Fig 1.
Landscape composition and configuration of the study region.
Location of the study region (a). The different land-cover types and their spatial configuration across the study region (b) and the landscape units resulting from the integration of composition (Comp) and configuration (Conf) categories are depicted. This figure was produced by JMH using QGIS Development Team, 2015. QGIS Geographic Information System. Open Source Geospatial Foundation.
Table 1.
Mean (±SE) values of landscape variables used for characterizing landscape composition (a) and configuration (b) classes.
Numbers in brackets represent minimum and maximum values. Dominant land-cover types are those which accounted for more than 25% of each 65 ha hexagonal cell (500-m radius) which were used to extract landscape structural variables and species distribution data.
Table 2.
Total number (N total) of cells belonging to each type of landscape unit and number of cells where bird censuses was carried out (Nsampled).
Expected frequency distribution, calculated as N sampled / N total × 100, and real frequency distribution for forest-specialist, farmland-specialist, and generalist species are also reported. The suitability of each landscape unit for each bird guild was determined by comparing the real frequency distribution to the expected frequency distribution. Any landscape unit with a real frequency between 0–5% above the expected frequency distribution for a guild was considered as passage, while those landscape units showing real values 5% above or below expected values were identified as suitable and unsuitable, respectively.
Fig 2.
Maps of landscape suitability.
Maps depicting the landscape suitability for (a) forest-specialists, (b) farmland-specialists and (c) generalist species. Different colors represent suitable (dark green), passage (light brown) and unsuitable (white) cells for each bird species group.
Table 3.
Connectivity metrics, mean number of connected cells as well as the number and mean size of connected clusters, for the three scenarios investigated in the present study: baseline (a), degradation (b) and restoration (c) scenarios.
Fig 3.
Spatial distribution and frequency selection of cells comprising corridors for the three guilds.
The spatial distribution and frequency selection in the baseline landscape scenario (a) and after applying the restoration approach (b). The restoration approach consisted of restoring an increasing number of cells into suitable landscape units for the three bird guilds in a simulation process (n = 15; see Fig 4). Key conservation areas are those cells consistently selected (n = 100). The color legend with shades of orange/brown shows the number of times a cell was selected in the randomization process.
Fig 4.
Variation of landscape connectivity metrics as a function of number of cells restored to suitable landscape units.
Upper panels (a) show the response in total number of connected cells and lower panels (b) show the number of connected clusters. Boxplots represent variation in values over 100 runs for each set of restored cells. Horizontal lines represent medians, the bottom and top box edges are the 1st and 3rd quartile, respectively, and whisker lines represent 1.5 times the inter-quartile range. In each scenario, the left-most (light blue) boxes represent scenarios in which unsuitable landscape units for the three bird guilds simultaneously were converted into suitable or passage landscape units. The right-most (dark blue) boxes represent scenarios where cells for restoration were chosen randomly from across the study area, irrespective of guild-specific landscape suitability. Note the different scales of the y-axes in each case.