Temporal trends in global reptile species descriptions over three decades
Year: 2024
Type: article
Abstract: Despite the urgency imposed by the current biodiversity crisis, many species remain undiscovered or undescribed, facing extinction before their formal recognition by science. Accelerating species descriptions is thus imperative. However, descriptions should be robust and based on good taxonomic practice, which may enhance long-term nomenclatural stability that is crucial for scientific research and conservation planning. Yet, few studies have assessed the robustness of species description. Here, we evaluated temporal trends in species descriptions of nearly 4,000 squamate reptiles spanning more than three decades (from 1990 to 2023). We observed an average increase of about 115% in the number of lines of evidence used in descriptions, such as pholidosis, morphometrics, and genes sequenced for molecular analysis. Type-series size decreased by nearly half for lizards but remained roughly constant for snakes, while page length dedicated to descriptions showed an upward trajectory for both taxa, increasing 42.7% on average over time. Furthermore, our study highlights the positive correlation between robust species descriptions (i.e., measured by the number of lines of evidence provided, type-series size, and number of pages) and taxonomic reviews, as well as the impact of collaborative efforts, with more authors on descriptions associated with greater lines of evidence and pages. Overall, descriptions of reptiles have become more thorough and robust over recent decades. The proportion of species descriptions that included molecular analysis has grown over time, with more than half of all species described annually in recent years having molecular data. Despite hundreds of new reptile species being described annually, the potential invalidation of many species in the future may jeopardise current conservation efforts. Therefore, thorough, accurate descriptions are important in addressing the biodiversity crisis and providing reliable data for biodiversity analyses and conservation planning. Keywords: Taxonomy, Linnaean shortfall, lizards, snakes, species descriptions, Squamata, taxonomic practice. README There are 2 files associated with this paper: (1) the raw data (Dataset.csv); and (2) the R code (Guedes_etal_reptile_description_trends.R). Description of the data and file structure (1) Dataset.csv: contains 24 columns, each one explained below:– Species: Binomial name.– Genus: Taxonomic genus to which a species belongs.– Family: Txonomic family to which a species belongs.– Order: Taxonomic order to which a species belong ("Crocodylia", "Sauria", "Serpentes", "Testudines").– Author: Author(s) involved in the species description.– Year_of_description: Year in which the species was formally described.– TaxonomicReview: Was the paper of description based on a Taxonomic review or not? (Yes or No answer).– Morphometrics: Binary variable informing whether morphometric data as available for the described species. – Pholidoses: Binary variable informing whether pholidoses was provided in the description.– Coloration: Binary variable informing whether colour data was provided in the description.– Hemipenis: Binary variable informing whether the hemipenis was described in the description.– Osteology: Binary variable informing whether osteology data was provided in the description.– Karyotype: Binary variable informing whether karyotype data was provided in the description.– InternalAnatomy: Binary variable informing whether data on internal anatomy was provided in the description.– Molecular: Binary variable informing whether the authors use molecular data as evidence for the species description.– MolMethod: Informs the molecular method used in the description (i.e., mtDNA, nucDNA, multiLoci, SNPs)– N.Genes: Informs the total number of genes sequenced when the description used molecular analysis.– N_authors: The number of authors per description.– SppRichByGenus: The per-genus species until the year of each species description.– Log10BodyMass_g: Maximum body mass per species (log10-transformed).– N_pages: The number of pages (METHODS and RESULTS sections only) of the article divided by the number of described species. The page is divided into 4 quadrants, meaning 1 page is composed of 4x0.25 parts.– N.TypeSeries: The number of type-specimens used in the description.– N_evidencesI: The number of evidences (e.g., pholidosis, number of morphometric measures, and number of genes) used in descriptions. The number of morphometric measures and the number of genes sequenced are treated as continuous characters, while the others are binary. Therefore, this represent an unequal-weight metric.– N_evidencesII: The number of evidences used. Morphology and number of genes as continuous variables, whereas the others as binary (see main text). All evidence types are treated as binary (0 or 1), thus having equal weight. (2) Guedes_etal_reptile_description_trends.R: this file contains the R code used to perform all analyses of our work as well as for creating the images and tables of the main text and supplementary material. The script is commented by the authors and intuitively structured, including needed packages and their respective versions. Code/Software All analyses were performed using the software R version 4.3.1. The R-code has 4 steps: 1. Load and understand the dataset. 2. Temporal trends in robustness of publications - through annual means. 3. Temporal trends in robustness of publications - through generalised linear models. 4. Check phylogenetic correlation in model residuals. Usage notes Software R is required to open the file Guedes_etal_reptile_description_trends.R.Microsoft Excel can be used to open the file Dataset.csv. Materials and methods Data collection We examined the original descriptions of 3,905 reptiles published from 1990 to 2023, spanning 34 years of reptile discovery. Taxonomic nomenclature follows the July 2023 version of the Reptile Database (Uetz et al., 2024), and we did not analyse redescriptions, revalidations, or other taxonomic works. Our dataset included 2,860 lizards and amphisbaenians, 976 snakes, 68 chelonians and one crocodile. We kept only squamate reptiles in our final dataset due to their higher sample size, totalling 3,836 species. For each species, we compiled data on taxonomic ranks, including: (1) binomial name, (2) genus, (3) family, (4) order, as well as (5) authority information. We used eight binary variables to inform if the species description included (Yes = 1) or not (No = 0) data related to (5) pholidosis, (6) coloration, (7) hemipenis, (8) osteology (data on dentition and vertebra count was also considered), (9), karyotype, (10) internal anatomy, (11) molecular analysis (e.g., mtDNA, nucDNA, multiLoci), and if the discovery was based on (12) taxonomic revision. We considered taxonomic revision as any work that explicitly mentioned the term “revision” or similar in its title, abstract, keywords, or main text. Other eight continuous variables registered the (13) number of morphometric measurements (e.g., snout-vent length, head length, head width), excluding proportions and mensuration of scales, bones, and other structures already accounted by their own binary categories, (14) number of sequenced genes, (15) year of publication, (16) number of authors; (17) maximum body mass (obtained from Meiri, 2024; Moura et al., 2024), (18) per-genus species richness based on the year of species descriptions (i.e., for each species, we counted the total number of species described in the same genus until their discovery year), (19) number of type-specimens (i.e., holotype plus all other types) used to obtain diagnostic traits, (20) number of pages dedicated to each description, with each page measured in units of 0.25 page and including only the methods and results sections. If the description involved the discovery of more than one species, then we divided the number of pages by the number of described species. An additional variable informed the (21) number of evidence types used in each species description. This variable was computed in two ways: (21.I) the first version sums the binary variables numbered above from 5 to 11, plus morphometric measurements treated as binary. In this approach, each line of evidence has the same weight in the compound metric. The second version (21.II) represents the sum of the binary variables from 5 to 10, plus morphometric measurements and the number of sequenced genes treated as continuous variables, thus, an unequal-weight metric. The variables from 19 to 21 were used as proxies for the robustness of species description. Statistical analyses Temporal trends in the proxies of the robustness of taxonomic descriptions (number of evidence, number of type-specimens, and number of pages) were initially investigated using Spearman’s (rs) correlation between the year of publication and annual means for the three metrics analysed across species described in each year. We also computed pairwise correlations among quality metrics to check how they covaried in relation to each other. We applied generalised linear models (GLMs) with a negative binomial error distribution to analyse the response variables: number of evidence and number of type-specimens, both of which are count data. Since the number of pages was calculated as a continuous variable (including non-integer values), we log10-transformed it to improve normality and modelled it using a GLM with a Gaussian error distribution. We accessed whether our response variables were affected by the year of species description, number of authors (log10-transformed), body mass (log10-transformed), per-genus species richness, taxonomic revisions, and taxonomic group. Continuous predictors were z-transformed to allow direct comparisons of their effect sizes. Due to varying amounts of missing data among response and predictor variables, GLMs were perform
Authors: Guedes, Jhonny J. M., Valentim, Hidalgo, Mendonça, Lucas R., Chen-Zhao, Rosana, Diniz-Filho, José Alexandre F., Moura, Mario R.
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W6893059798