Fitness, behavioral, and energetic trade‐offs of different migratory strategies in a partially migratory species

Abstract Alternative migratory strategies can coexist within animal populations and species. Anthropogenic impacts can shift the fitness balance between these strategies leading to changes in migratory behaviors. Yet some of the mechanisms that drive such changes remain poorly understood. Here we investigate the phenotypic differences, and the energetic, behavioral, and fitness trade‐offs associated with four different movement strategies (long‐distance and short‐distance migration, and regional and local residency) in a population of white storks (Ciconia ciconia) that has shifted its migratory behavior over the last decades, from fully long‐distance migration toward year‐round residency. To do this, we tracked 75 adult storks fitted with GPS/GSM loggers with tri‐axial acceleration sensors over 5 years, and estimated individual displacement, behavior, and overall dynamic body acceleration, a proxy for activity‐related energy expenditure. Additionally, we monitored nesting colonies to assess individual survival and breeding success. We found that long‐distance migrants traveled thousands of kilometers more throughout the year, spent more energy, and >10% less time resting compared with short‐distance migrants and residents. Long‐distance migrants also spent on average more energy per unit of time while foraging, and less energy per unit of time while soaring. Migratory individuals also occupied their nests later than resident ones, later occupation led to later laying dates and a lower number of fledglings. However, we did not find significant differences in survival probability. Finally, we found phenotypic differences in the migratory probability, as smaller sized individuals were more likely to migrate, and they might be incurring higher energetic and fitness costs than larger ones. Our results shed light on the shifting migratory strategies in a partially migratory population and highlight the nuances of anthropogenic impacts on species behavior, fitness, and evolutionary dynamics.


Section S3: Multievent Model Design
Matrix representations with departure states in rows and arrival states in columns are commonly used in multievent models (See for instance Sanz-Aguilar et al., 2012).The initial state probabilities corresponded to the probability that a newly marked individual was an adult with an active GPS logger (Aa), an adult that has lost the GPS logger or its signal (Ai), a recently dead individual with an active GPS logger (Ra) and a long dead individual (LD).Here initial state probability () was certainly known for every individual, as all individuals started as alive with an active GPS device deployed (ℎ,  = 1).     =  0 0 0 Matrix 1 We decomposed the transition between the state's probabilities into two steps: the first step corresponded to the probability of losing the GPS device or its signal (ψ, matrix 2) and the probability of survival (ϕ, matrix 3).The event probabilities (matrix 4) corresponded to the resighting probabilities (p) and the recovery probability (r).Since all recoveries were from individuals with active GPS loggers, we fixed recovery probability to one for all models (r=1).
Fig S1: Relationship between the migratory strategy and the seasonal displacement, during the (a) autumn transition, (b) the wintering period, (c) the spring transition, and (d) the breeding period.Black dots are predicted estimates from the LMM, vertical lines are the confidence intervals based on fixed-effect uncertainty, and grey dots are raw data.

Fig S2 :
Fig S2: Relationship between the migratory strategy and the seasonal mean ODBA, during the (a) autumn transition, (b) the wintering period, (c) the spring transition, and (d) the breeding period.Black dots are predicted estimates from the LMM, vertical lines are the confidence intervals based on fixed-effect uncertainty, and grey dots are raw data.

Fig S3 :
Fig S3: Relationship between the migratory strategy and the seasonal mean foraging ODBA, during the (a) autumn transition, (b) the wintering period, (c) the spring transition, and (d) the breeding period.Black dots are predicted estimates from the LMM, vertical lines are the confidence intervals based on fixed-effect uncertainty, and grey dots are raw data.

Fig S4 :
Fig S4: Relationship between the migratory strategy and the seasonal mean soaring ODBA, during the (a) autumn transition, (b) the wintering period, (c) the spring transition, and (d) the breeding period.Black dots are predicted estimates from the LMM, vertical lines are the confidence intervals based on fixed-effect uncertainty, and grey dots are raw data.

Fig S5 :
Fig S5: Relationship between the migratory strategy and foraging time, during the (a) autumn transition, (b) the wintering period, (c) the spring transition, and (d) the breeding period.Black dots are predicted estimates from the GLMM, vertical lines are the confidence intervals based on fixed-effect uncertainty, and grey dots are raw data.

Fig S6 :
Fig S6: Relationship between the migratory strategy and resting time, during the (a) autumn transition, (b) the wintering period, (c) the spring transition, and (d) the breeding period.Black dots are predicted estimates from the GLMM, vertical lines are the confidence intervals based on fixed-effect uncertainty, and grey dots are raw data.

Table S1 :
Tukey's contrasts multiple comparisons with adjusted p-values for the annual displacement per migratory strategy.

Table S2 :
Tukey's contrasts multiple comparisons with adjusted p-values for the annual mean ODBA per migratory strategy.

Table S3 :
Tukey's contrasts multiple comparisons with adjusted p-values for the annual mean foraging ODBA per migratory strategy.

Table S4 :
Tukey's contrasts multiple comparisons with adjusted p-values for the annual mean soaring ODBA per migratory strategy.

Table S5 :
Tukey's contrasts multiple comparisons with adjusted p-values for the proportion of resting time per migratory strategy.

Table S6 :
Tukey's contrasts multiple comparisons with adjusted p-values for the proportion of soaring time per migratory strategy.

Table S7 :
Tukey's contrasts multiple comparisons with adjusted p-values for the seasonal displacement per migratory strategy and season.

Table S9 :
Tukey's contrasts multiple comparisons with adjusted p-values for the seasonal mean foraging ODBA per migratory strategy and season.

Table S10 :
Tukey's contrasts multiple comparisons with adjusted p-values for the seasonal mean soaring ODBA per migratory strategy and season.

Table S11 :
Effects the GLMM for the foraging time per migratory strategy and season.

Table S12 :
Effects for the GLMM for the resting time per migratory strategy and season.

Table S13 :
Effects for the GLMM for the soaring time per migratory strategy and season.

Table S14 :
Effects for the GLMM for the flapping time per migratory strategy and season.

Table S14 :
Effects for the GLMM for the flapping time per migratory strategy and season.(continued)

Table S16 :
Tukey's contrasts multiple comparisons with adjusted p-values for the resting time per migratory strategy and season.