Species and sex specific chemical composition in a concealed gland from an internal gland-like tissue of an African frog family

# Species and sex specific chemical composition in a concealed gland in an internal tissue of an African frog family - Electronic Supplement Contains 6 files: includingone csv file with original GCMS data (1), one csv file with geospatial data (2), one excel file with additional peak information for the GCMS data file (3), and two high-resolution histological crosssection scans of the mandibular strand in a male and female *Odontobatrachus arndti (4), as well as the annotated R Code. Scans and Code are stored in Zenodo as Supplementary and Software respectively. Description of the data and file structure 1. GC_data.csv (the original GCMS data): depicts metadata and GC count data for the 161 strand samples (ZFG_XXX) from 1178 peaks and contains the following relevant columns: "ID"= sample identification, "species" = species identification "date"= day of collection, "season"= corresponding wet or dry season, "river"=name of river where sample was collected, "size"= bodysize (SVL) of speciemen in mm, "weight"= weight of speciemen in g, "sex" = sex of the collected frog specimen, "Detail" = either the microhabitat (slope, cascade, plain; see manuscript for details) or apparent developmental status of the gland (good, bad, and moderate) of collected specimen. "Peaks.1-1178" = relative abundance (area under the curve of respective peaks, normalized over the total ioncounts of entire sample) for the 161 individual samples (ZFG_XXX). 2. geodistance.csv (Geospatial data): depicts GPS coordinates for each of the 161 strand samples (ZFG_XXX) and contains the following columns: "ID"= sample identification, "latitude" = easting, "longitude" = northing 3. GC_RT_alligned.xlsx: depicts the retention times of the 1178 peaks and contains the following columns: "mean rt" = mean retention time for peaks 1-1178 derived from, "sd" = the respective standard deviation for the before-mentioned values, "min rt" = minimal retention time found for the respective for peaks, "max rt" = maximal retention time found for the respective for peaks. "samples with peaks"= the number of samples that contained the respective peak. File also constitutes a presence absence matrix for each peak (1-1178) in each of the 161 samples (ZFG_XXX). In this matrix 1 refers to presence (irrespective of the intensity of the signal) empty cells refer to the absence. 4. High resolution histological sections are based on individual scans at 20 fold magnification on a Zeiss Axioscop stereomicrope, merged manually using Microsoft Photoshop and Powerpoint. ## Sharing/Access information Original strand extracts and unprocessed GCMS data stored at University of Würzburg, Department of Animal Ecology and Tropical Ecology. Histological material based on museum specimen from the collection of the Museum für Naturkunde Berlin (ZMB): Voucher numers ZMB 78360 and ZMB 78361. For further details on geolocation of specimen as well as used museum specimen, see manuscript. ## Code/Software Annotaded R- Code for statistical and graphical analysis: GCMS data were ordinated by an NMDS using Bray-Curtis dissimilarity indices, respective group homogeneity was tested with permutational analyses of multivariate dispersions (PERMDISP) and permutational multivariate analysis of variance (PERMANOVA) were computed for differences in group means. For spatial analysis, we additionally performed Mantel tests to control for correlation between the dissimilarity matrix and a distance matrix computed from the localities of the respective samples. To identify important decisive peaks in the chemical profiles, we trained random forests on a labelled dataset and used these models (one for sex and one for species) to predict the assignment of sex and species of an unlabelled dataset. All data handling and analysis was computed with R software 4.2.1, using the following packages: caret, dplyr, geodist, gplot2, randomForest, recipes, vegan