Figure 1.
Conceptual diagram showing the three hierarchical scales of habitat preference by lizards.
Home-range preferences, derived from the characteristics of the grid-cells included in the predicted home-range (dark green) and those in the area accessible to lizards (circle enclosing light-grey and dark-green areas), are determined broad-scale habitat features (c. 2.5–10*103 m2). Within home-range preferences, defined as the presence (dark green) vs. absence (light grey) of telemetry locations in the grid-cells included within each home-range, reflect the preferential use of certain habitats within each home-range. Microhabitat preferences, based on surveys of lizard abundance at each habitat type, reflect the uneven use of space within each grid-cell. Grid-cells represent the spatial unit (‘grain’) of observation within our study area.
Figure 2.
Cumulative frequency of defecation of Ephedra seeds ingested by lizards over time (i.e., proportion of ingested seeds defecated during such time interval; left axis), and net displacement (i.e., maximum net distance from the last relocation to the release location) of radio-tagged lizards over time (right axis).
Lines represent accumulative log-normal fits. Symbols represent daily averages (±SE) each two hours of both variables (see material & methods); note, however, that fits were based on all measured values (not shown for clarity).
Table 1.
Results of Generalized Linear (Mixed) Model of (a) home-range, and (b) between-patch habitat preferences of lizards.
Figure 3.
Habitat preference by lizards and habitat cover in the survey area at Dragonera Islet.
Habitat preference by lizards (bars in black) was based on the proportion of visual censuses (n = 506 lizards) whereas habitat cover (bars in red) was calculated from a habitat map which only covers the area of visual censuses.
Figure 4.
Mantel correlograms of simulated seed rain (seed dispersal probability per grid-cell) across the study-area lattice, using different habitat-dependent scenarios: one ‘random’ scenario, three ‘single-scale’ scenarios (home-range, within home-range and microhabitat preferences; diagonal panels) and their two-scales (‘home-range+within home-range’, etc.; lower-left panels) and three-scales (upper-right panel) combinations.
At each distance, filled symbols indicate significant autocorrelations (Mantel correlation at a given distance matrix after sequential Bonferroni corrections), empty symbols non-significant ones. The ‘random’ scenario is depicted in grey and followed the same conventions as habitat-preference scenarios. At each panel, we also show the results of a partial-Mantel test comparing the similarity of the corresponding habitat-dependent scenario with the ‘random’ one (i.e., higher Mantel-r values indicate higher correlation between both scenarios, p<0.05 a significant departure from the null hypothesis of non-correlation). See Text S3 for further details.