2. Materials and Methods
2.1. Study areas
This study was carried out in three different areas within the city of Porto, Portugal (Figure 1): Contumil, 1.76 km2 (41º10.563’ N / 8º 33.977’ W to 41º 10.058’ N / 8º 35.257’ W); Campo Alegre, 1.10 km2 (41º 9.656’ N / 8º 38.928’ W to 41º 9.045’ N / 8º 37.870’ W); and Porto City Park, 1.03 km2 (41º 10.413 N / 8º 40.251 W to 41º 9.916 N / 8º 41.268 W); study area coordinates were obtained from Google Earth. Cat location coordinates were registered during surveys, in each place, with a GPS device (Garmin, eTrex®).
Figure 1. Location of all the study areas: Contumil (blue line), Campo Alegre (yellow line), and Porto City Park (green line) within the Porto city limits (red line; Copyright Google Earth 2017).
The landscape within Porto city is highly variable and includes croplands, green areas (wasteland), scrublands, house areas and others infrastructures, such as roads and subways (classification acording to Monterroso et al., 2009; see Annex Figure A1 for examples). In each area, the proportion of each type of landscape was assessed, according to the previous classification (Figure 2). As a City Park, the Porto City Park study area has higher percentage of green surface than the other two study areas; nevertheless, Contumil green surface is larger than Campo Alegre, which is the more urbanized area (Figure 2).
Figure 2. Habitat types available in each study area (habitat classification based on Monterroso et al. (2009); see Annex Figure A1 for examples). (A) Contumil; (B) Campo Alegre; (C) Porto City Park.
2.2. Free-roaming cats sighting surveys
Monthly cat sighting surveys, in all the extent of each study area, were conducted from November/2011 to April/2012, in Contumil study area, from November/2012 to April/2013, in Campo Alegre study area, and from October/2013 to April/2014, in Porto City Park study area. Path coverage was confirmed through a GPS tracking device (Telespial Systems Inc., Trackstick®). Sighting surveys were made monthly during three-day periods, of approximately equal duration, (morning: 09:00 to 13:00; afternoon: 14:00 to 18:00; and night: 20:00 to 24:00). Not all the area was thoroughly monitored because some gardens and wastelands were not accessible or cat sighting was difficult.
For each individual cat, or cat group, sighted, several parameters were registered (Table 2; Annex Figure A1) and each cat, or cat group, was photographed to allow posterior identification. The area occupied by each colony was estimated through Minimum Convex Polygon (MCP; Barratt, 1997), including the observations of all cats of the colony; these areas were then used to estimate cat’s colony densities. The density of free-roaming cat populations, in each of the different study areas, was estimated based on our observations and the respective area of each study area.
Table 2. Parameters registered during cat sighting walks and methodology adopted. Definitions for each classification are found in the respective reference. See Annex Figure A1 for example images from the classification.
2.3. Statistical analyses
All data sets were checked for normality of the distribution (Shapiro-Wilk test; Shapiro and Wilk, 1965) and homogeneity of variances (Levene’s test; Levene, 1961). Whenever data do not follow these assumptions, we used non-parametric tests.
Daily and monthly variability significance, throughout sampling periods, considering all the study areas, were searched with a Krustal-Wallis statistical test (Krustal and Wallis, 1952).
For colony data, we searched for differences among study areas, through a Student’s t-test (Zar, 1999), considering the total number of cats sighted per colony, the colony area, and the colony cats’ density. We also searched for colony differences between food distribution type and shelter availability type, among all the colonies found in the different study areas, with a Student’s t-test. Furthermore, we tested for colony differences between habitat types with a Krustal-Wallis statistical test. Finally, we searched for a relationship between the colony number of cats and the area occupied by each colony with a Spearman’s correlation test.
All the above statistical tests were made in IBM SPSS Statistics 24 (IBM Corp, 2016).