Morphology of nares associated with stereo-olfaction in baleen whales
Cite this dataset
Ryan, Conor et al. (2024). Morphology of nares associated with stereo-olfaction in baleen whales [Dataset]. Dryad. https://doi.org/10.5061/dryad.rn8pk0pjb
Abstract
The sensory mechanisms used by baleen whales (Mysticeti) for locating ephemeral, dense prey patches in vast marine habitats are poorly understood. Baleen whales have a functional olfactory system with paired rather than single blowholes (nares), potentially enabling stereo-olfaction. Dimethyl sulfide (DMS) is an odorous gas emitted by phytoplankton in response to grazing by zooplankton. Some seabirds use DMS to locate prey, but this ability has not been demonstrated in whales. For all 15 extant species of baleen whale, nares morphometrics (imagery from unoccupied aerial systems, UAS) was related to published trophic level indices using Bayesian phylogenetic mixed modelling. A significant negative relationship was found between nares-width and whale trophic level (β = -0.07, Lower 95% CI = -0.12, Upper 95% CI = -0.02), corresponding with a 36% increase in nares-width from highest to lowest trophic level. Thus, species with nasal morphology best suited to stereo-olfaction are more zooplanktivorous. These findings provide evidence that some baleen whale species may be able to localise odorants e.g., DMS. Our results helps direct future behavioural trials of olfaction in baleen whales, by highlighting the most appropriate species to study. This is a research priority, given the potential for DMS-mediated plastic ingestion by whales.
README: Morphology of nares associated with stereo-olfaction in baleen whales
https://doi.org/10.5061/dryad.rn8pk0pjb
The dataset contains one .csv file and the necessary code to reproduce our results using the data therein. The .csv file contains nares measurements, made using UAV (drone) imagery with some additional underwater, ship-bourne and airplane-bourne images for rarely observed whale species. This .csv data file also contains trophic index information derived from the literature and the citations to these sources are listed therein. Both .csv and .tre files described below are required to produce the Bayesian phylogenetic mixed model presented in the paper. We controlled for the expected similarity in trait values due to common descent among species using a distance matrix of the phylogenetic distance between these species. The code for this phylogenetic tree is provided in the .tre file, as described in the methods section of the paper. R (R Core Team, 2022) code to run the analysis is included as a .pdf file which also includes the sensitivity analysis, whereby nares measurements made from non-UAV platforms listed above are considered, in order to rule out the possibility that they are overly influential on the final model fit.
Description of the data and file structure
whale_blowhole_measures_1_12_2023.csv
The data comprise the following rows with the following defined column headers, i.e. row 1 in order of occurence:
- species: Categorical variable. Binomial species name.
- sp_common: Categorical variable. Common species name.
- contributor: Categorical variable. Initials of the UAV data contributor. Full names can be found in the author list or acknowledgements section of the paper.
- samplingfolder: Categorical variable. For filing purposes, used by the authors to track original file location.
- individual: UID, unique identifier. Assigned to each independent sample, i.e. each whale.
- bl: Continuous variable. Body length (m), from previous published studies (See methods for details).
- state: Categorical variable. Can be considered a binary variable which indicates whether the nares were open or closed when the corresponding measurements were made (See methods for definition of each state).
- bottom_spacing: Continuous variable. Distance (pixels) between a pair of nares at the caudal end of the nares.
- tl_fin: Continuous variable. Trophic index assigned to each species (See methods for details).
- tl_ref: Categorical variable. ISBN numbers, DOI numbers or weblinks to the source of tl_fin. Note that parent source, indicated as SeaLifeBase is found at https://www.sealifebase.ca/
- width_at_nares: Continuous variable. Width (pixels) of the whale's head in line with the caudal-end of the nares.
- bottom_width_ratio: Continuous variable. Ratio of bottom_spacing to width_at_nares.
MamPhy_BDvr_Completed_5911sp_topoCons_NDexp_v2_tree0000_(1).tre
Phylogentic tree file derived using results by Upham et al. (2019) in a text-based format. See methods section in the paper for how this has been truncated. The accompanying R code is used to read-in the phylogeny data using the read.tree command.
Code/Software
Ryan_et_al_code_fig_redacted.pdf
R code was written and annotated by Kevin Healy using R version 4.2.2 (R Core Team, 2022). It uses the accompanying .csv and .tre files described above to reproduce the Bayesian phylogenetic mixed models (BPMM) presented in the paper, using the MCMCglmm package (Hadfield, 2010). Resulting figures have been redacted from the pdf for copyright reasons, but they will be produced when the code is run by users in R.
References
Hadfield JD. 2010 MCMC Methods for Multi-Response Generalized Linear Mixed Models: The MCMCglmm R Package. J Stat Softw 33. (doi:10.18637/jss.v033.i02)
R Core Team 2022 R: A language and environment for statistical computing. Version 4.2.2. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Upham NS, Esselstyn JA, Jetz W. 2019 Inferring the mammal tree: Species-level sets of phylogenies for questions in ecology, evolution, and conservation. PLoS Biol 17, e3000494. (doi:10.1371/journal.pbio.3000494)
Methods
Morphometric measurements were made for free-ranging whales mostly using UAS or aircraft-based imagery, including stills from video footage. The selection criteria for images were that they were vertical, in focus and clearly showed the nares. Altitude was typically between 10 – 60 m, however accurate altimetry data were not available for all flights and therefore relative measurements (i.e., pixel counts) were used. Because images were compiled from a range of different studies, the UAS, camera, camera sensor, lens focal length, altitude and thus pixel resolution will have varied among sampling events . Similarly, lens distortion will vary in its effect on whale measurements due to differences among UAVs. To mitigate for possible effects from image distortion, only images where the subject of interest (head) was in the center of the frame (≥ 30% from all frame edges) were used .
Absolute measurements were not generated in this study, rather those previously published (total body length) estimates were used (tables S1 & S2). To compile a dataset of all 15 extant species of baleen whales, additional images for those rarely documented species were sourced from publications and reports . Two of these photographs were taken from a ship or underwater, but all clearly showed the nares. For each individual whale, the clearest available overhead image (i.e., photograph or video still) was used to measure the maximum width (pixels) of the open or closed nares and the maximum width of the body (pixels) at the nares using the straight-line tool in ImageJ Analyze and Measure (https://imagej.nih.gov/ij/index.html).
The maximum-nares-width measurements could vary depending on what point during the respiration cycle that the measurement was made, from fully closed through to fully open. We considered this by assigning two simplified nares states which were assigned to each set of measurements: open (ranging from fully to partially open: bubbles, vapour or respiratory tract visible) and closed (no bubbles, vapour or respiratory tract visible). The aim was to obtain a set of measurements for both states for each whale, which was not always possible.
To assign a trophic level to each whale species, we used available estimates based on diet from the SeaLifeBase database. There were no available estimates in the database for pygmy right (Caperea marginata), Rice’s (B. ricei) and Omura’s (B. omurai) whales. Instead, we determined the preferred prey for these whale species from published sources and applied trophic level estimates of the preferred prey plus 1. For phylogeny, we used the dated mammal phylogeny from Upham et al. which was truncated to include only those species present in our dataset. Because Rice's is missing from Upham et al., we added it as a sister species, Eden’s whale (B. edeni), based on its current phylogenetic positioning.