An empirical test of source–sink dynamics induced by hunting


Andrés J. Novaro, CONICET and Wildlife Conservation Society, CEAN, CC 7, Junín de los Andes, 8371 Neuquén, Argentina (e-mail


  • 1Under the source–sink model, persistence of populations in habitat sinks, where deaths outnumber births, depends on dispersal from high-quality habitat sources, where births outnumber deaths. The persistence of the regional population depends on the proportion of sink relative to source habitat.
  • 2Hunting that occurs in some parts of the landscape and not in others can create patches where deaths outnumber births. We tested whether hunting of culpeo foxes Pseudalopex culpaeus, which is patchily distributed in relatively homogeneous habitat in Argentine Patagonia, induces source–sink dynamics.
  • 3On Patagonian sheep ranches, culpeos are hunted for fur and to protect sheep, and on cattle ranches hunting is usually banned. We monitored culpeo densities using scent stations and estimated survival, fecundity and dispersal by radio-tracking 44 culpeos and analysing carcasses collected from hunters on two cattle and four sheep ranches between 1989 and 1997.
  • 4Survival of juvenile culpeos was lower on hunted than unhunted ranches, mainly as a result of hunting mortality. Reproduction could not compensate for high mortality on hunted ranches. Interruption of hunting led to an increase in juvenile survival, indicating that hunting and natural mortality were not compensatory. We concluded that sheep ranches were sinks because of the high mortality and that sink populations may be maintained by dispersal from cattle ranches.
  • 5We used a simulation model to assess implications of changes in the proportion of source and sink areas on population dynamics. The percentage of land on cattle ranches in the study area was 37%. Current hunting pressure on culpeos would not be sustainable if that percentage fell below 30%.
  • 6Synthesis and applications. Source–sink dynamics may occur in landscapes where hunting is intense and spatially heterogeneous. Wildlife management traditionally monitors demographic rates to evaluate the sustainability of hunting, but our results suggest that the size and spatial arrangement of areas with and without hunting should be considered as well. In regions where enforcement and monitoring are limited, securing large and regularly distributed source areas for hunted species may be more effective than trying to regulate harvest size.


The composition and spatial arrangement of habitat patches in the landscape often have a strong influence on the population dynamics of species (Dunning, Danielson & Pulliam 1992; Hanski & Gilpin 1997). One of the consequences of landscape complexity can be the source–sink dynamic produced when individuals in the same population occupy habitat patches of different qualities (Pulliam 1988). Productive patches, where reproduction exceeds mortality, act as sources of individuals that disperse to less productive sink patches. Subpopulations in sinks would go extinct without dispersers from sources. Empirical studies that have described populations with source–sink dynamics have focused on natural heterogeneity in habitat quality, or heterogeneity produced indirectly by human alteration of habitats (Vierling 2000; earlier studies reviewed by Pulliam 1996).

Hunting by humans affects animal populations in most landscapes. Hunting is usually concentrated in areas that are more accessible to humans or where land use and tenure make it feasible (Novaro, Redford & Bodmer 2000). Thus hunting may create spatial differences in mortality for game species and influence their population dynamics at a landscape scale. Harvest theory was originally built on an assumption of uniformly distributed populations (Beddington & May 1977), but this assumption has begun to be revised (Lundberg & Jonzén 1999; Delibes, Gaona & Ferreras 2001). Fisheries studies, for example, have shown that reserves can export fish into catchment areas and help maintain their productivity (McClanahan & Mangi 2000).

Intensively hunted populations of terrestrial vertebrates with high dispersal ability may also persist as a result of immigration from unhunted areas (Littel, Crowe & Grant 1993; Slough & Mowat 1996). Woodroffe & Ginsberg (1998) have suggested that the peripheries of protected areas may act as sinks for large carnivores that are killed by humans when they roam beyond reserve limits. Recent theoretical and empirical studies of harvested populations that are structured in sources and sinks, however, have addressed cases where this structure is produced by spatial differences in habitat quality (Lundberg & Jonzén 1999; Millner-Gulland, Coulson & Clutton-Brock 2000). Source–sink dynamics where the sinks are the result of hunting have not been tested empirically as a mechanism to explain the dynamics of hunted populations.

Source–sink dynamics of hunted populations can have ecological and management implications. Hunting can produce attractive sinks if dispersing individuals select good habitats with abundant resources but with high human-related mortality (Delibes, Gaona & Ferreras 2001). Dynamics of source–sink systems where sinks are attractive are highly sensitive to small changes in the proportion of sink habitat. Management of harvested populations with source–sink dynamics must be implemented at a regional or landscape scale and maintain the spatial heterogeneity of land use that allows the existence of unhunted sources.

We studied the population dynamics of culpeo foxes Pseudalopex culpaeus Molina in an area where the spatial distribution of hunting is heterogeneous. Culpeos are medium-sized canids that are hunted intensively in Argentinean Patagonia for their fur and because they prey on sheep (Novaro 1995). In north-western Patagonia hunting is banned by owners of cattle ranches because culpeos also prey on European hares Lepus europaeus Pallas that compete for forage with livestock. Cattle ranches are intermixed with sheep ranches where owners promote culpeo hunting. Most ranches range in size between 100 and 400 km2. Hunters, mostly ranch workers, remove up to 75% of culpeos each winter and simulations based on demographic data from sheep ranches suggested that culpeo populations should collapse (Novaro 1995). However, culpeos persist in both sheep and cattle ranches, and the number of culpeos hunted each year on sheep ranches has changed little during recent decades.

The purposes of this study were to test the source–sink model on a hunted culpeo population and analyse the implications of landscape changes on culpeo population dynamics. We hypothesized that the spatial distribution of hunting produces a mosaic of sources and sinks that allows culpeos to persist in hunted areas, where population recovery is mainly the result of dispersal of surplus individuals from unhunted areas. Increased reproduction or reduced natural mortality could occur in sinks because of reduced density-dependent effects (Caughley & Sinclair 1994) but would be insufficient to compensate for the high hunting mortality. We assessed the implications of landscape changes using a simulation model. Mosaics of hunted and unhunted sites are often dynamic because human decisions to hunt at a site frequently depend on factors such as market prices of wildlife and agricultural products. In Patagonia the decision by ranchers to raise sheep (and promote culpeo hunting) or cattle (and ban culpeo hunting) is based to a large extent on changing market prices of wool and beef (Novaro 1995).


study area

The study area was a mixed steppe of grass and shrubs in north-western Patagonia (40°S, 71°W), Argentina (see Appendix 1). Six ranches (80–400 km2) were chosen to represent dominant land uses in the region. Collun-Co and La Rinconada were cattle ranches where culpeo hunting was rare or non-existent, and Catan Lil, La Papay, Los Remolinos and Cerro de los Pinos were predominantly sheep ranches where hunting was intense. Hunting occurred primarily during late autumn and early winter (1 Ma.−31 August), although some occurred throughout the year. The total area of all six ranches was 1420 km2 (34% on cattle and 66% on sheep ranches). About 37% of land in the surrounding region (5236 km2) was cattle ranches, so our main study area was representative of the proportion of land with and without hunting in this region (see Appendix 1).

Biomass of culpeo prey, except for sheep, was similar on sheep and cattle ranches, and sheep represented 20% of the biomass of the culpeo diet on sheep ranches (Novaro, Funes & Walker 2000). Thus we did not expect differences in prey productivity between ranches to influence culpeo survival or fecundity.

Culpeos have been hunted in this area since the introduction of sheep in the early 1900s (Crespo & de Carlo 1963), sustaining high hunting pressure when fur demand increased during the 1960s and 1970s (Novaro 1995). Prior to 1994, all ranches studied had maintained consistent practices with respect to hunting and livestock for at least 30 years. As a result of declining sheep wool prices, the owners of Cerro de los Pinos and Los Remolinos began raising only cattle in early 1994 and banned culpeo hunting. This ban during our study (1989–97) allowed us to test for compensatory mortality (see below).

test of the sourcesink model

We tested the following predictions of the source–sink hypothesis.

Prediction 1: rates of increase

The rate of increase of the culpeo population based on schedules of survival and fecundity (rc; Caughley & Sinclair 1994) is significantly < 0 on hunted ranches and significantly > 0 on unhunted ones, but the observed rates of increase (ro) are not significantly different from 0 if the regional population level is stable (Hanski & Simberloff 1997). To evaluate which demographic parameter determined differences between rc and ro, we compared rates of survival and fecundity between ranches.

Prediction 2: dispersal direction

Most culpeos disperse from unhunted to hunted ranches.

Prediction 3: non-compensatory mortality

If hunting mortality is removed from a previously hunted ranch, culpeo survival should increase significantly because any increase in natural mortality would not be sufficient to compensate for hunting mortality (Caughley & Sinclair 1994).

ro for culpeos was estimated from population trends obtained at the beginning of the hunting season between 1989 and 1997 with scent stations (Roughton & Sweeny 1982). Forty scent stations (eight lines of five stations each) were operated along all internal roads and trails of each ranch during one day and night. We calculated culpeo densities from scent station indices (SSI) using a conversion factor [density = SSI/(32·8 ± 3·0)] obtained from a simultaneous calibration with line transect estimates (Novaro et al. 2000). Rates of increase between 1989 and 1994 were calculated with the Caughley & Sinclair (1994) linear regression method. Densities for 1995–1997 were not used because of the ban on hunting in two of the ranches in 1994.

rc was calculated with data from radio-telemetry and analysis of 362 culpeo carcasses collected during the hunting season between 1989 and 1994. We captured 47 culpeos on Collun-Co (unhunted control, hereafter UCC) and adjacent Cerros de los Pinos (hunted, HCP) between January and May of 1993 and 1994 using padded foot-hold traps (Victor 1.5 soft-catch; Woodstream Corp., Lititz, Pennsylvania). Ages of culpeos were estimated according to tooth wear (Zapata, Funes & Novaro 1997). A total of 23 adults and 21 juveniles, which weighed at least 50% of adult body mass (4 months old; Crespo & de Carlo 1963), were fitted with radio-collars (ATS Inc., Isanti, MN; weighing up to 5% of body mass). Fifty per cent of radio-collared culpeos were females. Twenty-seven culpeos were captured and radio-tracked in 1993 (11 at UCC and 16 at HCP). In 1994, 13 survivors from 1993 and 17 newly captured culpeos were radio-tracked (16 at UCC and 14 at HCP). Eighteen culpeos that survived throughout 1994 were radio-tracked until they died or until 15 March 1996. Culpeos were located during the daytime at least twice every week from a vehicle or from a fixed-wing aircraft. We did not locate culpeos at night because we did not attempt to describe habitat use or activity patterns. We interviewed hunters to establish when and where culpeos were killed. Carcasses of non-hunted culpeos that died were necropsied to determine cause of death.

Survival rates were estimated from dates of capture and death with the Kaplan–Meier method (White & Garrott 1990; Bechet et al. 2003) and cause-specific mortality rates with the Mayfield estimator using program micromort (Heisey & Fuller 1985). Survival of juveniles and adults were different, so they were analysed separately. Yearly survival of adults was estimated between 15 November (the mid-point of the birth pulse; Crespo & de Carlo 1963) of one year and 14 November of the following year. Survival of radio-tracked juveniles 4–12 months of age was estimated between 15 March and 14 November. Some culpeos dispersed from HCP and UCC to other ranches. Survival data on other ranches were grouped with data from HCP and UCC according to hunting pattern to estimate overall survival of hunted and unhunted populations. Survival rates were compared using the log–rank test in SAS (SAS Institute Inc. 1996).

Culpeo fecundity rate was the product of the proportion of breeding females and the number of female pups produced per female (Caughley & Sinclair 1994). Radio-tracked females were identified as breeding if they denned and if signs of pups (e.g. faeces, tracks) were observed. Survival of juveniles between 0 and 4 months of age was estimated by dividing the average litter size seen at 4 months by in-utero embryo counts. Nine pups from three litters were captured and ear tagged. We estimated litter size by combining counts of pups at dens and embryos in female reproductive tracts (Gese, Rongstad & Mytton 1989). Fecundity of culpeos that dispersed was analysed following the criteria for grouping ranches for survival analysis. Because not all pregnant females in the carcass sample might have given birth to live litters, we may have overestimated rates of fecundity and population increase for hunted ranches. This overestimation, however, would result in a more conservative test of the predictions. Fecundity rates were compared using a randomization test because of the small sample sizes and lack of independence among samples (because data for the same females on different years were grouped; Bruce, Simon & Oswald 1995).

We estimated the age of dead animals by counting cementum annuli on canine teeth (Zapata, Funes & Novaro 1997). Culpeos were grouped into yearly age classes up to 6 year olds; older culpeos were grouped into one age class. Sex ratio was compared with a 1 : 1 ratio using a chi-square test.

The rc of culpeos was calculated using the Lotka (1907a,b) iterative equation:

lx e − rx mx = 1

where lx and mx are the survival and fecundity, respectively, of age class x. We assumed constant fecundity rates for 1-year-old and older females. Rates of increase were also expressed as finite rates per year (λ = er).

We estimated dispersal rates from the proportion of radio-tracked culpeos that dispersed from one ranch to another. We recorded dates of departure from natal range and arrival at new range, dispersal distance, and type of ranch (unhunted or hunted) for the new range.

culpeo population dynamics and landscape changes

We studied population dynamics of culpeos using the RAMAS/Metapop simulation model (Akçakaya 1994). The model had a spatial structure defined by the geographical location of populations, dispersal among populations and correlation among their vital rates. To simulate dynamics on a continuous landscape, we modelled the populations of the six ranches studied and of the 17 additional ranches to the north and south, for which we recorded size and culpeo hunting pattern (see Appendix 1). The area was delimited to the west by the Andes Mountains and the east by the limit of culpeo distribution. More distant ranches to the south were included because two radio-collared culpeos dispersed 86 and 90 km south (see the Results).

Percentage change in population size was the dependent state variable. Input data were demographic parameters estimated on the six ranches. Initial age structures and abundances on the additional ranches were calculated using average age structures and densities from the former six ranches (0·49 ± 0·12 and 0·31 ± 0·09 culpeos km−2 on the four hunted and two unhunted ranches, respectively). We assumed age structures did not change during the simulation, modelled all individuals in the population (females and males, ratio 1 : 1), and used a matrix model with three stages (juveniles, 1 year olds and older). We used two adult age classes because survival of 1-year-old adults was lower and because some 1 year olds dispersed.

Vital rates were calculated following methods proposed by Caswell (1989) and Akçakaya (1994). Reproductive data were analysed as maternity rates. Transition matrices were constructed assuming age structures from a pre-breeding census because most carcasses were collected during winter, before the culpeo birth pulse (Caswell 1989). The transition matrix for populations on ranches without hunting, built with vital rates from UCC (see the Results),was inline image, and for ranches with hunting,with vital rates from HCP and other hunted ranches,was inline image. The matrix for hunted rancheswas assigned using the catastrophe feature of RAMAS, with local probability of 1.

Environmental stochasticity was simulated by choosing vital rates at random from a normal distribution with means from each transition matrix and standard deviations from a matrix calculated from annual variation of vital rates. The matrix of standard deviations of vitalrates was inline image. Demographic stochasticitywas included because it could be significant if there were sharp declines in abundance, which was likely on small hunted ranches.

Dispersal rates among ranches were calculated assuming that dispersal declined monotonically with distance (Akçakaya 1994), which was reasonable based on our radio-telemetry data. Dispersal rate between ranches i and j was:


where Dij was the distance between the geographical centre of ranches i and j. Maximum dispersal distance (D) was that recorded with radio-telemetry (see the Results).

We did not include density dependence in the simulations for two reasons. First, when populations are subject to ‘systemic’ pressures such as regular and intense hunting, inclusion of density dependence in simulations can lead to an underestimation of extinction risks (Ginzburg, Ferson & Akcakaya 1990). Secondly, we did not have information on density-dependence parameters of culpeo populations. Because hunting was decreasing in the area, we presumed that density dependence might become a more significant factor in the future, and therefore ran the simulations for only 8 years, the duration of the study of demographic parameters, assuming these parameters incorporated density dependence effects at current densities. We used 1000 replications in each simulation and validated the model by comparing output population trends with trends estimated between 1989 and 1997.

The implication of changes in the proportion of hunted and unhunted areas was analysed by simulating changes in the proportion between cattle and sheep ranches. To study the effect of a declining unhunted area, we switched unhunted to hunted ranches one at a time and ran the model with remaining parameters unchanged. We also evaluated a combination of changes in landscape proportions and vital rates. To assess which vital rates may have a stronger influence on population dynamics we did a sensitivity analysis. We changed survival, fecundity and dispersal rates by 10%, 20% and 30% and measured changes in population trend.


test of predictions from the sourcesink model

Rates of increase

Overall, culpeo densities at the beginning of each hunting season were similar on hunted and unhunted ranches (Fig. 1). Before hunting stopped at two of the ranches, densities remained stable on hunted ranches (ro1989−94= 0·09, λ = 1·096, not significantly different from 0, t = 1·757, P = 0·154) and increased slightly on unhunted ranches (ro1989−94 = 0·17, λ = 1·158, significantly > 0, t = 3·940, P = 0·017).

Figure 1.

Mean relative densities of culpeos (± 1 SE) on the four ranches with hunting and two without hunting. Densities are percentages of scent stations visited.

The survival rate of radio-tracked culpeos was lower on hunted than unhunted ranches. Juvenile survival on hunted ranches during 1993 (8-month rate = 0·083, 95% CI = 0·0–0·250, n = 10) was significantly lower than on unhunted ranches during 1993 and 1994 (0·800, 95% CI = 0·449–1·000, n = 11, χ2 = 6·1, P = 0·013; Fig. 2). Juvenile mortality on hunted ranches was mainly the result of hunting with shotguns and dogs (seven of eight deaths); the remaining death was a roadkill (Table 1). Only two juveniles that dispersed from HCP in 1993 survived. On unhunted ranches only one juvenile was killed by feral dogs in 1994 (Table 1).

Figure 2.

Daily survival rates of juvenile culpeos (a) radio-tracked in unhunted (n = 10) and hunted ranches (n = 11) between 15 March and 15 November of 1993 and 1994, and adult culpeos (b) in unhunted (n = 17) and hunted ranches (n = 9) between 15 March 1993 and 14 March 1996. Rates for juveniles are from 4 to 12 months of age; 95% confidence intervals are calculated after one or more animals die.

Table 1.  Cause-specific mortality rates of juvenile culpeo foxes radio-tracked on hunted (n = 10 culpeos) and unhunted ranches (n = 11) between March 1993 and November 1994
Interval*Days in intervalHunted ranchesUnhunted ranches
Transmitter daysMortality causeRate (deaths)Transmitter daysMortality causeRate (deaths)
  • *

    Intervals are between the 15th of the first month and the 14th of the second month.

  • Total number of days different culpeos were radio-tracked during interval.

  • Number of deaths during interval.

Autumn61222Hunt0·212 (1)283Hunt0·000 (0)
(March–May)  Other0·212 (1) Other0·000 (0)
Early winter61349Hunt0·505 (4)564Hunt0·000 (0)
(May–July)  Other0·000 (0) Other0·000 (0)
Late winter62164Hunt0·316 (1)562Hunt0·000 (0)
(July–September)  Other0·000 (0) Other0·000 (0)
Spring61 75Hunt0·559 (1)445Hunt0·000 (0)
(September–November)  Other0·000 (0) Other0·061 (1)
8-month rates and 95% CI
 Hunting0·702 (0·332–1·000)  0·000  
Other0·212 (0–0·579)  0·128 (0–0·363)  

Adult survival on hunted ranches during 1993–1995 (annual rate = 0·509; 95% CI = 0·259–0·998; n = 9) was not significantly different from unhunted ranches (0·693, 95% CI = 0·521–0·937, n = 17, χ2 = 1·3, P = 0·25; Fig. 2). Hunting was the only cause of adult mortality on hunted ranches (four of four deaths; Table 2). Adult mortality on unhunted ranches was mainly the result of predation by feral and domestic dogs (five of nine deaths) and puma Puma concolor Linnaeus 1771 (one death). Two adult culpeos were killed by poachers on unhunted ranches, and the remaining death was the result of unknown natural causes (Table 2).

Table 2.  Cause-specific mortality rates of adult culpeo foxes radio-tracked on hunted (n = 9 culpeos) and unhunted ranches (n = 17) between January 1993 and November 1995
Interval*Days in intervalHunted ranchesUnhunted ranches
Transmitter daysMortality causeRate (deaths)Transmitter daysMortality causeRate (deaths)
  • *

    Intervals are between the 15th of the first month and the 14th of the second month.

  • Total number of days different culpeos were radio-tracked during interval.

  • Number of deaths as a result of hunting or other causes during interval.

Early summer61183Hunt0·000 (0)1075Hunt0·052 (1)
(November–January)  Other0·000 (0) Other0·104 (2)
Late summer59278Hunt0·192 (1) 972Hunt0·000 (0)
(January–March)  Other0·000 (0) Other0·073 (1)
Autumn61443Hunt0·000 (0)1213Hunt0·000 (0)
(March–May)  Other0·000 (0) Other0·049 (1)
Early winter61484Hunt0·119 (1)1165Hunt0·000 (0)
(May–July)  Other0·000 (0) Other0·100 (2)
Late winter62403Hunt0·143 (1)1067Hunt0·056 (1)
(July–September)  Other0·000 (0) Other0·000 (0)
Spring61333Hunt0·168 (1) 977Hunt0·000 (0)
(Sep.–Nov.)  Other0·000 (0) Other0·061 (1)
Annual rates and 95% CI
 Hunting0·491 (0·148–0·835)  0·090 (0–0·210)  
Other0·000  0·317 (0·122–0·512)  

Fecundity rates of adult females living on hunted (2·27 pups female−1, SD = 0·84) and unhunted ranches (3·19, SD = 0·72) were not significantly different (randomization test, P = 0·182). The proportion of adult females that bred was 0·64 on hunted ranches (n = 11, from eight reproductive tracts and three radio-tracked females), 0·75 on unhunted ranches (n = 8 radio-tracked females) and 0·75 on previously hunted HCP in 1994 and 1995 (n = 8 radio-tracked females). Mean litter sizes of adult females on hunted and unhunted ranches were 3·60 pups (SD = 1·35, n = 20, from six reproductive tracts and 14 dens) and 4·25 pups (SD = 0·96, n = 14 dens), respectively. One of five juvenile females studied on hunted ranches bred, giving birth to three pups, vs. zero of six juvenile females on unhunted ranches. Overall pup survival between 0 and 4 months of age was 0·906 (n = 6 reproductive tracts and eight dens on hunted and unhunted ranches). The sex ratio of hunted and radio-collared culpeos was not significantly different from 1 : 1 (47% female, 53% male; χ2 = 1·542, P = 0·410, n = 352).

The rc based on culpeo survival and fecundity rates (see Appendix 2) was −0·36 on hunted ranches (29·9% decline expected each year) and 0·57 on unhunted ranches (76·8% increase expected). These expected trends contradicted the observed population trends and supported the prediction on rates of increase. Differences between calculated rates of increase were mainly the result of lower survival on hunted ranches.

dispersal direction

Most dispersing culpeos that were radio-tracked left from unhunted ranches (eight of n = 11 dispersers), but most dispersers established new ranges on hunted ranches (five of seven). Remaining dispersers (n = 4) were hunted on ranches with hunting before they established new ranges. Dispersal distances ranged between 12 and 90 km (mean 30·5 km, SD 29·4; Fig. 3). One culpeo tagged as a pup at a den dispersed 86 km. Dispersal rates of juveniles (n = 13 culpeos that survived to 1 year of age), 1 year olds (n = 10) and older culpeos (n = 13) were 0·692, 0·200 and 0·000, respectively.

Figure 3.

Dispersal distances of culpeos radio-tracked on hunted and an unhunted ranches. Shaded column represents dispersal within a ranch (n = 3) and unshaded columns represent dispersal between ranches (n = 8).

compensatory mortality

The interruption of hunting on HCP in 1994 led to a pronounced increase in culpeo survival, indicating that hunting mortality and natural mortality were not compensatory. Eight-month juvenile survival on HCP in 1994 (0·830, 95% CI = 0·377–1·000, n = 6) was significantly higher than in 1993 (0·083, χ2 = 6·0, P < 0·015). Survival of adults on HCP in 1994 was 0·854 (95% CI = 0·626–1·000, n = 7) but was not significantly higher than in 1993 (0·556, 95% CI = 0·285–0·832, n = 6, χ2 = 2·1, P = 0·19). Conversely, on control ranches juvenile (0·738, 95% CI = 0·406–1·000, n = 6) and adult survival in 1994 (0·540, 95% CI = 0·314–0·928, n = 11) were lower than in 1993, when both survival rates for radio-tracked culpeos were 1·000 (n = 5 juveniles and 7 adults). Two deaths occurred on the formerly hunted ranch in 1994 as a result of poisoning (one juvenile) and an unknown natural cause (one adult). Annual mortality rates for adults on the control ranches in 1994 were 0·078 (95% CI = 0·000–0·226) as a result of poaching (one death) and 0·382 (95% CI = 0·090–0·674) as a result of predation by dogs and puma (four deaths).

culpeo population dynamics and landscape changes

Our model predicted population trends that were similar to the 8-year population trend recorded using scent stations (Figs 1 and 4). The mean size of the simulated population increased by 68% in 8 years, while according to scent station results the population increased by 80%.

Figure 4.

Predicted population trend of culpeos using RAMAS/Metapop simulation model for 23 ranches in north-western Patagonia.

Adult survival was the rate to which the culpeo model was most sensitive (Fig. 5). A 10% decline in fecundity and adult survival led to population sizes in the eighth year of simulation that were 19% and 45% lower, respectively, than the size predicted with current rates. Dispersal had a smaller effect than fecundity and an opposite effect than adult survival and fecundity.

Figure 5.

Sensitivity analysis of changes in culpeo population size after 8 years of simulation in relation to changes in fecundity, survival and dispersal. Middle values correspond to current estimates. Values to the left and right correspond to rates that are 10%, 20% and 30% smaller and larger than current rates, respectively.

Culpeo populations collapsed under current hunting pressure when the percentage of area on unhunted ranches fell below 30% (Fig. 6). A 7% decline would occur if cattle were replaced by sheep on three small- to medium-sized ranches or on only one large ranch. Population collapses with reductions in unhunted area were more likely if adult survival rate was lower. When adult survival was reduced by 3%, the population declined with any reduction in unhunted area larger than 5%. If adult survival declined by more than 6%, even the current percentage of unhunted area would result in a population decline (Fig. 6).

Figure 6.

Change in population size of culpeos after 8 years of simulation in relation to changes in percentage of area in ranches without hunting and different rates of adult survival.


sourcesink vs. intrapatch processes

The source–sink hypothesis for culpeo populations was supported by the comparison between calculated and observed rates of increase, the lack of compensation between natural and hunting mortality, and the dominant dispersal direction from unhunted to hunted ranches. Many recent studies have suggested that source–sink dynamics of animal populations may be induced by hunting that is restricted to certain habitat patches (Doak 1995; Slough & Mowat 1996; Etheridge, Summers & Green 1997; Gaona, Ferreras & Delibes 1998; Hart 2000). To the best of our knowledge, however, our study is the first empirical test of source–sink dynamics where sinks result from hunting by humans.

Our radio-telemetry study was done at only two ranches, so our extrapolation of demographic rates to a mosaic of ranches may be questionable. The survival and dispersal pattern in this study, however, was consistent with density reductions and rapid recovery during the 1989 and 1990 hunting seasons (Novaro et al. 2000). Culpeo densities in 1989 declined by 78% (± 13%) on five hunted ranches, but declined only 8% (± 12%) at two unhunted ranches. Culpeo densities in all the hunted ranches in 1990 recovered to the 1989 pre-hunting levels (Novaro et al. 2000), as they did in HCP in 1994.

Natural mortality of culpeos was low during our study and hunting and natural mortality appeared to be additive. Additive hunting mortality has been reported for other carnivores including lynx Lynx lynx L. (review by Slough & Mowat 1996). High natural mortality of lynx during years of low prey availability may be a condition for compensatory hunting mortality. Low natural mortality of culpeos, perhaps because of high prey availability (Novaro, Funes & Walker 2000), may have precluded compensation of hunting mortality. Additionally, low culpeo mortality on the formerly hunted ranch in 1994–95 suggests that the assumption that prey productivity on sheep and cattle ranches did not affect culpeo survival differently was correct.

Based on estimated dispersal rates, directions, and distances, culpeo dispersal from unhunted ranches could be a mechanism that rebuilds hunted populations. Most culpeos that survived dispersed during their first year of life. Dispersal distances were large in relation to average distances among sources and sinks because of the intermixed spatial arrangement of ranches (Dunning, Danielson & Pulliam 1992). The longest distance to a hunted ranch from the centre of the largest cluster of unhunted ranches was 25 km (see Appendix 1). Thus, the probability of dispersing culpeos reaching and repopulating hunted ranches was high.

A change in sex ratio in addition to source–sink dynamics could allow culpeo populations to withstand hunting. Crespo & de Carlo (1963) estimated a sex ratio biased towards males (41% females; n = 254) in our study area, while our data show a ratio close to unity after a likely increase in hunting during the peak in fur prices of the 1970s and 1980s. A sex-ratio change towards unity was documented for coyote populations subject to intense hunting and results in compensation for high mortality (Knowlton 1972).

According to rates of increase calculated from survival and fecundity, hunted culpeo populations should decline and unhunted populations increase markedly every year. Based on the source–sink model, pronounced local declines and increases do not occur because of constant net dispersal from unhunted to hunted ranches. Watkinson & Sutherland (1995) have argued, however, that populations that would sustain themselves without immigration may appear to have negative rates of increase because dispersers depress fecundity or increase mortality through density dependence. The only ways to identify sinks and sources unequivocally would be to determine the nature of density-dependent processes or to isolate sinks experimentally. Hanski & Simberloff (1997) added that the confounding effect of density dependence can be avoided by measuring rates of increase at low density and in the absence of dispersal.

Dispersal occurred and culpeo population density was high (Novaro et al. 2000) in our study. A significant increase in natural mortality of hunted populations as a result of dispersal was unlikely, because natural mortality was generally low. Reproduction occurs after the hunting season, when densities have been drastically reduced. Thus dispersers into hunted areas are unlikely to reduce resource availability for breeding culpeos and prevent a density-dependent response in reproduction. Culpeo mortality on hunted ranches is simply too high to maintain populations only via local recruitment, regardless of a balanced sex ratio and any density-dependent response in reproduction. Some adult culpeos may survive in hunted ranches for several years and delay local extinction, but eventually these culpeos must be killed by hunters, as evidenced by the complete extirpation of culpeos from sheep ranches that are surrounded by extensive sheep raising areas in other parts of Patagonia (A. Novaro, personal observation). This anecdotal information further suggests that sheep ranches are true sinks.

culpeo population dynamics and landscape changes

The strong impact of culpeo adult survival on population trends and on the proportion of source area required for population stability confirm findings of other simulation studies (Gaona, Ferreras & Delibes 1998; Delibes, Gaona & Ferreras 2001). Theoretical (Howe, Davis & Mosca 1991) and empirical studies (reviewed by Pulliam 1996) have shown that sources can be small compared with sinks in regionally stable populations. Populations of red-winged blackbirds Agelaius phoenicus, L. 1766 (Vierling 2000), for example, can be stable with less than 4% of area in small, isolated sources, and Pulliam's (1996) review indicates that on average 10% of area in sources leads to stable populations. The 30% threshold of source area required to maintain stable culpeo populations may be the result of high mortality in sinks, and also of a high rate of dispersal from sources to sinks, which make a high proportion of the population susceptible to hunting. Finally, the attractive-sink characteristic of sheep ranches may also contribute to the high proportion of source area needed.

Culpeo hunting produces spatial heterogeneity that resembles the attractive-sink scenario of Delibes, Gaona & Ferreras (2001). Sheep ranches in Patagonia are probably attractive to dispersing culpeos because prey availability is high, culpeo density is low after the hunting season, and density of predators, pumas and feral dogs, is often low because of intensive control by sheep ranchers. Conversely, culpeos are probably unable to detect clues from the feature that makes sheep ranches sinks, namely increased mortality from humans. High culpeo densities on hunted ranches at the beginning of the hunting season support the proposition that sheep ranches may be attractive sinks. Furthermore, the curve of the change in culpeo population trend with a declining proportion of unhunted area was similar to Delibes, Gaona & Ferreras’ (2001) curve for attractive sinks and contrasted with curves obtained by Doak (1995) for random dispersal. A small reduction in the proportion of source area would result in a rapid decline in the culpeo regional population. In addition, because the interaction between the proportion of source area and adult culpeo survival is strong, the 30% threshold could shift dramatically if hunting pressure changed.

management implications

The designation of source areas for game may be more effective than quota-based systems for sustainable harvest of populations with source–sink dynamics like the culpeo, particularly in regions where enforcement and monitoring are limited (Joshi & Gadgil 1991; Rowcliffe, Cowlishaw & Long 2003). Game sources may be useful in tropical areas where hunting is probably sustainable because there are extensive unhunted areas that are sources of dispersers (Hart 2000; further review in Novaro, Redford & Bodmer 2000). In the case of the culpeo, permanent refugia could be designated on certain cattle ranches and in more inaccessible, rugged areas. To facilitate repopulation of hunted ranches, key refugia should be established in areas with a high concentration of sheep ranches so that a heterogeneous arrangement of landscape patches is maintained and distances among sources and sinks are minimized. Enforcement should concentrate on enhancing productivity in refugia, for example by limiting poaching and feral dog abundance. Spatial control for culpeos may be applicable only in areas with < 70% of land in sheep ranches and where refugia can be interspersed among the sheep ranches.

Where permanent refugia cannot be implemented, our simulations indicate that the proportion of area on source and sinks should be monitored closely. The mosaic of hunted and unhunted areas in most landscapes is dynamic. In the culpeo case, the source area could fall below the stability threshold if only a few cattle ranches switched to sheep husbandry, which can occur if wool prices increase and/or beef prices decline. Monitoring of source–sink area should be supplemented by landscape-specific simulation of population dynamics and by monitoring population trends. Landscape-specific simulation is necessary to predict population trends because the spatial arrangement and size of hunted patches determines specific threshold levels in the proportion of source and sink areas. Monitoring population trends is particularly important on unhunted sources, because population size may decline first on sources if hunted patches are attractive sinks (Delibes, Gaona & Ferreras 2001).


We thank ranch workers and owners for their co-operation, and O. Monsalvo, C. Rambeaud, E. Donadio, S. Di Martino and G. Sanchez for assistance with field work, and S. Zapata for lab work. K. Redford, L. Branch, C. S. Holling, M. Sunquist, S. Humphrey and two anonymous referees provided valuable comments on the manuscript. Financial support was provided by the Wildlife Conservation Society, the Lincoln Park Zoo, the Argentine Fur Association (CIP), the American Society of Mammalogists and Sigma-Xi. A. J. Novaro was supported by fellowships from Fulbright Commission, PSTC and TCD Programs at University of Florida, Buenos Aires University and Argentine Research Council (CONICET).