Computational protocol: High-resolution GPS tracking reveals sex differences in migratory behaviour and stopover habitat use in the Lesser Black-backed Gull Larus fuscus

Similar protocols

Protocol publication

[…] We limited our analysis to complete migration routes (i.e. successful arrival at the breeding or wintering area) for which we had at least 1 GPS location per hour. Using this criterion, a total of 150 migration routes of 48 individuals (22 males, 26 females) were retained between autumn 2013 and spring 2017: i.e. 31 autumn migration routes of 19 males, 39 autumn migration routes of 21 females, 35 spring migration routes of 22 males and 45 spring migration routes of 24 females.Linear mixed effects models were fitted to the data to test for differences in migration distance and timing of migration (i.e. start and end date) in relation to sex and body size. The day at which spring and autumn migration started or ended was expressed as a numeric value, the first of January being day 1. Migration distances were calculated as the Haversine distance between breeding and wintering area. Body size was expressed as a body size index, which explained 80% of the variation in body size in our study population (Fig. S):10.86×tarsuslength+0.95×headlength+0.90×billlength+0.88×gonyswidthAll measurements are expressed in mm. All models included body size and sex and the two-way interactions as fixed effects. Because migration distance may be confounded with timing and/or duration of migration, migration distance and the two-way interactions with body size and sex were added as additional fixed effects when testing for differences in timing of migration. Individual and tracking year were added as random effects to all full models. The level of significance of both random effects was assessed using likelihood-ratio tests, and only tracking year was retained in the final model.Similarity in habitat use between pairs of individuals was calculated based on the Bray-Curtis similarity index:2Similaritym,1−2=1−∑i|pm,i,individual1−pm,i,individual2|2Pm,i represents the fraction of GPS observations of an individual assigned to habitat type i (agricultural, urban, freshwater or marine) during a given month (m). A value of 1 thus corresponds to identical habitat use, whereas a value of zero corresponds to complete segregation. To test if intra- and inter-sex variation in habitat use changed throughout the year, a linear mixed effects model was fitted including month and type of comparison (i.e. male-female, male-male or female-female) and the two-way interaction as fixed effects. To account for statistical dependence, the combination of individuals being compared was included as a random effect and a continuous first order autocorrelation structure was fitted. Similarities in habitat use were pooled per month across years. However, to fit a continuous autocorrelation structure, each month between June 2013 and May 2017 was assigned a unique, consecutive numeric value (June 2013 being month 1). The level of significance of random effects and autocorrelation structures was assessed using likelihood-ratio tests. The fraction of time spent in agricultural, urban, fresh water and marine habitats was compared between males and females by fitting a linear mixed effects model that included month and sex as fixed effects, individual as a random effect and a first order continuous autocorrelation structure as described above.Only significant two-way interactions were retained in models, whereas main effects were always retained to avoid parameter estimation bias. Finally, normality, independence and homoscedasticity of model residuals was inspected (Figs S4–S12). All analyses were carried out in R using the nlme package,. Great circle distances were calculated using the distHaversine function in the geosphere package. Maps were created using the packages maps, mapData and OpenStreetMap–. […]

Pipeline specifications

Software tools lme4, nlme
Application Mathematical modeling
Organisms Homo sapiens, Gallus gallus