

Acknowledgments

 
We would like to thank Zoe Wood, Evan Hobson, and Charles Braman for helpful discussion and comments on the manuscript, Dr. Christopher Evelyn for statistical guidance, and Yolanda Diao for landmarking specimens. We also thank Luz Ceja for producing the lateral specimen images. We also thank reviewers for comments that improved the manuscript. Finally, we thank Jaime Pawelek for identification of bee specimens. This research was funded by a National Science Foundation (NSF) Award (DBI-2102006) to KCS and a University of California Santa Barbara College of Creative Studies Summer Undergraduate Research Fellowship to CNT.
 

Data Accessibility Statement

All data associated with this study is publicly available in Zenodo.

 

Supplementary Materials
 
Supplementary Table 1: Mean landmark coordinate values by species and population.
  Halictus tripartitus Halictus ligatus Halictus farinosus
Landmark Mainland population Island population    
1x 0.394 0.392 0.395 0.386
1y 0.153 0.153 0.127 0.136
2x 0.242 0.244 0.267 0.292
2y 0.256 0.253 0.249 0.275
3x -0.025 -0.031 -0.032 -0.024
3y 0.185 0.185 0.181 0.177
4x -0.121 -0.122 -0.119 -0.113
4y 0.206 0.207 0.202 0.197
5x -0.321 -0.320 -0.339 -0.338
5y 0.010 0.014 0.005 -0.005
6x -0.488 -0.488 -0.503 -0.496
6y -0.025 -0.023 -0.036 -0.046
7x 0.030 0.032 0.029 0.024
7y -0.175 -0.176 -0.159 -0.163
8x 0.100 0.106 0.109 0.093
8y -0.290 -0.293 -0.271 -0.274
9x 0.188 0.187 0.193 0.175
9y -0.321 -0.320 -0.298 -0.297