Ronald J. Sarno, Laboratory of Genomic Diversity, FCRDC/NCI, Building 560, Room 11–84, Frederick, MD 21702–1201, USA (fax 301 846 1299; e-mail firstname.lastname@example.org).
1. The Chilean National Forestry and Park Service is striving to implement a guanaco management programme of sustained-yield use. To achieve this, the rate, variation and causes of juvenile guanaco mortality must be understood thoroughly. Therefore, we monitored the survival of 409 radio-collared juvenile guanacos in Torres del Paine National Park, Chile, from 1991 to 1996.
2. The Kaplan–Meier product limit estimator of survival for staggered entry was calculated, and survival rates between juvenile males and females and among years were compared using the lifetest procedure in SAS. The Cox proportional hazards model was used to relate mortality rate to explanatory variables such as juvenile sex, birth weight, adult female aggression towards researchers during the capture and tagging of newborns, population density, and mean monthly winter snowfall.
3. Mean juvenile survival rate (Ŝ) was 0·38, but varied between 0·31 and 0·55. Survival rates between the sexes were not significantly different, although male survival was lower than that of females. Mortality rate was highest during the first 14 days after birth. Most deaths occurred between birth and 7 months of age.
4. The risk of mortality increased by almost 6% with every 1 cm increase in winter snowfall, whereas the risk of mortality decreased by almost 24% as adult female aggression increased towards researchers.
5. Current management objectives are aimed at the implementation of a rational harvest of guanacos on the Chilean side of the island of Tierra del Fuego. Our results provide improved and updated estimates of juvenile guanaco survival and will aid in the modelling of harvest rates of guanacos in southern Chile. Future proposed harvests from wild populations in southern Chile need to consider the rate and variation of this critical life-history parameter.
We investigated patterns of juvenile (from birth to age 1 year) guanaco Lama guanicoe (Muller) survival in southern Chile. Guanacos are sexually monomorphic and exhibit a resource defence polygyny mating system (Franklin 1983). The major social units are family groups (one adult male, adult females and young), solitary territorial males (defending a territory but usually without females), male groups (non-territorial males), female groups (adult females with or without young) and mixed groups (individuals of both sexes and all ages) (Franklin 1983; Ortega & Franklin 1995). Adult females are socially and spatially segregated from non-territorial males from September (spring) to March (autumn) (Wilson & Franklin 1985). Parturition occurs in November and early December, in which adult females produce one offspring (Franklin & Johnson 1994) after an 11·5-month gestation period (Franklin 1983).
Although guanacos are protected (Iriarte & Jaksic 1986), they have declined throughout their range due to poaching and agricultural practices (Cunazza 1992a). Despite their threatened status, however, guanacos continue to be an important local and regional economic resource (Franklin & Fritz 1991). Thus, it is a species for which a scientifically based managed harvest could contribute to its conservation (Franklin & Fritz 1991). The Chilean National Forestry and Park Service (CONAF) is currently striving to implement a guanaco management programme of sustained-yield use that is based upon sound and updated studies of population dynamics (Cunazza 1992b). Before the implementation of any rational harvesting programme, however, the rate, variation and causes of juvenile guanaco mortality must first be understood thoroughly.
The objectives of the study were to (1) determine the timing, causes and rate of juvenile guanaco mortality from birth to age 1; (2) determine if mortality rates are different between males and females; (3) compare mortality rates among years; and (4) relate possible explanatory variables to mortality.
The study was conducted from 1991 to 1996 in Torres del Paine National Park (51°3′S, 72°55′W), an International Man and Biosphere reserve located in the eastern foothills of the Andean mountain range of southern Chile. The Park encompasses 2400 km2 and provides almost undisturbed habitat for wildlife. The study area ranged from 200 to 400 m in elevation. The landscape was open with rolling hills, and vegetation was rarely > 1 m high, facilitating easy location of carcasses. Grasses (Festuca gracillana and Anarthrophyllum patagonium) and shrubs (Mulinum spinosum, Senecio patagonicus and Berberis buxifolia) dominated this pre-Andean steppe (Pisano 1974). The climate was generally windy and cool. Yearly mean temperatures ranged from 5·2 °C to 10·3 °C, while annual precipitation varied from 311 mm to 744 mm. The mean winter (June–August) temperature ranged from 0·2 °C to 4·7 °C, and mean monthly winter snowfall varied between 28 cm and 100 cm (data obtained from Park weather stations maintained by Ministerio de Obras Publicas, Punta Arenas, Chile).
We captured newborn chulengos (juvenile guanacos between birth and age 1), using a technique involving detection, approach, and the reuniting of offspring with their mothers (Franklin & Johnson 1994), between 18 November and 10 December each year from 1991 to 1995. Adult female aggression toward researchers was recorded while attempting to capture chulengos. Female aggression was classified as low (ear postures), medium (spitting, vocalizations), and high (kicking, charging and making physical contact with researchers). Chulengos were sexed, weighed and tagged in both ears with individually numbered plastic ear tags. We fitted most chulengos with radio-transmitters mounted on expandable collars (modified from Keister, Trainer & Willis 1988). Transmitters were equipped with motion sensors that were activated after approximately 60 min of inactivity. We tagged and placed collars on 98 chulengos in 1991, 91 in 1992, 94 in 1993, 66 in 1994, and 60 in 1995. Until 1 March of each year, we monitored collared chulengos 6–7 days per week using hand-held antennas and receivers (Advanced Telemetry Systems Inc., Bethel, MN); thereafter, they were monitored 3 days each week for the duration of the year.
When a chulengo died, the carcass was located by following the signal of the activated motion sensor. The probable cause of death was determined based upon field necropsy and site examination. Mortality was classified as puma Puma concolor (Linnaeus) predation when we found large canine punctures on the head and/or neck, fractured and/or chewed portions of the skull, large bones such as the femur and ribs broken in two, and/or carcass remains covered with vegetation. Although there were other potential predators of chulengos in Torres del Paine, such as the culpeo fox Dusicyon culpaeus (Smith) and Geoffroy's cat Felis geoffroyi (d’Orbiginy & Gervais), there was no indication that they preyed on them (Johnson & Franklin 1991, 1994). Disease was assigned as the cause of death upon discovering a prominent sign of infection or congenital defect. Miscellaneous causes of death included broken bones, drowning, entanglement in fences, starvation, and collisions with cars. Mortality was classified as unknown when we could not assign predation, disease or miscellaneous as the probable cause of death. In most cases mortality was detected within 1 day. If the interval was greater than 1 day but ≤ 14 days, the date of death was assigned as the midpoint between the date when we last received a live signal and first received a mortality signal (Pollock, Winterstein & Conroy 1989a).
We used the Cox proportional hazards model (procedure phreg; Allison 1995) to relate failure times (died or censored) to explanatory variables possibly associated with mortality. The model relates the hazard rate, which is equivalent to the instantaneous mortality rate, to explanatory covariates. We screened potential explanatory variables by measuring their correlation with the logarithm of failure times ln(T) before attempting to model the hazard. Based on this screening, chulengo sex, birth weight, adult female aggression, population density, mean winter temperature and mean monthly winter snowfall were considered.
We examined the correlation matrix among possible covariates so that we could reduce multicollinearity. We observed a significant negative correlation between mean winter temperature and mean monthly winter snowfall (R = –0·929, P < 0·001). Because winter snowfall has been demonstrated to lower juvenile survival in various populations of large mammals (Sauer & Boyce 1983; Jedrzejewski et al. 1992; Singer et al. 1997; Smith & Anderson 1998), we removed mean winter temperature as a possible covariate. Additionally, we did not find any references to winter temperature influencing juvenile survival. We then considered a complex model with five covariates and searched for a reduced model based on the smallest values of Akaike's (1973) Information Criterion (AIC = – 2lnL + 2p), which combines the lack of fit measured by the log-likelihood of the model (lnL) with the number of estimated parameters (p).
Survival of 409 radio-collared chulengos was estimated. One radio-collared chulengo was censored in 1992, one in 1993, six in 1994, and two in 1995. Four of the 10 chulengos were censored because of radio failure or collar loss. There were three cases in which we found intact collars with functioning radios in the field, and three instances in which we identified the general location of collars by triangulation, but were unable to retrieve them. We suspect that pumas dragged these chulengo carcasses and radios back to their dens, as they had done on previous occasions.
Mean survival of chulengos was Ŝ365 = 0·38, SE(Ŝ) = 0·02. Overall survival between the sexes was not significantly different (log rank χ2 = 1·84, d.f. = 1, P = 0·175). Although male survival (Ŝ = 0·36, SE(Ŝ) = 0·03, n = 121) was lower than that of females (Ŝ = 0·40, SE(Ŝ) = 0·03, n = (126), there was no consistent trend in each year (sign test P > 0·05; Table 1). There was a significant difference in survival among years (χ2 = 15·69, d.f. = 4, P = 0·004; Fig. 1 and Table 1). Population size and birth weight varied on a yearly basis during the study (Table 2). Puma predation was the leading cause of mortality in all years, followed by unknown, disease, and miscellaneous causes (Table 3).
Table 1. Survival estimates (Ŝ) of juvenile guanacos by year and sex from birth to 1 year of age in Torres del Paine National Park, Chile. n = number of individuals that survival rate was based upon for a given cohort
Table 2. Mean birth weight of juvenile guanacos born in the study area in Torres del Paine National Park, Chile, from 1991 to 1996. The population density (represented in animals km–2) of all guanacos in the study area is also presented for the same period
Population density (animals km–2)
Table 3. Percentages and sample sizes (n) of radio-collared juvenile guanacos dying by various causes between birth and 1 year of age in Torres del Paine National Park, Chile, from 1991 to 1996
Miscellaneous causes included broken bones (3), drowning (1), entangled in fence (4), struck by car (1) and starvation (5).
In each year, mortality was highest during the first 14 days after birth (Fig. 1). Twenty-three per cent of all radio-collared chulengos died during this period. Mortality appeared to occur at a more constant rate throughout the year for the 1992 and 1993 cohorts, while it increased between 200 and 220 days (early winter) after birth for the 1991 and 1994 cohorts. The 1995 cohort experienced the highest rate of mortality early in life. Virtually all of the deaths occurred between birth and 120 days of age.
We present a table of the fitted models along with their AIC values (Table 4). We rejected the global hypothesis of the model with five covariates (chulengo sex, birth weight, adult female aggression, population size, and mean winter precipitation) that all coefficients are zero. Only adult female aggression and mean monthly winter snowfall were significantly related to survival (Table 5). We next considered a reduced model with these two parameters, and estimated their βs along with their SEs and risk ratios (Table 5). The risk ratio (eβ) relates a unit increase in the covariate to an increase or decrease in the hazard function or instantaneous mortality rate. The parameter estimates for female aggression and mean monthly winter snowfall were slightly different than the estimates from the full model (Table 5). We chose the reduced model because the models with the lowest AIC value were the most parsimonious (Table 4).
Table 4. Model selection of survival data for juvenile guanacos in Torres del Paine National Park, Chile from 1991 to 1996. The reduced model was selected on the basis of the minimum Akaike's Information Criterion (AIC)
Log likelihood ratio value
Table 5. Parameters of a proportional hazards model relating covariates to juvenile guanaco survival from birth to 1 year of age in Torres del Paine National Park, Chile, from 1991 to 1996
Female aggression towards taggers
Mean monthly winter precipitation
Female aggression towards taggers
Mean monthly winter precipitation
The overall proportion of censored chulengos during the study period was 4%, well within the established level of robustness for the Kaplan–Meier method (Pollock et al. 1989b). Our annual estimates of juvenile guanaco survival are comparable with those reported for similar-sized northern temperate ungulates from North America, Europe and Africa (Linnell, Aanes & Andersen 1995 and references therein).
Our results do not support sex-differential juvenile mortality (sign test P > 0·05), although Behl (1992) reported that annual survival of male chulengos was consistently lower than that of females by approximately 10% (P = 0·086) in this same population. Sex-differential juvenile mortality in mammals has been observed principally in dimorphic species (Ralls, Brownwell & Ballou 1980; Clutton-Brock 1991), in which juvenile males apparently suffer increased mortality as a result of greater susceptibility to food shortage resulting from faster growth rates and greater nutritional requirements (Clutton-Brock, Albon & Guinness 1985). Equal birth weights (Franklin & Johnson 1994) and growth rates (Sarno & Franklin 1999) of male and female chulengos may explain similar survival rates of males and females.
Mean yearly survival rate of chulengos during this study (38%) was lower than those based on life-table analyses in the early 1980s (Franklin & Fritz 1991; Fritz & Franklin 1994) and mark–resighting studies in the late 1980s (Behl 1992), when mean yearly survival was 70%. What are the potential factors that have led to decreased chulengo survival over the past 10 years?
Our data suggest that puma density has increased in Torres del Paine since the Park was established in 1975 (Franklin et al. 1999). In the 1970s observations of pumas in the Park were rare. Pumas were occasionally seen in the early 1980s, but sightings increased dramatically in the mid- to late 1980s. This increase continued into the early 1990s, with three to 14 sightings per year. In the mid-1990s, puma sightings numbered 20–50 per year, with as many as nine individuals sighted in 1 day (Franklin et al. 1999; H. Miles, personal communication). In 1995, based upon sightings, natural markings and tracks, we estimated a minimum of 12 pumas in the 46-km–2 peninsula study area, or approximately 26 pumas 100 km–2 (Bank & Franklin 1998).
There is also strong evidence that puma predation on guanacos has increased in response to the growth in population size of guanacos. From 1975 to 1988 the number of guanacos increased 13-fold from 97 to 1276 animals. As the number of guanacos increased, pumas fed upon them more often. Between 1982–83 and 1987–88 the frequency of guanaco remains in puma scats increased from 9% to 29% (Iriarte, Johnson & Franklin 1991). Predation is cited as the leading cause of juvenile ungulate mortality in sites where predators occur (Linnell, Aanes & Andersen 1995 and references therein).
Variation in chulengo survival appears to be related to monthly winter snowfall. Although puma predation was the most important proximate factor influencing winter mortality (as during the rest of the year), climatic conditions could have directly or indirectly predisposed chulengos to puma predation. The risk ratio for mean monthly winter snowfall indicates that the risk of mortality increased by 5·7% for each centimetre of snowfall. Jedrzejewski et al. (1992) showed that predation was the leading cause of death for red deer Cervus elaphus (Linnaeus) and roe deer Capreolus capreolus (Linnaeus) during winter, while snow cover, which led to malnutrition and/or disease, was the leading cause of mortality for wild boar Sus scrofa (Linnaeus) during the same period at the same site. He made no mention, however, of the possible interaction between snow cover, malnutrition and predation. Deep snow probably makes it difficult for chulengos to feed, while cold temperatures may serve to weaken malnourished animals by causing an increase in metabolism to keep warm, while using up dwindling energy reserves.
Chulengos may have increased their susceptibility to puma predation by exhibiting risk-prone behaviour associated with foraging decisions. McNamara & Houston (1987) proposed that food-stressed individuals should accept higher predation risk to prevent starvation. Guanacos migrated to the extreme western portion of the study area in snowy winters (Ortega & Franklin 1995). This sector was characterized by higher vegetation and steeper and more rugged terrain, which may have enabled hunting pumas to approach guanacos more closely before detection (Bank & Franklin 1998).
Increasing levels of maternal defence toward researchers during capture of newborns increased offspring survival. The risk to mortality decreased by almost 24%. This behaviour may ultimately reflect a mother's ability to protect and defend her chulengo during its first year of life. More aggressive females may have been more vigilant while scanning for predators, and also may have protected their offspring in dangerous situations. Female age also may have influenced aggression, but there are insufficient data to test this hypothesis rigorously.
The World Conservation Strategy of the International Union of Nature and Natural Resources and the World Wildlife Fund both advocate the managed harvesting of wild animals if economic incentives are likely to be the most effective motivators for species conservation (IUCN 1981). Systematic harvesting of wildlife for meat and skins has been profitable in Zimbabwe (Hanks 1981), South Africa (Joubert 1985) and Latin America (Robinson & Redford 1991). Scientifically based rational harvest programmes hold high economic potential if based upon sound population parameters.
Current management objectives on the Chilean side of the island of Tierra del Fuego are aimed at the implementation of a rational harvest of guanacos. Although there are no pumas on Tierra del Fuego, guanaco populations on the island and in Torres del Paine exhibit remarkably similar survival rates (Franklin & Fritz 1991).
Franklin & Fritz (1991) and Fritz & Franklin (1994) originally proposed that 30–40% of males from male groups could be harvested without adversely disrupting reproduction and population size, but stated that this could occur only when estimates of juvenile survival and female natality were improved. We believe that the harvesting of males from male groups is feasible, but that harvest rates should be lower than was originally estimated to compensate for lower chulengo survival.
Harvesting guanacos unselectively for sustainable yield would be imprudent because of the rigid separation of the majority of adult males and females. Removal of adults from family groups could severely impede reproduction and might result in lower chulengo survival by increasing the proportion of orphaned animals (Franklin & Fritz 1991).
Although chulengo pelts have traditionally been the most coveted guanaco product (Franklin & Fritz 1991), we urge extreme caution in harvesting them because original harvest estimates (20%) were based upon a 70% chulengo survival rate (Franklin & Fritz 1991). As chulengo survival varies widely and can be as low as 31%, the proportion of harvested chulengos would need to be lowered substantially.
In conclusion, we believe that if the harvesting of guanacos is going to occur in southern Chile, adult males from male groups could be harvested without adversely affecting population size if juvenile mortality is considered carefully. The harvest of juveniles is more problematic for ethical reasons and because of the sensitivity of guanaco populations to juvenile mortality. Any harvesting of juveniles should be done sparingly, and with close monitoring of annual survival rates.
We thank the Chilean National Forestry and Park Service (CONAF) and the administration at Torres del Paine National Park for their assistance and collaboration, particularly G. Santana, H. Gonzales, J. Romero, J. Toro, J. Vargas, H. Yeager, L. Mercado, and F. Aro. A. Anderson, C. Bergman, C. Boggon, C. Caceres, T. Chladny, J. Cleckler, A. Engh, L. Farrell, E. Gaylord, K. Gaylord, K., Guderian, P. Heaven, M. Lederbuhr, W. Loya, K. Nielsen, I. O’Conell, J. Rathje, J. Reed, S. Shoemaker, C. Solek, B. Soppe, K. Stueckrath, T. Sulser, N. Varley and D. Whitaker, provided field assistance. M. Festa-Bianchet, J. M. Gaillard, B. Jedrzejewska and an anonymous referee provided many helpful comments on an earlier version of this manuscript. This study was supported by Patagonia Research Expeditions, National Science Foundation Grant No. BSR-9112826, and Organization of American States Grant no. 19104.
Received 17 August 1998; revision received 22 July 1999