Climate change is predicted to reduce sympatry among North American wood-warblers

ABSTRACT Anthropogenic climate change will dramatically alter species distributions. The rate and magnitude of range shifts, however, will differ among taxa, resulting in altered patterns of co-occurrence and interspecific interactions. We examined potential climate-mediated breeding range shifts among North American wood-warblers (Parulidae), a speciose avian family likely to be especially impacted by such changes due to high levels of interspecific competition and hybridization. We used publicly available species distribution model (SDM) range outputs to compare current ranges and patterns of sympatry among warbler species to future ranges and sympatry under 1.5°C, 2.0°C, and 3.0°C of average global warming. Range overlap among species and number of sympatric species are expected to decrease significantly in future warming scenarios, and unequal range shifts will alter the composition of warbler communities. On average, climate change will result in net decreases in the local species diversity; each warbler species is predicted to gain sympatry with approximately 1 new species and lose sympatry with approximately 2 species. Climate-mediated changes are predicted to differ among warblers in different regions of North America, with greatest impacts on eastern and boreal forest species. Our findings suggest that climate change will alter the diversity of wood-warbler communities during this century. Targeted monitoring of these changing interspecific relationships, especially for antagonistic interactions or hybridization between newly sympatric species, will be crucial for prioritizing particular species and regions in future conservation or management efforts. LAY SUMMARY Different rates and magnitudes of range shifts driven by climate change can lead to new patterns of range overlap among species. Changing patterns of range overlap can, in turn, change species interactions, resulting in altered patterns of competition and hybridization, for example. We compared current patterns of range overlap among North American wood-warblers (Parulidae) to predicted patterns with climate change using publicly available species distribution models. On average, each warbler species is predicted to gain sympatry with approximately one new species, while at the same time experiencing decreases in breeding range area, range overlap among species, and overall number of sympatric species, leading to changes in community composition. These results suggest that climate-induced changes in species interactions, especially among newly sympatric species, will be crucial to consider in future conservation efforts. RESUMEN El cambio climático antropogénico alterará drásticamente la distribución de las especies. Sin embargo, la tasa y la magnitud de los cambios de rango diferirán entre taxones, dando como resultado patrones alterados de co-ocurrencia y de interacciones interespecíficas. Examinamos los posibles cambios en el rango de reproducción mediados por el clima entre las especies de Parulidae de América del Norte, una familia de aves con muchas especies que probablemente se vea especialmente afectada por tales cambios debido a los altos niveles de competencia interespecífica e hibridación. Utilizamos rangos generados por modelos de distribución de especies disponibles públicamente para comparar los rangos actuales y los patrones de simpatría entre las especies de Parulidae con los rangos y los patrones de simpatría futuros bajo escenarios de calentamiento global promedio de 1,5°C, 2,0°C y 3,0°C. Se espera que la superposición de rangos entre las especies y el número de especies simpátricas disminuya significativamente en futuros escenarios de calentamiento, y los cambios de rango desiguales alterarán la composición de las comunidades de Parulidae. En promedio, el cambio climático resultará en disminuciones netas de la diversidad local de especies; se predice que cada especie de Parulidae aumentará su simpatría con aproximadamente 1 especie nueva y perderá simpatría con aproximadamente 2 especies, lo que provocará disminuciones netas en la simpatría. Se prevé que los cambios provocados por el clima difieran entre las especies de Parulidae en diferentes regiones de América del Norte, con mayores impactos en las especies de bosques orientales y boreales. Nuestros hallazgos sugieren que el cambio climático alterará la diversidad de las comunidades de Parulidae durante este siglo. El monitoreo específico de estas relaciones interespecíficas cambiantes, especialmente para las interacciones antagónicas o la hibridación entre nuevas especies simpátricas, será crucial para priorizar especies y regiones particulares en futuros esfuerzos de conservación o manejo.


INTRODUCTION
Anthropogenic climate change is causing a global redistribution of terrestrial species. Latitudinal and longitudinal geographic range shifts have been documented for a wide variety of terrestrial plants (Root et al. 2003, Parmesan and Yohe 2003, Kelly and Goulden 2008, Harsch et al. 2009) and animals (Hickling et al. 2006, Myers et al. 2009, Wu 2016. Virtually without exception, range shifts differ substantially among members of the same ecological community, often in both distance and direction (Lenoir andSvenning 2015, Fei et al. 2017).
Differences in species-specific range shifts are likely to increase as temperature and precipitation continue to change; consequently, patterns of species co-occurrence are likely to change substantially this century. This is exactly what most modeling studies predict. For example, Thuiller et al. (2005) were among the first to predict dramatic reorganization of ecological communities, with turnover of European plant communities this century estimated to range from 48% to 71% among biogeographic regions. Similarly, Bauer et al. (2016) predicted that, by 2,100, species composition of eastern U.S. forests will share only 45%-53% similarity with current species composition.
Novel community assemblages are expected to occur increasingly throughout this century (Williams and Jackson 2007, Reu et al. 2014, Radeloff et al. 2015, Ordonez et al. 2016. Novel communities will necessarily result in novel interspecific interactions, including competition over common resources, predator-prey interactions, hostparasite interactions, mutualisms, and in some cases hybridization. Shifts in species distributions are expected to bring previously allopatric species into contact while at the same time causing previously interacting species to become allopatric. These altered patterns of co-occurrence may have particularly dramatic consequences for taxa that are ecologically and/or genetically similar, such as species that are close phylogenetic relatives (Krosby et al. 2015).
Here, we explore climate-mediated changes in patterns of range overlap (sympatry) among the North American wood-warblers (family Parulidae), a songbird clade representing North America's most spectacular avian adaptive radiation Bermingham 1999, Lovette et al. 2010). Wood-warblers generally show little divergence in body size or shape or in general foraging preferences (MacArthur 1958, Morse 1989), yet they exhibit remarkable diversification in sexual signals such as plumage colors and song patterns (Shutler and Weatherhead 1990). Indeed, the wood-warblers family has been a model system for numerous studies of ecological and behavioral differentiation among species (MacArthur 1958, Shutler and Weatherhead 1990, Lovette and Bermingham 1999, Simpson et al. 2015, Simpson et al. 2020. Hybridization is relatively common in the family (Willis et al. 2014), including between genera (Toews et al. 2020). Furthermore, levels of sympatric overlap among wood-warblers species are significantly positively associated with levels of evolutionary trait divergence (Simpson et al. 2021), suggesting that species impose ecologically important selective pressures on each other where they co-occur.
Given the biological consequences of changes in sympatry, we sought to investigate the effects of anthropogenic climate change on potential future warbler communities. Accordingly, we calculated changes in the extent of range overlap between pairs of 47 warbler species (Supplementary Material Table S1) using projections generated by Bateman et al. (2020). In our analyses, we focused on breeding ranges and included all warblers with sufficient data to generate projections. These projections about climate-induced changes in the size and location of individual species' breeding ranges corresponded with global warming scenarios with average temperature increases of 1.5°C, 2.0°C, and 3.0°C. Because range shifts predicted by Bateman et al. (2020) differ among species, we expected that climate change would result in altered patterns of co-occurrence. We evaluated species distributions individually and in relation to each other under each climate scenario to estimate: (1) changes in breeding range area, (2) absolute and proportional gains and losses in sympatry among species, and (3) how these patterns will differ for species associated with different geographic regions of North America.

Species Distribution Models (SDM)
To assess the potential impacts of climate-mediated range shifts on warblers, we used SDM outputs provided by Bateman et al. (2020). These model outputs included projections for the breeding ranges of 47 wood-warblers species under greenhouse gas Representative Concentration Pathways (RCPs) 4.5 and 8.5 in 2041-2070 (the 2050s) and 2071-2100 (the 2080s). The RCPs across different decades corresponded with specific temperatures above pre-industrial levels, with RCP 4.5 in the 2050s corresponding to 1.5°C of average global warming, RCP 8.5 in the 2050s to 2.0°C, and RCP 8.5 in the 2080s to 3.0°C (IPCC 2014). To predict species ranges under each warming scenario, the SDMs incorporated species occurrence records from over 70 structured and unstructured avian survey datasets across Mexico, the United States of America, and Canada to create continuous projections of suitable habitat for 1 km 2 "pixels" across North America (see Bateman et al. 2020 supplement for occurrence data citations). The covariates used in these projections included modeled climate (AdaptWest Project 2015, Wang et al. 2016, projected vegetation type (Rehfeldt et al. 2012), and land cover (CCRS et al. 2013, Pekel et al. 2016 data. Furthermore, the models accounted for dispersal limitation by constraining suitable habitat areas based on mean natal dispersal (BirdLife International 2017) and generation time (Beauchamp 2010) of each species. We followed identical thresholding approaches presented by Bateman et al. (2020) to eliminate pixels that species are unlikely to occupy before creating predicted range maps for each model projection. We then used current and future modeled range maps to quantify predicted changes in overall range area, geographic range shifts, and co-occurrence among all warbler species pairs.
We further compared changes in overall breeding range area and patterns of sympatry among warbler species in four habitat affinity groups defined by Bateman et al. (2020): boreal forest (n = 14 species), eastern forest (n = 19), subtropical forest (n = 4), and western forest (n = 8) (North American Bird Conservation Initiative [NABCI] 2009). We did not include 2 of the 47 study species in these comparisons: one that was the only member of its habitat affinity group (Lucy's Warbler [Leiothlypis luciae]: arid lands) and one generalist species that did not belong to a specific habitat affinity group (Common Yellowthroat [Geothlypis trichas]).

Range Shift Distance and Direction Calculations
To calculate the distances of species range shifts, we measured the distance between the centroid of a species' current breeding range and the centroid of that range under each warming scenario (Pegan and Winger 2020). We weighted our centroid calculations in each warming scenario to prevent areas of low suitability from exerting a disproportionate effect on our estimations relative to their likelihood of occupation, based on the continuous 0 to 1 scale of habitat suitability generated by Bateman et al. (2020). This weighting was especially important for several species that are currently restricted to either eastern or western extents of North America, where SDMs predicted low suitability areas at opposite latitudinal extents that are much less likely to be occupied.
We calculated the distance by which species ranges are predicted to shift between each pairwise comparison of warming scenarios by converting the coordinates of each range centroid pair into a World Geodetic System 1984 (WGS84) projection. Furthermore, we used trigonometric functions to calculate the direction at which each centroid is predicted to shift in terms of a degree of angle bearing.

Species Range Overlap Analysis
To assess changes in sympatry among species, we compared range overlap between species' current breeding ranges to future overlap under each warming scenario. For each possible pair of species (n = 2,162 pairs) in each warming scenario, we used Python (version 3.7.6, Van Rossum and Drake 2009) for ArcGIS to overlay the range maps of each species in an Albers equal area conic projection and calculate the area of range overlap between the two species (see Figure  1). Using these results, we expressed this overlap area as a proportion of the overall breeding range of each species under a given warming scenario. We then calculated for each species the total number of other warbler species with which it is predicted to be sympatric over at least some of its breeding range. We considered a species as sympatric with another if it shared at least 10% of its total breeding range with the other species, based on inherent uncertainties in the distribution and climate models, as well as a conservative assumption that species pairs with range overlaps below this 10% threshold are less likely to have appreciable ecological effects on each other. Such effects are undoubtedly species-dependent and vary across habitats and seasons (Dhondt 2012). Ample resources may allow some species to broadly coexist, for example, while other species may intensely compete for limited foraging or nesting habitat (Martin and Martin 2001). We considered our 10% threshold to be a reasonable compromise.
To assess changes in community composition, we calculated the number of sympatric species gained and lost (currently allopatric species becoming sympatric and vice versa) for each species under each warming scenario, again defining the threshold between sympatric and allopatric as 10% of breeding range overlap. In practice, we found that different placements of this threshold (0%-20%) had little effect on relative changes in co-occurrence among taxa. We calculated gains and losses of sympatry as changes in proportional breeding range overlap between species and as changes in absolute numbers of sympatric species, both overall and within each of the four habitat affinity groups.
We compared our estimates of range shift, range area, range area overlap, proportional range overlap, number of sympatric species, and gains and losses of sympatry under each warming scenario to current values and to each other using paired t-tests. For each set of analyses examining variation in a response variable of instance (e.g., range area) across scenarios, we applied a Holm correction to adjust for multiple comparisons (Holm 1979). To examine patterns of change within habitat affinity groups, we additionally conducted these tests among each of these groups. After visual inspection to confirm assumptions of normality in differences of paired values (Leon 1998), we administered these tests in Python.

RESULTS
Wood-warbler breeding ranges are predicted to shift primarily northwards under all three warming scenarios (Figures 2 and 3), moving significantly farther under 3.0°C warming than under 1.5°C or 2.0°C warming (paired-t [hereafter t] = 9.07, 8.48, df = 46, P < 0.001 for both comparisons) and significantly farther under 2.0°C than 1.5°C warming (t = 9.07, P < 0.001). Similar climate-mediated range shifts are predicted to occur in each of the individual habitat groups, with higher warming causing greater shifts in species' breeding ranges (Figure 2, Supplementary Material Table S2).
Breeding range areas are not projected to decrease significantly for warblers as a whole, but species in boreal forest habitats are expected to experience significant decreases in breeding range area under 3.0°C warming (by 910,000 ± 200,000 km 2 on average, t = 3.44, P = 0.002; Figure 4A). Absolute and proportional levels of range overlap are projected to decrease significantly from current levels under all three warming scenarios, with higher levels of warming generally resulting in significantly greater change for warblers overall and within boreal and eastern forests ( Figure 4B and C, Supplementary Material Table  S3). Decreases in areas of overlap will, in turn, result in significant decreases in the average numbers of sympatric warbler species, and thus local warbler species diversity, under all warming scenarios (t = 1.81-3.17, P < 0.01 for all), especially among eastern forest species (t = 4.12-5.14, df = 18, P < 0.001 for all, Figure 4D, Supplementary Material Table S3).
These general decreases in local warbler species diversity reflect more complicated changes that will occur in community composition, in which species will gain new sympatric relationships with some species while losing sympatry with others. Currently, based on our 10% range overlap threshold for defining sympatry and allopatry, each of the 47 wood-warbler species in our study is sympatric with 19 other species on average (standard error [SE] = 1, range: 2-30 species, Figure 4D). With future warming, net decreases in sympatric species ( Figure 4D) are predicted to include novel sympatry with 1 new species on average along with simultaneous losses of sympatry (becoming allopatric) with 2 species at each level of warming ( Figure 5). Warbler species will experience significantly more losses of sympatry than gains across all 3 warming scenarios (t = 2.00-3.41, P < 0.05 for all, Supplementary Material Table S4), with more extreme levels of warming resulting in significantly greater net losses in sympatry and thus local species diversity under 2.0°C and 3.0°C scenarios ( Figure 5). Lost interactions among boreal species will be greatest under 2.0°C and 3.0°C warming (t = 2.47, df = 13, P = 0.01 for both comparisons). Furthermore, as noted above, net losses of sympatry will be especially dramatic in eastern forests, where species will lose approximately 5 times as many sympatric species as they gain (t = 4.66-5.42, df = 18, P < 0.001 for all warming scenarios, Figure 5, Supplementary Material Table S4).

DISCUSSION
Our study contributes to a growing body of evidence predicting that climate change will notably alter biological communities during this century, resulting in changes to existing species interactions that could pose a threat to the persistence of some species (Thuiller et al. 2005, Krosby et al. 2015, Bauer et al. 2016, Ordonez et al. 2016). North American wood-warblers are of particular interest in this context given their recent evolutionary divergence Bermingham 1999, Lovette et al. 2010), relatively frequent hybridization (at least 53% of species: Willis et al. 2014), high levels of sympatry (Simpson et al. 2021), and lack of morphological differentiation in traits related to foraging, such as bill size and shape (MacArthur 1958). We demonstrate that unequal range shifts among warblers are likely to result in general decreases in sympatry among species, while at the same time inducing novel contact between previously isolated species. Importantly, we further provide evidence that these changing interactions could differ in magnitude between species associated with different geographic regions Climate-induced range shifts will likely be accompanied by new sympatric contact between previously isolated species. In each of our warming scenarios, SDM outputs predicted that roughly half of North American wood-warbler species (23 of 47 species, or 48.9%) will become sympatric with at least one new species. These rates of secondary contact predicted here are much greater than that proposed for birds in general by Krosby et al. (2015), who projected that only 11.6% of currently isolated, closely related bird species across North, Central, and South America will experience future climate-induced range overlap. Given the ecological similarities and lack of morphological differentiation among most warbler species (MacArthur 1958, Morse 1989, Lovette and Bermingham 1999, as well as a high potential for hybridization across the family (Willis et al. 2014, Toews et al. 2020, currently allopatric wood-warbler species are likely to be especially impacted by secondary contact, which may have important consequences for warbler species persistence. This might have particularly far-reaching effects in subtropical and western forests, where our analyses predict that species could gain sympatric interactions with ~2 new species on average ( Figure 5). We speculate that the projected range shifts and relatively low current number of sympatric species among subtropical and western forest species facilitate greater increases in sympatry with future warming. However, our samples were limited (only 4 subtropical and 8 western forest species) for these groups.
At the same time, a net decrease in sympatric interactions among warbler species, measured as decreases in range overlap and average numbers of co-occurring taxa, is also likely to occur, largely driven by primarily northward shifts in breeding range areas. Range contractions are predicted to occur in boreal forest habitats, where the breeding range of nearly every warbler species is predicted to decrease, as well as in other habitat types. Species projected to have especially dramatic range contractions, with losses of 30%-87% of their breeding range areas, include Yellow-throated Warblers (Setophaga dominica), Hooded Warblers (S. citrina), and Pine Warblers (S. pinus) in eastern forests; Connecticut Warblers (Oporornis agilis) in boreal forests; and Common Yellowthroats, which are habitat generalists. Altogether, climate change will result in 94% (44 out of 47) of warbler species losing sympatry with at least one other species in each of our warming scenarios, leading to patterns of decreased local species richness. These lost interactions could have important consequences for species success. Newly allopatric warbler populations may benefit from reduced competition, especially among species with similar ecological niches Hochachka 2006, Rabosky andLovette 2008). However, species may also lose previous benefits from interspecific interactions . In migrants, for example, first time breeders may no longer be able to use the presence of certain other species in selecting profitable breeding habitat (Mönkkonen andForsman 2002, Szymkowiak et al. 2017). Species could also lose previous benefits of shared predator vigilance (Magrath et al. 2015).
Changes in community interactions, in which previously allopatric warbler species come into contact while previously sympatric species become allopatric, will affect nearly all North American wood-warbler species. In fact, apart from two species-Cape May Warbler (S. tigrine) and Bay-breasted Warbler (S. castanea), which are both boreal and almost entirely sympatric-all species in our study are predicted to gain or lose sympatric relationships, and often both (16-20 of 47 species), in one or more of our climate change scenarios.
Yet, not all climate-induced changes will be negative. The breeding ranges of some warbler species will not decrease and in fact are expected to expand under our warming scenarios. Eleven species across a range of nonboreal habitats are projected to experience range expansions greater than 25% of their current breeding range areas, including Cerulean Warblers (S. cerulea), Blackand-white Warblers (Mniotilta varia), and Prothonotary Warblers (Protonotaria citrea) in eastern forests, Blackthroated Gray Warblers (S. nigrescens) in western forests, Lucy's Warblers in aridlands, and Golden-cheeked Warblers (Vermivora chrysoparia) in subtropical forests. However, some factors pertaining to species-specific life histories that are not considered in this study, such as phenotypic plasticity and migration, could prevent these expansions and are important to consider alongside SDM projections (Gómez et al. 2016, Sofaer et al. 2018. For instance, the effects of climate change on warbler wintering grounds could limit breeding range expansion by inducing later arrival dates and other negative carry-over effects on breeding success (Rockwell 2012).

Implications for Conservation and Policy
The potential impacts of shifting breeding ranges and changing patterns of co-occurrence among warbler species are difficult to predict. Nevertheless, we illustrate through species of conservation concern (IUCN 2021) how our results may provide useful guidance for management to help warblers cope with some of the many threats to their persistence. The endangered Golden-cheeked Warbler, for example, is threatened primarily by reduction and fragmentation of current breeding habitat (Wahl et al. 1990). Our analyses suggest that its breeding range area could increase with global warming (by at least 38,000 km 2 , or 200% of its current range), yet it will also become newly sympatric with two other warbler species in each warming scenario: Painted Redstart (Myioborus pictus) and Redfaced Warbler (Cardellina rubrifrons). While it is unclear how interactions with new sympatric species could take shape, the expansion of favorable climates for Goldencheeked Warblers presents opportunities for population growth if management efforts target the conservation and restoration of habitat corridors or stepping-stone habitats that lead to large forest patches (Heller and Zavaleta 2009). Furthermore, predicted shifts in climatically optimal habitats for species of conservation concern could even require managers to shift their designations of high conservation value areas and consider the potential value of assisted dispersal in their species recovery plans.
In Cerulean Warblers, local fragmentation and loss of forests have also been the primary drivers of population decline (Robbins et al. 1992, Sauer et al. 2011, although habitat loss across wintering areas may also play a major role (Hamel et al. 2005). Thus, the notable expansion of suitable habitat predicted for this species in our study (by at least 160,881 km 2 , or >30%) could allow it to occupy new breeding habitats. While this expansion may not benefit overall population trends, our results also predict decreased areas of sympatry with a variety of other warbler species known to compete and sometimes replace Cerulean Warblers (Buehler et al. 2020), including American Redstarts (Setophaga ruticilla), Black-and-white Warblers, Hooded Warblers, and Golden-winged Warblers (V. chrysoptera). Such reduced competition could benefit Cerulean Warblers, though changing winter range conditions are also important to consider.
For Golden-winged Warblers, interactions with Bluewinged Warblers (V. cyanoptera) rather than habitat loss alone, have largely accounted for extirpation of local populations (Buehler et al. 2007). These two closely related species hybridize where they are sympatric, resulting in introgression of Blue-winged Warbler alleles and extensive range loss by Golden-winged Warblers (Confer et al. 2020). Our results predict that the breeding ranges of both species will expand and that overlap between them will decrease by 12.5%-17.2% in all scenarios except the most extreme warming scenario, where the range of Golden-winged Warblers is predicted to contract. However, both ranges will shift geographically and result in new areas of contact between these species in all scenarios. Frequent range shifts have occurred for this species in the past (Rosenberg et al. 2016), and based on historical interactions between these two species (Gill 1980), such novel areas of sympatry are likely to have disproportionately negative effects on Golden-winged Warblers.
General and targeted monitoring efforts will be crucial in predicting and responding to the impacts of climate change on many species. For example, Blackpoll Warblers (S. striata) have seen a population decline of at least 25% over the past decade (BirdLife International 2021), yet little information is available on the causes of this decline (DeLuca et al. 2020). Our results suggest that the breeding range of this species could either expand (by at least 150,000 km 2 ) or contract (by at least 340,000 km 2 ), depending on the warming scenario, and that these shifts would be accompanied by new sympatric contact with 3 other warbler species. As such, basic demographic monitoring is necessary to elucidate patterns and drivers of change, and targeted monitoring for antagonistic interactions as well as hybridization between newly sympatric species could be important in determining whether management interventions are necessary in novel warbler communities. Nevertheless, in terms of general trends in range areas and patterns of co-occurrence among warbler species, our study suggests that preventing higher levels of warming will result in significantly less change from current conditions. While our ultimate goal should be to entirely end anthropogenic warming of the planet, we emphasize that even partial mitigation of warming could be highly beneficial for species persistence.

Conclusions
Our results provide a framework for future work examining changes in patterns of competition and hybridization among North American warblers, adding to mounting evidence that species would benefit from immediate emissions mitigation to prevent substantial warming (Krosby et al. 2015, Bateman et al. 2020). Climate change is likely to differentially alter interactions among warbler species across multiple habitats through shifts in geographic distributions and patterns of species co-occurrence. Additionally, for species of conservation concern, differences among taxa in potential future changes highlight the need to consider each species individually in conservation planning. Finally, given that lower levels of warming will tend to result in significantly less change in breeding ranges and patterns of co-occurrence, our results emphasize that mitigation can minimize some of the potential impacts of climate change. Conservation efforts to help species adjust to the impacts of climate change will be crucial, and future research should focus on determining which species are most susceptible to impacts such as changes in ecological competition and introgression due to hybridization. Policies to mitigate and ultimately eliminate anthropogenic greenhouse gas emissions will be critical in ensuring the persistence of our ecosystems for generations to come.

SUPPLEMENTARY MATERIAL
Supplementary material is available at Ornithological Applications online.

ACKNOWLEDGMENTS
We first thank Dr. Brooke Bateman, Dr. Chad Wilsey, Lotem Taylor, Joanna Wu, Geoffrey LeBaron, and Dr. Gary Langham for developing and providing the model outputs necessary to our project. We also thank Dr. Steve Bertman, Alitzel Villanueva, Amelia Lewis, Elizabeth Clippard, and Raven Zellers for providing feedback throughout the project development process. Funding statement: This project was made possible by funding from the University of Michigan, National Science Foundation, and U.S. Department of Defense obtained through participation in the REU program "Climate Change in the Great Lakes Region" (AGS-1659338). Ethics statement: For this project, we used publicly available data obtained from an online repository.

Conflict of interest statement:
We have no conflicts of interest to declare. Author contributions: All authors contributed to establishing research questions, performing data analyses, and writing the manuscript. Data availability: Analyses reported in this article can be reproduced using the data provided by Bateman et al. (2020) and Pham et al. (2022).