Website visits can predict angler presence using machine learning

TL;DR

The study demonstrates that leveraging angler-generated data from online platforms, particularly lake-specific website visits, enables strong day-to-day prediction of boat presence at lakes in Ontario (about $78{-}82 ext{ %}$ accuracy depending on feature set), while predicting exact boat counts remains challenging, especially for unknown lakes ( $R^2$ ≈ 0.2–0.8 depending on lake familiarity). The most influential predictor across presence models is website visitation, with shoreline length and proximity to urban areas also contributing, whereas counts predictions rely more on spatial features. Integrating platform-derived signals with environmental and socio-ecological features yields marginal gains for presence but substantially improves boat-count predictions at known lakes; generalization to unknown lakes remains limited by data sparsity and temporal resolution. Overall, the work highlights the value and limitations of online angler data for informing fisheries management, suggesting that such signals can support near-real-time indicators of angler pressure and guide targeted management decisions, while acknowledging the need for finer temporal data and complementary ground-truth measures to extend applicability to new locations.

Abstract

Understanding and predicting recreational angler effort is important for sustainable fisheries management. However, conventional methods of measuring angler effort, such as surveys, can be costly and limited in both time and spatial extent. Models that predict angler effort based on environmental or economic factors typically rely on historical data, which often limits their spatial and temporal generalizability due to data scarcity. In this study, high-resolution data from an online fishing platform and easily accessible auxiliary data were tested to predict daily boat presence and aerial counts of boats at almost 200 lakes over five years in Ontario, Canada. Lake-information website visits alone enabled predicting daily angler boat presence with 78% accuracy. While incorporating additional environmental, socio-ecological, weather and angler-reported features into machine learning models did not remarkably improve prediction performance of boat presence, they were substantial for the prediction of boat counts. Models achieved an R2 of up to 0.77 at known lakes included in the model training, but they performed poorly for unknown lakes (R2 = 0.21). The results demonstrate the value of integrating data from online fishing platforms into predictive models and highlight the potential of machine learning models to enhance fisheries management.

Figures (7)

• Figure 1: The 187 lakes across Ontario considered for model training and testing. Black lines show borders of different management units and brown lines indicate roads.
• Figure S1: Frequency of fractions of absence of angling boats on observation days at the 187 lakes. At 45 lakes, there were no fishing boats detected over all observation days, and at 18 lakes, there were always boats present on the observation days.
• Figure S2: Start times of angling boat counts. All counts started between 9:00 and 16:00.
• Figure S3: Frequencies of number of flights (observation days) at the 187 lakes. 13 flights was the most likely number of flights for a lake.
• Figure S4: Frequencies of number of flights (observation days) at the lakes. Up to 41 lakes were observed on a specific day (three days). On most dates, nine lakes were observed (21 days).