Data from: Temperature seasonality drives taxonomic and functional homogenization of tropical butterflies
Cite this dataset
Hulshof, Catherine et al. (2024). Data from: Temperature seasonality drives taxonomic and functional homogenization of tropical butterflies [Dataset]. Dryad. https://doi.org/10.5061/dryad.ngf1vhhzz
Abstract
To understand the potential impact of climate change on butterfly assemblages across a tropical island, we used thousands of museum records of diurnal Lepidoptera to model current (1970–2000) and forecast future (2061–2080) species distributions and combined these to test for taxonomic and functional homogenization. We then quantified climatic-mediated effects on current and forecasted taxonomic and functional composition and, specifically, whether temperature was a primary driver, as predicted by the temperature-size rule and the thermal melanism hypotheses. Finally, we measured wing traits important in thermoregulation (size and color) and determined trait-mediated changes in forecasted species distributions over time. Our models projected an increase in taxonomic and functional richness over time, and a decrease in taxonomic and functional turnover – a signature of biotic homogenization. Under future climate scenarios, models projected a decrease in wing length and an increase in wing brightness at higher elevations. One variable, temperature seasonality, was the strongest predicted driver of both the current spatial distribution and the projected percent change over time for not only wing traits, but also taxonomic and functional richness and turnover. This dataset contains wing trait data for 62 butterfly species of Puerto Rico, included in the study, measured on digitized museum specimens. Generally, 10 individuals of each species were measured, allowing for measurements of inter- and intraspecific wing trait variation. Traits include: wing length and width, and color metrics: intensity, hue, saturation, and brightness.
README: Data from: Temperature seasonality drives taxonomic and functional homogenization of tropical butterflies
Description of the Data and file structure
We used the Stuart J. Ramos Collection to measure wing traits because it was (and is) the only digitized collection of Puerto Rico butterflies. The Stuart J. Ramos Collection (SJRC; 3410 records; SJRC) was donated by the late S. J. Ramos to the McGuire Center for Lepidoptera and Biodiversity, Florida Museum of Natural History (MGCL) at the University of Florida, in Gainesville, Florida, USA. We used image analysis software to measure dorsal forewing length as the length between the forewing tip and the site of wing attachment on the thorax, and forewing width as the widest section of the forewing. We also measured hue, saturation, and brightness (HSB) on a 1 cm2 region of the wing near the thorax, avoiding major veins and visible scale damage. Furthermore, we focused on females because 1) their parental investment and, thus, energetic requirements are often greater than those of males, and 2) male coloration can be biased by sexual selection. Of the 62 species, only four species (all Nymphalidae) had visually noticeable size differences, with females being larger than males. In a few instances in which images of females were not available, or the sex could not be determined from visual characteristics such as markings or size differences, we selected individuals randomly within species for trait measurements (Female:Unknown = 48:14). We aimed for a maximum of ten individuals per species when possible (median = 10; mean = 8.1 individuals per species). The speciesTraits.csv file contains the family and species name for 62 species in addition to wing traits: wing length (cm), wing width(cm), intensity, hue, saturation, and brightness. Additional columns describe the number of images for which traits were measured (num_images) and the number of those images whose specimen was identified as female (num_images_females). A blank cell indicates that it was not possible to identify the sex of the specimen based on visual characteristics alone. The column 'occurences_n' describes the number of occurrences for each species. The column 'percPoorAUC' describes the percentage of models per species exhibiting poor performance (using a mean AUC threshold greater than 0.6; Cheng & Bonebrake 2017). Finally, species range size in square kilometers is described in column 'rangeSize_km2'. Interestingly, the percentage of models exhibiting poor performance per species was positively correlated with species range size (Rsq = 0.58).
Sharing/access Information
The data are derived from digitized specimen images from the Stuart J. Ramos collection which are available in the Symbiota Collections of Arthropods Network (Gries et al., 2014; SCAN, 2019).
Gries, C., Gilbert, M. E. E., & Franz, N. M. (2014). Symbiota: A virtual platform for creating voucher-based biodiversity information communities. Biodiversity Data Journal, 2, e1114.
SCAN. 2019. Biodiversity occurrence data accessed through SCAN Data Portal, http//:scan-bugs.org/portal/index.php, accessed on 2018-05-01.
Methods
We used the Stuart J. Ramos Collection to measure wing traits because it was (and is) the only digitized collection of Puerto Rico butterflies. The Stuart J. Ramos Collection (SJRC; 3410 records; SJRC) was donated by the late S. J. Ramos to the McGuire Center for Lepidoptera and Biodiversity, Florida Museum of Natural History (MGCL) at the University of Florida, in Gainesville, Florida, USA. We used image analysis software to measure dorsal forewing length as the length between the forewing tip and the site of wing attachment on the thorax, and forewing width as the widest section of the forewing. We also measured hue, saturation, and brightness (HSB) on a 1 cm2 region of the wing near the thorax, avoiding major veins and visible scale damage. Furthermore, we focused on females because 1) their parental investment and, thus, energetic requirements are often greater than those of males, and 2) male coloration can be biased by sexual selection. Of the 62 species, only four (all Nymphalidae) had visually noticeable size differences, with females being larger than males. In a few instances in which images of females were not available, or the sex could not be determined from visual characteristics such as markings or size differences, we selected individuals randomly within species for trait measurements (Female:Unknown = 48:14). We aimed for a maximum of ten individuals per species when possible (median = 10; mean = 8.1 individuals per species).
Funding
Puerto Rico Science, Technology & Research Trust
National Science Foundation, Award: CAREER-2042453