Method selection affects the estimates of residency and site fidelity in bottlenose dolphins: testing sensitivity and performance of different methods using mark-resight data.

Residency (R) and site fidelity (SF) are important parameters in population ecology, yet their quantification poses challenges in marine mammals. Based on a previous review, this study used simulated and empirical mark-resight data to assess the variations and performance of the most used R ( = 8) and SF ( = 11) indices in peer-reviewed literature under different scenarios. We applied the Jolly-Seber model to simulate thousands of bottlenose dolphin populations varying resighting () and survival () probabilities, and performed calibration, sensitivity, and validation analyses. Our results underscore the effects of and on individual categorization within the diverse simulated conditions, representing the often-overlooked heterogeneity in residency classification for populations. All SF indices showed similar and consistent performance (>0.70 Gower's distance) across the simulated scenarios, even when compared to field study data from wild dolphin populations ( Savannah, USA, and Alvarado, Mexico); thus, SF should be a critical parameter for interstudy comparisons. Conversely, R indices were remarkably different based on their definitions and classification criteria. The different thresholds among definitions largely biased the proportion of residents and transient individuals (or occasional visitors) even leading to counterintuitive outcomes. This emphasizes the importance of considering trade-offs in R index selection aligned with project goals, specific sampling efforts, and population dynamics. For instance, the simplified binomial categorization of R defined by Conway (2017) (https://digitalcommons.coastal.edu/etd/10/) easier to interpret but R indices incorporating temporal components ( monthly, seasonal, and annual) outperformed (>0.70 Gower's distance) other R indices lacking such criteria. This allowed for a more detailed representation of the temporal structure of the population, and higher consistency and accuracy while classifying individuals. Also, although the residency categories proposed by Möller, Allen & Harcourt (2002) (DOI 10.1071/AM02011) did not perform as well, these seemed to fit better when dealing with data gaps across spatial and temporal scales. Our results contribute to the ongoing discussion on methodological implications for the interpretation of ecological patterns, facilitating a nuanced understanding of population dynamics, aiding scientists, and conservation agencies in making informed decisions for bottlenose dolphin populations worldwide.