## Mastozoología neotropical

##
*On-line version* ISSN 1666-0536

### Mastozool. neotrop. vol.15 no.1 Mendoza Jan./June 2008

**Comparison of two methods for estimation of abundance of mountain vizcachas ( Lagidium viscacia) based on direct counts**

**R. Susan Walker ^{1}, Andrés J. Novaro^{1,2}, and Pablo G. Perovic^{3} **

^{1} Wildlife Conservation Society, Centro de Ecología Aplicada del Neuquén, Curruhue y Río Chimehuín, (8371) Junín de los Andes, Neuquén, Argentina <swalker@wcs.org>.

^{2} CONICET, Centro de Ecología Aplicada del Neuquén, Curruhue y Río Chimehuín, (8371) Junín de los Andes, Neuquén, Argentina <anovaro@wcs.org>.

^{3} Museo de Ciencias Naturales, Universidad Nacional de Salta, Mendoza 2, (4400) Salta, Argentina <pgperovic@gmail.com>.

**ABSTRACT:** Mountain vizcachas (genus *Lagidium*) are large, rock-dwelling rodents of the Andes and the Patagonian steppe. We tested the bounded count and double observer methods for estimating mountain vizcacha population size. Estimates obtained with these two count-based methods and a Lincoln-Petersen mark-resight estimate were similar. However, estimates obtained with the double observer method were the most precise, and the method is easy to use and practical. We recommend the use of the double observer method because it incorporates an estimate of detection probability that reduces bias and allows comparisons between sites and across time.

**RESUMEN**: **Comparación de dos métodos para estimación de la abundancia de la vizcacha de la sierra ( Lagidium viscacia) en base a conteos directos.** Las vizcachas de la sierra (género

*Lagidium*) son roedores grandes de hábitats rocosos de los Andes y la estepa patagónica. Pusimos a prueba los métodos de conteo acotado y doble observador para estimar el tamaño poblacional de esta especie. Estimadores obtenidos con los dos métodos fueron similares al estimador de marcado y avistaje de Lincoln-Petersen. Sin embargo, los estimadores de doble observador fueron más precisos y el método fue más fácil de usar. Recomendamos el uso del método de doble observador porque incorpora un estimador de probabilidad de detección que reduce el sesgo y permite comparaciones espaciales y temporales.

**Key words.** Bounded counts; Detection probability; Double observer; Mark-resight; Population estimation.

**Palabras clave.** Conteos acotados; Doble observador; Estimación poblacional; Marcado-avistaje; Probabilidad de detección.

Mountain vizcachas, genus *Lagidium*, are three or four species (Redford and Eisenberg, 1992; Pearson, 1995; Spotorno et al., 2004) of large (1.5-3 kg), rock-dwelling rodents of Andean and Patagonian South America. Populations of mountain vizcachas are declining in parts of northern Patagonia (Walker et al., 2000a), and for most other regions, data on population trends are unavailable. The status of *L. viscacia* is considered unknown due to lack of data (Amori, 2000), and *L. wolfsohnii *is considered near threatened because of its restricted distribution (Contreras, 1996). Andean populations of the more widespread *L. viscacia* are the principal prey of the endangered Andean cat, *Leopardus jacobita* (Walker et al., 2007). Valid methods for estimating size or relative abundance of mountain vizcacha populations are necessary to monitor their conservation status and trends, and to evaluate prey availability for their predators.

Population estimations are often based on incomplete counts of individuals (Williams et al., 2002). To make comparisons of population sizes across time and space with incomplete counts, we must either assume that the probability of detection of individuals is constant among populations, which generally is not true, or incorporate ancillary information into the counts that can be used to calculate detection probability for each population sampled. The most common methods for incorporating detection probability into counts are distance sampling, mark-recapture or markresight methods, double sampling, multiple observers, and specific detection probability models for the species.

Mountain vizcachas are restricted to rock patches, the size and structure of which vary greatly. In many areas, mountain vizcachas are easy to observe during the day, but detectability may be affected by habitat structure, weather, human activities, and other, unknown factors. The patchy spatial distribution of mountain vizcachas and the structure of their habitat are not amenable to distance sampling. Capturing and marking mountain vizcachas to use mark-resight methods is feasible in some, but not all populations (Walker et al., 2000a).

We present a preliminary evaluation of two methods for estimation of abundance of mountain vizcachas based on direct counts of individuals- the bounded count (Robson and Whitlock, 1964) and double observer methods (Lancia et al., 1994). The bounded count method has been used with rock hyraxes, *Procavia capensis*, in Africa (Fairall and Crawford, 1981), which like mountain vizcachas are colonial, observable during the day, and occupy discrete patches of rocky habitat. The only assumption required is that there is some probability (greater than 0) that all animals may be seen during any one count. Population size is estimated as = 2n_{max} - n_{max} _{- 1}, where n_{max} is the largest count and n_{max} _{- 1} is the next largest count. The lower confidence limit is the largest number counted- the minimum number known to be present. The approximate upper confidence limit N_{u} is given as N_{u} = n_{max} + (1-a/a)(n_{max} - n_{max} _{- 1}), where a is the significance level. This method does not take detection probability into account, and without simultaneously collecting information on detection probability or calibrating bounded count estimates to known population sizes, it cannot be used to make comparisons among populations (Williams et al., 2002).

The double observer method (Lancia et al., 1994) requires that counts of individuals be carried out simultaneously and independently by two different observers. Detection probability and population size can be calculated using the Lincoln-Petersen estimator, with the individuals counted by the first observer considered to be the first sample (*n*_{1}), those counted by the second observer considered to be the second sample (*n*_{2}), and those seen by both observers considered to be the marked individuals (*m*).

We estimated population size of mountain vizcachas (*L. viscacia*) at two rock outcrops- Peña Baja and Peñon Solitario-in altiplano habitat at 4600 m in Vilama, Jujuy Province, Argentina (22.63o S and 66.97o W), and one at Puesto La Agüita, in steppe habitat at 1700 m, near the Payunia Provincial Reserve in Mendoza (35.98o S and 69.10o W). Peña Baja was a low, vertical cliff, 3-5 m high and 400 m long, and Peñón Solitario was a high outcrop, 50 meters high and 200 m long. The outcrop at La Agüita ran along both sides of a small stream, and was approximately 10 to 15 m high and 200 m long.

We used the bounded count method to estimate population size at all three sites. Observations were done at various times of day between 08:00 and 19:00, by one or more of the authors, over a two-week period, and with consecutive observations lasting approximately 30 minutes separated by at least five hours. Nine counts were conducted at Peña Baja, 12 at Peñón Solitario, and six at La Agüita.

Because there is no method for independently estimating detection probability for bounded count estimations, we used the average proportion of the estimate that was seen on each occasion ( = n_{i} / , where n_{i}= the number seen on occasion i, and = the estimated population size) to estimate detection probabilities. These estimated detection probabilities were biased to the unknown degree to which each of the bounded count estimates were biased, but can provide an indication of the relative detection probabilities among the sites.

Numbers of mountain vizcachas observed at each site varied greatly from count to count, estimated detection probability varied among sites, and 95% confidence intervals obtained with the bounded count estimate were wide (**Table 1**). With the bounded count method, the greater the difference between the highest and second-highest counts, the lower the precision. Increasing the number of counts may increase the precision of the estimate, by increasing the likelihood of obtaining two similar, high counts. The precision of the estimate was highest at Peñón Solitario, as reflected by the lower 95% upper confidence limit, than at the other sites, perhaps due to a higher detection probability and greater number of counts than at the other sites, and lowest at La Agüita, where the fewest counts were carried out.

**Table 1**

Numbers of mountain vizcachas counted on different occasions at three sites, their means and standard deviations, bounded count estimates of population size with 95% and 90% Upper Confidence Limits (U.C.L.), and estimated mean detection probability.

To compare bounded count estimates to Lincoln-Petersen estimates based on resightings of marked animals, we captured and marked mountain vizcachas at Peña Baja (Williams et al., 2002). Tomahawk traps, baited with oats *Avena* spp. rolled in butter, were placed at the base of the cliff for three days. Four mountain vizcachas were captured and sedated with tiletamine/zolazepam at 15 mg/kg via intramuscular injection. Vizcachas were marked with colored plastic disks on each ear with rabbit ear tags, and released at the capture site after recovery. Three days after the last capture, two marked animals were resighted in a total count of five individuals. We used the unbiased Lincoln-Petersen estimator (Chapman, 1951) to calculate a population size of nine, and Seber's (1982) variance estimator to determine 95% confidence limits of five to 13 (Lancia et al., 1994), with a detection probability of 0.50. Traps were also placed at Peñón Solitario for three days, but no mountain vizcachas were caught there, probably because they were being hunted during the same period by local boys and their dogs.

We used the double observer method for estimating population size of mountain vizcachas on two occasions at La Agüita. One observer walked slowly along the top of the cliff on one side of the stream, and the other walked at the bottom of the cliff along the opposite side of the stream. A third person followed the two observers and recorded their observations, which were transmitted by handheld radios. This person recorded how many mountain vizcachas were seen by each observer and which were seen by both or only one of the two observers. Using one of the observer's counts as the primary sample, and the other observer's counts as the secondary, we considered every animal seen by both observers to be a re-sighting of a marked individual. Based on these data we used the unbiased Lincoln-Petersen estimator (Chapman, 1951) to estimate population size, and Seber's (1982) variance estimator to determine 95% confidence limits for the estimate (Lancia et al., 1994).

Estimates of population size based on the double observer method were similar on the two occasions (**Table 2**). Detection probability was 0.5 for both occasions, similar to the estimated detection probability based on the bounded count method. The double observer estimates were higher than the bounded count estimate for La Agüita, and the confidence intervals were much smaller, indicating greater precision.

**Table 2**

Numbers of mountain vizcachas counted by two observers on two occasions at La Agüita, Argentina, the number of mountain vizcachas counted by both observers, estimated population size, and 95% lower and upper confidence limits (L.C.L. and U.C.L.) of the estimates.

Although based on few repetitions of these methods, we can draw some useful conclusions from our results. Population-size estimates obtained with the bounded count method at Peña Baja and La Agüita were within 95% confidence intervals of estimates obtained with both the mark-resight and double observer methods, which incorporate detection probabilities. This indicates that bounded counts may provide a valid, though less precise, alternative for estimating population size of mountain vizcachas, when neither captures nor the double observer method are possible. Nevertheless, bounded counts require more time than the double observer method, and cannot be used to compare population sizes across time and space. Both the bounded count and mark-resight methods require a minimum of one to two weeks at each site, but the bounded count method is non-invasive, can be done at more than one site each day, and uses an empirical lower confidence limit.

The double observer method appears to be the most precise and easiest method to use for estimating abundance of mountain vizcachas among the methods we tried. Both the markresight and the double observer methods should provide accurate estimates that are comparable across time and space. However, the markresight method is more time-consuming and disruptive to mountain vizcacha populations. Capture success may be so low at some sites and times that the method is impossible to use due to lack of captures, as happened at Peñón Solitario.

The double observer method has been criticized for being biased in bird counts because the sampling frame may not completely cover the population of interest, and not all birds may be in the study area during the counts (Bart et al., 2004). Because mountain vizcachas do not venture far from their rocky shelters (Walker et al., 2000b), they are not likely to be out of the study area during counts. The greatest potential source of bias for both direct count methods used here is probably the lack of complete coverage of the population due to habitat structure. Many rock outcrops and cliffs have portions that are inaccessible or not visible from an accessible place, and mountain vizcachas using those areas would not be detected. Visibility of mountain vizcachas for observers should be evaluated at each rock outcrop before attempting counts, as accurate estimates of population size are not possible if large portions of an outcrop are not visible to either observer.

Our results confirm that a single count of mountain vizcachas at a site cannot be used as an index of relative abundance, as has been done for this species (Lucherini et al., 1999). The detection probability at each site varied from day to day according to unknown variables, probably including weather conditions and recent human activities. The occasion on which the count at La Agüita was 0 was in the afternoon of a hot summer day. In addition, detection probabilities varied greatly among the sites, perhaps due to some extent to habitat structure. For example, a larger proportion of the population was observed with greater frequency at Peñón Solitario than at Peña Baja, within the same area. At the high Peñón Solitario cliff, mountain vizcachas could safely expose themselves on rocky shelves out of reach of predators. Because Peña Baja was so low, there were few, if any such unreachable places where it was safe for a mountain vizcacha to sit in plain sight.

Both bounded and double observer counts will be less biased when care is taken to maximize detection probability. This means that counts should be done at the time of day when mountain vizcachas are most likely to be out of their dens and active (which may vary according to the site and time of year), and observers must walk slowly, carefully searching for mountain vizcachas. For the double observer method, counts of the two observers must be independent, so observers should not communicate with each other about observations. In some areas, at some times of year, mountain vizcachas are nocturnal. In that case, counts would have to be done with a spotlight, and the methods would need to be tested under those conditions. It is probably not possible to use the double observer method with spotlights because the spotlight may direct one observer's attention to animals detected by the other observer.

These count-based methods for estimation of population size of mountain vizcachas are likely to be feasible for many habitat types, perhaps with some local adaptations, and also useful for other species that are easy to observe. Although the double observer method can be done with a single count, it may be prudent to repeat counts, making adjustments to methodology to maximize detection probability. In areas where the double observer method is not possible, because habitat structure does not allow survey by two independent observers, it may be possible to use the bounded count method. Based on our results, we recommend repeating the count at least 10 times for more precise results and to reduce the bias that may be introduced by weather and other factors that affect detectability. Nevertheless, both of these count methods should be useful for surveys, as counts at several outcrops may be done in one day.

**ACKNOWLEDGEMENTS**

This research was supported by the Wildlife Conservation Society. A vehicle for work in Vilama was provided by the Instituto de Biología de Altura (INBIAL), Universidad de Jujuy. We thank the Sagal family of Kiñe for their hospitality at La Agüita. Roxana Sagal and Jimena Novaro assisted in the fieldwork at La Agüita.

**LITERATURE CITED**

1. AMORI G. 2000. *Lagidium viscacia*. In: IUCN 2004. 2004 IUCN Red List of Threatened Species. http://www.redlist.org. [ Links ]

2. BART J, S DROEGE, P GEISSLER, B PETERJOHN, and CJ RALPH. 2004. Density estimation in wildlife surveys. Wildlife Society Bulletin 12:1242-1247. [ Links ]

3. CHAPMAN DG. 1951. Some properties of the hypergeometric distribution with applications to zoological sample censuses. University of California Publications in Statistics 1:131-160. [ Links ]

4. CONTRERAS L. 1996. *Lagidium wolffsohni*. In: IUCN 2004. 2004 IUCN Red List of Threatened Species. http://www.redlist.org. [ Links ]

5. FAIRALL N and RJM CRAWFORD. 1981. Application of the Robson-Whitlock technique to estimate hyrax Procavia capensis numbers. South African Journal of Wildlife Research 13(1):25-26. [ Links ]

6. LANCIA RA, JD NICHOLS, and KH POLLOCK. 1994. Estimating the number of animals in wildlife populations. Pp. 215-253, in: Research and management techniques for wildlife and habitats (TA Bookout, ed.). Fifth ed. The Wildlife Society, Bethesda, MD. [ Links ]

7. LUCHERINI M, D SANA, and D BIROCHIO. 1999. The Andean mountain cat (*Oreailurus jacobita*) and the other wild carnivores in the proposed Anconquija National Park, Argentina. Società Zoologica "La Torbiera" Scientific Reports, 5: 3-31. [ Links ]

8. PEARSON OP. 1995. Annotated keys for identifying small mammals living in or near Nahuel Huapi National Park or Lanin National Park, southern Argentina. Mastozoologia Neotropical 2:99-148. [ Links ]

9. REDFORD K and JE EISENBERG. 1992. Mammals of the Neotropics: the Southern Cone. Vol. 2. University of Chicago Press, Chicago, IL. 430 pp. [ Links ]

10. ROBSON DS and JH WHITLOCK. 1964. Estimation of a truncation point. Biometrika 51:33-39. [ Links ]

11. SEBER GAF. 1982. The estimation of animal abundance and related parameters. Second ed. Macmillan Publ. Co., Inc., New York, N.Y. [ Links ]

12. SPOTORNO AE, JP VALLADARES, JC MARIN, RE PALMA, and C ZULETA R. 2004. Molecular divergence and phylogenetic relationships of Chinchillids (Rodentia: Chinchillidae). Journal of Mammalogy 85:384-388. [ Links ]

13. WALKER RS, V PANCOTTO, J SCHACHTERBROIDE, G ACKERMANN, and AJ NOVARO. 2000a. Evaluation of a fecal-pellet index of abundance for mountain vizcachas (*Lagidium viscacia*) in Patagonia. Mastozoología Neotropical 7:89-94. [ Links ]

14. WALKER RS, G ACKERMANN, J SCHACHTERBROIDE, V PANCOTTO, and AJ NOVARO. 2000b. Habitat use by the mountain vizcacha (*Lagidium viscacia* Molina) in the Patagonian steppe. Zeitschrift für Säugetierkunde 65: 293-300. [ Links ]

15. WALKER RS, AJ NOVARO, P PEROVIC, R PALACIOS, E DONADIO, M LUCHERINI, M PIA, and MS LÓPEZ. 2007. Diets of three species of Andean carnivores in high-altitude deserts of Argentina. Journal of Mammalogy 88:519-525. [ Links ]

16. WILLIAMS BK, JD NICHOLS, and MJ CONROY. 2002. Analysis and management of animal populations: Modeling, estimation, and decision making. Academic Press, San Diego, CA, USA. [ Links ]

**Recibido** 2 agosto 2006.

**Aceptación final** 15 julio 2007.