Estimating the population density of Iberan wild goat Capra pyrenaica and mouflon Ovis aries in a Mediterranean forest environment

Aim of study: To manage and conserve wild populations effectively, a good understating of population density is critical. During 2010, the density of Iberian wild goat Capra pyrenaica and mouflon Ovis aries were estimated. Area of study: The area is situated in Muela de Cortes Game Reservation (Spain), a Mediterranean forest plateau, after a mange Sarcoptes scabiei outbreak that affected both species. Material and methods: To measure the abundance, sex ratio and productivity of the Iberian wild goat and mouflon. Field work was conducted during spring (after parturition) and autumn (during rut) by walking along itineraries, using a Distance Sampling approach. Main results: Based on DS, the best relative fit of model and adjustment term for Iberian wild goat was hazard-rate cosine, based on the lowest AIC score. The average density for Iberian wild goat was 4 km (95% CI: 2,3-6,9) (after parturition) and 3,6 km (95% CI: 2-6.6) (during rut). Average estimation was 1,422 goats (95% CI: 813-2,487) after parturition and 1,308 during rut (95% CI: 7252,362). Mouflon best relative fit of model and adjustment term was uniform cosine after parturition, based on the lowest AIC score. The best relative fit of model and adjustment term for mouflon was hazard-rate cosine, based on the lowest AIC score. The average density was 6.8 mouflon km (95% CI: 4.7-9,9) after parturition and 7,4 mouflon km (95% CI: 4,4 -12,5) during rut. Average estimation was 2,440 mouflon after parturition (95% CI: 1,673-3,558) and 2,678 during rut (95% CI: 1,589-4,515). Research highlights: The area represents one of the largest continental free-living populations of mouflon in Europe and a relevant area for Iberian wild goat, where it has survived for centuries and spread into the East Iberia. This study suggests that the survey methods used are suitable and sustainable with available f ield personnel for quantifying changes in wild goat and mouflon populations, particularly in rugged forest environments. Monitoring should be continued and be part of the development of a comprehensive management programme for Iberian wild goat and mouflon.

different management actions (Yoccoz et al., 2001).However, the choice of a specific method depends on the ecology of the species of interest, the addressed question, the habitat type, the management goals to be achieved, and the sustainability in time of monitoring with the available resources.
The Iberian wild goat Capra pyrenaica is endemic to the Iberian Peninsula (Cabrera, 1911).Of the four original subspecies, C. p. lusitanica became extinct during the XIX century; C. p. pyrenaica became extinct at the end of the XX century (García-González and Herrero, 1999) and the two extant, C. p. victoriae and C. p. hispanica, are currently undergoing expansion and recovery (Pérez et al., 2002).Such expansion is mainly due to habitat changes, game management translocations (Gortázar et al., 2000) and probably a decrease of hunting pressure (Garrido, 2004).
Mouflon Ovis aries is a feral breed originated in Sardinia and Corsica (Shackleton, 1997), as genetical, paleontological and arqueological studies show (e.g., Manceau et al., 1999;Kahila bar-Gal et al., 2003).It was translocated to continental Europe, crossed with domestic sheep and today is present in a number of countries, even if normally populations are small, fenced or heavily managed (Apollonio et al., 2010).
Detailed data on host density population is needed to predict the spread of mange within a given population area (Gilbert and Dodds, 1992).Such information is essential to develop prevention and control programmes.Based on the previous, the aim of this study was to estimate Iberian wild goat and mouflon densi-ties, their sex ratio and productivity after the mange outbreak, and discuss implications for conservation and management of both populations.

Study area
The study was conducted during 2010 in the MC (39°11' 15'' N, 0°54' 13'') (MC), (east Iberian Peninsula, Valencia region, Spain) (Fig. 1) a calcareous plateau massif of 36,009 ha, with important slopes in its perimeter.MC is part of the Natura 2,000 network.The altitude ranges between 300-750 m.The climate is Mediterranean, with hot and dry summers and mild winters.Precipitation varies between 300-1,000 mm.Annual average temperature is between 14-16°C.Vegetation is typically Mediterranean and heavily influenced by fires.Forests of Aleppo pine Pinus halepensis, Maritime pine Pinus pinaster, Holm oak Quercus rotundifolia and shrubs (Pistacia lentiscus, Rhamnus oleoides, Juniperus oxicedrus Quercus coccifera, Rosmarinus officinalis, heather Erica sp., Genista scorpius, Stipa sp.) predominate.Human density is 6.2 km -2 however all villages are located outside the MC.Main economical activities inside the MC are agriculture (cereals, almond trees, and vineyards) and livestock of sheep (15.5 km -2 ) and domestic goats (9.2 km -2 ).Fallow deer Dama dama, red deer Cervus elaphus and wild boar Sus scrofa are also present although at low densities.

Survey design
We established 59 itineraries, placed systematically to provide complete coverage of the study area (Fig. 1).They were placed using a fine-scale map of the estate and aerial photographs and then they were digitalized.A strict protocol for the fieldwork was designed and followed: itineraries were walked slowly and in silence by one or two rangers and the team, after dawn, with good visibility (no precipitation, fog, or strong winds), and slowly, trying not to make noise.Each observer was equipped with binoculars, ad hoc maps, and datasheets.Itineraries were performed after the parturitions periods (June) and during rut (end of November), during a period of one week each.When wild goat and mouflon groups were detected, the position to the cen-Estimating the density of Iberial wild goat and mouflon in a Mediterranean forest environment tre of the group was recorded, using ad hoc cartography and binoculars.The following parameters were also recorded: size of the group, sex, age categories whenever possible (e.g., adult male, adult female, kids of the year and kids of one year), allowing the estimation of basic demographic parameters: sex ratio (males per females), productivity (kids per female) and age structure of males.

Distance sampling
Each group seen during the itinerary was marked in the ad hoc map and its perpendicular distance to the itinerary was calculated using a GIS program.Population density, the detection function and the average cluster size of groups were estimated.Data were righttruncated to eliminate 5% of the furthest observations (Buckland et al., 2001).Half-normal, uniform and hazard rate models for the detection function were fitted against the data using cosine, Hermite polynomial and simple polynomial adjustment terms, fitted sequentially.The selection of the best model and adjustment term were based on Akaike's Information Criterion (AIC), low coefficient variation (CV), low number of parameters and non significant c 2 .Statistical significance was assumed wherever p < 0.05.To calculate wild goat and mouflon density per km 2 , the program multiplied the group density calculated, with the average group size.We used the smallest average group size to minimize the skew produced with heterogeneous aggregations i.e. the average size of observed groups and the average size calculated by Distance Sampling using a regression.To transform this density in an absolute abundance, the program multiplied it with the total GRM surface.We used R (2.12.2) and Distance Sampling version 6.0 (Thomas et al., 2010) for the analyses.

Results
A total of 243 km of itineraries were surveyed within the Muela de Cortes Game Reservation.Analyses of the survey results using Distance 6.0 suggested that the most conservative model that fitted the data for Iberian wild goat, after parturition, was a hazardrate cosine model based on the lowest AIC score (Ta- Iberian wild goats per km 2 (CI 95%: 2.3-6.9).Distance software calculated an average group size of 3.4 wild goats km -2 .Twenty-three percent (23%) of the Iberian wild goat was solitary.During rut, the best relative fit of model and adjustment term was hazard-rate cosine based on the lowest AIC score (Table 1).The detection function histograms of the perpendicular distance are shown in Fig. 2.During rut, the average density of Iberian wild goat was 3.6 km -2 (CI 95%: 2-6.6).Distance software calculated an average group size of 5. Twenty-six percent (26%) of the Iberian wild goat groups were of single animals.The variation coefficient (VC) was 29% during parturition and 31% during rut.Sex ratio was 0.23 after parturition and 1.09 during rut.Productivity was 0.32 after parturition and 0.62 during rut.
Regarding mouflon, after parturition, the best relative fit of model and adjustment term was uniform cosine based on the lowest AIC score (Table 1).The detection function histograms of the perpendicular distance are shown in Fig. 2.After parturition, the average density was 6.8 mouflon km -2 (CI 95%: 4.7-9.9).Distance software calculated an average group size of 4.8.Fourteen percent (14%) of the groups were of single animals.During rut, the best relative fit of model and adjustment term was hazard-rate cosine based on the lowest AIC score (Table 1).The detection function histograms of the perpendicular distance are shown in Fig. 2.During rut, the average density was 7.4 mouflon km -2 (CI 95%: 4.4-12.5).Distance software calculated an average group size of 4.5.Thirteen percent (13%) of the groups were of single animals.The variation coefficient (VC) was 19% during parturition and 27% during rut.Sex ratio was 0.50 after parturition and 0.98 during rut.Productivity was 0.35 after parturition and 0.25 during rut.

Discussion
One important change regarding wild animal populations monitoring has been the link between research and management, allowing the application of scienti-   fic methods to population estimations, which have increased the quality of results and therefore, potentially, the sustainable management of wild populations.Shifting from total counts in low visibility and rugged terrains to systematic sampling, allied with Distance Sampling, has been an important step forward.This has been used successfully in a number of cases to estimate animal density (Brown and Boyce, 1998;Lynch and Rusydi, 1999;Palomares, 2001;McConkey and Chivers, 2004) and is increasingly been used for direct and indirect (pellet group counts) estimations of wild ungulates abundance (Marques et al., 2001;Smart et al., 2004;Focardi et al., 2001Focardi et al., , 2005Focardi et al., , 2006;;Franzetti et al., 2011) and specifically on mountain ungulates (Garin and Herrero, 1997;Granados et al., 2001;Herrero et al., 2011).
Monitoring was designed considering the number and dedication of field personnel available, basis of the long term monitoring.The calculated coefficient of variation (CV) can be considered acceptable (≤ 31%) (Focardi et al., 2006) however this is greater than the CV < 20% recommended for estimates of density (Buckland et al., 1993).The precision of the estimate (CV) would improve with increased sample size, which translates into sampling effort.The sampling effort needed to accomplish a desired precision would be unpractical in our study area and by the limiting number of rangers available.For example, a reduction of 10% in the CV, with the current mountain ungulate abundance, should require a sampling of a total of 1,490 km, which corresponds to 20 rangers working 15 days (3 weeks), something that cannot be afforded in the present situation.It is important to bear in mind that the precision varies with species and location, abundance, variation in group size and detectability (Varman and Sukumar, 1995).In general, a substantial effort is needed to achieve a < 20% CV.
Iberian wild goat densities are consistent with estimates from other studies and were within the actual (Alados and Escós, 2003;Pérez et al., 2002).Our results show that the population structure and parameters of the wild goat are similar to those of other areas from east Iberia (Herrero et al., 2011;Prada et al., 2011).The wild goat is a gregarious animal, living in groups of different size and composition throughout the year (Alados and Escós, 2003).The group size varies depending on the population density and the characteristics of the vegetation.During parturition (late spring) females isolated to give birth.After this, females aggregate into small groups, usually the family.Our results show a sex ratio quite unbalanced in favour of females, during the autumn and winter of 2009 and parturition 2010.Previous studies report a strong variability in this parameter ranging between 0.5 and 2.5 (Sánchez-Hernández, 2002;Moço et al., 2006).In our study area this can be due to sex-biased mortality (males) during mange outbreak (Sarasa et al., 2010) combined with the low overall extraction, which is mainly focused on males, but certainly a relevant parameter to monitor.Human harvesting may differ significantly  from those derived from natural mortality (Solberg et al., 2002).Even though both sexes are hunted, wild goat males are preferentially hunted since they are Iberian exclusive trophy-hunting species, with an high price in the market (Hofer, 1999).Our productivity values are close to those described for other populations (Alados and Escós, 2003;Moço et al., 2006;Prada et al., 2011).There is a gap of information regarding mouflon densities, both in Europe (Apollonio et al., 2010) and specifically in the Iberian Peninsula.Blanco (1998) estimated densities of 3.2 mouflon km -2 (Sierras de Cazorla, Segura y Las Villas Nature Park), 1.6 mouflons km -2 for Teide National Park (Tenerife) and 1.6 km -2 for the Serranía de Cuenca.Despite some reviews (Spain: Rodríguez Luengo et al. 1992; Italy: Apollonio andMeneguz, 2003, andFrance: Bon et al., 1991), there are few studies estimating mouflon abundance (Tsaparis et al., 2008), which reveal the lack of attention that is dedicated to the study and monitoring of this species.Tsparis et al. (2008) estimated mouflon densities as 22.1 km -2 however it was a confined area.Considering mouflon population parameters, we emphasize a balanced sex ratio and an overall low productivity.This may be due, as in the wild goat case, to the overlap of the counting period with the calving season, where females are less conspicuous.Our results show that our study area contains one of the most relevant continuous and non-fenced continental European mouflon populations, in a context were fences, artificial feeding, bad adaptation and small populations are common (Apollonio et al., 2010).The overall density of all ungulates in MC can be considered high (over 35 ungulates km -2 ) (presence of Iberian wild goat, wild boar, domestic ungulates, small densities of red and fallow deer and the possible future expansion of roe deer e.g.Gortázar et al., 2000, and probably Barbary sheep Ammotragus laervia).This has leaded to prioritize the most relevant species and the control of the rest, taking in account the recent mange outbreak and the experiences of its avoidance (Meneguz et al., 1996).Particularly sanitary status of domestic goats should be carefully monitored, due to the unanimous view that blames them of being the origin of outbreaks in Cantabrian chamois Rupicapra pyrenaica parva (Meneguz et al., 1996) and Iberian wild goat (León-Vizcaíno et al., 1999).Domestic goats are also the populations were prevention can take place.
The management of the MC has been deeply influenced by the mange epidemic, its demographic con-sequences for the wild goat population and the demanded management measures to reduce the outbreak effects (Sánchez-Isarria et al., 2008).
The use of itineraries as an adaptation of line transect for the Distance Sampling software met the objectives laid out in this study.In fact, this study highlights the potential of using different and easy to implement methods hat can be used by rangers and wildlife managers as useful and affordable tools for population monitoring in both wild goat and mouflon management, requiring therefore minimal equipment and man-power.
The results of our analysis indicate that an adequate monitoring program, using an adaptation of Distance Sampling, to detect population trends is feasible.In fact, using wildlife rangers (and possibly volunteers) is viable to monitor population trends using Distance Sampling techniques.Although the methodology presented in this paper was applied in the context of mountain ungulates, it can be applied to other animals.

Figure 1 .
Figure 1.Map of the study area with the distribution of the sampling plots.

Figure 2 .
Figure 2. Detection function g(y) illustrating the probability of detection of Iberian wild goat by perpendicular distance from the itinerary in the parturition (a) and in the rut period (b).

Figure 3 .
Figure 3. Detection function g(y) illustrating the probability of detection of mouflon by perpendicular distance from the itinerary in the parturition (a) and in the rut period (b).

Table 1 .
Estimating the density of Iberial wild goat and mouflon in a Mediterranean forest environment 39 Estimation of the density of individual(s) of Iberian wild goat and mouflon in MC : hazard-rate.cos: cosine.GOF Chi-p: goodness-of-fit chi-square test probability.D: density of individuals per km 2 .D LCL: density of individuals analytic lower conf.limit.D UCL: density of individuals analytic upper conf.limit.D CV: density of individuals analytic coeff.Of var.DS: density of clusters.CS: mean cluster size.Sz Bias CS: expected cluster size, correcting for size bias.N: number of individuals. Hz

Table 2 .
Results of Iberian wild goat and mouflon counts in the MC