Merging computational fluid dynamics and machine learning to reveal animal migration strategies

The calculated voyage: benchmarking optimal strategies and consumptions in the Japanese eel's spawning migration

Eels migrate along largely unknown routes to their spawning ground. By coupling Zermelo's navigation solution and data from the Japan Coastal Ocean Predictability Experiment 2 (JCOPE2M), we simulated a range of seasonal scenarios, swimming speeds, and swimming depths to predict paths that minimize migration duration and energy cost. Our simulations predict a trade-off between migration duration and energy cost. Given that eels do not refuel during their migration, our simulations suggest eels should travel at speeds of 0.4-0.6 body-length per second to retain enough energy reserves for reproduction. For real eels without full information of the ocean currents, they cannot optimize their migration in strong surface currents, thus when swimming at slow swimming speeds, they should swim at depths of 200 m or greater. Eels swimming near the surface are also influenced by seasonal factors, however, migrating at greater depths mitigates these effects. While greater depths present more favorable flow conditions, water temperature may become increasingly unfavorable, dropping near or below 5 °C. Our results serve as a benchmark, demonstrating the complex interplay between swimming speed, depth, seasonal factors, migration time, and energy consumption, to comprehend the migratory behaviors of Japanese eels and other migratory fish.

When to Go Against the Flow: Examining Patterns of Performance Over Multiday Migration Events in the Hawaiian Stream Fish, 'O'opu Nōpili (Sicyopterus stimpsoni)

Many animals migrate across regions of their geographic range as part of extended events, with groups of individuals proceeding through areas of travel on several successive days. Early migrating individuals may have an advantage over late migrating individuals by gaining early access to the resources at the eventual destination. For situations where early access to resources would provide an advantage, specific sets of locomotor traits might be found among individuals that are earlier migrators. We tested for associations between migration timing and traits related to escape responses, climbing, and morphology in the amphidromous Hawaiian stream goby, 'o'opu nōpili (Sicyopterus stimpsoni). In this species, juvenile fish migrate in pulses over several days immediately following flash floods. We collected daily measurements of escape responses and waterfall climbing from juvenile fish arriving at streams from the ocean. We found that escape performance showed mainly stochastic variation across migrating individuals tested on successive days. In contrast, some metrics of climbing performance decrease over successive pulses during a migration event. We also found more variation in body shape among fish from early pulses during migration events compared to later in pulses. These results could have implications for guiding conservation efforts, identifying critical time windows for protection as periods with the greatest likelihood of successful migrants.

Machine learned daily life history classification using low frequency tracking data and automated modelling pipelines: application to North American waterfowl

BACKGROUND

Identifying animal behaviors, life history states, and movement patterns is a prerequisite for many animal behavior analyses and effective management of wildlife and habitats. Most approaches classify short-term movement patterns with high frequency location or accelerometry data. However, patterns reflecting life history across longer time scales can have greater relevance to species biology or management needs, especially when available in near real-time. Given limitations in collecting and using such data to accurately classify complex behaviors in the long-term, we used hourly GPS data from 5 waterfowl species to produce daily activity classifications with machine-learned models using "automated modelling pipelines".

METHODS

Automated pipelines are computer-generated code that complete many tasks including feature engineering, multi-framework model development, training, validation, and hyperparameter tuning to produce daily classifications from eight activity patterns reflecting waterfowl life history or movement states. We developed several input features for modeling grouped into three broad categories, hereafter "feature sets": GPS locations, habitat information, and movement history. Each feature set used different data sources or data collected across different time intervals to develop the "features" (independent variables) used in models.

RESULTS

Automated modelling pipelines rapidly developed easily reproducible data preprocessing and analysis steps, identification and optimization of the best performing model and provided outputs for interpreting feature importance. Unequal expression of life history states caused unbalanced classes, so we evaluated feature set importance using a weighted F1-score to balance model recall and precision among individual classes. Although the best model using the least restrictive feature set (only 24 hourly relocations in a day) produced effective classifications (weighted F1 = 0.887), models using all feature sets performed substantially better (weighted F1 = 0.95), particularly for rarer but demographically more impactful life history states (i.e., nesting).

CONCLUSIONS

Automated pipelines generated models producing highly accurate classifications of complex daily activity patterns using relatively low frequency GPS and incorporating more classes than previous GPS studies. Near real-time classification is possible which is ideal for time-sensitive needs such as identifying reproduction. Including habitat and longer sequences of spatial information produced more accurate classifications but incurred slight delays in processing.