Name: Santos D

Biodeserts supervisors: Brito JC

Co-supervisor:  Gonçalves D

Title: Patterns of habitat use by an insular endemic bird, the Azores woodpigeon (Columba palumbus azorica)

Institution: University of Porto

Status: Completed




About 17% of the world’s bird species are present on islands, and many are local endemics. The Azores archipelago is home to an endemic subspecies of woodpigeon (Columba palumbus), the Azores woodpigeon (C. p. azorica). The Azores woodpigeon breeds on all the islands of the archipelago, but is more common on Pico, Graciosa and Terceira islands. Despite being relatively common in the archipelago, little is known about the biology of this endemic bird.
The objectives of this work were to investigate patterns in the distribution of abundance and habitat use of the Azores woodpigeon on Terceira island. In particular, it was aimed to: i) investigate seasonal and daily variations in habitat use to understand the movement dynamics of the species; ii) map the distribution of abundance and identify environmental factors that may be related with observed distribution of the Azores woodpigeon abundance; iii) model the potential distribution of the Azores woodpigeon abundance and compare the results of three different measures of abundance.
Abundance data of the Azores woodpigeon (number of pigeons / 10 minutes of observation) was collected on Terceira island between November 2012 to August 2014, in 100 sampling points, visited fortnightly during mornings and afternoons. Three abundance datasets were generated: All data, Truncated data (excluding zero observations) and Maximum data (maximum recorded abundance). Differences in the distribution of abundance of the Azores woodpigeon were investigated according to daily variations (Morning vs. Afternoon), seasonal variation (Breeding vs. Non-breeding) and yearly variations (Year 1 vs. Year 2). Significant differences were only found between Breeding and Non-breeding seasons. Given the observed correlation among environmental variables (habitat, topography and climate), ArcGIS 10.1 was used to perform a principal component analysis (PCA). Using R 3.3.2 we performed multiple regressions models for the Breeding season and Non-breeding season to assess relationships between the observed abundances and the components of the PCA.
The Truncated dataset was the data transformation that exhibited the strongest relationship with the three PCA components (PC01, PC02 and PC03) retained in the models. PC01 shows a negative correlation in Terrain ruggedness index (TRI) and accumulated rainfall, PC02 shows a positive correlation in percentage of coniferous forests and PC03 shows a positive correlation in percentage of natural vegetation and a negative correlation in percentage of eucalyptus plantations. The observed abundance varies positively with the percentages of coniferous forests and eucalyptus
plantations, whereas the relation with TRI, accumulated rainfall and natural vegetation is inverse. The regions predicted with high values of abundance corresponded to agriculture fields, whereas areas with natural vegetation and social and urban areas exhibited lower values of expected abundance.
The models developed could be further improved if other variables, such as food distribution across space and normalized difference vegetation index, were included. This is one of the first studies of this nature for this species, not only in Azores but also in Europe and much more work need to be done.