Data from: Microhabitat patchiness structures benthic biodiversity in the Western Antarctic Peninsula
Data files
Apr 08, 2026 version files 229.72 KB
-
README.md
2.05 KB
-
tango2_abundance_raw.csv
227.67 KB
Abstract
In the accelerating global biodiversity crisis, it is imperative to document biodiversity patterns along with their underlying drivers and the processes driving species distributions. Our objective was to study how benthic community composition varies across spatial scales and how fine-scale patchiness contributes to larger-scale biodiversity in the Western Antarctic Peninsula. Using underwater imagery, we quantified benthic community composition at three different scales using a nested sampling design across. We used multivariate analyses (CLUSTER, NMDS, ANOSIM, SIMPROF, and SIMPER analysis) to detect patterns in benthic community variability. NMDS showed that images were clustered according to their community composition into 31 significant groupings, or microhabitat types, revealing high variability. These microhabitats were not restricted by geographic location, substrate type, or macroalgal cover, as initially presumed. Instead, the differences in faunal assemblage seem to be influenced by the dominant macroalgae species and biogenic habitat complexity. We found that these microhabitats emerge as ecologically meaningful units that offer a practical scale at which to detect patterns and early ecological shifts. Changes in the distribution or frequency of these microhabitats may occur before shifts in species composition become detectable at site or station level, which is crucial for ecosystem monitoring.
Dataset DOI: 10.5061/dryad.stqjq2cj3
Description of the data and file structure
Since the identification level varied among taxa (species, genus, or family), we use the term "morphotaxa" to refer to the identified biota throughout this study. Given the distance, resolution of images, habitat type, and size and colour of some taxa, many taxa will not have been observable or fully accounted for. The example images (Table S2) demonstrate that sub-centimetre sized organism, those with camouflage or cryptic habitat choices, or any infauna are unlikely to be well represented, if at all. Whilst other collection methods were used during the second TANGO expedition, only organisms that could be confidently observed from overhead photography were analysed for this study.
For more common morphotaxa, we were able to confidently assign species, as some were hand-picked by SCUBA divers. However, when labels lacked the precision required for this study, we chose to omit the corresponding images. The resulting abundance dataset comprised 336 annotated photos which included 2410 individual annotations, 91 identified CATAMI categories (of which 13 were substrate features). Individual animals were kept as counts, but the abundance of macroalgae and colonial animals was converted into percentage (%) cover.
The data were compiled into a matrix for each image and the abundance or coverage of each morphotaxa: tango2_abundance_raw.csv. The matrix is structured with the rows representing 336 different images and the 81 columns representing the abundance or percentage (%) cover of the morphotaxa identified in those photographs. The first column gives the image number and subsequent columns give morphotaxa names in alphabetical order.
Access information
Other publicly accessible locations of the data:
The BlueROV was deployed from either the main vessel or a zodiac and flown in linear transects at a constant altitude (approximately 1 meter from the bottom, adjusted for visibility, using the ROV’s ping sonar altimeter) and a very slow pace (< 1 m/s) to maintain a high video quality. The total area of each analysed photograph was ~ 0.8 m2. Depending on the topography and deployment vessel, transects were either evenly spaced out in the sampling site, in a zig-zag pattern, or started from the same point and went in different directions, but never overlapping in both cases. Transect length varied from 25 to 110 meters.
For each video transect, still images were extracted using FFmpeg. And a set of evenly spaced-out images was selected for annotation from each transect, with a minimum of 30 images per site. The selection of the images was done automatically based on time-separation, but if an image was blurry, the closest non-blurry image was manually selected.
Image annotation was performed in BIIGLE. In each image, all visible animals, algae and substrate features were recorded using the CATAMI label tree with supplementary branches specific to Antarctic biota.