Material and methods
Study area.– The field study was conducted in the Granada Depression and nearby areas, within a region of approximately 3000 km2 in the south-eastern Iberian Peninsula (36°55’-37°20’N, 3°30’-4°15’W), where altitudes range between 450-1200 m asl. Mean annual temperature range between 12.5-14.3°C, and the average yearly rainfall between 355.4-448.0 mm (data from the Cartuja weather station [37°12’N, 3°36’W]). The study area is currently characterised by a mosaic of habitats dominated by cultivated land (olive orchards and cereal crops), mixed areas of evergreen forest and scrubland (Quercus ilex) and, to a lesser extent, pine plantations (Pinus halepensis, P. pinaster).
Sampling.– Field sampling was conducted from 1993 to 2000, within the framework of a larger study on the snake fauna of the region (details in Feriche, 1998). We made searches 3-4 field days per month (c. six hours each), throughout all months of the year. Specimens killed by local people and road-kills were collected (n = 320). Although the species is difficult to find and catch, some free-ranging live specimens were also captured (n = 25), processed for food items and analysed for getting morphology data (only tail injury, not maxillary teeth count). Specimens from the collections of the Estación Biológica de Doñana, Seville (EBD; n = 20) and Museo Nacional de Ciencias Naturales, Madrid (MNCN; n = 3), collected in the study area, were also analysed. In total, 368 specimens (202 males, 166 females) were examined.
Data collection on morphology and feeding.– In all specimens snout-vent length (SVL) was measured with a cord (± 1 mm) and body mass with an electronic balance (± 0.1 gr.). We checked for tail completeness and classified individuals with undamaged and damaged tails. In specimens collected by the authors, we were sure that tail breakage was the consequence of natural events and not a result of tail grasping during capture. We chose to classify the sub-sample from collections in a conservative way, and thus we did not consider that a tail was damaged if the break-point was not healed. Although it is known that the frequency of tail breakage varies among age classes and between sexes in snakes (Mendelson, 1992), we have for this study pooled the population. Our sample size for analysing this trait was sufficient in R. scalaris and it included individuals of all sexes and sizes, so we did not consider it necessary to perform sex or size-corrected comparisons.
Individuals were sexed by dissection in voucher specimens or according to Feriche et al. (1993) in live specimens. Stomach contents were checked by making a mid-ventral incision. Live snakes were gently palpated in the fore abdomen to force regurgitation in the case of recently ingested food (not in the rear abdomen to avoid damage of reproductive organs). All reptilian and bird eggs, as well as bird and mammal nestlings, were considered as one item because they were assumed to represent a single feeding event (Rodríguez-Robles and Greene, 1999). Whenever possible, prey items were identified to species level, measured (SVL; ± 1 mm), weighed (± 0.1 g) and the direction of ingestion of the prey (inferred from its orientation in the gut) recorded. Body length and weight of partially digested prey were estimated by comparing it with conspecifics from the study area, considering both traits at the time of ingestion. For the analysis of the ontogenetic shift in diet, four artificial SVL-categories for the snake were used, based upon the need for an analogous sample size and corresponding to approximately the following ontogenetic stages: i) newborn and immature snakes, ii) males just matured and females close to sexual maturity (males mature at 450 mm SVL, females at 660 mm SVL [Pleguezuelos and Feriche, 2006]), iii) medium-sized adults and iv) large adults (see Table 1 for SVL-range for each category). Distributions of data were checked for normality prior to analyses, and in this article mean values are followed by ± one standard deviation with alpha set at 0.05. Statistics were performed by STATISTICA 6.0 for WINDOWS PC.