Chapter 7
POPULATION GENETICS AND SPATIAL STRUCTURE IN TWO ANDEAN CATS (THE PAMPAS CAT, LEOPARDUS PAJEROS AND THE ANDEAN MOUNTAIN CAT, L. JACOBITA) BY MEANS OF NUCLEAR AND MITOCHONDRIAL MARKERS AND SOME NOTES ON SKULL BIOMETRICS
Manuel Ruiz-García1,, Daniel Cossíos2, Mauro Lucherini3, José Yáñez4, Myreya Pinedo-Castro1 and Bernard Angers2,5 1Laboratorio de Genética de Poblaciones Molecular-Biología Evolutiva, Unidad de Genética, Departamento de Biología, Facultad de Ciencias, Pontificia Universidad Javeriana, Bogotá DC, Colombia 2Department of Genetics and Evolution, University of Geneva, Switzerland 3GECM, Universidad Nacional del Sur-CONICET, Bahia Blanca, Argentina 4Museo Nacional de Historia Natural, Santiago, Chile 5Departement de Sciences Biologiques, Université de Montreal, Montreal, Canada
ABSTRACT
In this chapter, we show complementary results to the works of Cossíos et al., (2009, 2012), on the genetic structure and phylogenetics of two small Andean cats, the Pampas cat (Leopardus pajeros) and the Andean mountain cat (Leopardus jacobita). In the present study we increased the samples sizes to 235 individuals for L. pajeros and 115 individuals for L. jacobita, effectively making these samples the largest to date for these two species. We analyzed five microsatellites for L. pajeros and seven microsatellites for
For correspondence: [email protected], [email protected]. 2 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
L. jacobita as well as the hypervariable domain 1 (HVS-1) of the mtDNA control region for both species. The main results obtained were as follows: 1- The levels of gene diversity for L. pajeros with microsatellites were considerable higher than in L. jacobita (average H = 0.73 vs. 0.42, respectively), with the first similar to other Neotropical felids but the second one lower than other Neotropical felids and many other Neotropical mammals analyzed from this point of view. The same was recorded for mtDNA sequences, with the Pampas cat ( = 0.0513) presenting more than 10 times higher nucleotide diversity than the Andean mountain cat ( = 0.0047). The sample which could represent the putative morphological subspecies, L. p. budini, was that which yielded the highest levels of gene diversity. This could mean that this is the original L. pajeros form from which the other forms derived. Alternatively, the northern area of Argentina, where L. p. budini occurs, could be a hybridization zone among several L. pajeros forms. 2- Microsatellite heterogeneity for the Pampas cat was significant but it was relatively low with regard to the high genetic heterogeneity found for L. jacobita for microsatellites. For mtDNA, the genetic heterogeneity was very high and similar for both species. This could indicate that for the Pampas cat the gene flow is male biased, meanwhile the Andean mountain cat populations are hardly isolated in the high land deserts of the Andes and the gene flow is more restricted for both males and females. Also this analysis puts in doubt that L. pajeros pajeros and L. pajeros crucinus are two different subspecies. Furthermore, this analysis revealed that if the different gene pools determined in L. pajeros are classified as different subspecies, then four different subspecies, or at least, four different evolutionary lineages must be consider in L. jacobita. 3- The assignation analyses presented relatively low percentages of correct assignation for L. pajeros, while the percentages of adequate assignation for L. jacobita were very high. This is related with the fact that gene flow estimates among the populations of the Pampas cat are considerably higher than for the populations of the Andean mountain cat for nuclear markers. 4- L. pajeros presented more evidence of population expansions during its history for microsatellites than did L. jacobita. For mtDNA, both species did not reveal traces of population expansions and L. jacobita showed a trend indicative of a moderate bottleneck. 5- Both species showed 4-5 % of mutations with multiple steps and different mutation rates for the microsatellites employed. 6- The effective number estimates were around 10 times higher for L. pajeros than for L. jacobita independently of the procedures employed. The effective sizes for L. pajeros ranged from 80,000 to 330,000 and for L. jacobita ranged from 12,000 to 38,000. However, these estimates seem to be higher than the current census sizes. The procedures of Hill (1981) and Pudovkin et al., (1996) were not useful for effective number estimations in this case. 7- Both species presented significant spatial structure related with isolation by distance and monotonic clinal trends, but this spatial structure was more developed in L. jacobita. Around 35 % of the genetic differences were explained by the geographical distances among the populations in L. pajeros, while around 64 % of the genetic differences were explained by geographical distances in L. jacobita. 8- The northern Chilean Pampas cat population seems to be an extension of the Peruvian and north Bolivian L. p. garleppi in contradiction with García-Perea (1994), who denominated that population as a new subspecies L. colocolo wolffsohni. Nevertheless, more samples of that region are needed to have total clarity of what Pampas cat is living there. Additionally, in Bolivia, we determined the existence, at least, of two putative subspecies (garleppi and steinbachi). Finally, although molecular conclusive studies are needed, the first molecular studies indicate that the existence of a unique Pampas cat species is more probably than three different species such as García-Perea (1994) proposed.
Population Genetics and Spatial Structure in Two Andean Cats … 3
Keywords: Pampas cat, Andean Mountain cat, Lynchailurus, Oreailurus, Leopardus pajeros, Leopardus jacobita, DNA microsatellites, mDNA control region, spatial structure, genetic structure, phylogenies, skull biometrics
INTRODUCTION
Two small and elusive cats are found in the major part of the Andes cordillera. One of them is the Pampas cat (Leopardus pajeros) and the other is the Andean mountain cat (Leopardus jacobita). These cats live in high-altitude deserts of the Andes, although the Pampas cat also lives in open grassland, evergreen forests, dense shrub lands and humid forests. However, although they frequently live in sympatry, Lucherini et al., (2009) demonstrated that the Pampas cat showed the greatest proportion of nocturnal activity, while the Andean cat was more diurnal. Such a separation of activities supports the temporal niche segregation hypothesis between these two Andean felids. Aditionally, Napolitano et al., (2008) showed, in Chile, low levels of prey partitioning between both cat species because there was a wide overlap in diet composition (82 %), with the mountain viscacha (Lagidium viscacia) being the most important prey species for both (93.9 % for the Andean cat and 74.8 % for the Pampas cat, respectively). They also determined that the Andean cat’s scats increased with altitude and slope of the mountains, but there was a substantial geographic overlap in the distribution of both felines. The Pampas cat was included within the Lynchailurus genus (-Severtzov, 1858-) and it is distributed from Ecuador to southern Patagonia in Argentina and Chile. This species seems to be closely related to Leopardus tigrinus by means of the mtDNA genes 16SrRNA, ATP8 and NADH5 within the ocelot linage (Johnson et al., 1998, 1999). How many species or subspecies are included within Lynchailurus is a matter of debate. For instance, Cabrera and Yepes (1940) speculated about the possible existence of two species, the pajeros cat (Lynchailurus pajeros), basically distributed throughout Argentina, and another the “kudmu” or Molina colocolo cat (Lynchailurus colocolo), from Chile and northern Argentina, throughout Matto-Grosso, and Peru up until Ecuador. However, these authors also speculated with the possibility that they were unique species and that the study of the exemplars from Catamarca and northern Argentina could be determinant in demonstrating that they are effectively unique species. More recently, García-Perea (1994) proposed the existence of three species within Lynchailurus using meticulous skull morphology, biometrics and skin analyses: 1) L. pajeros- Desmarest 1816-, which occurs in the Andes high elevation steppes of Ecuador, Peru, Bolivia, Argentina (eastern slope of the Andes) to southern Patagonia in Chile and Argentina in lowland levels, shrub land and dry forests. The type locality was based on Azara’s account in Buenos Aires province between 35-36°; 2) L. braccatus (Cope 1889) found in southern and south western Brazil, Paraguay and Uruguay in humid and warm grassland and forested areas. The type locality is the state of Rio Grande do Sul. Allen (1919) restricted it to Chapada at Matto Grosso in Brazil. 3) L. colocolo (Molina 1782) was found at middle elevations in central Chile (with the type locality in Valparaiso province) and in high elevation steppes in northern Chile at the Andes western slopes. 4 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Within each one of these species, García-Perea (1994) determined several subspecies. L. pajeros has seven subspecies (1a through 1f ): 1a) L. pajeros thomasi -Lönnberg 1913- has a type locality “near Quito” and has had its samples obtained from the Napo and Pichincha provinces in Ecuador. 1b) L. p. garleppi- Matschie 1912- has a type locality from Cuzco and the Apurimac area in southern Peru with a distribution in highland steppes in the eastern Peruvian Andes. 1c) L. p. steinbachi- Pocock 1941- with type locality in Tiraque, Cochabamba with distribution in the highlands steppes of eastern Bolivian Andes. In this country, this species is found in the Departments of Cochabamba, Beni, La Paz, Pando, Santa Cruz, Chuquisaca y Tarija including the National Reserves of Manuripi-Heath and in Ulla Ulla. 1d) L. p. budini- Pocock 1941-with the type locality in Mount Sola in Salta, northwestern Argentina; 1e) L. p. crespoi- Cabrera 1957- with the type locality in Aguaray and also in Salta within northwestern Argentina; 1f) L. p. pajeros- Desmarest 1816- (with the type locality is the Azara’s account of the species) with distribution in La Pampa province in central Argentina and 1g) L. p. crucinus- Thomas 1901- with the type locality in Santa Cruz in Argentina and distribution in the southern half of Argentina and in the Chilean Patagonia. L. braccatus has 2 listed subspecies (2a and 2b): 2a) L. braccatus braccattus- Cope 1889- has a distribution in Matto Grosso and Matto Grosso do Sul in Brazil and in Paraguay; 2b) L. b. munoai- Ximenez 1961- with the type locality in Arroyo Perdido, Department of Soriano, Uruguay and a distribution in Rio Grande do Sul in Brazil and in Uruguay. L. colocolo is listed as having two subspecies: 3a) L. colocolo colocolo- Molina 1782- with the type locality in Valparaiso province and a distribution from Coquimbo to Concepción in central Chile; 3b) L. c. wolffsohni- García-Perea 1994- with type locality in Rio Camarones, Tarapacá province in the western Andes slopes in northern Chile. The geographic distribution of the Pampas cat could reach the southern area of Colombia (Ruiz- García et al., 2003). In our study, we basically analyzed the molecular systematics of L. pajeros sensu García-Perea (1994). As it happens with the major part of all the wild cats, the human population growth has negative effects on their distributions. For example, Cabrera and Yepes (1940) affirmed, more than seven decades ago, that the Pampas cat population were reduced in the Buenos Aires and the Santa Fe provinces and drastically reduced in Uruguay. In other countries, as Bolivia, the Pampas cat was classified as in undetermined state (I) (Ergueta and Morales, 1996) and DD by IUCN (1994). IUCN (2006) classified this species as “near threatened”. It is classified in the Appendix II of CITES. The Pampas cat was extensively hunted in Argentina for the fur trade. Since 1976-1980, 78,000 skins were exported (Mares and Ojeda, 1984). In 1987, the last international shipment contained 10,000 pelts. Thus, although, its conservation status seems to be not critical, in countries as Argentina and Uruguay, its situation could be dangerous. The other cat is the Andean cat, which was classified in its own genus as Oreailurus jacobita by Cabrera (1940) because the ectotympanic of the auditory bullae is much larger than the entotympanic in contrast to other small cat species. It’s a very rare species. For instance, García-Perea (2002) commented that only three skulls (Kuhn, 1973) and 14 skins (Scrocchi and Halloy, 1986) had been studied and photographs of three individuals in their natural habitat (Sanderson, 1999; Scrocchi and Hally, 1986; Ziesler, 1992). Furthermore, of these three skulls, one was lost at the beginning of 20th century and all were of sub-adult individuals. In her study, she examined 44 specimens (three skulls, 39 skins and two skulls plus skins). This species is highly related to the cluster formed by Leopardus pardalis and L. Population Genetics and Spatial Structure in Two Andean Cats … 5 wiedii within the ocelot linage (Johnson et al., 1998, 1999) and does not belongs to the lineage of the rusty-spotted cat (Prionailurus rubiginosa) as was proposed by Salles (1992). The geographical and ecological distribution of this species is more restricted than that explained of the Pampas cat. Its distribution goes from Peru to northern Chile and central- western Argentina. In Peru, its presence is clearly determined in the Departments of Ayacucho (Pampas Galera National Reserve), Arequipa (Salinas and Aguada Blanca National Park), Puno (Aymara Lupaca Reserved Zone), Tacna and Cuzco (Machupicchu Historic Sanctuary and Manu National Park) and it is very probable in the Departments of Moquegua (Salinas and Aguada Blanca National Park), Apurímac (Ampay National Sanctuary), Huancavelina, Junín (Chacamarca Historic Sanctuary, Junín National Reserve and Yauyos- Cochas Landscape Reserve), Huanuco (Cordillera Huayhuash Reserved Zone), Ancash (Cordillera Huayhuash Reserved Zone and Huascaran National Park), Lima (Cordillera Huayhuash Reserved Zone, and Yauyos-Cochas Landscape Reserve) and Pasco (Huayllay National Sanctuary) (Cossíos and Madrid, 2003). This species, in Bolivia, is located in the Departments of Cochabamba (Cordillera Turani National Park, Carrasco National Park and Altamachi Departamental Park), Tarija (Cordillera de Sama Biologic Reserve), La Paz (Cotapata National Park and Apolobamba National Park), Potosí (Eduardo Avaroa National Andean Fauna Reserve) and Oruro (Sajama National Park). In Chile, the presence of the Andean cat is confirmed in Region I (Tarapacá; Las Vicuñas National Park, Lauca National Park, Salar de Surine National Monument and Volcán Isluga National Park), Region II (Los Flamencos National Reserve and Llullaillaco National Park) and in Region III (Atacama; Nevado Tres Cruces National Park), all in northern Chile. In Argentina, its presence is reported in Jujuy (Laguna Los Pozuelos Biosphere Reserve, Olaroz-Cauchari Provincial Reserve, Laguna Vilama Provincial Reserve), Salta (Los Andes Provincial Reserve, Los Cardones National Park), Tucuman (Campo de Los Alisos Nacional Park), Catamarca (Laguna Blanca Provincial Reserve), La Rioja (Laguna Brava Provincial Reserve), San Juan- Mendoza (San Guillermo National Park) and in Neuquen. One interesting feature is that no morphological subspecies has been recognized for this species in total contrast with the case of the Pampas cat as we aforementioned. This species was classified as rare (R) by IUCN (1990), unknown (K) by IUCN (1994) and endangered (EN) by IUCN (2002). In this last classification, the Andean mountain cat was categorized as “C2a”, which means the estimated population size is less than 2,500 adult individuals, has a diminishing trend and lacks any subpopulations that have more than 250 mature individuals. IUCN (2006) classified this species as “Endangered with small populations in decline”. CITES recorded in 1976, 84 skins of this very rare species exported to Spain (Broad, 1987). Although the genus Lynchailurus and Oreailurus have been traditionally employed, more recently some authors (Werdelin 1996; Yensen and Seymour, 2000; Salazar-Bravo et al., 2003; Johnson et al., 2006) have enclosed them in the genus Leopardus. For this reason, we employed the scientific names of Leopardus pajeros and L. jacobita. Only a very limited number of genetics papers have been published on these two Andean cats. Three works have been published regarding the Pampas cat (Johnson et al., 1999; Napolitano et al., 2008 and Cossíos et al., 2009) and three have also been published concerning the Andean cat (Johnson et al., 1998; Napolitano et al., 2008 and Cossíos et al., 2012). However, especially, in the works of Johnson et al., (1998) (9 samples of Andean cat), Johnson et al., (1999) (22 samples of Pampas cat) and Napolitano et al., (2008) (27 samples of Andean cat and 51 samples of Pampas cats for complete sequences), the sample sizes 6 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al. studied were small or very small. Cossíos et al., (2009, 2012) analyzed larger sample sizes for a more complete distribution of both species. However, in this chapter, we report the largest sample sizes analyzed to date for both species: 235 exemplars of the Pampas cat and 115 Andean cats. The current aims of the present work, are complementary to the works of Cossíos et al., (2009, 2012), and are as follows: 1- To determine the gene diversity levels of microsatellites and a mtDNA marker for both cat species and their putative subspecies, at least, for the Pampas cat, within the populations studied; 2- To determine the levels of gene heterogeneity and gene flow among populations and subspecies of the Pampas cat and among the populations of the Andean mountain cat for both kinds of molecular markers; 3- To determine levels of population and subspecies assignment in the Pampas cat and of population assignment in the Andean mountain cat as well as to determine migrants of first generations among subspecies or populations within each one of these species by means of microsatellite markers; 4- To determine possible demographic changes (bottlenecks and population expansions) by means of microsatellites and mtDNA; 5- To analyze the mutation molecular evolution of microsatellites in both cat species; 6- To estimate historical effective numbers for both species as well as for population and subspecies by using the microsatellite information and different mathematical procedures; 7- To determine possible spatial genetic structure for the populations studied of both cat species by means of different procedures (Mantel’s test, isolation by distance tests and spatial autocorrelation) by using the microsatellite data, and 8- To estimate possible skull biometric differences between two populations of the Pampas cat sampled in Bolivia and Chile, respectively.
MATERIAL AND METHODS
Number of samples and localities. A big fraction of the samples herein studied are those analyzed for other population genetics procedures by Cossíos et al., (2009, 2012). However, in the current work we have added a considerable number of samples, especially, for L. jacobita. In the case of L. pajeros, in the work of Cossíos et al., (2009), 198 individuals were analyzed proceeding from Peru [3 from Lambayeque, 19 from Huaraz (Huari), 19 from Yauyos (Canchayllo), 12 from Junin National Reserve, 8 from Ayacucho, 6 from Arequipa, 19 from Tacna-Puno (Tarata, Mazocruz, Condoriri, Chichillapi, Capaso, Collao)], Bolivia [14 from Departamento de La Paz, 20 from Cochabamba-Potosi (Sud Lipez)-Tarija] and Argentina [28 from Jujuy-Salta (Coranzuli, Salar Diablillo and Nevado Acay), 30 from Catamarca (Antofalla), 5 from Tucuman, 7 from Mendoza (Caserones, Puerto Panorama), 5 from Buenos Aires’ province and 3 from Pilcaniyeu-Patagonia]. However, in the current chapter a total number of 235 individuals of L. pajeros was analyzed because we added 12 new individuals for the Peruvian area of Tacna-Puno, 15 new individuals for the La Paz Department, six new individuals for the Potosi-Tarija area both in Bolivia and four new individuals for the area of Arica in northern Chile. However, these last four individuals were included in the Tacna-Puno population because as we later explain they were undifferentiated from the genotypes of the animals of that Peruvian area. In the case of L. jacobita, Cossíos et al., (2012) studied 56 exemplars proceeding from Peru (11 from North-Central Peru-Cuzco- Ayacucho-Arequipa and 10 from Tacna-Puno), Bolivia (10 from La Paz Department and 7 Population Genetics and Spatial Structure in Two Andean Cats … 7 from Potosi-Sud Lipez) and Argentina (7 from Jujuy, 5 from Catamarca-Tucuman and 6 from Mendoza). Herein, we analyzed a total of 115 exemplars of L. jacobita because we enclosed 59 new exemplars: seven for the Tacna-Puno area, 18 for the La Paz Department, 20 for the Potosi-Tarija area and 12 for the Jujuy area. All these new samples for both Andean cats were little pieces of skin and teeth of hunted animals. The samples used to obtain DNA were basically made up of hairs with roots, muscle, bones, teeth and pieces of skin from animals killed by hunters and fecal samples collected opportunistically in the field. For pieces of muscle, skin, teeth and bones, the phenol-chloroform procedure was employed with several modifications of standard techniques. The DNA extraction from the hairs with follicle was carried out by using 10-20 % Chelex 100 resin (Bio-Rad, USA), with several modifications from Walsh (1991). We followed the procedures used by Cossíos et al., (2009, 2012) for DNA extraction from scat.
Molecular Markers Employed
Microsatellites. Five microsatellites (Fca24, Fca31, Fca45, Fca96 and Fca294) for L. pajeros and seven microsatellites (Fca08, Fca24, Fca31, Fca96, Fca173, Fca176 and Fca294) were respectively examined. All these microsatellites were dinucleotide repeats
(CA)n or (GT)n. The primer sequences for these microsatellites were obtained by Menotti- Raymond and O’Brien (1995) and Menotti-Raymond et al., (1999). When DNA was extracted by means of phenol-chloroform, a polymerase chain reaction (PCR) was performed in a 25 l volume. In this case, PCR reaction mixtures included 2.5 l of 2.5 mM MgCl2, 2.5
l of a 10x Buffer, 1 l of 1 mM dNTPs, 10 pmol of each primer, 14.5 l of H2O, 2 l of DNA (50-100 ng/l) and one unit of Taq polymerase. DNA extracted from Chelex resin was subjected to PCR in a 50 l PCR reaction volume, containing twice the quantity of all the above reagents and 20 l of DNA. PCR reactions were carried out in a Geneamp PCR System 9600 thermocycler (Perkin Elmer, Wellesley, Massachusetts) and in an iCycler Thermal Cycler (Bio-Rad, Hercules, California). The temperatures employed were as follows: 95 °C for 5 minutes, 35 cycles of 1 minute at 95 °C, 2 minutes at 55 °C for all the microsatellites employed and 2 minutes at 72 °C. A final extension for 5 minutes at 72 °C was used. Amplification products were kept at 4 °C until used. PCR products were electrophoresed in denaturing 6 % polyacrylamide gels and visualized in a Hoefer SQ3 sequencer vertical camera. Allele sizes were obtained by comparison with the molecular weight 174 DNA digested with Hind III and Hinf I. A molecular weight marker was loaded every four lanes. The PCR amplifications were performed in triplicate to ensure the accuracy of the genotypes obtained. In nearly 94 % of the cases, the observed genotypes were the same in the three replicates. Thus, preferential allele amplification had minimal effects on our results. MtDNA marker. The hypervariable domain 1 (HVS-1) of the mtDNA control region was analyzed because it is a non-coding sequence with a very high polymorphism within felid species. The primers CH3F-H1rev and H2for-CH3R were employed to amplify PCR products smaller than 300 bp (Freeman et al., 2001; Cossíos et al., 2009). The PCR conditions for obtaining these sequences are described by Cossíos et al., (2009). 8 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Microsatellite Population Genetics Analyses for the Two Andean Cats Studied
Gene diversity and genetic heterogeneity. The expected heterozygosity (H) (Nei, 1973) was calculated for the different populations studied in L. pajeros and L. jacobita and differences among estimates were statistically analyzed with a Student t test. The heterozygosity data were arcsine transformed prior to analysis, as proposed by Archie (1985). Several strategies were used to calculate genetic heterogeneity among population pairs and globally. Firstly, the Wright F-statistics (1951), with the Michalakis and Excoffier (1996)’s procedure, was applied. The standard deviations of the F-statistics and the confidence intervals (95 and 99 %) were calculated with jackknifing and bootstrapping over loci, respectively. The significance of FST (genetic heterogeneity) was calculated with the log- likelihood G test (5,000 genotype randomizations, random mating not assumed) (Goudet et al., 1996). Secondly, genic and genotypic frequencies were employed to determine genetic heterogeneity by means of exact tests with Markov chains (10,000 dememorizations, 100 batches at 5,000 iterations per batch). Thirdly, the RST statistic (Slatkin, 1995; Rousset, 1997; Goodman, 1997) was also employed to measure genetic heterogeneity among the Andean cat populations studied. Several indirect gene flow estimates were obtained with the n- dimensional island model (Takahata 1983; Crow and Aoki 1984; Ruiz-García and Álvarez 2000). Futhermore, another gene flow estimate was obtained by using the private allele method (Slatkin, 1985; Barton and Slatkin, 1986). Genetic assignment. We used the Geneclass 2 program (Piry et al., 2004) to develop diverse assignment analyses for populations and for putative subspecies in the case of L. pajeros and for populations in the case of L. jacobita. Different strategies were performed by using two Bayesian procedures (Rannala and Mountain, 1997; Badouin and Lebrun, 2000) and one frequency procedure (Paetkau et al., 1995). The assignation analyses were carried out without simulations and estimating probabilities of belonging, or exclusion, of the individuals to the original groups or subspecies where they were “a priori” assigned (P < 0.05). Also some assignation analyses were performed with 10,000 resampling simulations by means of the Monte Carlo technique and with the Paetkau et al., (2004) and Rannala and Mountain (1997)’ procedures. Additionally, we estimated the possible existence of first generation migrants in the different Andean cat groups considered. In order to do this, we used the Bayesian, frequency and genetic distance procedures commented above without simulations. To determine this, we considered the relationship: L = Lhome/Lmax, which is the ratio of the likelihood computed from the population where the individual was sampled (Lhome) over the highest likelihood value among all population samples including the population where the individual was sampled (Lmax) (Paetkau et al., 2004). Demographic changes. The first analysis to detect recent bottleneck events was that generated by Cornuet and Luikart (1996), and Luikart et al., (1998). The species, which experienced a recent bottleneck, simultaneously decreases the allele number and the expected levels of heterozygosity. Nevertheless, the allele number (ko) is reduced faster than the expected heterozygosity. Therefore, the value of the expected heterozygosity calculated through the allele number by a coalescence procedure (Heq) is lower than the obtained expected heterozygosity estimated directly from allele frequencies (He). For neutral markers, Population Genetics and Spatial Structure in Two Andean Cats … 9 in a population in gene mutation drift equilibrium, there is an equal probability that a given locus has a slight excess or deficit of heterozygosity in regard to the heterozygosity calculated from the number of alleles. In contrast, in a bottlenecked population, a large fraction of the loci analyzed will exhibit a significant excess of the expected heterozygosity. Three mutation models were employed: infinite allele model (IAM), two phase model (TPM) and step-wise mutation model (SMM). To measure this probability, four diverse procedures were used as follows: sign test, standardized difference test, Wilcoxon´s signed rank test and graphical descriptor of the shape of the allele frequency distribution. A population, which did not suffer a recent bottleneck event, will yield an L-shape distribution (such as expected in a stable population in mutation-gene drift equilibrium), whereas a recently bottlenecked population will show a mode-shift distribution. The Wilcoxon´s signed rank test probably has its greatest power when the number of loci analyzed is low, such as in the current case. The BOTTLENECK program was used to test for recent bottlenecks. Two tests to detect possible population expansions were carried out. Kimmel’s et al., test
(1998) is based on the principle that two different estimates of (= 4Ne) could be obtained
[one is v = V , with V being the variance of the tandem repeat size, whose expression is V = 2 (2 I = 1….n(Xi – X) )/(n – 1), where n is the number of chromosomes analyzed, Xi is the number of repeats of each allele found and X is the average repeat number of all the alleles 2 found in a microsatellite. The second one is Po = (1/(Po – 1))/2 (estimator of the 2 homozygosity), where Po = (n k = 1…..n (p k – 1))/(n -1), with pk being the allele frequency of the kth allele.] An imbalanced index could be defined as: (t) = E(v)/E(Po) = V(t)/[((1/ 2 2 Po ) – 1)/2 or by the expression: ln (t) = ln v – ln Po = ln (V) – ln (((1/ Po ) – 1)/2). If a population is in equilibrium, has a constant demographic size, and isn’t suffering an expansion, = 1 (ln = 0). On the contrary, if a population has suffered an expansion coming from a mutation-drift equilibrium situation (constant population size), < 1 (ln < 0). If a population has experienced an expansion coming from a previous bottleneck , > 1 (ln > 0). This last value will be present for a long time (several thousand generations) before showing the signature of a population expansion (< 1 (ln < 0)). There is an exception to this general rule, when a bottleneck is so intense the population becomes monomorphic before the demographic expansion, in which case, < 1 (ln < 0) all the time. All these values are consistent in stepwise, logistic or exponential population growth and are not especially affected in diverse mutation models (Kimmel et al., 1998). We used empirical distributions of ln from 500 coalescence simulations with a = 5 to determine the statistical significance of (ln ) In this case, a 95 % confidence interval was determined (- 0.18, 0.23). The second test employed was that from Zivothovsky et al., (2000), which calculates an 2 expansion index: Sk = 1 – ((K – (RkV/2)/5V ), where K and V are the unnormalized kurtosis (fourth central moment) and the allele size variance is estimated from a sample and corrected 2 for sampling bias, respectively, whereas Rk = km/ m (they are the kurtosis and the variance in the repeat number mutational changes). The expressions used to estimate V and K are: V = i 2 4 = 1…n pi (Xi – X) and K = i = 1…n pi (Xi – X) , where X = i = 1…k pi k, and k represents the alleles in a locus given and pi, the allele frequencies. All the other terms were defined in the previous analysis. We used 6.3 for Rk because it was the value obtained for dinucleotide microsatellites by Dib et al., (1996), and because dinucleotide microsatellites were used in the 10 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al. current study. Feldman et al., (1999), used the same data and a geometrical distribution of 2 mutational events, and obtained an estimated m of 2.5, which is basically the same as what was obtained by using a truncated Poisson distribution (Zhivothovsky et al., 2000) and by 2 ourselves in the present work ( m = 2.45). The value of Sk is expected to be 0 in a general symmetric stepwise mutation model for a population in equilibrium and of constant size (this was derived by Zhivotovsky and Feldman, 1995). The Sk is positive if an expansion affected the population and is contrarily negative if a bottleneck affected the population. To obtain demographic conclusions of this analysis, the within-population variance and the expansion index are averaged for all the microsatellites studied within each population and their dynamics are compared. Zhivotovsky et al., (2000) showed that a significant correlation existed between V and Sk (r = 0.58) for a human data set, but this correlation was moderate and, in fact, both statistics could react differently to the changes in population size and have different patterns in different populations. Noteworthy consequences could be extracted from the differential behavior of both of these statistics in a given population. To measure the statistical significance of the Sk values, a jackknife procedure was performed to obtain the variance of Sk and, with this variance, a Student’s t test and a 95 % confidence interval were estimated. Microsatellite evolution and effective numbers. A maximum likelihood procedure with a Markov chain recursion method (Nielsen, 1997) was used to estimate the
statisticewhere e is the effective population size and the mutation rate per generation) for the microsatellites applied to the Andean cats, to determine the microsatellite mutation model of these markers (one-step versus multi-step mutations) and to calculate historical effective number for the two Andean cat species studied assuming a mutation rate of 2.1 x 10-5, which is a typical mutation rate per generation for wild cats (Ruiz-García et al., 2012). Probable values were calculated using the Nielsen’s (1997) model. A one step mutation model, that was typical of microsatellites, was also adopted (Nielsen, 1997). Mathematical expressions by Ohta and Kimura (1973) and Wehrhahn (1975) were used to calculate the probability that an allele chosen at random was m repetitions higher than other allele chosen at random. The recursivity of the coalescence theory was applied to obtain the likelihood functions of for samples of a determined size. The MISAT program (Nielsen, 1997) was used to estimate the likelihood surfaces for . The 5 % confidence interval was calculated by multiplying the log likelihood of the maximum likelihood value by two. A grid size of 40 with a previous calculation and method of the moments (0) in a mutation one- step model with 1,000,000 Markov chains was used. The value with the least negative log likelihood is the estimate of the maximum likelihood. From this value, Ne was calculated for the two Andean cat species studied. In addition, the largest possible multi-step mutation percentages (ranging from 0 to 0.5) were calculated through the maximum likelihood of by means of 3,000,000 Markov chains. We analyzed the different possible mutation rates that affected each one of the microsatellites studied for these two Andean cats. We tested the hypothesis 1 = 2 = (the values of for two different microsatellites) using a likelihood 2 ratio test with the expression -2 log[L1()L2()]/[L(1,2)], following a with one degree of freedom. A probability lower than = 0.05 indicates that both microsatellites have different mutation rates. Likewise, we measured if the multi-step mutation models’ estimates were significant improvements over the uni-step mutation models. The likelihood ratio of -2 log Population Genetics and Spatial Structure in Two Andean Cats … 11
[L(, p = 0)/L(, p)] was applied to the maximum likelihood obtained multi-step p percentage. Large samples have a value of 2 with one degree of freedom with the null hypothesis p = 0. A probability lower than = 0.05 indicates that the multi-step mutation percentage is significantly different from the uni-step mutation model and this last model is then rejected. We used the step-wise mutation model (H = [1 – (1 – H)2/(8 (1 – H)2)], as a second method, to estimate the effective average expected heterozygosity, where is the mutation rate per generation; Ohta and Kimura, 1973). Additionally, the methods of Hill (1981) and Pudovkin et al., (1996) were applied to determine possible effective numbers in these two Andean cat species by means of the Neestimator software. We also used the method by Griffiths and Tavaré (1994) to estimate effective numbers, in this case, for the subspecies and populations studied in pajeros and in populations studied in jacobita and to evaluate the likelihood functions based on a Monte Carlo method. This allowed us to estimate the possible effective numbers from the likelihood functions obtained for each microsatellite studied. A mutation rate of 2.1 x 10-5 was employed in this analysis. Spatial genetic structure. Several strategies were applied to determine if L. pajeros and L. jacobita presented some significant spatial genetic trend because this could help to understand the evolutionary events that have determined the natural history of these two species. These strategies were as follows:
1. A Mantel’s test (Mantel, 1967) was used to detect possible overall relationships between a genetic matrix among populations within each species (Nei, 1972 genetic distance) and the geographic distance matrix among the populations analyzed. In this study, Mantel’s statistic was normalized according to Smouse et al. (1986). This procedure transforms the statistic into a correlation coefficient. The geographic distances were measured with the Spuhler’s (1972) procedure, where D = arcos (cos
X(i) . cos X(j) + sin X(i) . sin X(j) cos |Y(i) – Y(j)|) , where X(n) and Y(n) are the latitude and longitude of the nth individual sampled, respectively. The significance of the correlations obtained was tested using a Monte Carlo simulation with 1,000 permutations. 2. To determine possible isolation by distance among the populations within these two Andean cat species, the IBD version 1.5.1 software (Bohonak, 2002) was employed. In this analysis, we used the quoted genetic distance against the geographical distance among the populations sampled. The intercept and the slope of this relationship was calculated using Reduced Major Axis (RMA) regression (Sokal and Rohlf, 1981; Hellberg, 1994). Ten thousand randomizations (jackknife over population pairs and bootstrapping over independent population pairs) were executed to determine 95 and 99 % confidence intervals. The calculations were completed with non-transformed data and with log transformed data (genetic distance and geographical distance) jointly and separately. 3. A spatial autocorrelation analysis (Sokal and Oden 1978ab; Sokal and Wartenberg 1983; Sokal et al., 1986, 1987, 1989; Sokal and Jacquez 1991; Epperson 1990, 1993; Ruiz-García 1998, 1999 and Ruiz-García and Jordana 1997, 2000) was applied among the different populations of both species (15 populations for L. pajeros and 7 populations for L. jacobita) for the 25 alleles with the highest frequencies. 12 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Autocorrelation coefficients and correlograms were estimated. For this, the Moran's I index and the Geary's c coefficient (Moran, 1950; Durbin and Watson, 1950) were employed in the current study. In the case of L.pajeros, three distance classes (DC) with a relatively equal number of point pairs, among distance classes (3 DC: 0-830 km; 830-1,650 km; 1,650-3,640 km) and four DC with identical geographic intervals (4 DC: 0-910 km; 910-1,820 km, 1,820-2,730 km; 2,730-3,640 km) were estimated. For L. jacobita, we estimated three distance classes (DC) with a relatively equal number of point pairs, among distance classes (3 DC: 0-1,010 km; 1,010-2,490 km; 2,490-3,250 km) and four DC with identical geographic intervals (4 DC: 0-812 km; 812-1,624 km, 1,624-2,436 km; 2,436-3,248 km). To use these statistics, individuals must be connected using some type of network, which simulates as realistically as possible, the relationships existing between them. In this case, three network connections were used. The first method was binary, with all pairs of populations connected at different specified distance classes. These connections were determined by using the possible gene flow routes between the individuals considered (Trexler, 1988). Also, the Gabriel-Sokal network (Gabriel and Sokal, 1969; Matula and Sokal, 1980) and the Delaunay's triangulation with elimination of the crossing edges (Ripley, 1981; Upton and Fingleton, 1985; Isaaks and Srivastava, 1989) were used. However, the results were very similar in each case. The Bonferroni (Oden, 1984), Oden's Q and the Kooijman’s tests were calculated with SAAP 4.3 software to determine the statistical significance of the individual autocorrelation coefficients and for the overall correlograms.
Mitochondrial Gene Analyses for the Two Andean Cats Studied
Gene diversity and genetic heterogeneity. We used the number of polymorphic sites (S), the number of haplotypes, the haplotypic diversity (Hd), the nucleotide diversity (), the average number of nucleotide differences (k) and the statistic by sequence to determine genetic diversity within the different populations and subspecies of L. pajeros and for the different populations of L. jacobita analyzed. We used the following tests to measure genetic heterogeneity and to determine possible gene flow estimates at the HVS-I of the mtDNA control region locus: HST, KST, KST*, Z and Z* (Hudson et al., 1992a,b), Snn (Hudson, 2000), chi-square (on the haplotypic frequencies with permutation tests of 10,000 replicates), GST
(from the haplotypic frequencies), ST, NST, FST (Hudson et al., 1992a; from the nucleotide sequences), Kxy (mean proportion of nucleotide differences between taxa) and Da (net number of nucleotide substitutions per site among taxa). The analyses were performed with DNAsp 4.56 and Arlequin 3.1 software. Demographic changes. We used the following procedures to determine possible historical population changes for the two Andean cats studied by means of the mtDNA sequences. 1- The mismatch distribution (pairwise sequence differences) was obtained following the method of Rogers and Harpending (1992) and Rogers et al., (1996) where two theoretical curves were obtained (population growth and bottleneck show characteristic signatures in histograms yielding the relative frequencies of pairs of individuals who differ by i nucleotide sites). One curve assumed a constant population size and another assumed a Population Genetics and Spatial Structure in Two Andean Cats … 13
population expansion. The population size before expansion is represented by 0 (= 2Ne0 with Ne0 equal to the female effective number before growth and equal to the mutation rate per generation. The population after expansion was represented by = 2Ne1where Ne1 equals the number of females after growth). Finally, = 2t (t = number of generations), represents the time elapsed from the population expansion on a mutational temporal scale. We compared these curves to the empirical observed distribution. We used the raggedness rg statistic (Harpending et al., 1993; Harpending 1994) and the R2 statistic of Ramos-Onsins and Rozas (2002) to determine the similarity between the observed and the theoretical curves. 2- For the second procedure we used nucleotide segregating sites (polymorphism) (frequency spectrum) derived from the work of Tajima (Tajima 1989b, p 594), based on the comparison of the observed and the expected frequency spectrums of polymorphic variation. 3- We used the Fu and Li D* and F* tests (Fu and Li, 1993), the Fu FS statistic (Fu, 1997) and the Tajima D test (Tajima 1989a) (originally created to detect natural selection affecting DNA sequences) to determine possible population size changes in the L. pajeros and the L. jacobita populations we analyzed (Simonsen et al., 1995; Ramos-Onsins and Rozas, 2002).
Craniometric Analysis
We analyzed nine biometric skull variables in two sets of Pampas cats; one of them was composed of 16 skulls from the central area of Chile (thus, representing L. colocolo colocolo following García-Perea, 1994) and the other was composed of 11 skulls from Corocoro in the La Paz Department (representing by our molecular analyses as L. p. garleppi). In addition we estimated the greatest length of skull along the medial plane (GLS), craneal width (CW), length of rostrum (LRO), zygomatic width (ZW), postorbital width (POW), Foramen Magnum length (FML), Foramen Magnum width (FMW), nasal height (NH) and nasal width (NW). A Student t test was employed to determine possible significant differences between the two sets of the Pampas cat skull.
RESULTS
Gene Diversity
Microsatellites. The Pampas cat gene diversity by populations (14) ranged from 0.520 + 0.179 (Lambayeque-Peru) to 0.810 + 0.054 (Mendoza, Argentina), with an average value of 0.727 + 0.067. By putative subspecies (5), these values oscillated from 0.667 + 0.104 (L. p. pajeros) to 0.828 + 0.049 (L. p. budini) with an average value of 0.741 + 0.064 (Table 1a). The four jacobita population sets considered showed average H values ranging from 0.216 + 0.195 (Central Peru) to 0.468 + 0.263 (South Peru-North Bolivia) (Table 1b), with an average H for all the species of 0.423 + 0.137. These are low, or very low, levels of gene diversity for markers as microsatellites. Therefore, it was clear that the gene diversity of the Pampas cat was considerably higher than that estimated for jacobita.
mtDNA. For pajeros as a whole, the number of different haplotypes was 41 and Hd = 0.932 + 0.007, = 0.0513 + 0.0016 and k = 16.925 + 7.557. For the pajeros population set, 14 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
the values ranged from Hd = 0.626 + 0.058, = 0.0080 + 0.0010 and k = 2.642 + 0.945
(Huaraz-Peru) and Hd = 0.813 + 0.060, = 0.0077 + 0.0005 and k = 2.532 + 0.865 (Yauyos-
Peru) to Hd = 0.924 + 0.054, = 0.0527 + 0.0011 and k = 17.382 + 1.023 (Jujuy-Salta, northern Argentina). For the pajeros subspecies (crucinus was added to pajeros in this analysis), the budini subspecies had the highest diversity, 23 different haplotypes and Hd = 0.903 + 0.002, = 0.0526 + 0.0013 and k = 17.375 + 5.587. The northern subspecies, garleppi, had the lowest levels of genetic diversity, 16 different haplotypes and Hd = 0.840 + 0.004, = 0.0153 + 0.0009 and k = 5.058 + 2.136, being similar to that of pajeros (Table 1c).
Table 1. Gene diversity statistics found in the putative Leopardus pajeros’s subspecies at DNA microsatellite loci (A). Gene diversity statistics found in the four different Leopardus jacobita populations analyzed at DNA microsatellite loci(B). Gene diversity statistics found in the putative Leopardus pajeros’s subspecies at mtDNA (C). Gene diversity statistics found in the four different Leopardus jacobita populations analyzed at mtDNA (D)
A H + SD MNA + SD L.p.garleppi 0.758 0.103 10.60 2.51 L.p.steinbachi 0.690 0.161 7.80 1.64 L.p.budini 0.828 0.049 12.20 2.65 L.p.pajeros 0.667 0.104 4.00 0.71 L.p.crucinus 0.753 0.166 3.40 1.14
B H + SD MNA + SD Northern Peru 0.216 0.169 1.72 0.49 South Peru-North Bolivia 0.438 0.217 3.29 1.25 South Bolivia-North Argentina 0.436 0.176 3.43 0.98 Mendoza-Argentina 0.407 0.230 2.43 0.79
C Statistics L.p.garleppi L.p.steinbachi L.p.budini L.p.pajeros-crucinus S 48 51 68 14 50 51 71 14 h 16 8 23 4
Hd 0.841 ± 0.019 0.837 ± 0.043 0.903 ± 0.023 0.810 ± 0.129 π 0.0153 0.0408 0.0525 0.0188
k 5.058 ± 2.657 13.449 ± 6.247 17.375 ± 7.814 6.381 ± 3.442 θ(per sequence) 9.243 13.365 14.734 5.714
D Statistics North Peru S.Peru-N.Bolivia S.Bolivia- Mendoza- N.Argentina Argentina S 0 4 2 2 0 5 2 2 Population Genetics and Spatial Structure in Two Andean Cats … 15
h 1 3 2 2
Hd 0.000 ± 0.000 0.637 ± 0.064 0.189 ±0.108 1.000 ± 0.129 π 0.0000 0.0049 0.0011 0.0059 Statistics North Peru S.Peru-N.Bolivia S.Bolivia- Mendoza- N.Argentina Argentina k 0.000 ± 0.000 1.700 ±1.037 0.379 ±0.374 2.000 ± 1.000 θ(per sequence) 0.000 1.409 0.564 0.789 H = Mean expected heterozygosity, MNA = Mean number of alleles, SE = Standard deviation, S =
Number of polymorphic sites, = Number of mutations, h = Number of haplotypes, Hd = Haplotype diversity, = Nucleotide diversity, k = Average number of nucleotide differences
between sequences, = 2Ne (Ne = Number of females and = mutation rate per generation).
As it was found for microsatellites, the diverse gene diversity statistics for mtDNA sequences showed very limited diversity in jacobita. For the species as a whole, five haplotypes were found and Hd = 0.557 + 0.061, = 0.0047 + 0.0006 and k = 1.617 + 0.974. For the four population sets considered, the values were zero for all of these statistics (North
Peru) to Hd = 0.677 + 0.069, = 0.0049 + 0.0006 and k = 1.637 + 0.994 (South Peru-North
Bolivia) and Hd = 1.000 + 0.128, = 0.0059 + 0.0059 and k = 2.000 + 1.000 (Mendoza) (Table 1d). Thus, similar to what was observed for the microsatellites, pajeros showed noteworthy higher mtDNA gene diversity than jacobita.
Genetic Heterogeneity and Gene Flow
Microsatellites. The genotypic differentiation with exact tests for pajeros’ populations was significant for the five microsatellites individually and together (2 = infinite, 10 df, P = 0.000000) (Table 2a). Genic differentiation with exact G tests was also significant for all of the microsatellites individually and together (2 = infinite, 10 df, P = 0.000000) (Table 2b). The exact tests for pajeros population pairs (Table 3a) clearly showed that the major part of the populations considered presented significant heterogeneity among them. Only four population pairs out 91 population comparisons were not significant. These were the cases of the population of Arequipa and Tacna-Puno in Peru and the three populations of the southern Argentina (Mendoza-San Juan, Buenos Aires and Patagonia). The genotypic differentiation with exact tests for pajeros’ subspecies was significant for the five microsatellites employed individually and all them taken together (2 = infinite, 10 df, P = 0.000000) (Table 2c) as well as the exact G tests for genic differentiation for all the microsatellites and all of them taken together (2 = infinite, 10 df, P = 0.000000) (Table 2d). Therefore, the classification in populations or subspecies showed the same power of genetic discrimination in pajeros. The exact tests for pajeros subspecies pairs (Table 3b) showed that only one subspecies comparison was not significant. This comparison was between the two most southern subspecies in Argentina, L. p. pajeros and L. p. crucinus. The genotypic differentiation with exact tests for jacobita was significant for the seven microsatellites employed individually and all of them taken together (2 = infinite, 14 df, P = 0.000000) (Table 2e) as well as the exact G tests for genic differentiation for all the microsatellites and all of them taken together (2 = infinite, 14 df, P = 0.000000) (Table 2f). 16 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Table 2. Exact G tests to determine genotypic heterogeneity at five DNA microsatellites for Leopardus pajeros analyzing 14 populations (A). Exact G tests to determine genic heterogeneity at five DNA microsatellites for Leopardus pajeros analyzing 14 populations (B). Exact G tests to determine genotypic heterogeneity at five DNA microsatellites for Leopardus pajeros analyzing four subspecies (C). Exact G tests to determine genic heterogeneity at five DNA microsatellites for Leopardus pajeros analyzing four subspecies (D). Exact G tests to determine genotypic heterogeneity at seven DNA microsatellites for Leopardus jacobita analyzing four populations (E). Exact G tests to determine genic heterogeneity at seven DNA microsatellites for Leopardus jacobita analyzing four populations (F). * P < 0.05; ** P < 0.01; df = degree of freedom
A P-value across all loci (Fisher's method) Locus P-Value FCA31 0.0000** FCA24 0.0000** FCA45 0.0000** FCA96 0.0000** FCA294 0.0000** All loci: = Infinity (df = 10), P = highly significant**.
B P-value across all loci (Fisher's method) Locus P-Value FCA31 0.0000** FCA24 0.0000** FCA45 0.0000** FCA96 0.0000** FCA294 0.0000** All loci: = Infinity (df = 10), P = highly significant**.
C P-value across all loci (Fisher's method) Locus P-Value FCA31 0.0000** FCA24 0.0000** FCA45 0.0000** FCA96 0.0000** FCA294 0.0000** All loci: = Infinity (df = 10), P = highly significant**.
D P-value across all loci (Fisher's method) Locus P-Value FCA31 0.0000** FCA24 0.0000** Population Genetics and Spatial Structure in Two Andean Cats … 17
FCA45 0.0000**
P-value across all loci (Fisher's method) Locus P-Value FCA96 0.0000** FCA294 0.0000** All loci: = Infinity (df = 10), P = highly significant**.
E P-value across all loci (Fisher's method) Locus P-Value FCA08 0.0348* FCA31 0.0000** FCA24 0.0000** FCA96 0.0004** FCA176 0.0000** FCA294 0.0002** FCA173 0.0000** All loci: = Infinity (df = 14), P = highly significant**.
F P-value across all loci (Fisher's method) Locus P-Value FCA08 0.0186* FCA31 0.0000** FCA24 0.0000** FCA96 0.0000** FCA176 0.0000** FCA294 0.0000** FCA173 0.0000** All loci: = Infinity (df = 14), P = highly significant**.
The exact tests for jacobita population pairs (Table 3c) clearly showed that the seven populations studied could be clustered in four population sets: Northern-Central Peru, southern Peru-northern Bolivia, southern Bolivia-northern Argentina and Mendoza (Argentina).
The genetic heterogeneity statistics for the populations of pajeros were significant: GST = 0.155, GST’ = 0.165, FST = 0.158 + 0.016 (jackknife), FST = 0.119-0.192 (99% confidence interval- bootstrap), RST = 0.303 (weigthed) and RST = 0.306 (Goodman’s procedure) (Table 4a). All the individual markers yielded significant heterogeneity. Additionally, all the individual and collective FIS and FIT statistics were significant. The genetic heterogeneity statistics for the pajeros subspecies were significant and similar to those obtained with populations: GST = 0.125, GST’ = 0.152, FST = 0.132 + 0.023 (jackknife), FST = 0.078-0.182 (99% confidence interval- bootstrap), RST = 0.233 (weighted) and RST = 0.227 (Goodman’s procedure) (Table 4b). All the individual markers yielded significant heterogeneity.
Additionally, all the individual and collective FIS and FIT statistics were significant. The gene flow estimates among pajeros populations was Nm = 1.052-1.851 (99 % confidence interval from FST) and Nm = 1.437 (private alleles), and for pajeros subspecies was Nm = 1.124-2.955 (99 % confidence interval from FST) and Nm = 1.882 (private alleles). 18 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Table 3. Exact G tests to determine genetic heterogeneity among the population pairs of Leopardus pajeros analyzed at five DNA microsatellites (A). Exact G tests to determine genetic heterogeneity among the subspecies pairs of Leopardus pajeros analyzed at five DNA microsatellites (B). Exact G tests to determine genetic heterogeneity among the population pairs of Leopardus jacobita analyzed at seven DNA microsatellites (C)
A P-value for each population pair across all loci (Fisher's method) Population pair 2 df P-Value 1 & 2 52.90172 10 0.00000** 1 & 4 Infinity 10 0.00000** 2 & 4 Infinity 10 0.00000** 1 & 3 Infinity 10 0.00000** 2 & 3 Infinity 10 0.00000**. 4 & 3 40.239615 10 0.00001** 1 & 5 45.516259 10 0.00000** 2 & 5 Infinity 10 0.00000** 4 & 5 30.710753 10 0.00065** 3 & 5 25.700539 10 0.00417** 1 & 7 34.876800 10 0.00013** 2 & 7 25.626947 10 0.00428** 4 & 7 31.398317 10 0.00050** 3 & 7 24.569678 10 0.00622** 5 & 7 23.553570 10 0.00888** 1 & 8 Infinity 10 0.00000** 2 & 8 56.590428 10 0.00000** 4 & 8 Infinity 10 0.00000** 3 & 8 53.659549 10 0.00000** 5 & 8 43.401192 10 0.00000** 7 & 8 3.415053 10 0.96990 1 & 9 63.849457 10 0.00000** 2 & 9 Infinity 10 0.00000** 4 & 9 Infinity 10 0.00000** 3 & 9 53.519185 10 0.00000** 5 & 9 Infinity 10 0.00000** 7 & 9 24.298440 10 0.00685** 8 & 9 29.982076 10 0.00086** 1 & 11 Infinity 10 0.00000** 2 & 11 Infinity 10 0.00000** 4 & 11 Infinity 10 0.00000** 3 & 11 Infinity 10 0.00000** 5 & 11 Infinity 10 0.00000** 7 & 11 Infinity 10 0.00000** 8 & 11 Infinity 10 0.00000** 9 & 11 Infinity 10 0.00000**
Population Genetics and Spatial Structure in Two Andean Cats … 19
P-value for each population pair across all loci (Fisher's method) Population pair 2 df P-Value 1 & 14 Infinity 10 0.00000** 2 & 14 Infinity 10 0.00000** 4 & 14 Infinity 10 0.00000** 3 & 14 Infinity 10 0.00000** 5 & 14 Infinity 10 0.00000** 7 & 14 Infinity 10 0.00000** 8 & 14 Infinity 10 0.00000** 9 & 14 Infinity 10 0.00000** 11 & 14 24.407700 10 0.00659** 1 & 16 85.924521 10 0.00000** 2 & 16 Infinity 10 0.00000** 4 & 16 Infinity 10 0.00000** 3 & 16 Infinity 10 0.00000** 5 & 16 Infinity 10 0.00000** 7 & 16 Infinity 10 0.00000** 8 & 16 Infinity 10 0.00000** 9 & 16 Infinity 10 0.00000** 11 & 16 Infinity 10 0.00000** 14 & 16 Infinity 10 0.00000** 1 & 17 55.351156 10 0.00000** 2 & 17 Infinity 10 0.00000** 4 & 17 Infinity 10 0.00000** 3 & 17 Infinity 10 0.00000** 5 & 17 68.351124 10 0.00000** 7 & 17 70.332904 10 0.00000** 8 & 17 Infinity 10 0.00000** 9 & 17 Infinity 10 0.00000** 11 & 17 Infinity 10 0.00000** 14 & 17 Infinity 10 0.00000** 16 & 17 41.327505 10 0.00001** 1 & 18 64.178371 10 0.00000** 2 & 18 Infinity 10 0.00000** 4 & 18 Infinity 10 0.00000** 3 & 18 Infinity 10 0.00000** 5 & 18 67.028112 10 0.00000** 7 & 18 Infinity 10 0.00000** 8 & 18 Infinity 10 0.00000** 9 & 18 Infinity 10 0.00000** 11 & 18 Infinity 10 0.00000** 14 & 18 Infinity 10 0.00000** 16 & 18 Infinity 10 0.00000** 17 & 18 9.207952 10 0.51248 1 & 19 38.269764 10 0.00003** 2 & 19 50.239330 10 0.00000** 4 & 19 72.373658 10 0.00000** 20 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Table 3. (Continued)
A P-value for each population pair across all loci (Fisher's method) Population pair 2 df P-Value 3 & 19 51.905934 10 0.00000** 5 & 19 36.109899 10 0.00008** 7 & 19 34.552319 10 0.00015** 8 & 19 62.543053 10 0.00000** 9 & 19 48.582902 10 0.00000** 11 & 19 70.717117 10 0.00000** 14 & 19 60.955348 10 0.00000** 16 & 19 30.242314 10 0.00078** 17 & 19 3.019964 10 0.98095** 18 & 19 12.118636 10 0.27719**
B P-value for each population pair across all loci (Fisher's method) Subspecies pair 2 df P-Value garleppi & steinbachi Infinity 10 0.00000** garleppi & budini Infinity 10 0.00000** steinbachi & budini Infinity 10 0.00000** garleppi & pajeros Infinity 10 0.00000** steinbachi & pajeros Infinity 10 0.00000** budini & pajeros 51.05309 10 0.00000** garleppi & crucinus 72.150453 10 0.00000** steinbachi & crucinus 75.178055 10 0.00000** budini & crucinus 25.724385 10 0.00412** pajeros & crucinus 12.199280 10 0.27194
C P-value for each population pair across all loci (Fisher's method) Population pair 2 df P-Value N-C.Peru & S.Peru 41.33585 14 0.00016** N-C.Peru & N.Bolivia 30.022111 14 0.00758** S.Peru & N.Boliv 6.611909 14 0.94863 N-C.Peru & S.Bolivia Infinity 14 0.00000** S.Peru & S.Bolivia Infinity 14 0.00000** N.Bolivia & S.Bolivia Infinity 14 0.00000** N-C.Peru & N.AR-Jujuy Infinity 14 0.00000** S.Peru & N.AR-Jujuy Infinity 14 0.00000** N.Bolivia & N.AR-Jujuy Infinity 14 0.00000** S.Bolivia & N.AR-Jujuy 3.341006 14 0.99823 N-C.Peru & N.AR-Tucuman Infinity 14 0.00000** S.Peru & N.AR-Tucuman 64.237583 14 0.00000** N.Bolivia & N.AR-Tucuman Infinity 14 0.00000** Population Genetics and Spatial Structure in Two Andean Cats … 21
P-value for each population pair across all loci (Fisher's method) Population pair 2 df P-Value S.Bolivia & N.AR-Tucuman 3.952223 14 0.99574 N.AR-Jujuy & N.AR-Tucuman 7.164274 14 0.92819 N-C.Peru & Mendoza-AR Infinity 14 0.00000** S.Peru & Mendoza-AR Infinity 14 0.00000** N.Bolivia & Mendoza-AR Infinity 14 0.00000** S.Bolivia & Mendoza-AR Infinity 14 0.00000** N.AR-Jujuy & Mendoza-AR 51.256774 14 0.00000** N.AR-Tucuman & Mendoza-AR 54.434305 14 0.00000** * P < 0.05; ** P < 0.01; df = degree of freedom. 1 = Lambayeque (Peru), 2 = Huaraz-Huari (Peru), 3 = Yauyos-Canchayllo (Peru), 4 = Junin National Reserve (Peru), 5 = Ayacucho (Peru), 7 = Arequipa-Cuzco (Peru), 8= Tacna-Puno (Peru) and Arica (Chile), 9 = La Paz Department-Sajama National Park (Northern Bolivia), 11 = Sud Lipez-Potosi-Tarija (Southern Bolivia), 14 = Salta (Salar Diablillo, Nevado Acay)-Jujuy (Coranzuli) (Argentina), 16 = Catamarca (Antofalla)- Tucuman (Argentina), 17 = Mendoza-San Juan (Pto. Panorama) (Argentina), 18 = Buenos Aires province (Argentina), 19 = Pilcaniyeu-Patagonia (Argentina), N-C = Northern-Central, S = Southern, N = Northern, N.AR = Northern Argentina, AR = Argentina.
These gene flow estimates showed that historically the pajeros populations or subspecies have not been totally disconnected from a reproductive point of view. The different genetic heterogeneity statistics obtained showed very high levels of significance for the four jacobita population sets: GST = 0.347, GST’ = 0.415, FST = 0.360 + 0.078 (jackknife), FST = 0.197-0.569 (99% confidence interval- bootstrap), RST = 0.538 (weigthed) and RST = 0.417 (Goodman’s procedure) (Table 4c). All the individual markers yielded significant heterogeneity. On the contrary, all the FIS statistics showed no significant values, with the exception of Fca96 (p = 0.037-0.018), which agrees quite well with no endogamy and no fragmentation within each one of the populations considered. Obviously, all the FIT statistics were significant due to the FST’s values, with the exception of Fca08. The gene flow estimates were very small (from FST: Nm = 0.189-1.019 for 99 % confidence interval; private alleles: Nm = 0.439). mtDNA. For pajeros populations all the genetic heterogeneity statistics estimated were highly significant (P < 0.001). For instance, GST = 0.106, ST = 0.491, NST = 0.527 and FST = 0.521. Thus, both microsatellites and mtDNA markers revealed significant differences for the pajeros populations. Nevertheless, the degree of genetic heterogeneity was higher for mtDNA than for microsatellites. The gene flow estimates were clearly very small (Nm = 0.45-0.52) (Table 4d). For pajeros subspecies, the degree of heterogeneity is similar to that obtained among populations. All the statistics were significant and GST = 0.085, ST = 0.428, NST = 0.473 and FST = 0.468. As in the previous case, the gene flow estimates were clearly very small (Nm = 0.57-0.67) (Table 4e). For jacobita, all the statistics employed showed significant genetic heterogeneity (P <
0.001). For instance, GST = 0.302, ST = 0.442, NST = 0.568 and FST = 0.567. Therefore, there was a high coincidence between the microsatellites and the mtDNA elevated gene heterogeneity for jacobita. The gene flow estimates were clearly very small (Nm = 0.38-0.63) (Table 4f). For microsatellites, the genetic heterogeneity for jacobita was considerable higher than for pajeros, while for mtDNA the genetic heterogeneity was similar for both species. 22 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Table 4. Genetic diversity analysis for two species of Andean cats by means
of the Nei’s and the Wright’s F statistics applied to microsatellites (Ho = observed gene diversity in the total population; HS = average gene diversity within the subpopulations; HT = expected gene diversity in the total population; DST = absolute gene differentiation among subpopulations; GST = relative genetic differentiation among subpopulations with regard to the total gene diversity; GST’= the same as the previous statistics but corrected by sample size; FIS - FST - FIT = Wright’s F-statistics; RST = Slatkin [1995]’s genetic heterogeneity statistic measured by allele variances)
A
Loci Ho Hs Ht Dst Gst Gst' FIS FST FIT RST FCA31 0.680 0.747 0.891 0.145 0.162 0.172 0.096** 0.169** 0.248** 0.489 FCA24 0.593 0.625 0.772 0.147 0.191 0.202 0.088* 0.204** 0.274** 0.304 FCA45 0.631 0.747 0.888 0.141 0.159 0.169 0.158** 0.176** 0.306** 0.281 FCA96 0.655 0.744 0.869 0.125 0.144 0.153 0.140** 0.117** 0.240** 0.305 FCA294 0.628 0.730 0.832 0.102 0.123 0.131 0.109** 0.123** 0.218** 0.144 Overall 0.637 0.718 0.850 0.132 0.155 0.165 0.119** 0.158** 0.258** 0.304 Jackknife 0.119 0.158 0.258 +0.013 +0.016 +0.015 Bootstrap 99% Confidence Interval 0.092 0.119 0.228 0.150 0.192 0.294
B
Loci Ho Hs Ht Dst Gst Gst' FIS FST FIT RST FCA31 0.638 0.783 0.895 0.112 0.125 0.151 0.160** 0.142** 0.279** 0.279 FCA24 0.645 0.711 0.805 0.094 0.116 0.141 0.148** 0.196** 0.315** 0.166 FCA45 0.625 0.763 0.880 0.117 0.133 0.16 0.209** 0.162** 0.338** 0.228 FCA96 0.671 0.806 0.890 0.084 0.094 0.115 0.194** 0.074** 0.253** 0.314 FCA294 0.539 0.686 0.817 0.131 0.161 0.193 0.164** 0.084** 0.234** 0.141 Overall 0.624 0.750 0.857 0.108 0.125 0.152 0.176** 0.132** 0.285** 0.225 Jackknife 0.176 0.132 0.285 +0.011 +0.023 +0.019 Bootstrap 99% Confidence Interval 0.153 0.078 0.244 0.201 0.182 0.329
C
Loci Ho Hs Ht Dst Gst Gst' FIS FST FIT RST FCA08 0.200 0.231 0.256 0.025 0.097 0.125 0.017 0.091** 0.106 0.101 FCA31 0.423 0.486 0.684 0.198 0.290 0.352 0.128 0.286** 0.378** 0.443 FCA24 0.500 0.473 0.776 0.303 0.391 0.461 -0.051 0.394** 0.363** 0.821 FCA96 0.305 0.387 0.477 0.090 0.190 0.238 0.269* 0.171** 0.394** 0.144 FCA176 0.347 0.464 0.685 0.221 0.323 0.388 0.162 0.329** 0.438** 0.196 FCA294 0.346 0.448 0.646 0.198 0.307 0.371 0.127 0.230** 0.328** 0.206 FCA173 0.191 0.171 0.548 0.377 0.687 0.745 -0.092 0.763** 0.741** 0.789 Overall 0.330 0.380 0.582 0.202 0.347 0.415 0.097* 0.357** 0.419** 0.386 Jackknife 0.098 0.360 0.420 +0.046 +0.078 +0.061 Bootstrap 99% Confidence Interval -0.016 0.197 0.295 0.196 0.569 0.590 Population Genetics and Spatial Structure in Two Andean Cats … 23
D Statistics Value Prob Statistics Nm 746.043 0.0000** Gst = 0.1063
Hst 0.10579 0.0000** st = 0.4908 0.52 Kst 0.46800 0.0000** Nst = 0.5277 0.45 Kst* 0.28946 0.0000** Fst = 0.5212 0.46 Z 7,317.139 0.0000** Z* 8.51601 0.0000**
Snn 0.30540 0.0000*
E Statistics Value Prob Statistics Nm 378.072 0.0000** Gst = 0.0856
Hst 0.07849 0.0000** st = 0.4284 0.67 Kst 0.42002 0.0000** Nst = 0.4737 0.56 Kst* 0.25675 0.0000** Fst = 0.4683 0.57 Z 7,800.191 0.0000** Z* 8.60071 0.0000**
Snn 0.82267 0.0000**
F Statistics Value Prob Statistics Nm 34.993 0.0000** Gst = 0.3020
Hst 0.31517 0.0001** st = 0.4421 0.63 Kst 0.40028 0.0000** Nst = 0.5682 0.38 Kst* 0.36601 0.0000** Fst = 0.5673 0.38 Z 419.9267 0.0000** Z* 5.89034 0.0000**
Snn 0.48396 0.0001** * P < 0.05; ** P < 0.001. A- For 14 populations of Leopardus pajeros; B- For 5 subspecies of Leopardus pajeros; C- For 7 populations of Leopardus jacobita. Also, in this table, are shown the gene heterogeneity statistics for the mtDNA: D- For 14 populations of Leopardus pajeros; E- For 5 subspecies of Leopardus pajeros; F- For 7 populations of Leopardus jacobita. Nm = Gene flow.
Assignment Analyses
The diverse procedures employed showed a discrete efficacy to differentiate the pajeros individuals to the 14 populations analyzed. The procedures of Baudouin and Lebrun (2000), Paetkau et al., (1995) and Rannala and Mountain (1997) showed from 49.2 % to 51.8 % of animals correctly assigned. The procedures with 1,000 individuals simulated offered 55.3 % (procedure of Cornuet et al., 1999), 46.7 % (procedure of Paetkau et al., 2004), and 49.2 % (procedure of Rannala and Mountain, 1997) of animals correctly assigned. The analysis to detect recent migrants between the populations considered showed 96 possible migrant individuals for the Baudouin and Lebrun (2000), 101 possible migrants for the Paetkau et al., (1995) and 101 possible migrants for the Rannala and Mountain (1997)’s procedures without simulations, whereas the 1,000 individuals simulation procedures (Cornuet et al., 1999; Paetkau et al., 2004; Rannala and Mountain, 1997) showed 43, 17 and 38 migrant animals, respectively. It is noteworthy that the migrant analysis showed a considerably lower number 24 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al. of first migrant animals with simulations than without simulations. Therefore, the individual assignment for populations of pajeros did not discriminate well and many possible migrants among populations were determined. For subspecies, the assignment procedures revealed a better classification of animals to their respective subspecies than for populations. The Baudouin and Lebrun (2000), Paetkau et al., (1995) and the Rannala and Mountain (1997)’ procedures showed from 71.4 % to 74.4 % of animals correctly assigned. The procedures with 1,000 individuals simulated offered 82.4 % of animals correctly assigned for the Cornuet et al., (1999)’ procedure, 69.8 % for the Paetkau et al., (2004)’s procedure and 79.4 % for the Rannala and Mountain (1997)’s procedure. The analysis to detect recent migrants between the subspecies considered showed 57 possible migrant individuals for the Baudouin and Lebrun (2000), 55 possible migrants for the Paetkau et al., (1995) and 51 possible migrants for the Rannala and Mountain (1997)’s procedures without simulations, whereas the 1,000 individuals simulation procedures (Cornuet et al., 1999; Paetkau et al., 2004; Rannala and Mountain, 1997) showed 42, 10 and 38 migrant animals, respectively. Again the procedure with simulations did detect a lower number of first migrants than the procedure without simulations. The diverse procedures we used showed a high efficacy to discriminate the jacobita individuals in the four populations detected. The procedures of Baudouin and Lebrun (2000), Paetkau et al., (1995) and the Rannala and Mountain (1997) showed 96.1 % of animals correctly assigned (one animal from Central Peru was assigned to the southern Peru and northern Bolivia and three animals from southern Peru and northern Bolivia were assigned to Central Peru). The procedures with 1,000 individuals simulated offered 92.2 % (procedure of Cornuet et al.,1999), 96.1% (procedure of Paetkau et al., 2004) and 98 % (procedure of Rannala and Mountain, 1997) of animals correctly assigned. The analysis to detect recent migrants between the populations showed two migrant individuals (one from Central Peru to southern Peru-northern Bolivia and the other from southern Peru-northern Bolivia to Central Peru) with the procedures of Baudouin and Lebrun (2000), Paetkau et al., (1995) and Rannala and Mountain (1997) without simulations. The 1,000 individual simulation procedures of (Cornuet et al., 1999; Paetkau et al., 2004; Rannala and Mountain, 1997) showed only one migrant animal (from southern Peru-northern Bolivia to Central Peru). These results were the same as that of the genetic heterogeneity statistic analyses. The degree of genetic heterogeneity is higher in jacobita than in pajeros.
Demographic Changes
Microsatellites. For the global pajeros sample, two markers yielded results in agreement with a recent bottleneck (Fca96 and Fca294) for the IAM model. However, when the TPM and the SMM models were employed, no markers showed any individual trend in favor of a bottleneck. For the Wilcoxon test, a significant result which agrees quite well with a bottleneck was obtained for the IAM model (P = 0.0156), but this is the less probable mutation model for microsatellites. The sign test and the standardized differences test were significant for the SMM model (P = 0.0129 and 0.00026), contrary to a bottleneck explanation (population expansion or population subdivision). The allele frequency distribution was a normal L-shaped one, which agrees quite well with a stable demographic population. Thus, at a global level for species, there was no clear evidence of a recent Population Genetics and Spatial Structure in Two Andean Cats … 25 bottleneck for pajeros. For pajeros populations, the situation was as follows: 1- some populations did show evidence against a recent bottleneck (population expansion or population subdivision). These were the cases of Huaraz (standardized differences tests for TPM, p = 0.00001, and SMM, p = 0.00000), Ayacucho (Sign tests for TPM, p = 0.0085, and SMM, p = 0.0107, and standardized differences test for TPM, p = 0.0202, and SMM, p = 0.0077), Tacna (sign test for SMM, p = 0.011), La Paz (standardized differences test for SMM, p = 0.0332), Potosi (Sign tests for TPM, p = 0.0109, and SMM, p = 0.0112, and standardized differences test for TPM, p = 0.00000, and SMM, p = 0.00000), Jujuy (Sign tests for TPM, p = 0.0113, and SMM, p = 0.0119, and standardized differences test for TPM, p = 0.00000, and SMM, p = 0.00000), Catamarca (Sign tests for TPM, p = 0.0113, and SMM, p = 0.0118, and standardized differences test for TPM, p = 0.0009, and SMM, p = 0.00001), Buenos Aires (Sign test for SMM, p = 0.0378) 2- some populations showed some trends in agreement with a bottleneck. These were the cases of Lambayeque, Buenos Aires and Patagonia, which showed a shifted mode distribution. However, these three populations were those with the lowest sample sizes and thus these results in favor of a bottleneck could be spurious. 3- Some populations did not show any significant trend, and were probably stable population (Junin, Yauyos, Arequipa, and Tucuman). Therefore, two results are relevant in this case. Some populations showed a striking population subdivision (or population expansion) and all these populations are in the central distribution range of this species: southern Bolivia and northern Argentina (Jujuy and Catamarca). The possible population subdivision is more restricted in the extreme areas of the distribution of this species. On the other hand, no population revealed clear evidences of bottlenecks, which have important repercussions on conservation politics. For pajeros subspecies, garleppi showed significant subdivision or population expansion for the sign tests for TPM, p = 0.0124, and SMM, p = 0.0115, and for the standardized differences test for TPM, p = 0.0022, and SMM, p = 0.0000 as well as steinbachi (sign tests for TPM, p = 0.0117, and SMM, p = 0.0119, and for the standardized differences test for TPM, p = 0.0000, and SMM, p = 0.0000), and also budini (standardized differences test for IAM, p = 0.0459, TPM, p = 0.0275, and SMM, p = 0.0002). For pajeros pajeros and crucinus, the results are the same that for the populations of Buenos Aires and Patagonia. Thus, for the three subspecies where sample sizes are large enough, no evidence of recent bottlenecks was detected, but possible population division or expansion was determined to be significantly detected. The test of Kimmel et al., (1998) did detect some significant demographic changes for the pajeros subspecies (Table 5a). All subspecies, less budini, showed an initial bottleneck followed by a population expansion: garleppi (ln = 0.510, t = 3.998, 4 df, p < 0.02), steinbachi (ln = 0.193, t = 2.911, 4 df, p < 0.05), pajeros (ln = 1.373, t = 9.475, 4 df, p < 0.001) and crucinus (ln = 0.715, t = 4.905, 4 df, p < 0.01). Especially strong was this initial bottleneck followed by a population expansion in the two most southern subspecies (pajeros and crucinus). The test of Zhivotovsky et al. (2000) yielded some evidence of population expansion for pajeros taken as a whole (Sk = 0.721, t = 2.999, 4 df, p < 0.05) (Table 5a). Subspecies, garleppi (Sk = 0.291, t = 0.987, 4 df, NS) and steinbachi (Sk = 0.219, t = 0.548, 4 df, NS) did not show evidence of demographic changes, but the three southern subspecies in Argentina revealed clear symptoms of a significant population expansion: budini (Sk = 0.611, t = 2.858, 26 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
4 df, p < 0.05), pajeros (Sk = 0.429, t = 2.754, 4 df, p < 0.05) and crucinus (Sk = 0.616, t = 2.974, 4 df, p < 0.05). Therefore, the three Argentinean subspecies revealed clearer evidence of population expansion.
Table 5. Statistics calculated for the imbalance index (Kimmel et al., 1998)
and for the test of Zhivotovsky et al. (2000) (Sk) to detect demographic changes applied to microsatellites in two species of Andean cats. t = Student´s t test
A Statistics garleppi steinbachi budini pajeros crucinus _ ^ V 9.7766 9.0598 25.6105 10.7888 10.3333 _ ^
Po 0.2802 0.2505 0.1243 0.3933 0.3000 ^ 1.6658 1.2132 0.8035 3.9481 2.0439 ^ ln 0.5103 0.1932 -0.2187 1.3732 0.7149 t 3.998** 2.911* 1.924NS 9.475**** 4.905*** Kimmel et al., (1998)’s imbalance index.
Statistics garleppi steinbachi budini pajeros crucinus total Var 6.1561 5.3760 11.6552 6.3747 6.0333 19.9915 _ K 153.6735 129.7173 300.9839 136.0246 88.9111 619.8681
Sk 0.2913 0.2195 0.6109 0.4294 0.6159 0.7213 t NS NS 2.098* 1.999* 2.224* 2.789* Zhivotovsky et al., (2000)’ test.
B Statistics N-C. Peru S.Peru-N.Bolivia N.AR Mendoza-AR Total _ ^ V 0.2444 1.7206 1.8471 3.5470 2.8776 _ ^
Po 0.7841 0.4982 0.5648 0.5931 0.3901 ^ 0.7281 1.0468 1.7309 3.8491 1.0329 ^ ln -0.3173 0.0457 0.5486 1.3478 0.0321 t 1.747NS 2.512NS 4.154** 9.475**** 2.478NS Kimmel et al., (1998)’s imbalance index
Statistics N-C. Peru S.Peru-N. Bolivia N.AR Mendoza-AR Total Var 0.3829 4.9111 1.3286 1.1077 5.1070 _ K 0.7633 303.6807 4.9049 3.8760 201.6916 Population Genetics and Spatial Structure in Two Andean Cats … 27
Statistics N-C. Peru S.Peru-N. Bolivia N.AR Mendoza-AR Total
Sk 1.6041 -1.3602 0.9184 0.9369 -0.4233 t 6.416*** -5.441** 3.673* 3.748* 1.693NS Zhivotovsky et al., (2000)’ test. * Statistically significant value at P < 0.05, ** P < 0.02, *** P <0.01; ****P < 0.001, NS = not significant. A- For Leopardus pajeros subspecies. B- For the four different gene pools found in Leopardus jacobita. N-C = Northern-Central, S = Southern, N = Northern, N.AR = Northern Argentina, AR = Argentina.
For the global jacobita sample, no clear evidence of significant recent bottleneck was detected independently of the mutation microsatellite model employed. One marker, Fca08, showed a significant value for the TPM and SMM mutation models, but this value was negative which indicates the contrary of a bottleneck or population fragmentation. The sign test, the Wilcoxon test (for TPM and SMM) and the normal L-shaped distribution did not show any evidence of bottleneck. The standardized differences test showed a significant negative value for the TPM and the SMM models, and thus also does not agree with a bottleneck or population fragmentation. The unique evidence of a possible recent bottleneck was shown through the Wilcoxon test with the IAM model. However, this is the mutation model less probable in microsatellites. Thus, there was no strong evidence for a bottleneck in jacobita as a species. For jacobita populations, the situation was as follows: For Central Peru, no individual loci nor the sign, standardized differences and Wilcoxon tests offered any evidence of a recent bottleneck. Nevertheless, the graphic method showed a shifted mode, which is compatible with a recent bottleneck. For southern Peru-northern Bolivia, no test yielded any evidence of bottleneck. For southern Bolivia-northern Argentina, no test detected any evidence of bottleneck. However, the sign and the standardized differences tests for TPM and SMM showed strong evidence of population expansion. For the Mendoza population no test showed any evidence of a recent bottleneck. The Kimmel et al., (1998)’s test did not detect any significant demographic change for the total jacobita sample (ln = 0.032, t = 2.478, 6 df, NS) (Table 5b). The same was detected for the North-Central Peru and the southern Peru-northern Bolivia populations (ln = -0.317, t = 1.747, 6 df, NS and ln = 0.045, t = 2.512, 6 df, NS). In contrast, the southern Bolivia-northern Argentina and the Mendoza populations showed a significant expansion after a bottleneck (ln = 0.549, t = 4.154, 6 df, P < 0.05 and ln = 1.348, t = 9.237, 6 df, P < 0.01). The test of Zhivotovsky et al. (2000) yielded different results than previous tests (Table 5b). For the total jacobita sample some evidence of a bottleneck was obtained, although it was not significant (Sk = -0.423, t = 1.693, 6 df, NS). The southern Peru-northern Bolivia population was that with the highest average variance, but it was also the population that showed the most clear evidence of a bottleneck (Sk = -1.360, t = -5.441, 6 df, P < 0.05). On the other hand, the other three populations (North-Central Peru, southern Peru-northern
Bolivia and Mendoza) showed significant evidence of population expansion (Sk = 1.604, t = 6.416, 6 df, P < 0.01; Sk = 0.918, t = 3.673, 6 df, P < 0.05, and Sk = 0.937, t = 3.748, 6 df, P < 0.05). Therefore, there is no clear indication of bottleneck events with all of the analyses for the jacobita populations studied, but there is a clear indication that the two most southern populations studied (southern Bolivia-northern Argentina and Mendoza) experienced population expansions during its natural history. 28 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Leopardus pajeros garleppi
Constant Size.
Demographic Change.
Leopardus pajeros steinbachi
Constant Size.
Population Genetics and Spatial Structure in Two Andean Cats … 29
Demographic Change.
Leopardus pajeros budini
Constant Size.
Demographic Change.
30 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Leopardus pajeros pajeros-crucinus
Constant Size.
Demographic Change. (A)
Leopardus jacobita population 2
Constant Size. Population Genetics and Spatial Structure in Two Andean Cats … 31
Demographic Changes.
Leopardus jacobita Population 3
Constant Size.
Demographic Change. (B)
Figure 1. Mismatch distribution (pairwise sequence differences) for Leopardus pajeros’ subspecies assuming a constant population size and assuming a population in expansion (A). Mismatch distribution (pairwise sequence differences) for Leopardus jacobita’ populations assuming a constant population size and assuming a population in expansion (B). Pop2 = Southern Peru-Northern Bolivia; Pop3 = Southern Bolivia-Northern Argentina.
Table. 6. Demographic statistics applied to: the total sample and the subspecies analyzed of Leopardus pajeros for the mtDNA control region (A). the total and the four gene pools analyzed of Leopardus jacobita for the mtDNA control region (B). However, two of the L. jacobita populations did not present mitochondrial gene diversity and therefore these analyses were not possible
A garleppi Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 P[D < -1.261] = P[Fs < P[D* < -3.038] = 0.0070** P[F* < -2.755] = 0.0110* P[rg < 0.041] = 0.5190 P[R2 < 0.065] = 0.2310 0.0690 1.111] = 0.7100 [-3.211, 1.951] [-3.242, 2.375] [0.0076, 0.2334] [0.0284, 0.1794] [-1.562, 2.076] [-12.807, 15.566] steinbachi Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 P[Fs < P[R2 < 0.1352] = P[D < 0.024] = 0.5880 P[D* < -0.399] = 0.3140 P[F* < -0.311] = 0.3640 P[rg < 0.1034] = 0.9840 7.202] = 0.9840 0.6880 [-2.199, 2.331] [-3.556, 1.575] [-3.846, 1.874] [0.0092, 0.0921] [-9.104, 8.661] [0.0624, 0.2105] budini Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 P[Fs < P[D < 0.604] = 0.7860 P[D* < 1.013] = 0.8960 P[F* < 1.021] = 0.8950 P[rg < 0.0296] = 0.9420 P[R2 < 0.1263] = 0.8220 2.872] = 0.8220 [-2.134, 2.461] [-4.028, 1.659] [-4.539, 2.415] [0.0034, 0.0593] [0.0407, 0.1771] [-13.797, 19.638] pajeros- Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 crucinus P[Fs < P[D < -0.642] = 0.7570 P[D* < 0.875] = 0.8260 P[F* < 0.905] = 0.8030 P[rg < 0.263] = 0.8154 P[R2 < 0.204] = 0.6346 2.289] = 0.8548 [-1.700, 2.162] [-1.787, 1.598] [-1.926, 1.808] [0.0294, 0.8367] [0.0958, 0.3495] [-4.358, 6.231]
Total Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 sample P[Fs < P[D < 0.322] = 0.6950 P[D* < 0.418] = 0.6630 P[F* < 0.447] = 0.7380 P[rg < 0.056] = 0.7930 P[R2 < 0.098] = 0.8070 1.613] = 0.7100 [-1.951, 2.454] [-4.302, 1.902] [-3.819, 2.600] [0.0021, 0.0427] [0.0334, 0.1576] [-23.969, 34.575]
B L.jacobita- Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 Pop2 P[Fs < P[D < 0.631] = 0.7660 P[D* < 1.186] = 0.9360 P[F* < 1.189] = 0.9030 P[rg < 0.361] = 0.8651 P[R2 < 0.212] = 0.9244 2.669] = 0.9149 [-2.047, 2.266] [-2.627, 1.501] [-3.124, 1.797] [0.0187, 0.7724] [0.0786, 0.2553] [-4.172, 7.572]
L.jacobita- Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 Pop3 P[F* < 0.488] = P[Fs < P[D < -0.769] = 0.2120 P[D* < 0.866] = 0.8210 P[rg < 0.739] = 0.9692 P[R2 < 0.095] = 0.0671 0.6470 0.909] = 0.7454 [-1.723, 2.139] [-2.386, 1.186] [0.0394, 0.8300] [0.0947, 0.2632] [-2.784, 1.559] [-3.052, 3.557]
Total Tajima D Fu and Li D* Fu and Li F* Fu’s Fs raggedness rg R2 P[Fs < P[D < 1.052] = 0.8870 P[D* < 1.102] = 0.8930 P[F* < 1.271] = 0.9160 P[rg < 0.372] = 0.8860 P[R2 < 0.202] = 0.9884 3.746] = 0.9569 [-2.016, 2.994] [-3.626, 1.510] [-3.387, 2.177] [0.0185, 0.839] [0.0416, 0.2389] [-5.201, 9.119] * P < 0.05 and ** P < 0.01 mean significant population expansions. P[] = Probability of the amount found for each one of the statistics analyzed ; in [] = 99 % confidence intervals for each one of the statistics analyzed. Pop2 = Southern Peru-Northern Bolivia; Pop3 = Southern Bolivia-Northern Argentina.
34 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
A. Leopardus pajeros
Population Genetics and Spatial Structure in Two Andean Cats … 35
B. Leopardus jacobita
Figure 2. Tajima (1989b)’s procedure (segregating sites-frequency spectrum) to determine demographic changes for the global Leopardus pajeros population analyzed (A). Tajima (1989b)’s procedure (segregating sites-frequency spectrum) to determine demographic changes for the global Leopardus jacobita population analyzed (B).
mtDNA. The mismatch distribution (pairwise sequence differences) and their respective * associated statistics (rg and the R2 statistics), as well as the Tajima D, Fu and Li D , Fu and Li F* and Fu’s Fs tests and the spectrum analysis, did not show any evidence of population expansion for the total pajeros sample (Figure 1a, 2a). Huaraz, Yauyos, Ayacucho, Arequipa, Tacna-Puno, Jujuy-Salta, Tucuman, Mendoza, and Buenos Aires-Patagonia (these both populations were put together in this analysis) did not show evidence of demographic changes, with apparent constant sizes. However, Junin (for Fu and Li D* and Fu and Li F* tests), La Paz (for Fu and Li D* test) and Potosi (for Fu’s Fs and rg tests) showed some possible evidences in favor of a bottleneck. No results agree with a population expansion, which disagree with the results obtained for microsatellites in this species. For subspecies, garleppi showed some evidence of population expansion (for Fu and Li D* and Fu and Li F* tests), while for steinbachi, some evidence of bottlenecks was discovered (for Fu’s Fs and rg tests). Budini and pajeros-crucinus did not show any evidence of demographic change (Figure 1b, Table 6). The mismatch distribution (pairwise sequence differences) and their respective associated * * statistics (rg and the R2 statistics), as well as the Tajima D, Fu and Li D , Fu and Li F and Fu’s Fs tests, did not show any evidence of population expansion for the total jacobita sample (Figure 1c, Table 6). In fact, two tests and the spectrum analysis (Figure 2b) showed evidence of bottlenecks (Fu’s Fs and R2 tests) and the others showed trends in favor of a bottleneck globally for jacobita. The southern Peru-northern Bolivia population did not show any significant analyses, but all of them yielded a trend in favor of a bottleneck (Figure 1c, Table 6). A similar trend, but less noteworthy, was observed for the southern Bolivia-northern Argentina population with the rg statistic being significantly in favor of a bottleneck (Figure 1c, Table 6). The Central Peru and the Mendoza populations did not show genetic variability or enough sequences analyzed for this analysis. Thus a differential trend was observed between nuclear microsatellites and mtDNA for jacobita showing with the first, no clear evidences of bottleneck and some clear evidences of population expansions for the southern 36 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al. populations whereas mtDNA showed constant evidence of bottleneck for all the populations studied.
Microsatellite Mutation Models
Four noteworthy results were obtained after applying the maximum likelihood method of Nielsen (1997) to analyze the microsatellite mutation models for pajeros and jacobita as species. Firstly, the average estimates were practically identical for uni-step and multi-step mutations models for pajeros (Uni-step = 26.59 + 13.45 vs. Multi-step = 27.56 + 14.31) and also for jacobita (Uni-step = 3.18 + 1.97 vs. Multi-step = 2.88 + 2.37; Table 7).
Table 7. Analysis of the microsatellite mutation models in Leopardus pajeros (A) and Leopardus jacobita (B) by using the maximum likelihood of Nielsen (1997)
A. Leopardus pajeros UNI-STEP M.M MULTI-STEP M.M 2 Microsatellites u log m log % likelihood likelihood MSM Fca24 5.424 8.299 -38.4772 11.391 -38.8448 3.45% NS Fca31 23.405 40.257 -75.8036 44.704 -77.9281 10.8% 4.25* Fca45 35.942 34.504 -72.4796 34.504 -71.1482 1.0% NS Fca96 28.831 33.155 -69.3298 33.155 -74.0481 5.9% 9.44** Fca294 24.822 16.754 -58.7118 14.396 -57.3393 1% NS Mean 23.658 26.594 27.560 4.43% +11.308 +13.452 +14.315 +4.01%
Ne 316,592 328,095
B. Leopardus jacobita UNI-STEP M.M MULTI-STEP M.M 2 Microsatellites u log m log % MSM likelihood likelihood Fca08 0.444 1.311 -11.8512 1.437 -11.7707 1.0% NS Fca24 8.134 6.263 -24.7554 7.809 -24.9536 1.0% NS Fca31 5.993 3.476 -21.3529 3.476 -21.5262 1.0% NS Fca96 1.636 2.036 -17.2288 1.259 -16.7964 15.7% NS Fca173 0.581 0.889 -8.5185 0.889 -8.5581 0.0% NS Fca176 5.897 5.100 -27.0817 2.299 -22.2497 13.25% 9.66** Fca294 1.581 3.019 -13.6223 3.019 -13.7304 0% NS Mean 3.466 3.181 2.884 4.56% +3.121 +1.972 +2.369 +6.82%
Ne 37,869 34,333
Ne estimated by the moment method; u = uni-step estimated assuming a uni-step mutation model; m = multi-step estimated assuming a multi-step mutation model; % multi-step (MSM) = most probable percentage of multi-step mutations; test to measure if the percentage of multi- step mutations is significantly better than the uni-step mutation model; NS = Non significant
values; * P < 0.05, ** P < 0.01; df = degree of freedom; Ne = Effective Numbers.
Population Genetics and Spatial Structure in Two Andean Cats … 37
Thus, the estimation of was practically the same independently of the mutation model employed and consequently the estimation of other parameters from , as the effective numbers or the mutation rates per generation, are not affected by the microsatellite mutation model employed. Secondly, the percentage of multi-step mutations was small and very similar for both species (4.43 % + 4.09 % for pajeros and 4.56 % + 6.82 % for jacobita). In pajeros, two out of five microsatellites presented a significant multi-step mutation model (Fca31 with a percentage of 10.8 %, 2 = 4.25, 1 df, P < 0.05, and Fca96 with a percentage of 5.9 %, 2 = 9.44, 1 df, P < 0.01) and in jacobita, one out of seven microsatellites showed a significant multi-step mutation model (Fca176 with a percentage of 13.25 %, 2 = 9.66, 1 df, P < 0.01). Therefore, the major part of the microsatellites studied did not significantly departure from a uni-step mutation model in both species. Thirdly, within each species, each microsatellite showed significantly different mutation rates. For pajeros, the order of the mutation rates from highest to lowest values was Fca31 > Fca45 > Fca96 > Fca294 > Fca24, and for jacobita, the same order was Fca176 > Fca24 > Fca31 > Fca294 > Fca08 > Fca173. Fourthly, all the estimations, independently of the procedures employed, showed that the values for pajeros were substantially higher than for jacobita. Similar to the previous analyses, pajeros showed noteworthy higher gene diversity than jacobita.
Effective Numbers
We estimated possible historical (long-term) effective population sizes for both cat species studied. The first method was throughout the Nielsen’s (1997) maximum likelihood procedure with a mutation rate per generation of 2.1 x 10-5. For pajeros, the effective numbers were 316,592 individuals for a uni-step mutation model and 328,095 individuals for a multi- step mutation model. For jacobita, the same values were 37,869 and 34,333 individuals, respectively. Thus, pajeros showed an effective number around eight times higher than jacobita. The same was observed for other three Neotropical wild cats: jaguar (270,056), puma (589,911) and ocelot (464,286). Jacobita yielded values that were 7, 16 and 12 times lower than these wild cat species, respectively. The average expected heterozygosity was used as another method to estimate historical effective numbers for microsatellite markers. For the same mutation rate, as before, the effective numbers for jacobita and pajeros were 11,926 individuals and 82,782 individuals respectively. Thus, pajeros yielded effective numbers seven times higher than jacobita. The other three cats showed the following effective numbers: jaguar (155,180), puma (238,639) and ocelot (449,887). This means that jaguar, puma and ocelot have an effective number 13, 20 and 38 times higher than jacobita. Additionally, our analyses showed that the methods of Hill (1981), by linkage disequilibrium, and Pudovkin et al., (1996), by heterozygote excess, were not appropriate for our data or for other data we have generated from other wild cats such as the jaguar and the ocelot. The second method offered average values of 8.93 and 4.99 (confidence interval: infinite, infinite) for the jaguar and the ocelot and negative values for pajeros and jacobita. Thus, these values have no biological meaning. The values of the Hill (1981) method also did not make sense biologically, but it is interesting to remark on one aspect.
38 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Population Genetics and Spatial Structure in Two Andean Cats … 39
(A)
40 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
(B)
Figure 3. Effective numbers estimated with the Griffiths and Tavaré (1994)’s procedure from the distribution of for the pajeros subspecies (A) and for the four jacobita populations (B) determined by this study. Population 1 = North-Central Peru; Population 2 = Southern Peru-Northern Bolivia; Population 3 = Southern Bolivia-Northern Argentina; Population 4 = Mendoza-Argentina.
The jaguar (average value: 213 [ci: 167, 290]) and the ocelot (average value: 170 [ci: 138, 217]) showed higher values than pajeros (average value: 71 [ci: 63, 80]) and all of them presented substantially higher values than jacobita (average value: 13 [ci: 10, 16]). This is expected for the geographical distribution of these species. For example, the jaguar could have nearly 3 times the effective size of pajeros and around 16 times the effective size of jacobita, and pajeros could have an effective number 5.5 times higher than jacobita. Henceforth, independently of the procedure employed, jacobita showed considerable lower effective numbers than other Neotropical wild cats. We also estimated the effective numbers with the Griffiths and Tavaré (1994)’s procedure for the pajeros subspecies and for the four jacobita populations determined by this study (Figure 3). For pajeros, pajeros pajeros showed the lowest effective number (109,549 + 48,769), while budini yielded the highest effective numbers (638,031 + 372,618). For jacobita, North-Central Peru showed the lowest effective number for this species (12,981 + 3,522), while southern Peru-northern Bolivia (50,799 + 40,780) and southern Bolivia- northern Argentina (50,582 + 37,255) yielded the highest effective numbers. As in previous analyses, the effective numbers of pajeros was around 10 times higher than those of jacobita. However, this procedure seems to overestimate the effective numbers in comparison with the other procedure employed. Such as Waples (1989) commented, these procedures could be more useful for obtaining effective numbers for species as a whole than for populations within species. Thus, the effective numbers obtained with the Nielsen (1997) and the expected average heterozygosity could show more realistic effective numbers than the procedure by Griffiths and Tavaré (1994) for these two species.
Spatial Genetic Structure
The Mantel test showed a very high significant relationship between the genetic distances and the geographic distances among the pajeros populations (r = 0.588; approximate Mantel Population Genetics and Spatial Structure in Two Andean Cats … 41 t-test, t = 4.926, P = 0.0000). This means that the geographical distances explained 34.57 % of the genetic distances found among pajeros populations. Thus, the geographical distance is a relevant fact in the genetic differentiation of the pajeros populations. The spatial genetic analysis was not done for pajeros subspecies because the subspecies number was very small. The pajeros populations significantly fit with an isolation by distance pattern by means of the IBD program [lineal model: intercept = 0.003303 + 0.00949, slope = 0.0000708 + 0.00000573, R2 = 0.325; 99% confidence intervals with 10,000 bootstraps over all the individuals: intercept = (-0.01648, 0.002231), slope = (0.0000561, 0.00008774), R2 = (0.155, 0.535)]. These results were obtained with normal data, but the results with the log genetic or log geographical distances or both log data simultaneously were very similar (Table 8a). The spatial autocorrelation analysis for pajeros showed a monotonic clinal pattern for this species for 3DC for both Moran’s and Geary’s indexes (0.12, -0.05, -0.28 and 0.73, 0.98, 1.29, respectively) as well as for 4 DC (0.11, -0.10, -0.25, -0.34 and 0.76, 0.99, 1.29, 1.36, respectively). For the 3 CD analysis, 68 % (17/25) of the correlograms were significant and 49.33 % (37/75) of the autocorrelation coefficients were significant with the Moran’s I, with these proportions significantly higher than the 5 % type I error. For Geary’s c, these percentages were 44 % (11/25) and 38.67 % (29/75), respectively, and also significantly higher than the 5 % type I error. For the 4 CD analysis, 68 % (17/25) of the correlograms were significant and 40 % (40/100) of the autocorrelation coefficients were significant with the Moran’s I, with these proportions significantly higher than the 5 % type I error. For Geary’s c, these percentages were 28 % (7/25) and 33 % (33/100), respectively, and also significantly higher than the 5 % type I error. Henceforth, pajeros populations showed a significant spatial structure, of which about 30-35 % could be explained by an isolation-by- distance model. The Mantel test showed a very high significant relationship between the genetic distances and the geographic distances among the jacobita populations (r = 0.794; approximate Mantel t-test, t = 3.392, P = 0.0003). This significant correlation was even greater than that found for pajeros. The jacobita populations clearly fit with an isolation by distance pattern by means of the IBD program [lineal model: intercept = -0.02223 + 0.01058, slope = 0.000077 + 0.0000107, R2 = 0.637; 99% confidence intervals with 10,000 bootstraps over all the individuals: intercept = (-0.04826, 0.00136), slope = (0.0000528, 0.00009732), R2 = (0.115, 0.896)]. These results were obtained with normal data, but the results with the log genetic or log geographical distances or both log data simultaneously were very similar (Table 8b). These results showed that the isolation-by-distance was clearly more intense in jacobita than in pajeros. The spatial autocorrelation analysis for jacobita showed a clear monotonic clinal pattern for this species for 3DC for both Moran’s and Geary’s indexes (0.06, -0.15, -0.41 and 0.65, 0.70, 1.65, respectively) as well as for 4 DC (0.06, -0.12, -0.23, -0.63 and 0.65, 0.76, 1.22, 1.98, respectively). For the 3 CD analysis, 32 % (8/25) of the correlograms were significant and 26.67 % (20/75) of the autocorrelation coefficients were significant with the Moran’s I, with these proportions significantly higher than the 5 % type I error. For Geary’s c, these percentages were 40 % (10/25) and 33.33 % (25/75), respectively, and also significantly higher than the 5 % type I error. For the 4 CD analysis, 36 % (9/25) of the correlograms were significant and 17 % (17/100) of the autocorrelation coefficients were significant with the Moran’s I, with these proportions significantly higher than the 5 % type I error.
42 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Table 8. Isolation by distance (IBD) analyses by means of reduced major axis regression. For the Leopardus pajeros populations analyzed (A). For the Leopardus jacobita populations analyzed (B)
A Normal Data INTERCEPT SLOPE R2 Linear model: estimate 3.303e-03 7.084e-05 0.325 + SD 9.494e-03 5.735e-06 Jack: 3.399e-03 7.066e-05 + SD 7.298e-03 5.934e-06 99% CI: lin.model: (-2.163e-02, 2.823e-02) (5.578e-05, 8.590e-05) Jack: (-1.576e-02, 2.256e-02) (5.507e-05, 8.624e-05) Boot: (-1.648e-02, 2.231e-02) (5.608e-05, 8.774e-05) (0.155, 0.535)
Log Geographical Distance INTERCEPT SLOPE R2 Linear model: estimate -0.4661 0.1872 0.333 + SD 0.0460 0.0151 Jack: -0.4640 0.1866 + SD 0.0514 0.0169 99% CI: lin.model: (-0.5869,-0.3453) (0.1477, 0.2268) Jack: (-0.5989,-0.3291) (0.1421, 0.2311) Boot: (-0.6100,-0.3480) (0.1494, 0.2324) (0.183, 0.508)
Log Genetic Distance INTERCEPT SLOPE R2 Linear model: estimate -1.676 4.072e-04 0.284 + SD 0.056 3.395e-05 Jack: -1.676 4.061e-04 + SD 0.068 3.917e-05 99% CI: lin.model: (-1.823,-1.528) (3.180e-04, 4.963e-04) Jack: (-1.854,-1.497) (3.032e-04, 5.089e-04) Boot: (-1.854,-1.525) (3.216e-04, 5.127e-04) (0.121, 0.474)
Log Geographical Distance-Log Genetic Distance INTERCEPT SLOPE R2 Linear model: estimate -4.374 1.076 0.330 + SD 0.265 0.087 Jack: -4.366 1.074 + SD 0.312 0.096 99% CI: lin.model: (-5.069,-3.678) (0.848, 1.304) Population Genetics and Spatial Structure in Two Andean Cats … 43
INTERCEPT SLOPE R2 Jack: (-5.185,-3.547) (0.821, 1.327) Boot: (-5.256,-3.696) (0.865, 1.346) (0.137, 0.553)
B Normal Data INTERCEPT SLOPE R2 Linear model: estimate -0.02223 7.737e-05 0.637 + SD 0.01058 1.069e-05 Jack: -0.02283 7.840e-05 + SD 0.00845 7.819e-06 99% CI: lin.model: (-0.05251, 0.00805) (4.679e-05, 1.080e-04) Jack: (-0.04689, 0.00122) (5.615e-05, 1.006e-04) Boot: (-0.04826, 0.00136) (5.282e-05, 9.732e-05) (0.115, 0.896)
Log Geographic Distance INTERCEPT SLOPE R2 Linear model: estimate -0.3617 0.1416 0.518 + SD 0.0651 0.0225 Jack: -0.3632 0.1424 + SD 0.0771 0.0267 99% CI: lin.model: (-0.5481,-0.1754) (0.0771, 0.2061) Jack: (-0.5825,-0.1440) (0.0664, 0.2184) Boot: (-0.5900,-0.2129) (0.0875, 0.2174) (0.112, 0.817)
Log Genetic Distance INTERCEPT SLOPE R2 Linear model: estimate -2.479 1.074e-03 0.509 + SD 0.171 1.727e-04 Jack: -2.477 1.057e-03 + SD 0.202 1.978e-04 99% CI: lin.model: (-2.968,-1.989) (5.799e-04, 1.568e-03) Jack: (-3.053,-1.902) (4.945e-04, 1.620e-03) Boot: (-2.923,-1.841) (5.755e-04, 1.626e-03) (0.136, 0.866)
Log Geographical Distance-Log Genetic Distance INTERCEPT SLOPE R2 Linear model: estimate -7.192 1.965 0.554 + SD 0.870 0.301 Jack: -7.185 1.962 + SD 0.792 0.259 99% CI: lin.model: (-9.682,-4.702) (1.103, 2.827)
44 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Table 8. (Continued)
INTERCEPT SLOPE R2 Jack: (-9.439,-4.931) (1.224, 2.700) Boot: (-9.757,-4.770) (1.163, 2.839) (0.112, 0.881) SD = Standard Deviation; Jack = Jackknife over all the points; CI = Confidence Interval; Boot = 1,000 bootstraps.
For Geary’s c, these percentages were 32 % (8/25) and 22 % (22/100), respectively, and also significantly higher than the 5 % type I error. Henceforth, a very strong spatial genetic structure related with isolation by distance and monotonic clinal trends is present in jacobita.
Craniometric Differences among L. colocolo colocolo and L. pajeros garleppi
Out of nine biometric skull variables measured in two collections of Pampas cat, one belonging to Central Chile (L. colocolo colocolo sensu García-Perea, 1994) and other belonging to northern Bolivia (L. pajeros garleppi), only one variable, FML, presented significant differences (t = 4.26, P < 0.01). The difficulty in obtaining skulls of these small Andean cats and hence small sample sizes made it difficult detect significant differences between these two putative species, even though García-Perea (1994) determined qualitative differences among the skulls of these taxa.
DISCUSSION
Relevant Genetics Aspects of the Pampas and the Andean Mountain Cat Populations
It was clear that the gene diversity of the Pampas cat was considerably higher than that estimated for jacobita, both for microsatellites as well as mtDNA. Our work, which is complementary to that of Cossíos et al., (2009), disagree with the work of Johnson et al., (1999), who found very moderate levels of gene diversity for the Pampas cat. This was motivated by the very small sample analyzed by these authors. For example, the levels of gene diversity of pajeros for microsatellites (H = 0.73) is somewhat lower than some Neotropical felines (L. pardalis, H = 0.92; L. wiedii, H = 0.85; Panthera onca, H = 0.85; Ruiz-García, 2001; Ruiz-García et al., 2006), but similar or higher than other Neotropical mammals including other felines (Puma concolor, H = 0.75; Puma (= Herpailurus) yaguaroundi, H = 0.62; Mazama americana, H = 0.64; Odocoileus virginianus, H = 0.61; Hydrochoerus hydrochaeris, H = 0.61; Ruiz-García, 2001; Ruiz-García et al., 2009; Maldonado et al., 2011). In contrast, the gene diversity level of jacobita is one of the lowest recorded for a Neotropical mammal (H = 0.42) but similar to another Andean carnivore, the Andean o spectacled bear (Tremarctos ornatus) (H = 0.45-0.55; Ruiz-García, 2003, 2007, 2012; Ruiz-García et al., 2003, 2005). In this topic, our study reflected similar findings to that reported by Napolitano et al., (2008). For mtDNA, they found two haplotypes for the Andean Population Genetics and Spatial Structure in Two Andean Cats … 45 cat and 17 haplotypes for the Pampas cat. Our number of mtDNA haplotypes was higher with five and 41 haplotypes respectively, because our sample is considerable larger as was the geographic area we sampled. In pajeros, the most northern population sampled (Lambayeque), was that which showed the lowest gene diversity for both microsatellites and mtDNA. Other northern-central Peru populations (as Huaraz and Yauyos) showed similarly low values of mtDNA gene diversity. Thus, the northern and Central Peruvian Pampas cat populations could be the most recent established Pampas cat populations losing gene diversity by gene drift during the colonization process. The Pampas cat populations from northern Argentina are those with the highest levels of gene diversity for microsatellites (Mendoza) and for mtDNA (Salta-Jujuy). In the same way, the budini subspecies (which inhabits the northern Argentina) showed the highest levels of gene diversity for both microsatellites and mtDNA. Thus, the northern Argentinean Pampas cat populations could be the original populations from which this species expanded. Dobzhansky (1971) demonstrated that the original populations are those which conserved the highest levels of gene diversity. An alternative explanation for these highest gene diversity levels is that another different subspecies of Pampas cat (L. p. crespoi sensu García-Perea, 1994) exists in the northern area of North Argentina, and that we mixed samples of budini and crespoi in a unique sample. This could explain why the Jujuy-Salta population had the highest mtDNA gene diversity. But this does not explain why the Mendoza population was that with the highest microsatellite gene diversity because the putative crespoi subspecies does not reach this area of northern Argentina. Therefore, we agree more with the first hypothesis. That is, the northern Argentinean Pampas cat population is the original one. It is also possible that the Pampas cat hybridizes with other small cats (as with L. tigrinus; Johnson et al., 1999) and this could augment the levels of gene diversity in that areas where hybridization occurs. For the case of L. jacobita, also the Central Peru population was that with the lowest gene diversity both for microsatellites and mtDNA. Thus, it is possible that a similar colonization process in the northern distribution area affected both species of Andean cats, but the gene drift process affected jacobita more intensely than pajeros. However, the jacobita population with the highest levels of gene diversity was southern Peru-northern Bolivia, which could indicate that this population is the original one. If this is certain, then the focus of dispersion for jacobita was more northern compared to the focus of dispersion for pajeros. Our microsatellites results yielded clear significant genetic heterogeneity among all putative pajeros subspecies studied. Therefore, there is significant molecular evidence for the morphological pajeros subspecies proposed by García-Perea (1994) and other authors. Only in one case, does this molecular differentiation not agree with the putative morphological subspecies proposed by García-Perea (1994). There is evidence that the southern Argentine populations or putative subspecies are probably a unique genetic pool. That is L. p. pajeros and L. p. crucinus could be the same subspecies, although we analyzed a limited number of exemplars of these putative subspecies. More samples of these populations should be studied to determine if our affirmation is true or not. Within the subspecies, there is also significant heterogeneity among the populations, which is evidence of the action of gene drift affecting these populations. Nevertheless, the estimates of gene flow obtained with microsatellites showed that historically, the pajeros populations or subspecies have not been totally disconnected reproductivelly. Also these results are relevant showing that the population levels of genetic heterogeneity and gene flow are similar independently to consider 46 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al. populations or morphological subspecies for L. pajeros. Conversely, the mtDNA results showed a considerably higher degree of genetic heterogeneity among populations and subspecies of pajeros and the gene flow estimates were considerably lower than with microsatellites. This could indicate that female gene flow was more restricted than the male gene flow in pajeros. The levels of genetic heterogeneity were considerably higher (and the gene flow values were considerably lower) for jacobita than for pajeros by using microsatellites, making it an interesting point to compare genetic heterogeneity and gene flow estimates between pajeros and jacobita. Therefore, if the major part of the pajeros subspecies morphologically proposed by García-Perea (1994) are accepted, we must also define, at least, four molecular subspecies for jacobita because the degree of isolation among populations of this last species was considerably higher than for pajeros. Another interesting difference between pajeros and jacobita is that in this last species the degree of genetic heterogeneity for microsatellites and mtDNA is similar, meanwhile in pajeros, as we explained, the degree of genetic heterogeneity of microsatellites was considerably lower than mtDNA. However, for mtDNA, the degree of heterogeneity between pajeros and jacobita is extremely similar. Thus, the capacity of dispersion and/or the habitat constrictions are low and similar for males and females of jacobita and for the females of pajeros, meanwhile the males of pajeros have had a higher power of gene flow. This fact, revealed with microsatellites, was clearly observed with the assignation analyses. A relatively high percentage of animals were misclassified in their respective populations and subspecies as well as the number of migrants of first generation was very elevated. This means that the gene flow due to males is considerable important in pajeros. Nevertheless, it was clear that the classification by subspecies assigned pajeros individuals better than by population. Thus, for conservation planning it is probably better to use subspecies rather than populations in pajeros. In contrast to that found in pajeros, the assignation analyses correctly classified the major part of the jacobita individuals, which means that the genetics isolation among jacobita populations is higher than among pajeros populations. If the subspecies of pajeros are maintained, then the four different jacobita populations detected must be elevated to the subspecies status. The procedure of Cornuet and Luickart (1996) did not detect any evidence of recent bottlenecks for pajeros by species, subspecies or populations. The other tests applied showed that the major part of the subspecies of pajeros crossed an initial bottleneck followed by a strong population expansion, which eliminated the possibility to detect bottleneck effects in these subspecies and populations. The three putative subspecies of Argentina are those which showed the most conspicuous population expansions. L. p. budini, was the unique subspecies, which probably did not cross an initial bottleneck, which agrees quite well with the hypothesis that this subspecies could be the original one from which the other current pajeros subspecies or populations derived. The comparisons of the demographic change results with microsatellites between pajeros and jacobita revealed some interesting differences and one noteworthy similarity. For pajeros, there was not any relevant trend in favor of some bottleneck, however, some trends revealed some possible bottleneck in the global jacobita population as well as in the southern Peru-northern Bolivia population. Thus, the effects of a possible bottleneck is more clear for jacobita, while in pajeros there was more clear evidence of population expansion. Almost all Population Genetics and Spatial Structure in Two Andean Cats … 47 the pajeros subspecies (exception budini), showed an initial bottleneck followed with a strong population expansion. For jacobita, this pattern was only present in the two southern populations (southern Bolivia-northern Argentina and Mendoza), but not for the two northern populations, which disagree with that observed in pajeros. Moreover, in part of the same area where there was an initial bottleneck in jacobita followed by a population expansion, in pajeros (budini), there was only population expansion without initial bottleneck. For pajeros, as a whole, there was clear evidence of population expansion but not for jacobita. Nevertheless, in both species, the most southern differentiated populations showed the most marked population expansion. Contrarily, the demographic change results with mtDNA did not reveal any evidence of population expansions for pajeros or for jacobita. Some evidence of possible bottlenecks was observed in both species but more marked in jacobita than in pajeros. This is related with the difference of genetic heterogeneity found in pajeros for microsatellites and mtDNA. L. pajeros males are probably responsible for gene flow among populations of this species as well as responsible for the historical population expansion detected in this species. However, the females of both species probably have more restricted mobility (phylopatric females). Their effective numbers are lower than the total effective sizes and thus the effects of possible bottlenecks were much easy to be recorded in mtDNA than in microsatellites in both species. Although jacobita showed clearer evidence of bottlenecks than did pajeros, the low gene diversity levels in jacobita could be constant over time and not be the effects of some drastic or strong bottleneck effects in some determinate moment of its natural history. This could agree with that determined for other Andean carnivores such as the Andean or spectacled bear, which also shows low-moderate levels of gene diversity for both microsatellites and mtDNA. In this aspect, Johnson et al., (1998) also showed that the pairwise Kimura distances among the nine Andean cats that they analyzed oscillated from 0.4-1.5 % which indicated that this species did not suffer extreme reductions or bottlenecks during its evolution. Thus, the low effective numbers and the low-moderate gene diversity of jacobita are attributed to the high degree of fragmentation of the high-altitude deserts of the Andes and the limited abundance of prey that live in these extreme conditions which limited the growth of the jacobita populations. The microsatellite mutation models for pajeros and jacobita were similar to that recorded for other felids for % of multi-step mutations (8.4 % for ocelot, 7.2 % for European wild cat, 7.8 % for puma and 6 % for jaguar; Ruiz-García et al., 2012) and for the different mutation rate for each microsatellite. The average value of was considerable higher in pajeros than in jacobita in total agreement with the other gene diversity statistics found. This is related with the fact that the effective sizes of the pajeros populations are around 10 times higher than the populations of jacobita following the results obtained through the procedures of Nielsen (1997) and that of the mean expected heterozygosity, which seem to be the best procedures to estimate long term historically effective numbers. Other procedures applied, such as those of Hill (1981), Pudovkin et al., (1996) and Griffith and Tavare (1994), did not adequately estimate the effective population sizes of these two species, although in all of the cases, the pajeros’ numbers were always higher than the jacobita values. The historical effective numbers are probably higher than the current population sizes of these species because the most recent human impact from habitat destruction and hunting may not yet affect the genomes of these species (Ruiz-García et al., 2007). The effective numbers of pajeros (80,000-300,000) are relatively similar to those found in other Neotropical wild cats, and are 48 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al. not extremely far from the ecological estimates of around 50,000-70,000 reproductive animals made by Nowell and Jackson (1996) or from the estimates that could be obtained from the data of Silveira et al., (2005). They estimated an average density from 2 to 10 adults per 100 km2 in the Emas National Park in Brazil or 15 individuals per 250 km2 in the Chilean regions of Arica and Parinacota estimated by Napolitano et al., (2008). However, the effective numbers of jacobita are considerable lower, which shows that this species deserves special measures of conservation. This agrees quite well with its classification as being in a dangerous situation (EN) by UICN (2002, 2006). The spatial structure was not identical between pajeros and jacobita, although both Andean cat species presented significant spatial structure, which means that in the Andes Cordillera, the geographical barriers affecting gene flow are important for both species. For the Moran’s I index, the percentage of significant autocorrelation coefficients and correlograms were higher for pajeros. This fact was not so obvious for the Geary’s c coefficient. However, jacobita showed a more marked genetic dissimilarity at the last CD. This means that short, or middle, geographical distances could provide more spatial structure in pajeros, but the extreme geographical distant populations are genetically more differentiated in jacobita than in pajeros. Globally, the isolation-by-distance pattern was more important for jacobita (geographic distances explained around 65 % of genetic distances) than for pajeros (geographic distances explained around 33 % of genetic distances). This ratifies that the different jacobita populations detected in this study must be named as subspecies.
Skull Morphology, Biometrics, Genetics and L. pajeros and L. colocolo Sensu García-Perea (1994)
The comparison of nine biometric skull variables between two different morphological putative species of Pampas cats (L. colocolo colocolo and L. pajeros garleppi sensu García- Perea, 1994), presented only one variable with significant differences. This could be incompatible with the fact that these two Pampas cat populations belong to different species as proposed by García-Perea (1994). In the bullar region, the ectotympanic bone has variation depending of the Pampas cat population analyzed. Colocolo was classified as Type 1 because it has the largest ectotympanic (around 50 % of bullar volume), which is inflated ventrally and posteriorly, whereas in pajeros (Type 2), the ectotympanic is moderately developed and not so conspicuously inflated as in colocolo, although García-Perea (1994) found one individual of pajeros with the character type 1. Colocolo shows a poorly developed caudal entotympanic whereas pajeros has a more developed one as well as a poor posteriorly developed mastoid process separated from the paroccipital process by a wide notch. The notch reveals the surface of the bulla between them in pajeros (also in braccatus), but the mastoid process is well developed posteriorly and covers the bulla in colocolo, although with differences between colocolo colocolo and the putative colocolo wolffsohni. In the orbital region, the inferior oblique muscle fossa is located on the lacrimal-palatine-maxilla suture in pajeros and braccatus (type 1), while in colocolo the fossa is placed on the lacrimal-palatine suture (type 2). In the palatal region, the shape of the notch for the pospalatine vein is wide and comparatively shallow in both colocolo and pajeros (type 1), while it’s narrow and deep in braccatus. In braccatus, the shape of the posterior margin of the palate is U shaped, with a medial notch in pajeros and braccatus (type 1). Also, 30 % of the exemplars of colocolo and Population Genetics and Spatial Structure in Two Andean Cats … 49
70 % of colocolo have no medial notch. For teeth, 33% of pajeros have upper secondary premolars meanwhile in pajeros no individual analyzed by García-Perea (1994) presented this character. Additionally, the shape of the upper third premolar with its narrow paracone, long in lateral view, giving the tooth an acutely pointed appearance is presented both in colocolo and in the major part of the populations of pajeros, and to a lesser extent in L. pajeros thomasi. However, in braccatus and L. pajeros thomasi the paracone is short and wide in lateral aspect. Furthermore, García-Perea (1994) commented that specimens of colocolo are the largest and have the most developed sagittal crests, while the populations of pajeros are variable, with the northern populations (thomasi) being the smallest and those from Patagonia (crucinus), the largest (similar to those of colocolo). However, recall that morphological characters easily submit to natural selection due climatic conditions and prey coevolution. Our molecular data showed several interesting new results in contradiction with the morphological analyses of García-Perea (1994). We analyzed four new individuals (skins) for the area of Arica in northern Chile. These samples come from the Tarapacá province in the western slope of the Andes. Following García-Perea (1994), this northern Chilean Pampas cat population constitutes a new subspecies within the species L. colocolo (L. colocolo wolffsohni). The cranial characteristics were typical for L. colocolo, with a large ectotympanic chamber, mastoid process well developed posteriorly. The posterior edge of the palate was U- shaped and the sagittal crests were well developed occupying most or the entire parietal suture. An upper premolar was present, although new specimens differed from L. colocolo colocolo in their type of coat pattern, having smaller ectotympanics and having less extensive mastoid processes. The mtDNA haplotypes and the microsatellite alleles found in these four exemplars were typical of L. pajeros garleppi, the pajeros subspecies of Peru and northern Bolivia. For this, these samples were enclosed in the Tacna-Puno Pampas cat populations for analysis. Thus, we concluded that the Pampas cat population in northern Chile could be an extension of the eastern Andes Pampas cat population (L. pajeros garleppi) and not a new subspecies of L. colocolo (L. colocolo wolffsohni). In fact, the coat pattern of the four skins molecularly analyzed is almost identical to L. pajeros garleppi, with large and reddish brown rosettes with the borders darker, running in oblique chains along the flanks. There is also a spinal crest and tail rings with the same color as flank spots and tail ringed from the base to the tip (usually eight rings) and stripes on the legs (three darker stripes on front legs and three to five more clear stripes on hind legs) and ventral markings, almost black, on a white background. Thus, the Andean cordillera should not be a geographical barrier for the eastern Andes Pampas cat populations in southern Peru and northern Bolivia to colonize the western slopes of the Andes in northern Chile in contradiction with the findings of García-Perea (1994). In fact, Johnson et al., (1999) concluded that the Paraguay and Rio de la Plata rivers could be more effective geographical barriers between Argentine Pampas cat populations and those from southern Brazil and Uruguay than the Andes cordillera between populations of Argentina and Chile. Thus, our molecular results seems to be closer to the opinion of Osgood (1943), who considered the northern Chilean Pampas cat population as garleppi and in contradiction with Cabrera (1961), who considered this population related to budini or Mann (1945), who included this population within colocolo as did García-Perea (1994). Another possibility is that this population could be a hybrid one between colocolo colocolo and pajeros garleppi and for this reason presented characters more similar to the first taxa (craniometric characters) or to the second taxa (coat pattern). Nevertheless, the results of Johnson et al., (1999), although with a very limited number of samples, showed the same 50 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al. trend for three different mitochondrial genes that we have presented herein: two samples from central Argentina (budini) clustered with samples of colocolo colocolo (80 % bootstrap) and one sample from Bolivia (garleppi) clustered with one sample from northern Chile (90 % bootstrap). These results were identical to that obtained by us with a considerably higher number of samples. Thus, our results are in favor that the northern Chilean Pampas cat population is an extension of pajeros garleppi and that colocolo and pajeros are probably taxa of the same species. Quintana et al., (2009) was also in agreement. However, the results of Napolitano et al., (2008) were more ambiguous because the authors detected a characteristic set of haplotypes from northern Chile, which could be in agreement with the hypothesis of García-Perea (1994) of a specific Pampas cat taxa in northern Chile. But these haplotypes were closely related to the central Chilean Pampas cat, but also to the Argentine budini subspecies. Moreover, one haplotype was closely related with Brazilian braccatus haplotypes. Thus, although the northern Chilean Pampas cat population could be a special one, the existence of three different species of Pampas cat proposed by García-Perea (1994) seems to not be in agreement with the molecular data obtained. Two of the morphological characters studied by García-Perea (1994) agree quite well with the molecular results that we present here in reference to the different molecular gene pools detected in pajeros. One of them was the ectotympanic bone geographic variation observed by her in the type 2 with the specimens in the northern half range of pajeros distribution with a bullar volume of 25-30 % and the animals of the southern half range with a volume of 30-35 %. The second one is the gradation in the level of expression of markings in the coat pattern. She defined three sub-types (2A, 2B and 2C) with 2A with the most intense markings, 2B with more dilute markings and 2C with only extremely dilute signal markings. The northern population is well defined molecularly speaking (garleppi) belonging to type 2A. Between 20° and 38° S, there is type B, although there are also exemplars with types 2A and 2C. This coincides with steinbachi, pajeros and especially budini gene pools, which are those with the highest gene diversity as well as the highest degree of color and coat pattern variation. In the southern parts (up 47 ° S), there is type 2C which coincides with the pajeros- crucinus gene pool. Another interesting fact is that García-Perea (1994) only determined one Pampas cat subspecies in Bolivia (steinbachi), whereas Anderson (1997) considered that the subspecies living in Bolivia was garleppi following Cabrera (1958) from the analysis of one individual from Tiraque (3,200 mt above sea level). Cabrera (1958) considered that L. pajeros steinbachi was synonymous to Felis colocolo garleppi. Salazar-Bravo et al., (2003) classified the pajeros form of Bolivia as spp. However, our molecular analysis clearly showed for both, microsatelites and mtDNA, that in Bolivia there are at least two pajeros subspecies. All the samples from the La Paz Department belonged to garleppi, whilst the samples from Cochabamba and more southern Bolivian localities belonged to steinbachi. Anderson (1997) described four localities where this species was recorded in Bolivia from six specimens studied (one from Los Totumu, Beni, one from Tirique, Cochabamba, one from Comanche, La Paz, one from Siranai, Santa Cruz and two individuals from unknown Bolivian origin). It is important for the reconstruction of evolutionary history and the phylogenetics relationships of these two small cats, and for conservation purposes, to obtain more samples and to apply additional molecular markers. The following aspects must be studied: 1- To determine the fine micro-geographic structure of populations of both species, especially in the high altitude deserts of the Andes (Puna); 2- To analyze MHC gene sequences in both species Population Genetics and Spatial Structure in Two Andean Cats … 51 but especially in the Pampas cat because it lives in very different habitats; 3- To analyze samples from the putative more northern Pampas cat population in Ecuador (thomasi) and to determine the constant presence of this taxa also in Colombia; 4- To obtain more samples of the Pampas cat from northern Chile to determine if this population is really a new taxa as maintained by García-Perea (1994), or is an extension of garleppi (as maintained by Johnson et al., 1999 and in this work), or if this is a hybridization area; 5- To analyze more samples of the Pampas cat in southern Bolivia and northern Argentina to determine the exact differences between the two putative steinbachi and budini and to look for possible hybridization between both populations. Also, it would be interesting to determine if the northern Argentina is another gene pool which agrees well with the putative subspecies named crespoi; 6- To analyze more samples of specimens listed tentatively as pajeros pajeros and pajeros crucinus to determine if they are two different subspecies or if they are really of the same gene pool as we have shown with a small sample size in this study; 6- To determine the real relationship between the two putative braccatus subspecies (bracattus bracattus and braccatus munoai); and 7- To analyze with a wide variety of molecular markers the relationship among these three main Pampas cat populations (colocolo, pajeros and braccattus sensu García-Perea, 1994) and to determine the real relationship among them and if they are really three different species or if they are a unique species, as the molecular results seem incipiently to show.
ACKNOWLEDGMENTS
Manuel Ruiz-García thanks Colciencias (Grant 1203-09-11239), the Fondo para la Accion Ambiental (US-Aid) (Grant 120108-E0102141) and the Pontificia Universidad Javeriana for financial support. Special thanks to Dr. Diana Álvarez, Pablo Escobar-Armel, Ariel Rodriguez, Nathalí Romero, Luisa Fernanda Castellanos-Mora, Kelly Luengas, Nicolás Lichilín, Dr. Clara Saldamando (Colombia), Nataniel Mamani and Dr. Volga Iñiguez (Bolivia), who all helped to obtain samples of Pampas and Andean cats throughout Peru, Bolivia and Chile.
REFERENCES
Allen, J. A. (1919). Notes on the synonymy and nomenclature of the smaller spotted cats of Tropical America. Bulletin of the American Museum of Natural History. Vol. XLI: 341-419. Anderson, S. (1997). Mammals of Bolivia: taxonomy and distribution. Bulletin of the American Museum of Natural History, 231, 1-652. Archie, J.W. (1985). Statistical analysis of heterozygosity data: independent sample comparisons. Evolution 39: 623–637. Badouin, L. and Lebrun, P. (2000). An operational Bayesian approach for the identification of sexually reproduced cross-fertilized populations using molecular markers. En: Dore C, Dosba F and C Baril C (Eds.), Proceedings of the International Symposium on Molecular Markers for characterizing genotypes and identifying cultivars in Horticulture: 6-9. Montpellier, France. 52 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Barton, N.H. and Slatkin, M. (1986). A quasi-equilibrium theory of the distribution of rare alleles in a subdivided population. Heredity 56: 409-416. Bohonak, A,J. (2002). IBD (Isolation by Distance): A program for Analyses of isolation by distance. Journal of Heredity 93: 153-154. Broad, S. (1987). International trade in skins of Latin American spotted cats. Traffic Bulletin, 9, 1-8. Cabrera, A. (1940). Notas sobre carnívoros sudamericanos. Notas del Museo de La Plata (Zoologia), 5, 1-22. Cabrera, A. (1958). Catálogo de los mamíferos de América del Sur. I (Metatheria, Unguiculata, Carnívora). Revista del Museo Argentino de Ciencias Naturales “Bernardo Rivadavia,” Zoología 4:1-307. Cabrera, A. (1961). Los félidos vivientes de la República de Argentina. Revista del Museo Argentino de Ciencias Naturales (Zoologia), 6, 161-247. Cabrera, A., and Yepes, J. (1940). Historia Natural Ediar. Mamíferos Sud-Americanos. Compañía Argentina de Editores, Buenos Aires, Argentina. Cornuet, J. M., and Luikart, G. (1996). Description of power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144: 2001-2014. Cornuet, J.M., Piry, S, Luikart, G, Estoup, A and Solignac, M. (1999). New methods employing multilocus genotypes to select or exclude populations as origins of individuals. Genetics 153:1989-2000. Cossíos, E.D., Lucherini, M., Ruiz-García, M., and Angers B. (2009). Influence of ancient glacial periods on the Andean fauna: the case of the Pampas cat (Leopardus colocolo). BMC Evolutionary Biology 9: 68-79. Cossíos, E.D, Walker S, Lucherini M, Ruiz-García M, and Angers B. (2012). Between high- altitude islands and high-altitude corridors. The population structure of the Andean cat (Leopardus jacobita). Endangered Species Research , 16, 283-294. Crow, J.F. and Aoki, K (1984). Group selection for a polygenic behavioral trait: estimating the degree of population subdivision. Proceedings of National Academy of Sciences USA 81: 6073-6077. Cuadras C. M. (1991). Métodos de Análisis Multivariante. Promociones y Publicaciones Universitarias. Barcelona. Dib, C., Faure, S., and Fizames, C. (1996). A comprehensive genetic map of the human genome based on 5,264 microsatellites. Nature, 380, 152-154. Dobzhansky, T. (1971). Evolutionary oscillations in Drosophila pseudoobscura.. In Ecological Genetics and Evolution (Ford EB, ed.), pp 109–133. Oxford, Blackwell Scientific. Durbin, J., and Watson, G.S. (1950). Testing for serial correlation in least squares regression. Biometrika 37: 409-428. Epperson, B.K. (1990). Spatial autocorrelation of genotypes under directional selection. Genetics 124: 757-771. Epperson, B.K. (1993). Recent advances in correlation studies of spatial patterns of genetic variation. Evolutionary Biology 27: 95-155. Ergueta, P., and Morales, C. (1996). Libro rojo de los vertebrados de Bolivia. CDC, La Paz, Bolivia. Population Genetics and Spatial Structure in Two Andean Cats … 53
Feldman, M. W., Kumm, J., and Pritchard, J. K. (1999). Mutation and migration in models of microsatellite evolution. In Goldstein, D. G., and Schlötterer, C. (Eds.), Microsatellites: evolution and applications. Oxford, United Kingdom: Oxford University Press. Freeman, A. R., MacHugh, D. E., McKeown, S., Walzer, C., McConnell, D. J., and Bradley, D. G. (2001). Sequence variation in the mitochondrial DNA control region of wild African cheetahs (Acinonyx jubatus). Heredity, 86, 355-362. Fu, Y. X. (1997). Statistical tests of neutrality against population growth, hitchhiking and background selection. Genetics, 147, 915-925. Fu, Y. X., and Li, W. H., 1993. Statistical tests of neutrality of mutations. Genetics, 133, 693-709. Gabriel, K.R., and Sokal, R.R. (1969). A new statistical approach to geographic variation analysis. Systematic Zoology 18, 259-278. Garcia-Perea, R. (1994). The Pampas cat group (Genus Lynchailurus Severtzov, 1858) (Cranivira: Felidae), a systematic and biogeographic review. American Museum Novitates 3096, 1-36. García-Perea, R. (2002) Andean mountain cat, Oreailurus jacobita: morphological description and comparison with other felines from the altiplano. Journal of Mammalogy, 83, 110–124.
Goodman SJ (1997). RST Calc: a collection of computer programs for calculating estimates of genetic differentiation from microsatellite data and determining their significances. Molecular Ecology 6,881–885. Goudet J., Raymond M., Demeeus T., and Rousset F. (1996). Testing differentiation in diploid populations. Genetics 144, 1933–1940. Griffiths, R.C., and Tavaré, S. (1994). Simulating probability distributions in the coalescent. Theoretical Population Biology, 46, 131-159. Harpending, H., (1994). Signature of ancient population growth in a low resolution mitochondrial DNA mismatch distribution. Human Biology, 66, 591-600. Harpending, H. C., Sherry, S. T., Rogers, A. R., and Stoneking, M. (1993). Genetic structure of ancient human populations. Current Anthropology, 34, 483-496. Hellberg, M.E. (1994). Relationships between inferred levels of gene flow and geographic distance in a philopatric coral, Balanophyllia elegans. Evolution 48, 1829-1854. Hill, W. G. (1981). Estimation of effective population size from data on linkage disequilibrium. Genetical Research, 38, 209-216. Hudson, R.R. (2000). A new statistic for detecting genetic differentiation. Genetics 155, 2011-2014. Hudson, R.R., Boss, D.D., and Kaplan, N.L. (1992a). A statistical test for detecting population subdivision. Molecular Biology and Evolution 9, 138-151. Hudson, R.R., Slatkin, M. and Maddison, W.P. (1992b). Estimations of levels of gene flow from DNA sequence data. Genetics 132, 583-589. Isaaks, E.H., and Srivastava, R.M. (1989). An introduction to applied geostatistics. Oxford University Press, New York. Pp 1-561. IUCN (1990, 1994, 2002, 2006) 1990, 1994, 2002, 2006 IUCN Red List of threatened species. www.iucnredlist.org. Johnson, W.E., Culver, M., Iriarte, J.A., et al. (1998). Tracking the evolution of the elusive Andean mountain cat (Oreailurus jacobita) from mitochondrial DNA. Journal of Heredity, 89, 227–232. 54 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Johnson, W.E., Eizirik, E., Pecon-Slattery, J. et al. (2006). The Late Miocene radiation of modern Felidae: a genetic assessment. Science, 311, 73–77. Johnson, W.E., Pecon Slattery, J., Eizirik, E. et al. (1999). Disparate phylogeographic patterns of mitochondrial DNA variation in four closely related South American small cat species. Molecular Ecology, 8, S79–S94. Kimmel, M., Chakraborty, R., King, J. P., Bamshad, M., Watkins, W. S., and Jorde, L. B. 1998. Signatures of population expansion in microsatellite repeat data. Genetics, 148, 1921-1930. Kuhn, H. J. (1973). Zur Kenntnis der Andenkatze, Felis (Oreailurus) jacobita Cornalia, 1865. Saugetierkundliche Mitteilungen, 21, 359-364. Lucherini, M., Reppucci, J. I., Walker, R. S., Villalba, L., Wurstten, A., Gallardo, G., Iriarte, A., Villalobos, R., and Perovic, P. (2009). Activity pattern segregation of carnivores in the high Andes. Journal of Mammalogy, 90, 1404-1409. Luikart, G., Allendorf, F., Sherwin, B., and Cornuet, J. M. (1998). Distortion of allele frequency distribution provides a test for recent population bottleneck. The Journal of Heredity, 86, 319-322. Mann, G. (1945). Mamíferos de Tarapacá. Observaciones realizadas durante una expedición al Alto Norte de Chile. Biológica, 2, 23-138. Mantel, N.A. (1967). The detection of disease clustering and a generalized regression approach. Cancer Research 27: 209-220. Mares, M. A., and Ojeda, R. A. (1984). Faunal commercialization and conservation in South America. Bioscience, 34, 580-584. Matula, D.W., and Sokal, R.R. (1980). Properties of Gabriel graphs relevant to geographic variation research and the clustering of points in the plane. Geographic Analysis 12, 205-222. Menotti-Raymond, M.A., David, V. A., Lyons, L.A., Schaffer, A. A., Tomlin, J. F., Hutton, M. K., and O’Brien, S. J., (1999). A genetic linkage map of microsatellites in the domestic cat (Felis catus). Genomics, 57, 9-23. Menotti-Raymond, M.A., and O´Brien, S. J. (1995). Evolutionary conservation of ten Microsatellite loci in four species of Felidae. Journal of Heredity, 86, 319-322. Michalakis Y, and Excoffier L. (1996). A generic estimation of population subdivision using distances between alleles with special reference to microsatellite loci. Genetics 142: 1061–1064. Moran, P.A.P. (1950). Notes on continuous stochastic phenomena. Biometrika 37, 17-23. Napolitano, C., Benett, M., Johnson, W. E., O’Brien, S. J., Marquet, P. A., Barría, I., Poulin, E., and Iriarte, A. (2008). Ecological and biogeographical inferences on two sympatric and enigmatic Andean cat species using genetic identification of faecal samples. Molecular Ecology, 17, 678-690. Nei M. (1972). Genetic distance between populations. American Naturalist 106: 283-292. Nei M. (1973). Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences of the USA 70: 3321–3323. Nielsen, R. (1997). A likelihood approach to population samples of microsatellite alleles. Genetics, 146, 711-716. Nowell, K., and Jackson, P. (1996). Wild Cats. Status survey and conservation action plan. IUCN/SSC Cat Specialist Group. Pp. 1-382. Population Genetics and Spatial Structure in Two Andean Cats … 55
Oden, N. (1984). Assessing the significance of a spatial correlogram. Geographical Analysis 16, 1-16. Ohta, T., and Kimura, M. (1973). A model of mutation appropriate to estimate the number of electrophoretically detectable alleles in a finite population. Genetical Research, 22, 201-204. Osgood, W.H. (1943) The Mammals of Chile. Field Museum Natural History Zoology Series, 30, 1–268. Paetkau D., Calvert, W., Stirling, I and Strobeck, C. (1995) Microsatellite analysis of population structure in Canadian polar bears. Molecular Ecology 4: 347-354. Patkeau D., Slade, R., Barden, M and Estoup, A. (2004). Genetic assignment methods for the direct, real-time estimation of migration rate: a simulation based exploration of accuracy and power. Molecular Ecology 13: 55-65. Piry S, Alapetite, A., Cornuet, J. M., Paetkau, D., Baudouin, B and Estoup, A. (2004). GENECLASS2: a software for genetic assignment and first-generation migrant detection. Journal of Heredity 95: 536-539. Pudovkin, A. I., Zaykin, D. V., and Hedgecock, D. (1996). On the potential for estimating the effective number of breeders from heterozygote-excess in progeny. Genetics, 144, 383-387. Quintana, V., Yáñez, J., Valdebenito, M., and Iriarte, A. (2009). Orden Carnivora. In: Mamíferos de Chile, Muñoz, A., and Yáñez, J, (Eds.), pp. 193–230. CEA Ediciones, Valdivia, Chile. Ramos-Onsins, S. E., and Rozas, J., (2002). Statistical properties of new neutrality tests against population growth. Molecular Biology and Evolution, 19, 2092-2100. Rannala B. and Mountain, J. L. (1997) Detecting immigration by using multilocus genotypes. Proceedings of the National Academy of Sciences USA 94: 9197-9201. Ripley, B.D. (1981). Spatial statistics. John Wiley and Sons, New York. Rogers, A. R., Fraley, A. E., Bamshad, M. J., Watkins, W. S., and Jorde, L. B. (1996). Mitochondrial mismatch analysis is insensitive to the mutational process. Molecular Biology and Evolution, 13, 895-902. Rogers, A. R., and Harpending, H. C. (1992). Population growth makes waves in the distribution of pairwise genetic differences. Molecular Biology and Evolution, 9, 552-569. Rousset F (1997). Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics 145: 1219–1228. Ruiz-Garcia, M. (1998). Genetic structure of different populations of domestic cat in Spain, Italy, and Argentina at a micro-geographic level. Acta Theriologica 43: 39-66. Ruiz-García, M. (1999). Genetic structure of different cat populations in Europe and South America at a microgeographic level: Importance of the choice of an adequate sampling level in the accuracy of population genetics interpretations. Genetics and Molecular Biology 22: 493-505. Ruiz-García, M. (2001). Diversidad genética como herramienta de zonificación ambiental: Estudios moleculares (microsatélites) en el caso de Primates y Félidos neotropicales comportan una nueva perspectiva. In Defler, T., and Palacios, P. A., (Eds.), Zonificación Ambiental para el Ordenamiento Territorial en la Amazonía Colombiana, (pp 85-108). Bogotá DC: Instituto Amazónico de Investigaciones, Imani and Instituto de Ciencias Naturales, Universidad Nacional de Colombia. 56 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Ruiz-García, M. (2003). Molecular population genetic analysis of the spectacled bear (Tremarctos ornatus) in the Northern Andean Area. Hereditas 138, 81-93. Ruiz-García, M. (2007). Genética de Poblaciones: Teoría y aplicación a la conservación de mamíferos neotropicales (Oso andino y delfín rosado). Boletin de la Real Sociedad Espanola de Historia Natural Seccion Biologica, 102 (1-4), 99-126. Ruiz-García, M. (2012). The genetic demography history and phylogeography of the Andean bear (Tremarctos ornatus) by means of microsatellites and mtDNA markers. In Molecular Population Genetics, Phylogenetics, Evolutionary Biology and Conservation of the Neotropical Carnivores. Ruiz-García, M., Shostell, J. (Eds.). Nova Science Publishers. New York. Ruiz-Garcia, M. and Jordana, J. (1997). Spatial genetic structure of the “Gos d' Atura” dog breed in Catalonia (Spain). Brazilian Journal of Genetics, 225-236. Ruiz-Garcia, M. and Jordana, J. (2000). Spatial structure and gene flow from biochemical markers in the “Pyrenean Brown” breed, a rare cattle race in Catalonia (Spain). Biochemical Genetics 38, 341-352. Ruiz-García, M. and Alvarez, D. (2000) Genetic microstructure in two Spanish cat populations I: Genic diversity, gene flow and selection. Genes and Genetic Systems 75: 269-280. Ruiz-García, M., Orozco-terWengel, P., Payán, E., and Castellanos, A. (2003). Genética de Poblaciones molecular aplicada al estudio de dos grandes carnívoros (Tremarctos ornatus – Oso andino, Panthera onca- jaguar): lecciones de conservación. Boletin de la Real Sociedad Espanola de Historia Natural Seccion Biologica, 98, (1-4): 135-158. Ruiz-García, M., Orozco-terWengel, P., Castellanos, A., and Arias, L. (2005). Microsatellite analysis of the spectacled bear (Tremarctos ornatus) across its range distribution. Genes and Genetics Systems 80, 57-69. Ruiz-García, M., Payán, E., and Hernández-Camacho, H. (2003). Possible records of Lynchailurus in southwestern Colombia. Cat News, 38, 35-37. Ruiz-García, M., Payán, E., Murillo, A., and Alvarez D. (2006). DNA Microsatellite characterization of the Jaguar (Panthera onca) in Colombia. Genes and Genetics Systems 81, 115-127. Ruiz-García, M., Murillo, A., Corrales, C., Romero-Aleán, N., and Alvarez-Prada, D. (2007). Genética de Poblaciones Amazónicas: La historia evolutiva del jaguar, ocelote, delfín rosado, mono lanudo y piurí reconstruida a partir de sus genes. Animal Biodiversity and Conservation, 30, 115-130. Ruiz-García, M., Martínez-Aguero, M., Alvarez, D., and Goodman S. (2009). Variabilidad genética en géneros de Ciervos neotropicales (Mammalia: Cervidae) según loci microsatélitales. Revista de Biología Tropical. International Journal of Tropical Conservation and Biology. 57, 879-904. Ruiz-García, M., García-Perea, R., Corrales, C., Murillo, A., Alvarez, D., Pinedo-Castro, M., and Shostell, J. M. (2012). Determination of microsatellite DNA mutation rates, mutation models and mutation bias in four main Felidae lineales (European wild cat, Felis silvestris; Ocelot, Leopardus pardalis; puma, Puma concolor, jaguar, Panthera onca). In Molecular Population Genetics, Phylogenetics, Evolutionary Biology and Conservation of the Neotropical Carnivores. Ruiz-García, M.,and Shostell, J. (Eds.). Nova Science Publishers. New York. Population Genetics and Spatial Structure in Two Andean Cats … 57
Salazar-Bravo, J., Tarifa, T., Aguirre, L. F., Yensen, E., and Yates, T. L. (2003). Revised checklist of Bolivian Mammals. Occasional Papers of the Museum of Texas Tech University, 220, 1-27. Salles, L. O. (1992). Felid phylogenetics: extant taxa and skull morphology (Felidae, Aeluroidea). American Museum Novitates, 3047, 1-67. Sanderson, J. (1999). Andean mountain cat (Oreailurus jacobita) in northern Chile. Cat News, 30, 25–26. Scrocchi GJ, Halloy SP (1986) Notas sistemáticas, ecológicas, etológicas y biogeográficas sobre el gato andino (Felis jacobita, Cornalia) (Felidae, Carnivora). Acta Zoológica Lilloana, 23, 157–180. Silveira, L., Jacomo, A. T. A., and Malzoni Furtado, M. (2005). Pampas cat ecology and conservation in the Brazilian grasslands. Project of the Month, Cat Specialist Group Website.http://www.catsg.org/catsgportal/projectmonth/02_webarchive/grafics/sept2005. pdf. Simonsen K. L., Churchill, G. A., and Aquadro C. F. (1995). Properties of statistical tests of neutrality for DNA polymorphism data. Genetics, 141, 413-429. Slatkin, M (1985). Rare alleles as indicators of gene flow. Evolution 39: 53-65. Slatkin, M. (1995). A measure of population subdivision based on microsatellite allele frequencies. Genetics, 139, 457-462. Smouse, P.E., Long, J.C., and Sokal, R.R. (1986). Multiple regression and correlation extension of the Mantel test of matrix correspondence. Systematic Zoology, 35, 627-632. Sokal, R.R., Harding, R.M., and Oden, N.L. (1989). Spatial patterns of human gene frequencies in Europe. American Journal of Physical Anthropology 80, 267-294. Sokal, R.R., and Jacquez, G.M. (1991). Testing inferences about microevolutionary processes by means of spatial autocorrelation analysis. Evolution 45, 155-172. Sokal, R.R., and Oden, N.L. (1978a). Spatial autocorrelation in Biology. 1. Methodology. Biological Journal of Linnean Society 10, 199-228. Sokal, R.R., and Oden, N.L. (1978b). Spatial autocorrelation in Biology. 2. Some biological implications and four applications of evolutionary and ecological interest. Biological Journal of Linnean Society 10, 229-249. Sokal, R.R., Oden, N.L., and Barker, J.S.F. (1987). Spatial structure in Drosophila buzzatii populations: simple and directional spatial autocorrelation. American naturalist 129, 122-142. Sokal, R. R., and Rohlf, F.J. (1995). Biometry. 3rd edition. W.H. Freeman and Co., New York. Sokal, R.R., Smouse, P.E., and Neel, J.V. (1986). The genetic structure of a tribal population, the Yanomama Indians. Genetics. XV. Patterns inferred by autocorrelation analysis. Genetics 114, 259-287. Sokal, R.R., and Wartenberg, D.E. (1983). A test of spatial autocorrelation using and isolation by distance model. Genetics 105, 219-237. Sphuler, J.N. (1972). Genetic, linguistic, and geographical distances in native North America. In The assessment of population affinities in man. Wiener, J. S., Huizinga, J (Eds.). Oxford University Press, Oxford. Tajima, F. (1989a). Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics, 123, 585-595. 58 Manuel Ruiz-García, Daniel Cossíos, Mauro Lucherini et al.
Tajima, F. (1989b). The effect of change in population size on DNA polymorphism. Genetics, 123, 597-601. Takahata, N. (1983). Gene identity and genetic differentiation of populations in the finite island model. Genetics 104: 497-512. Trexler, J.C. (1988). Hierarchical organization of genetic variation in the sailfin molly, Poecilia latipinna (Pisces: Poeciliidae). Evolution 42, 1006-1017. Upton, G., and Fingleton, B. (1985). Spatial data analysis by example. Vol 1: Point pattern and quantitative data. John Wiley and Sons, Chichester. Walsh, P.S., Metzger, D. A., and Higuchi, R. (1991). Chelex 100 as a medium for simple extraction of DNA for PCR-based typing from forensic material. BioTechniques, 10, 506-513. Waples, R. S. (1989). A generalized approach for estimating effective population size from temporal changes in allele frequency. Genetics, 121, 379-391. Wehrhahn, C. (1975). The evolution of selectively similar electrophoretically detectable alleles in finite natural populations. Genetics, 80, 375-394. Werdelin, L. (1996). The history of felid classification. In Wild Cats. Status survey and conservation action plan. Nowell, K., and Jackson, P., (Eds.), IUCN/SSC Cat Specialist Group. Wright S (1951). The genetical structure of populations. Annals of Eugenics 15: 323–354. Yensen, E., and Seymour, K. (2000) Oreailurus jacobita. Mammalian Species, 644,1-6. Zhivotovsky, L. A., and Feldman, M. W. (1995). Microsatellite variability and genetic distances. Proceedings of the National Academy of Sciences, 92, 11549-11552. Zhivotovsky, L. A., Bennett, L., Bowcock, A. M., and Feldman, M. W. (2000). Human population expansion and microsatellite variation. Molecular Biology and Evolution, 17, 757-767. Ziesler, G. (1992). Souvenir d’un Chat des Andes. Animan: Nature et Civilisations, 50, 68–79.