Study area
The study was conducted in suburban areas of the city of Temuco in southern Chile, located in the central depression of the Araucanía Region, 619.5 km south of Santiago de Chile. Temuco was founded in 1881; it had 10,000 inhabitants by 1893, 72,000 in 1960, 189,000 in 1982, 304,000 in 2002 and 410,520 in 2017, making it one of the fastest-growing cities in Chile in recent years.
The connected suburban areas included in the study were: (a) Chivilcan meadows (38º41'-38º43' S, 72º35'-72º36'W) covering 1,700 ha, altitude 112 m a.s.l., on the slopes of Huimpil Ñielol mountain (González and Cerda 1989); and (b) the Ñielol Hill protected area, adjacent to Chivilcan (38º43'S-72º35'W), 89.5 ha, altitude between 115 and 322 m a.s.l, with steep topography (Fig. 1). This second area is considered urban for administrative purposes, but contains no urban infrastructure, so for the purposes of this study it was considered part of the suburban area of the city. Suburban areas are defined as peripheral areas of a city (urban area), with housing density of 2.5–10 houses/ha. The climate is transitional between mediterranean climate and temperate rainforest (Di Castri and Hajek 1976), with a short, marked dry season (Inzunza 2003). The average annual temperature is 12ºC, relative humidity 80%, and the average annual rainfall is 1324.8 mm, with a dry summer period of two months. Five environments were identified from a land chart generated in GIS: (a) Meadows with winter flood regimes and mostly allochthonous vegetation (57.9%), predominantly hemicryptophytes (54%) and remains of the original swamp forest vegetation cleared for pasture (González and Cerda 1989); (b) dense scrub of Ulex europeus (an introduced species considered a weed) with fragments of Chusquea quila and Aristotelia chilensis surrounding a few native trees, mainly mature Nothofagus obliqua; (c) exotic tree plantations of Pinus radiata and Eucalyptus globulus, the area of which has been increasing over the years (Locher 2002), with contemporary monostratified plant coverage about 15 m high, without undergrowth; (d) regeneration, areas of young forest dominated by N. obliqua, Peumus boldus and Cryptocarya alba about 10 m high; and (e) dense forest containing clerophyllous species from the mediterranean area combined with Valdivian forest from the temperate rainy zone, formed by Nothofagus-Perseetumboldetosum (Oberdofer 1960); Boldo-Cryptocaryetum (Oberdofer 1960) and Lapagerio-Aetoxiconetum (Oberdofer 1960), with multi-layered, multi-aged plant cover containing trees between 20 and 45 m high, in open and closed canopies respectively (Hauenstein et al. 1988a).
Methodology
The sampling stations were selected in the field and then geographically referenced to prepare a chart with the census design (Fig. 1). Information on the richness and abundance of diurnal raptors was obtained using three complementary methodologies: (a) Acoustic census with three hearing stations from which acoustic decoys were emitted to record the responses of raptors (Fuller and Mosher 1987). We used a digital portable player connected to a megaphone, which emitted the calls of diurnal raptors obtained from a study performed with acoustic lures in agricultural ecosystems in six locations in southern Chile (Harris' hawk Parabuteo unicinctus, Cinereous Harrier Circus cinereus and American kestrel Falco sparverius) (Contreras and Gonzalez, 2007). The calls were emitted alternately for one minute with five-minute waiting periods for each species, to provoke a territorial defence response or contact (Ralph et al. 1996). The acoustic recordings were played from 08.00 to 10.00h, 12.00 to14.00h and 18.00 to 20.00h; (b) Fixed observation points on vantage points, with good visibility to the naked eye and using 10 x 50 binoculars and a 20-60X telescope, together with photographic records when possible. These points were manned for two 3-hour periods per day (08.00 to 11.00h and 15.00 to 18.00h); (c) Vehicle-borne runs to record sightings along two transects on roads (gravel, with a medium level of traffic). The vehicle passed through the environments of the study area at a speed of 20-40 km per hour with two observers. Observation stops were made on both sides of the transect, recording the time, type of habitat, and activity (e.g. resting, in flight) for each bird, in the mornings from one hour after sunrise to 10.00h, at noon (12.00-14.00h), and from 18.00 to 20.00h to detect twilight species. Each transect was sampled only once to avoid double counting and pseudo-replication.
The surveys were conducted from September 2010 to May 2011 on days with favourable weather conditions (persistent rainy days were discarded), and using a template for sightings (Márquez et al. 2004).
We determined the following parameters: (a) Richness of species (S), defined as the number of species in a sample; (b) Relative abundance (AB%), defined as the percentage fraction of all birds of prey (sensu Krebs 1985) from which we could identify the species with low representation (low abundance); (c) α diversity (intra-environment), considering the specific richness and structure. The latter was determined by the Shannon and Wiener diversity Index, according to the function:
H'=-Σ (pi x log2 pi), (1)
where pi is the proportion of the total number of individuals of the species in the sample, with values which vary between zero when there is only one species and the maximum (H'max) corresponding to log2 S. In addition, we calculated the Pielou equity index (J) according to the equation: J = H'/ H'max, to measure the contribution of equity to the total diversity observed. The values varied between 0 (low heterogeneity) and 1 (maximum heterogeneity, when the species are equally abundant) (Magurran 1998). To test the null hypothesis that the diversity H’ of diurnal raptors of the different environments are the same we followed the procedure of Hutcheson (1970) described in Zar (1996), consisting of a t test calculating the weighted diversity index: Hp = (NlogN) - (αfi log fi) / N), including the calculation of variance for each environment according to: SH’2 = [Σ fi log2 fi - (αfi log fi) 2 / N] / N2 (d) β diversity (between environments) was represented by the species turnover or through a cluster similarity/dissimilarity dendrogram (between sectors) based on the Bray-Curtis Index (1957), using the BioDiversity Professional program (McAleece 1998); (e) g diversity, represented by all the species of diurnal raptors recorded in the Araucanía region (31,842 km2) in a sample taken from five national parks covering 137,138 ha (4.3%), supplemented by published (Pavez, 2000; Silva-Rodríguez et al., 2008) and unpublished information (CONAF, T. Rivas-Fuenzalida, C. Gonzalez-Bulo, V. Raimilla, R. Reyes-Carrasco, P. Caceres, A. Jaramillo pers. comm.). For the ranges of abundance we followed Jaksic et al. (2001): abundant = > 5 individuals detected (seen or heard) per day; common = 1-5 individuals detected daily; frequent = 1 individual detected weekly; low = 1 individual detected monthly; rare = <5 individuals detected annually.
Data analysis
Spatial information was constructed and analysed with a land use chart prepared from aerial photographs and previous studies (Hauenstein et al. 1988b) and processed in a Geographic Information System (GIS) in the software ArcGIS 9.3. The vegetation structure of each land use was verified in the field, characterizing the diversity of vegetation strata to build up a chart of the environments by grouping disaggregated uses into more homogeneous structural environments. The occupancy attributes of each environment were assigned from records and these attributes were incorporated into a final chart of richness of raptors.
To contrast the environments, the plant species associated with native forest were analysed for Temuco municipal district (466 km2), homologating the territory to the environments of the study area; ecological and visual techniques with GIS tools were used for this analysis. We identified the spatial structure of the landscape mosaic, recognizing the elements present and reclassifying vegetation coverage in order to obtain a minimum of classes of fragments. The categories were used as classes to represent landscape heterogeneity, regrouped in the environments described above. We then carried out general dissolution of the polygons adjacent to mature forest in order to analyse the polygons with core areas.
We also conducted an analysis of the spatial patterns present, using the following landscape metrics (sensu Botequilha et al. 2006): shape (based on the morphometric characteristics of the fragments), metric area (calculated based on the area corresponding to each of the fragments), edge metrics (calculated to estimate the amplitude of the edge habitat in relation to the interior habitat), proximity analysis (to calculate distance to the nearest fragment of the same class within a given search radius), and analysis of core areas (which calculates the interior habitat area of each fragment). The metrics were applied at patch level (discrete basic spatial unit, polygons), by class (typological category, classifiable feature) and to the entire landscape according to importance (Forman and Godron 1986). The landscape metrics were analysed in vector format, consisting of a representation of spatial forms in points, lines and polygons based on vectors (nodes and lines).
ArcGIS 9.3 extensions, Fragstat v3.3, v3.0 and Patch Analyst vLATE, were used for spatial analysis. Finally, for mature native forest species the structural and functional connectivity of the landscape was analysed. Structural analysis was applied to adjacent, physically connected native forest polygons (McGarigal and Marks 1995), whereas the functional study analysed dependencies by distance based on connection with ecological phenomena. Thus distance matrices were first analysed and then evaluated on the basis of distance thresholds, to reflect the probability of connection of the different fragments at a certain distance (McGarigal and Marks 1995).