Does malaria infection increase the risk of predation-related mortality during bird migration?

We assessed whether birds infected with avian malaria parasites are more likely to be blown off course or fall prey to predators along their migration route. We sampled 357 passerines from 11 species during their autumn migration to wintering grounds in two different locations: (i) at a stopover in southern Spain, and (ii) in the Canary Islands, where they were at risk of being preyed upon by Eleonora's falcons while en route to the southern Sahara. The sampling took place from August 18th to October 9th, 2019, simultaneously at a stopover site just before crossing to Africa (Palomares del Río, Seville, southwestern Spain) and at an accidental arrival site in the middle of the flyway (Alegranza islet, Canary Islands; 1050 ha, 289 m above sea level). Molecular detection of infections by haemosporidian parasites of the genera Plasmodium, Haemoproteus, and Leucocytozoon was performed on the collected bird samples.

The data used for the statistical analyses and a figure are provided in three different .txt files:

1) The file dataset_prevalence_cbind was used to test for differences in the prevalence of each parasite genus in birds between the Iberian Peninsula and the Canary Islands using Generalized Linear Mixed Models GLMMs (cbind approach) that utilized a two-column matrix to represent the number of infected birds of each species in each site in relation to the number of uninfected birds. Plasmodium, Haemoproteus, and Leucocytozoon prevalence was individually modeled with binomial error and a logit link function, inherently controlling for sample size. Firstly, we fitted a GLMM for each parasite genus including all birds, where the bird species was included as a random effect and the site as the only fixed factor. Secondly, to reach better statistical fit and reliable estimates of the prevalence of infection, we fitted similar GLMMs where only those bird species with at least 25 individuals sampled at each locality were included. This is a .txt file with 13 columns:

• sp: Latin name of the 11 bird species included in the study
• loc: sampling locality. Factor with 2 levels: Peninsula Iberica (IB) and the Canary Islands (CI)
• tested_plam: number of individuals tested for Plasmodium infection
• infected_plasm: number of individuals infected with Plasmodium
• non_infected_plasm: number of individuals tested negative for Plasmodium
• tested_haem: number of individuals tested for Haemoproteus infection
• infected_haem: number of individuals infected with Haemoproteus
• non_infected_haem: number of individuals tested negative for Haemoproteus
• tested_leuco: number of individuals tested for Leucocytozoon infection
• infected_leuco: number of individuals infected with Leucocytozoon
• non_infected_leuco: number of individuals tested negative for Leucocytozoon
• tested: total number of individuals tested for each bird species
• infected: total number of individuals infected by any parasite for each bird species

2) The file GLMM_ind was used to assess the effect of birds’ sex, age, and date of sampling on the probability of infection. we fitted GLMMs for each parasite genus with the binomial response (1/0) representing individual infection data. In these models, the locality was also included as a fixed factor and bird species as a random term. This a .txt file with 11 columns:

• ID: sample unique identifier
• loc: sampling locality. Factor with 2 levels: Peninsula Iberica (IB) and the Canary Islands (CI)
• ring: ring number (only for birds sampled in IB)
• date: date of sampling
• julian: date of sampling in julian format
• sp: Latin name of the 11 bird species included in the study
• age: estimated age of the birds sampled according to EURING. Factor with 2 levles: EURING codes 3 and 4
• sex: factor with 2 levels: male (1) and female (2)
• plasm: factor with 2 levels: whether the individual was infected by Plasmodium (1) or uninfected (0)
• haem: factor with 2 levels: whether the individual was infected by Haemoproteus (1) or uninfected (0)
• leuco:  factor with 2 levels: whether the individual was infected by Leucocytozoon (1) or uninfected (0)

3) The file fid_syl_plot was used to create a bar plot showing the prevalence data + SE of the two bird species with the largest sample size (Ficedula hypoleuca and Sylvia communis) sampled at the two localities. This is a .txt file with 6 columns:

• species: Ficedula hypoleuca and Sylvia communis
• locality: sampling locality. Factor with 2 levels: Peninsula Iberica (IB) and the Canary Islands (CI)
• parasite: Plasmodium , Haemoproteus, and Leucocytozoon
• tested: number of individuals tested for each parasite genus
• infected: number of individuals infected by each parasite genus
• non-infected: number of individuals uninfected