Search for: All records

Body condition is a crucial and indicative measure of an animal’s fitness, reflecting overall foraging success, habitat quality, and balance between energy intake and energetic investment toward growth, maintenance, and reproduction. Recently, drone-based photogrammetry has provided new opportunities to obtain body condition estimates of baleen whales in one, two or three dimensions (1D, 2D, and 3D, respectively) – a single width, a projected dorsal surface area, or a body volume measure, respectively. However, no study to date has yet compared variation among these methods and described how measurement uncertainty scales across these dimensions. This associated uncertainty may affect inference derived from these measurements, which can lead to misinterpretation of data, and lack of comparison across body condition measurements restricts comparison of results between studies. Here we develop a Bayesian statistical model using known-sized calibration objects to predict the length and width measurements of unknown-sized objects (e.g., a whale). We use the fitted model to predict and compare uncertainty associated with 1D, 2D, and 3D photogrammetry-based body condition measurements of blue, humpback, and Antarctic minke whales – three species of baleen whales with a range of body sizes. The model outputs a posterior predictive distribution of body condition measurements and allows for the construction of highest posterior density intervals to define measurement uncertainty. We find that uncertainty does not scale linearly across multi-dimensional measurements, with 2D and 3D uncertainty increasing by a factor of 1.45 and 1.76 compared to 1D, respectively. Each standardized body condition measurement is highly correlated with one another, yet 2D body area index (BAI) accounts for potential variation along the body for each species and was the most precise body condition metric. We hope this study will serve as a guide to help researchers select the most appropriate body condition measurement for their purposes and allow them to incorporate photogrammetric uncertainty associated with these measurements which, in turn, will facilitate comparison of results across studies. 

The considerable power needed for large whales to leap out of the water may represent the single most expensive burst maneuver found in nature. However, the mechanics and energetic costs associated with the breaching behaviors of large whales remain poorly understood. In this study we deployed whale-borne tags to measure the kinematics of breaching to test the hypothesis that these spectacular aerial displays are metabolically expensive. We found that breaching whales use variable underwater trajectories, and that high-emergence breaches are faster and require more energy than predatory lunges. The most expensive breaches approach the upper limits of vertebrate muscle performance, and the energetic cost of breaching is high enough that repeated breaching events may serve as honest signaling of body condition. Furthermore, the confluence of muscle contractile properties, hydrodynamics, and the high speeds required likely impose an upper limit to the body size and effectiveness of breaching whales. 

Researchers can investigate many aspects of animal ecology through noninvasive photo–identification. Photo–identification is becoming more efficient as matching individuals between photos is increasingly automated. However, the convolutional neural network models that have facilitated this change need many training images to generalize well. As a result, they have often been developed for individual species that meet this threshold. These single‐species methods might underperform, as they ignore potential similarities in identifying characteristics and the photo–identification process among species.

In this paper, we introduce a multi‐species photo–identification model based on a state‐of‐the‐art method in human facial recognition, the ArcFace classification head. Our model uses two such heads to jointly classify species and identities, allowing species to share information and parameters within the network. As a demonstration, we trained this model with 50,796 images from 39 catalogues of 24 cetacean species, evaluating its predictive performance on 21,192 test images from the same catalogues. We further evaluated its predictive performance with two external catalogues entirely composed of identities that the model did not see during training.

The model achieved a mean average precision (MAP) of 0.869 on the test set. Of these, 10 catalogues representing seven species achieved a MAP score over 0.95. For some species, there was notable variation in performance among catalogues, largely explained by variation in photo quality. Finally, the model appeared to generalize well, with the two external catalogues scoring similarly to their species' counterparts in the larger test set.

From our cetacean application, we provide a list of recommendations for potential users of this model, focusing on those with cetacean photo–identification catalogues. For example, users with high quality images of animals identified by dorsal nicks and notches should expect near optimal performance. Users can expect decreasing performance for catalogues with higher proportions of indistinct individuals or poor quality photos. Finally, we note that this model is currently freely available as code in a GitHub repository and as a graphical user interface, with additional functionality for collaborative data management, via Happywhale.com.