Mitochondrial molecular procedures

DNA was extracted and isolated from either hair, skin, teeth, or muscle samples using the QIAamp DNA Micro Kit (Qiagen, Inc.) following the manufacturer's protocol. Mitochondrial genomes were sequenced by long-template PCR, which minimizes the chance of amplifying mitochondrial pseudogenes from the nuclear genome (numts) (Thalmann et al. 2004, Raaum et al. 2005). PCR amplification of mitochondrial DNA was carried out using a LongRange PCR Kit (Qiagen, Inc.), with a reaction volume of 25 µl and a reaction mix consisting of 2.5 µl of 10× LongRange PCR Buffer, 500 µM of each dNTP, 0.6 µM of each primer, 1 unit of Long-Range PCR Enzyme, and 50-250 ng of template DNA. Cycling conditions were as follows: 94 °C for 5 min, followed by 45 cycles denaturing at 94 °C for 30 s, primer annealing at 50-57 °C (depending on primer set) for 30 s, and an extension at 72 °C for 8 min, followed by 30 cycles of denaturing at 93 °C for 30 s, annealing at 45-52 °C (depending on primer set) for 30 s, and extension at 72 °C for 5 min, with a final extension at 72 °C for 8 min. Four sets of primers were used to generate overlapping amplicons from 3,687 to 4,051 bp in length, thereby enabling a quality test for genome circularity (Bensasson et al. 2001, Thalmann et al. 2004). Both mt DNA strands were sequenced directly using BigDye Terminator v3.1 (Applied Biosystems, Inc.). Sequencing products were analysed on an ABI 3730 DNA Analyzer system (Applied Biosystems, Inc.). Sequences were then assembled and edited using Sequencher 4.7 software (Gene Codes, Corp., Ann Arbor, MI). Overlapping regions were examined for irregularities such as frameshift mutations and premature stop codons. A lack of such irregularities indicates an absence of contaminating numt sequences.

The alignments with all the genes (16,114 bp) were concatenated after removing problematic regions using Gblocks 0.91 (Talavera & Castresana 2007) under a relaxed approach. This software removes all poorly aligned regions and is particularly effective in phylogenetic studies including highly divergent sequences (Castresana 2000, Talavera & Castresana 2007). The individual alignments were then concatenated by means of the SequenceMatrix v1.7.6 software (Vaidya et al. 2011) to create a master alignment. The GenBank accession numbers of the coati specimens analysed are from MT587713 to MT587788, MW410859 to MW410908, and MW419814 to MW419853.

Phylogenetical analyses to determine the relationships between the genera Nasua and Nasuella by using mitochondrial sequences. jModeltest v2.0 (Darriba et al. 2012), Kakusan4 (Tanabe 2011) and MEGA X 10.0.5 software (Kumar et al. 2018) were used to determine the best evolutionary mutation model for the sequences analysed for each individual gene, for different partitions and for all the concatenated sequences. Akaike information criterion (AIC; Akaike 1974, Posada & Buckley 2004) was used to determine the best evolutionary nucleotide model.

Fig. 1.

Map with the geographical origins and sample sizes of specimens of three coati species (Nasua nasua, Nasua narica and Nasuella olivacea) for which mitogenomes were sequenced (n = 205) throughout the Neotropics.

Phylogenetic trees with mitogenomes were constructed using two procedures: Maximum Likelihood tree (ML), and Bayesian Inference tree (BI). The ML tree were obtained using the RA × ML v8.2.X software (Stamatakis 2014) implemented in CIPRES Science Gateway (Miller et al. 2010). The GTR + G + I model (General Time Reversible model + gamma distributed rate variation among sites + proportion of invariable sites; Lanave et al. 1984) was used to search for the ML tree because it was the best model for the major part of the mitochondrial genes. We estimated support for nodes using the rapid-bootstrapping algorithm (–fa –x option) for 1,000 non-parametric bootstrap replicates (Stamatakis et al. 2008). The groups of coatis were considered significant when bootstraps were higher than 70% (lax limit; Hillis & Bull 1993). The BI tree was also performed using a GTR + G + I model for mitogenomes. This tree was completed with the BEAST v2.5.1 program (Drummond et al. 2012, Bouckaert et al. 2014). Four independent iterations were run using three data partitions (codon 1, codon 2, codon 3) with six Markov Chain Monte Carlo (MCMC) chains sampled every 1,000 generations for 30 million generations after a burn-in period of six million generations. Evidence of convergence and stationarity of model parameter posterior distributions was assessed based on ESS values > 200 and examination of trace files in Tracer v.1.7 (Rambaut et al. 2018). The burn-in was set at 20% and separate runs were assembled using LOGCOMBINER v.2.5.1 and TREEANNOTATOR v.2.5.1 (Rambaut et al. 2018). A Yule speciation model and a relaxed molecular clock with an uncorrelated log-normal rate of distribution (Drummond et al. 2006) was used. Posterior probability values provide an assessment of the degree of support of each node on the tree. Majority-rule consensus trees were constructed for each Bayesian analysis. Following Erixon et al. (2003), nodes supported by posterior probability (pp) ≥ 0.95 were considered strongly supported. Trees were visualized in the FigTree v1.4 software (Rambaut 2012).

To determine whether N. olivacea is nested within Nasua, we consider the mitogenome data set but also a data set with only three mt genes (ND5, Cytb and D-loop) with more specimens analysed (particularly for critical geographic areas) and with a wider geographical range (345 specimens), which unfortunately did not amplify for all the mitogenome. We obtained ten different trees (we show them in a simplified version), with or without different outgroups to determine the relationships between Nasuella and Nasua. We wanted to see the influence of different outgroups in the relationships of both genera, Nasua and Nasuella, as well as the presence or absence of outgroups and its influence on the relationships of taxa with relatively recent phylogenetic splits (Ho et al. 2008). We also reconstruct the possible relationships among the haplotypes of Nasuella and Nasua with a Median Joining Network (MJN) with Network v4.6.0.1 software (Fluxus Technology Ltd.) (Bandelt et al. 1999) with the mitogenome data set.

Genetic distances

The Kimura 2P genetic distance (Kimura 1980) was applied to determine the percentage of genetic differences among the different groups detected in the three species of coatis analysed for the mitogenome data set. The Kimura 2P genetic distance is a standard measurement for barcoding tasks (Hebert et al. 2003, 2004). Kartavtsev (2011) analysed sequences of COI from 20,731 vertebrate and invertebrate animal species and obtained 0.89% ± 0.16% for populations within species, 3.78% ± 1.18% for subspecies or semi-species, and 11.06% ± 0.53% for species within a genus. At COII, Ascunce et al. (2003), and Ruiz-García et al. (2014) reported an average genetic distance of around 8% among species within a genus, and around 2-5% for subspecies. Bradley & Baker (2001) and Baker & Bradley (2006) claimed for Cytb that values < 2% would equal intra-specific variation, values between 2% and 13% would merit additional study, and values > 13% would be indicative of specific recognition. Therefore, we take as an average for mitochondrial genes values above 3-5% for possible subspecies, and values around 12-13% for different species of the same genus. For species of different genera, this value should be above 16-18% (Kartavtsev 2011).

Spatial genetic analyses

Three Mantel tests (Mantel 1967) were used to detect possible overall relationships between the genetic matrices (Kimura 2P genetic distance) among specimens of each one of the three coati taxa and their respective geographic distance matrices among the specimens analysed for each one of these three taxa. Both genetic distances and geographical distances were log transformed. In this study, Mantel's statistic was normalized according to Smouse et al. (1986). This procedure transforms the statistic into a correlation coefficient.

Three spatial autocorrelation analyses were carried out for each of the three coati species. This analysis utilized the Ay statistic (Miller 2005) for each distance class (DC), where Ay = Σi = 1, n Σj > i, n (Dijwyij)/Σi = 1, n Σj > i, n wyij, where n is the number of individuals in the data set, and Dij is the genetic distance between observations i and j. Elements of a binary matrix, wyij, take on values of 1 if the geographical distance between observation i and j fall within the boundaries specified for a specified DC and are 0 otherwise. Ay can be interpreted as the average genetic distance between pairs of individuals that fall within a specified DC. Ay takes on a value of 0 when all individuals within a DC are genetically identical and takes on a value of 1 when all individuals within a DC are completely dissimilar. The probability for each DC is obtained using 1,000 randomizations. For this analysis there were ten defined DCs for both N. nasua and N. narica (N. nasua: 0-210 km; 210-368 km; 368-475 km; 475-614 km; 614-774 km; 774-1,001 km; 1,001-1,288 km; 1,288-1,718 km; 1,718-2,256 km; 2,256-3,514 km; N. narica: 0-26 km; 26-83 km; 83-183 km; 183-270 km; 270-337 km; 337-381 km; 381-547 km; 547-730 km; 730-1,089 km; 1,089-2,298 km) and six defined DCs for N. olivacea (0-46 km; 46-144 km; 144-206 km; 206-268 km; 268-429 km; 429-753 km). These DCs were the best to contain approximately the same number of comparisons in each DC. The size of each DC differs for each of the species analysed because the geographical range sampled for each species is also different, but the number of DCs is relatively similar to compare the shape of the correlograms of each species analysed. This analysis was carried out with AIS software (Miller 2005).

Fig. 2.

Maximum Likelihood tree based in complete mitogenomes with 179 haplotypes for three species of coatis (Nasua nasua, Nasua narica and Nasuella olivacea) sampled in Latin America. Nodes are labelled with bootstrap percentages. H144 corresponded to a specimen “a priori” classified as N. nasua that might represent the first confirmed record of N. olivacea in Peru (the River Urubamba, Cuzco).

Karyotype procedures

Cultures were carried out following the method described by Moorhead et al. (1960). Heparinized blood was sowed (1 mL) in a supplemented culture RPMI 1640 (SIGMA) with 10% bovine foetal serum (GIBCO) and with 1% antibiotic (streptomycin 100 µg/mL, and penicillin 100 UI) with a final volume of 10 mL. The mitogene phytohemagglutinin (SIGMA) was added (100 µL). The culture was incubated at 37.5 °C for 72 hours. To obtain R-replicative bands (RGB), 100 µL of 5-bromodeoxyrudine (BrdU) with a concentration of 2 mg/mL was added at 66 hours to obtain the metaphase chromosomes. After 71 hours, 100 micro litres of colcemid with a concentration of 10 µg/mL was added. After culturing, cultures were transferred to 15 mL centrifuge tubes and centrifuged at 1,000 rpm for seven minutes. The supernatant was discarded and the lymphocyte pellet was resuspended. KCl (0.075 M) was added to achieve a volume of 7 mL and the sample was incubated at 37 °C for seven minutes. It was again centrifuged, the supernatant as discarded, and the remaining pellet was resuspended. A methanol-acetic acid (3:1) fixing solution was vigorously added to reach a volume of 7 mL. The sample was again centrifuged and the supernatant discarded and a fixation solution added. This procedure was repeated until a translucent supernatant was obtained (Spowart 1994). Samples were dripped into clean plates with alcohol and then cooled. Alcohol was added (tincture) to the plates to reveal the RGB bands (Camargo & Cervenka 1982, López & Márquez 2002). The extended chromosomes were evaluated at 100x magnification using a ZEISS optical microscope. We analysed fifty cells undergoing mitosis for each of the cases. Chromosome size and centromere ubication were considered in preparation of the karyotype.

Mitochondrial phylogenetic procedures and their consequences on the systematics of the coatis

The most probable nucleotide substitution model considering the complete mitogenomes (all concatenated sequences; 16,114 bp, n = 205) was GTR + G + I (–Ln = 150, 195, 765, AIC). The mitogenome data set indicated a total of 179 coati haplotypes. The ML tree (Fig. 2) did not recover the three “a priori” species as monophyletic. In the clade of N. nasua, one haplotype of N. olivacea (H81) appeared from San José del Palmar (Chocó, Colombia). Between the clades of N. nasua and N. narica, one haplotype of N. nasua (H107) appeared from the PN Tamá (Norte de Santander, Colombia). Within the clade of N. olivacea, we found H144, which corresponded to a specimen “a priori” classified as N. nasua by its geographical origin (it was a road kill specimen and, therefore, its phenotype was not clearly recognized, although some traits seemed to be of N. olivacea) but, it could be the first real register of N. olivacea in Peru (the River Urubamba, Cuzco), and four haplotypes of N. narica (H95, H96, H140 and H115), which belonged (three of them) to southern Costa Rica, Nombre de Dios (Colón, Panama), and Arboletes (Antioquia, Colombia), and the other to western Ecuador (San José Cruz, Pichincha). These were undoubtedly specimens with N. narica‘s phenotype (they were alive) and in a geographical area where only N. narica lives but with mitogenomes of N. olivacea. The BI tree (Fig. S1) yielded the same inconsistences as the previous tree with two additions: the presence of the H51 (one specimen of N. narica from Costa Rica) within the clade of N. nasua and the H107 (N. nasua) within the N. narica clade and not in an intermediate position between N. nasua and N. narica as in the previous tree. Therefore, it is clear that no reciprocal monophyly existed among the three traditional species of coatis, nor between the two genera considered (Nasua and Nasuella) when mitogenomes were used. Note that most of the specimens that indicate monophyly were living or museum samples and thus there was no ambiguity regarding their identification.

Ten different phylogenetic trees (with different procedures and different outgroups) based on the three mt gene data set are shown in Fig. S2. Sixty percent of these trees (ML tree with only Bassarycion neblina as the outgroup, ML tree with all of the Bassarycion species as the outgroup, ML tree without an outgroup, NJ (neighbour-joining, Saitou & Nei 1987) tree with only B. neblina as the outgroup, NJ tree with all of the Bassarycion species as the outgroup, and NJ tree without an outgroup) yielded N. olivacea + N. narica with N. nasua as the most differentiated taxon. In contrast, 40% of the trees indicated N. nasua + N. narica and N. olivacea as the most differentiated taxon (ML tree with Procyon cancrivorus as outgroup, ML tree with P. cancrivorus + all the species of Bassarycion as the outgroup, NJ tree with P. cancrivorus as the outgroup, and NJ tree with P. cancrivorus + all the species of Bassarycion as the outgroup) but without reciprocal monophyly among these putative species.

Fig. 3.

Median Joining Network on mitogenomes of 205 specimens of three species of coatis (Nasua nasua, Nasua narica and Nasuella olivacea) sampled in Latin America. Haplogroups are shown with different colours; Bassarycion neblina (outgroup); 1) orange circles = N. olivacea; “transition-intermediate” haplotypes in the Colombian (Cauca, and Nariño Departments) and Ecuadorian (Carchi Province) Andean Cordilleras; 2) dark green circles = N. olivacea from the Eastern Colombian Andean Cordillera, including Norte de Santander, Boyacá, and Cundinamarca Departments; 3) light green circles = N. narica from southern Costa Rica, Panama and northern Colombia (Antioquia Department) introgressed with mtDNA from N. olivacea; 4) navy blue circles = N. olivacea from Western and Central Colombian and Ecuadorian Andean Cordilleras; Colombian Caldas, Risaralda, Chocó, and Tolima Departments, and Ecuadorian Pichincha, and Cotopaxi Provinces; 5) green circles = N. nasua from the Colombian and Ecuadorian Andean Cordilleras (one N. olivacea, H81, was included in this group); 6) yellow circles = N. nasua from the Ecuadorian and Colombian Amazon, Colombian Cundinamarca, Meta and Valle del Cauca Departments, and trans-Andean and Pacific Ecuador; 7) grey circles = N. nasua from southern Brazil, Paraguay, and Uruguay; 8) greenish-blue circles = N. nasua from different areas of the Peruvian Andes and Amazon, Colombian and western Brazilian Amazon); 9) lilac circles = N. nasua from southern Peru and Bolivia, different areas of the Peruvian Amazon and central Brazilian Amazon; 10) dark blue circles = N. narica from northern Costa Rica, Nicaragua, El Salvador, Honduras and Guatemala; 11) greenish-brown circles = N. narica from trans-Andean and cis-Andean Ecuador; 12) fuchsia-pink circles = N. narica from Guatemala, Belize, southern Mexico and Yucatan. Red circles indicate missing intermediate haplotypes. Some haplotypes have an asterisk (*). These are the cases of H40, 51, 81, 95, 96, 107, 115 and 144. All belonged to specimens with the morphotype of a determined species but with mitogenomes belonging to a different species. H144 corresponded to a specimen “a priori” classified as N. nasua that might represent the first confirmed record of N. olivacea in Peru (the River Urubamba, Cuzco).

Table 1.

Kimura (1980) 2P genetic distances among the main groups of Nasua nasua (five groups), Nasua narica (four groups), and Nasuella olivacea (three groups), together with Bassarycion medius as out-group, using the mitogenome data set. 1) N. nasua haplogroup from Colombian and Ecuadorian Amazon and Eastern Colombian Llanos; 2) N. nasua haplogroup from the Colombian and Peruvian Amazon; 3) N. nasua haplogroup from southern Peru and Bolivia; 4) N. nasua haplogroup from Colombian and Ecuadorian Andes; 5) N. nasua haplogroup from southern Brazil, Paraguay, and Uruguay; 6) N. narica haplogroup from southern Central America (southern Costa Rica and Panama) and northern Colombia introgressed by mtDNA of N. olivacea; 7) N. narica haplogroup from southern Mexico and part of Guatemala; 8) N. narica haplogroup from part of Guatemala and Belize; 9) N. narica haplogroup from part of Guatemala, Honduras, El Salvador, Nicaragua, and Costa Rica; 10) N. olivacea haplogroup from Colombian and Ecuadorian Andean Cordilleras more related to N. nasua; 11) N. olivacea haplogroup from Western-Central Colombian and Ecuadorian Andean Cordilleras; 12) N. olivacea haplogroup from Eastern Colombian Cordillera; 13) B. medius. Standard deviations are not shown because they were practically 0. Genetic distances in %.

The MJN for mitogenomes is shown in Fig. 3. The results obtained for this analysis were insensitive to the outgroup employed. Some haplotypes of N. olivacea were the first to appear (H88, H89, H93). From these, two pathways developed. The first gave rise to the remaining haplotypes of N. olivacea (with the exception of one haplotype). These were the first to derive from the Eastern Colombian Andean Cordillera (H67, H79, and related haplotypes) and, later, the Western and Central Colombian and Ecuadorian Andean Cordilleras (H56, H60, and related haplotypes). One group of N. narica in southern Central America (H95, H96, H140) was introgressed by mtDNA from N. olivacea. This haplogroup is an intermediate group between both main groups of N. nasuella. The second pathway first gave rise to the majority of the Andean Colombian and Ecuadorian N. nasua haplotypes (H154 and related haplotypes) together with a haplotype of a specimen of N. olivacea (Chocó Department, Colombia; H81). This group showed high internal heterogeneity. These Andean coati haplotypes (both N. nasua and N. olivacea) are the origin of the group of N. nasua distributed mainly within the Colombian and Ecuadorian Amazon and some Colombian Andean Departments. This genetic result confirmed the existence of sympatry in the Andes of N. nasua and N. olivacea, as was demonstrated by ecological analyses by González-Maya et al. (2015). This Colombian and Ecuadorian group of N. nasua is the origin of N. narica and, as well as all the other differentiated groups of N. nasua. These differentiated groups of N. nasua are: 1) southern South America (southern Brazil, Paraguay, and Uruguay; H83, H97, and related haplotypes), 2) the Peruvian Amazon and Peruvian Andes (H18, H176, and related haplotypes), and 3) the Andean Peru, Peruvian Amazon, including the Madre de Dios River basin (southern Peru), Bolivia, and central Brazilian Amazon (the River Negro) (H131, H132, H135, and related haplotypes). The first haplogroup to appear on the branch of N. narica was from the middle area of Central America (northern Costa Rica, Nicaragua, El Salvador, Honduras and certain areas of Guatemala; H51, H118, and related haplotypes). From this, the most northern haplotypes in Guatemala, Belize and southern Mexico (H25, H48, and related haplotypes) originated. The Ecuadorian specimens of N. narica were a derived haplogroup from this last Central American group. With the MJN analysis, a group of N. narica from the Yucatan Peninsula (Mexico) was not clearly discriminated from the northern Central American N. narica group, whereas in the phylogenetic trees, the Yucatan group was differentiated (H110 and H112). Therefore, we observed ancestral haplotypes of N. olivacea more related to those of the Andean haplogroup of N. nasua compared to most derived N. olivacea. Furthermore, the most southern Central America and northern Colombian N. narica showed mtDNA from N. olivacea because there was introgression of mtDNA of this last species in the first around 0.9-0.7 millions of year ago (Ruiz-García et al. 2020). Thus, the haplotypes of Nasuella and Nasua were not completely isolated and, in many cases, were mixed in some evolutionary trajectories. Thus, no reciprocal monophyly existed between Nasuella and Nasua.

Genetic distances among coati taxa

Kimura (1980) 2P genetic distances among the main groups of N. nasua (five groups), N. narica (four groups), and N. olivacea (three groups), together with B. medius, were estimated using the mitogenome data set (Table 1). The genetic distance among the coati group and B. medius was around 20%, which is within the range obtained by Kartavtsev (2011) (higher than 16-18% for species of different genera). The highest genetic distance values obtained among coati groups were around 15.4% (N. nasua from Colombian and Peruvian Amazon vs. N. narica introgressed by N. olivacea, and N. nasua from southern Brazil, Paraguay, and Uruguay vs. N. narica introgressed by N. olivacea). The two main groups of N. olivacea vs. the five groups of N. nasua showed genetic distances ranging from 12.3% to 15.3%, whilst the same two groups of N. olivacea vs. the three groups of N. narica (excluded the group of N. narica introgressed by N. olivacea) varied from 10.4% to 12.3%. The genetic distances among the five groups of N. nasua and these three groups of N. narica were of the same magnitude, ranging from 9.4% to 12%. All of these values were clearly lower than 16-18% identified by Kartavtsev (2011) as a threshold for species of different genera. Moreover, the smaller group of N. olivacea yielded lower genetic distances with reference to the five groups of N. nasua (2.9-11.1%) than with the two main groups of N. olivacea being “a priori” the same species (13.1-13.6%). In fact, the genetic distances of the smaller group of N. olivacea with reference to the group of N. nasua from the Colombian and Ecuadorian Andes (2.9%), or the group of N. nasua from the Colombian and Ecuadorian Amazon and Colombian Eastern Llanos (6.6%), were lower than some genetic distances among the five groups of N. nasua, which ranged from 1.6% to 8.9%. For instance, the genetic distances among the group of N. nasua from the Colombian and Ecuadorian Amazon and Colombian Eastern Llanos vs. the group of N. nasua from the southern Peru and Bolivia or the group of N. nasua from southern Brazil, Paraguay, and Uruguay were 8.6% and 7.7%, respectively. The smaller genetic distances were found among the groups of N. narica (excluding the group introgressed by N. olivacea), ranging from 0.9% to 1.7%.

Thus, the magnitude of the genetic distances between the groups of Nasuella and Nasua were lower than that expected for species of different genera based on Kartavtsev (2011) using two mt genes. Additionally, the genetic distances of the groups of N. narica vs. the groups of N. nasua and vs. the groups of N. olivacea were of the same magnitude, with one group of N. olivacea showing lower genetic distances with some groups of N. nasua than the genetic distances among many groups of N. nasua. These results question the designation of Nasuella as a different genera from Nasua.

Comparative spatial structure among N. nasua, N. narica and N. olivacea

The three taxa showed similar results with Mantel tests (Fig. 4). In the case of N. nasua, the relationship between the geographical distances and the genetic distances was significant (r = 0.44, P < 0.001). Geographic distance explains about 19.7% of the genetic distance. Similarly, for N. narica, the relationship between the geographical distances and the genetic distances was significant (r = 0.44, P < 0.001). About 19.6% of the genetic distance is explained by geographical distance. For N. olivacea, the relationship between the geographical distances and the genetic distances was also significant (r = 0.46, P < 0.001) with the geographical distance explaining about 20.9% of the genetic distances. Thus, the three taxa showed a similar overall genetic structure, which is symptomatic of species closely related phylogenetically and, probably, not from different genera.

Fig. 4.

Mantel test (log transformed) between the geographic and genetic distances for the entire mitogenomes of the specimens of Nasua nasua, Nasua narica and Nasuella olivacea studied. A) N. nasua; B) N. narica; C) N. olivacea.

Fig. 5.

Spatial autocorrelation analyses for specimens of Nasua nasua, Nasua narica and Nasuella olivacea with their entire mitogenomes sequenced. A) N. nasua with ten Distance Classes (DC); B) N. narica with ten Distance Classes (DC); C) N. olivacea with six Distance Classes (DC).

Fig. 6.

Karyotype of a female (XX) of Nasuella olivacea with RBG bands.

For the spatial autocorrelation analyses (Fig. 5), the situation is also similar for the three taxa. Using 10 DCs, for N. nasua, the overall correlogram is significant (V = 0.014, P < 0.001). The first DC (0-210 km, P < 0.001), second DC (210-368 km, P = 0.014), and fourth DC (475-614 km, P < 0.001) were all significantly positive. In contrast, the fifth (614-774 km, P = 0.0159), sixth (774-1,001 km, P < 0.001), eighth (1,288-1,718 km, P < 0.001) and tenth DCs (2,256-3,513 km, P < 0.001) were all significantly negative. Therefore, for N. nasua, we found a patch diameter of around 600 km and later a structure of isolation by distance at around 3,500 km.

For N. narica, the overall correlogram of the 10 DCs was also significant (V = 0.015, P < 0.001). The first (0-26 km, P < 0.001), second (26-83 km, P = 0.026), third (83-183 km, P = 0.018), fourth (183-270 km, P = 0.034), and fifth DCs (270-337 km, P < 0.001) were all significantly positive. In contrast, the ninth (730-1,089 km, P = 0.007) and tenth DCs (1,089-2,298 km, P = 0.002) were significantly negative. Therefore, for N. narica, we found a patch diameter of around 340 km and later a structure of isolation by distance, or clinal pattern, of around 2,300 km.

Finally, the overall correlogram representing six DCs for N. olivacea was positive (V = 0.012, P < 0.001). The first (0-46 km, P < 0.001), second (46-144 km, P = 0.023), and third DCs (144-206 km, P = 0.039) were all significantly positive. The fifth (268-429 km, P < 0.001), and sixth DCs (429-753 km, P = 0.003) were significantly negative. Therefore, for N. olivacea, we found a patch diameter of around 200 km and later a structure of isolation by distance, or clinal pattern, of around 750 km.

Although the geographical range distribution for the three taxa are unequal, their geographical structures are similar (patches and later clinal pattern), which means that different evolutionary forces were acting upon these taxa in a similar fashion. In turn, this revealed strong phylogenetic relationships among the three taxa, which contradicts with the species belonging to different genera. As we will show in brief, other procyonids more differentiated phylogenetically from the coatis, also has more differentiated spatial genetic patterns.

Fig. 7.

Karyotype of a male (XY) of Nasuella olivacea with RBG bands.

Karyotype

We found 2n = 38, the FN = 72 (Figs. 6 and 7), and there were two metacentric, ten submetacentric, four acrocentric, and two subtelocentric autosomic chromosome pairs. The X chromosome was submetacentric and the Y chromosome was subtelocentric. As such, the morphology of the chromosomes of N. olivacea was indistinguishable from that of N. nasua (Wurster & Benirschke 1968). Additionally, the morphology of chromosome 15 was identical for both N. olivacea and N. nasua. However, our banding pattern was not comparable with other studies, because we obtained RGB and the banding patterns previously reported for N. nasua and N. narica were G-banded.

The relative length of the 19 chromosome pairs expressed as an average ± standard deviation with the Centromeric Index are shown in Table 2. Based on Fig. 6 and 7, and the correlation of the CI, several chromosomes in the two studies specimens had polymorphisms. Chromosome pairs 5 and 11 presented chromosome polymorphisms in the p arm, and chromosome 15 showed a polymorphism in the centromere of the p arm. This last one should be a marker chromosome due to the differential behaviour between males and females. This hypothesis requires confirmation with the analysis of additional specimens.

The sex chromosomes were easily identified with the RBG procedure. The X chromosome represented about 5% of the total genome length, a characteristic in mammalian genomes. The Y chromosome showed subtelocentric morphology with a high heterochromatin content.

Table 2.

Relative length of the chromosomes found in the karyotype of Nasuella olivacea. RL = Relative Length; CI = Centromeric Index. SD = Standard Deviation.

How many genera of coatis are there by studying mitochondrial genes?

All of the phylogenetic trees that we generated did not show reciprocal monophyly between Nasua and Nasuella. The results obtained showed that within the clades with haplotype characteristics of N. nasua, there were haplotypes of specimens with typical phenotypes (and also typical geographical distributions) of N. narica (H51, Fig. S1) and N. olivacea (H81; Fig. 2, Fig. S2). Furthermore, within the clades with haplotype characteristics of N. narica, some specimens appeared with the full characteristics of N. nasua (H107, Fig. S1), and within of the main clade of N. olivacea, there were specimens with full phenotypes of N. narica (H95, H96, H140 and H115). In fact, specimens of N. narica within the main clade of N. olivacea represent two different introgression or hybridization events. The first is a clear old introgression event from N. olivacea into N. narica (H95, H96, H140; for time splits, see Ruiz-García et al. 2020), which affected all the specimens of N. narica distributed in southern Central America (southern Costa Rica and Panama) and the northern Colombian frontier with Panama. This was the case for all specimens of N. narica studied in this area by Nigenda-Morales et al. (2019), as well as all specimens herein studied from this geographical area. Additionally, these three haplotypes comprised a homogeneous haplogroup within the N. olivacea clade but clearly differentiated from other haplogroups of N. olivacea (Figs. 2, 3; Fig. S1), which agrees well with the fact that this introgression event was enough old to differentiate the mtDNA of this introgressed N. narica haplogroup from other haplotypes and haplogroups of N. olivacea. However, the specimen of N. narica from north-western Ecuador (H115) with a haplotype of N. olivacea seems to be a case of recent hybridization because its haplotype is similar to the current N. olivacea haplotypes from Ecuador and its morphology, although nearest to N. narica, has some traits similar to N. olivacea. These results were typical of taxa that have relatively small genetic differences among them and that have a typical reticulated evolution, with introgression or hybridization at different times (Ruiz-García et al. 2018, 2019b). Therefore, the interchange of genes among specimens of these three putative taxa is not consistent with species belonging to different genera. One alternative hypothesis is that these specimens were misclassified when the samples were obtained. Nevertheless, this seems unlikely because, at least for the specimens of N. narica, all of them were alive when they were sampled and they had the unmistakable physical characteristics of N. narica. Additionally, N. olivacea is not distributed in the frontier between Colombia and Panama, nor in Panama and southern Costa Rica, where only N. narica occurs. In the case of N. nasua-N. olivacea, it should be possible that the specimen of “a priori” N. olivacea with mtDNA of N. nasua (H81) would represent one specimen of N. nasua which morphologically evolved by convergent adaption to a similar morphotype to that shown by N. olivacea through occupation of the same Andean biome. However, the skull, mandible, and teeth of this exemplar were typically of N. olivacea. Furthermore, with the three mt gene data set, as well as in Ruiz-García et al. (2021), more specimens of N. olivacea were nested inside N. nasua and they conformed to homogeneous haplogroups within this last species but clearly differentiated from other haplogroups of N. nasua. Additionally, the skulls, mandibles, and teeth of these specimens were typically of N. olivacea. Introgression, recent hybridization, and intermediate haplotypes among the three species of coatis seem more likely than misclassifications or morphological convergent adaption (possible case of N. nasua from the Andean mountains of Colombia and Ecuador) and, therefore, this correlates well with there being no genetic differences among the three species of coatis. Furthermore, the haplogroup of N. narica introgressed with mtDNA of N. olivacea showed lower genetic distances with the main haplogroups of N. olivacea relative to haplogroups of its own species. The existence of introgression indicates no reproductive barriers between the ancestors of the current N. narica and N. olivacea. In fact, it correlated well with a possible scenario based on biogeographic grounds (Toews & Brelsford 2012) and the introgressed descendent expanded through northern Colombia, Panama, and southern Costa Rica. Henceforth, these introgressed specimens were highly successful showing no genetic incompatibilities between N. olivacea and N. narica. This is an improbable outcome for specimens of fully differentiated genera.

Another relevant result obtained here is in relation to the haplotype H144 found within N. olivacea. This specimen was “a priori” classified as N. nasua (Fig. 2) because in the geographical area where it was sampled there was no record of the presence of N. olivacea. However, the mitogenome obtained is typical of N. olivacea and a detailed analysis of the morphotype of the individual revealed that it probably corresponded with a “true” N. olivacea. Therefore, it could be the first real register of N. olivacea in Peru (the River Urubamba, Cuzco).

With the mitogenome tree analyses, we only discovered one relationship: ((N. nasua + N. narica) + N. olivacea) with the percentages of bootstraps considerably higher than those obtained with the three mt genes, potentially supporting the maintenance of two traditional genera of coatis. We, however, think that a unique genus (Nasua) should be considered as the preferential option because with mitogenomes no reciprocal monophyly was observed among the three coati taxa, which was consistent with the karyotype analysis with no relevant differences between Nasuella and Nasua. Also, for the three mt gene data set, the major part of the trees obtained with different outgroup species showed a major relationship between N. narica and N. olivacea that was greater than either of these two taxa with N. nasua, as Helgen et al. (2009) found for mtCytb. It is interesting to note that when the sequence of P. cancrivorus was present as an outgroup, with or without the other outgroup species, N. olivacea was differentiated from N. nasua + N. narica. In contrast, if P. cancrivorus was excluded (whether or not other species of out-groups were included), then, the relationship was N. olivacea + N. narica, with N. nasua more differentiated. It is curious that with the three mt genes, Nasua + Nasuella yielded a stronger relationship with P. cancrivorus than with Bassarycion. This finding agrees well with morphological studies (Baskin 2004), but contradicts the molecular relationships recovered by Koepfli et al. (2007), who showed that the sister species of Nasua + Nasuella was Bassarycion.

The absence of differentiation of Nasuella from Nasua is likely because the evolutionary trajectory for the coatis is a continuous process and not a discrete one. This is more apparent in the MJN than in phylogenetic trees. We consider this true for intra-specific relationships, or for closely related species (such as in this case). A MJN better reflects the evolution of taxa than do traditional phylogenetic bifurcating trees (PBT) for four reasons (Freeland et al. 2011). 1) Population genealogies are frequently multifurcated. In our case, MJN allowed multifurcated events, whereas PBT did not. 2) Within species, or among closely related species, genetic similarity can be generally high, or very high. Whilst MJN can reconstruct genealogies with restricted genetic variability, PBT requires more differentiated characters to discriminate among the taxa analysed. 3) At an intra-specific level, or among closely related species, ancestral and derived haplotypes can coexist within populations or closely related taxa. MJN allows for both original and descendant haplotypes, whereas PBT assumes that ancestral haplotypes no longer exist. 4) At the intra-specific or closely related species level, hybridization and recombination can occur often and be important. MJN can easily reveal hybridization and with some procedures, recombination (nuclear genes) as well. This is much more limited for PBT.

The MJN carried out here showed that one haplogroup of N. olivacea followed some Andean N. nasua haplotypes that were basal. In fact, some N. nasua, living at the Colombian and Ecuadorian Andean Cordilleras, were more related with one haplogroup of Nasuella than with other haplogroups of N. nasua. However, Ruiz-García et al. (2020, 2021) showed the most basal haplogroup to be coatis of the Colombian and Ecuadorian Andean N. nasua haplogroup, followed by one haplogroup of N. olivacea. Therefore, the mitogenome data set (with few specimens studied, but longer sequences) and the three mt data set (with a greater sample size and more diversified geographical origins, but shorter sequences) did not offer the same conclusion about which of the current coati haplotypes are basal. More Andean coatis (both N. nasua and N. olivacea) should be analysed with both mitochondrial and nuclear genes to resolve this question. Nevertheless, both mt data sets showed that the origin of the current coatis seem to have originated in the Andean cordilleras from north-western South America (current Colombia and Ecuador). This process could have begun around 13-10 MYA, during the Miocene (Ruiz-García et al. 2021) and from here the ancestors of the current coatis expanded to southern South America and Central America. Support in favour of the origin of the current coatis in north-western South America is the fact that the majority of introgression and hybridization cases were in the territory of Colombia and Ecuador. This result agrees well with the findings of Nigenda-Morales et al. (2019). The two S-DIVA and BBM biogeographic analyses conducted by these authors identified South America as an area of distribution for the most recent common ancestor of Nasua and Bassaricyon. They estimated the split between the ancestors of N. nasua and N. narica to have occurred around 6 MYA, which is compatible with that reported by Ruiz-García et al. (2021) and with the results shown here.

We detected intermediate haplotypes between N. nasua and N. narica. For instance, the genetic distance between a haplogroup of N. nasua, in the Colombian and Ecuadorian Amazon and Eastern Colombian Llanos, and the most basal haplogroup of N. narica was 7.7% for the mitogenome data set. This value is lower than the genetic distances of different haplogroups of N. nasua (for instance, 8.6% between this Colombian and Ecuadorian Amazon and Eastern Colombian Llanos one, and one haplogroup from southern Peru and Bolivia, or 8.9% between this last haplogroup and one haplogroup from the Colombian and Ecuadorian Andes). This result is consistent with colonization from northern South America into Central America. Indeed, Nigenda-Morales et al. (2019) detected asymmetric patterns of colonization, with migration from Panama into northern Central-American populations to be greater than in the opposite direction, which is the reverse of the traditional paleontological viewpoint (Soibelzon & Prevosti 2013).

Finally, the genetic distances of the most differentiated haplogroups of N. olivacea in relationship with N. nasua and N. narica were 12.3-15.3% and 10.4-12.3%, respectively. However, these values were not of the order of 16-18% or higher (Kartavtsev 2011), which is expected among species of well differentiated genera. As such, we suggest there are sufficient reasons to consider that all of the coatis are part of a single genus.

Total agreement in the spatial genetic structure of N. nasua, N. narica and N. olivacea

The results for the Mantel tests and those of the spatial autocorrelation showed similar structures for the three coati taxa studied although the geographical extent of each species was different (for instance, for N. nasua, we sampled specimens over a distance of more than 3,500 km, whereas, this distance was around 750 km for N. olivacea) as well as the geographical barriers and biomes where the three coati taxa occur are different.

Generally speaking the few spatial genetic studies carried out with Procyonids have detected significant spatial structure. Cullingham et al. (2008b), with Procyon lotor, detected that some geographical barriers could enhance significant genetic differences between populations of raccoons in North America. In the Niagara region, two genetically different raccoon populations were identified corresponding to either side of the River Niagara. However, for the St. Lawrence region, spatially congruent clusters were not identified, despite the presence of the intervening St. Lawrence River. Cullingham et al. (2008a) sequenced, for the mt control region, specimens from four putative morphological subspecies of P. lotor that occur along the eastern seaboard of North America through to the central United States. They showed three distinct lineages. One of them was found primarily in Florida, one along the eastern seaboard, and the third predominantly to the west of the River Mississippi. A SAMOVA analysis indicated that different barriers contributed to differentiate these three lineages (river-mountains at the east of the studied area, river-mountains at the west of the studied area, and by regions). However, there was considerable lineage mixing across the eastern seaboard and to the west of the River Mississippi. Rioux Paquette et al. (2014) analysed several microsatellites for raccoons in southern Quebec and they detected that the genetic distance among the raccoon males was strictly a function of geographic distance, while dispersal in raccoon females was significantly reduced by the presence of agricultural fields. Thus, females were more affected by barriers than males, which could agree with that reported here with coatis based on mtDNA. Biedrzycka et al. (2014) examined the microsatellite and mitochondrial diversity of raccoon populations recently introduced in Central Europe (Germany, Poland and Czech Republic). They detected two genetically different groups with isolation-by-distance showing a significant but weak positive relationship between geographic and genetic distance. Thus, procyonids seems to easily develop spatial structuring like we showed here for coatis. Nevertheless, the spatial genetic structures observed in P. lotor could not be compared with those shown here for the three coati species because correlograms with the A_y distance were not employed in these studies. The unique result obtained for a Procyonid with the same procedure employed here was the case of the kinkajou, Potos flavus (Ruiz-García et al. 2019a).

Diverse microevolutionary processes can differentially affect genomes if they are in some degree different and therefore develop different spatial structuring. In contrast, if genomes are similar they can respond to geography in a similar way (Sokal & Jacquez 1991). Sokal & Wartenberg (1983) and Sokal et al. (1989b) showed, in metasimulations, that stochastic generating processes produced genetic surfaces with characteristics that were a function of parameters such as parent vagility and neighbourhood size. Different simulations with identical parameters generated identical, or very similar, spatial correlograms, including different kinds of migration or selection. We wish to show that the spatial correlogram of N. olivacea is significantly more similar to those of N. nasua and N. narica than to the correlogram of other procyonids of other genera, such as P. flavus.

To demonstrate this, we generated a correlogram with the same number of DCs (ten) for the four Procyonid taxa with the size of each DC being as similar as possible. The correlograms were later compared by computing average Manhattan distances (Sneath & Sokal 1973) between pairs of correlograms over the ten DCs constructed. Sokal et al. (1986, 1987, 1989a) demonstrated that spatial correlograms generated by the same microevolutionary forces affecting identical genomes showed Manhattan distances among their correlograms of 0.1-0.2. The Manhattan distances between the correlogram of P. flavus and those of N. nasua, N. narica and N. olivacea were 0.354, 0.619 and 0.488, respectively (significantly different to 0.2; Fisher exact test, P < 0.001; Everitt 1992). Potos flavus also has a significant spatial structure like the three species of coatis, but its spatial structure was higher than that detected in the coatis and its correlogram was significantly different to the correlograms of the three coati taxa. Thus, the microevolutionary processes that affected kinkajous were different to those that affected the coatis. In contrast, the Manhattan distances between N. olivacea vs. N. nasua and N. narica were 0.147 and 0.183, respectively, and they did not differ from 0.2 (Fisher exact test, P > 0.6). Hence the three coatis have mitogenomes similar enough to be affected by the same microevolutionary processes in an identical way. In fact, the correlograms of N. olivacea and N. nasua were more similar (affected more similarly by identical evolutionary processes, 0.147) than the correlograms between N. nasua and N. narica (affected by less similar evolutionary processes, 0.265). This finding suggests that the reproductive and the migratory behaviours of the coatis are more relevant for this spatial structure than the geographical features in the distribution range of each taxon. Coatis are highly gregarious, forming social groups of up to 20-40 females and associated juveniles. Males are typically solitary and disperse once they reach sexual maturity, with brief contact with the female groups only during the mating period. Females are highly philopatric and their home ranges generally include their birth area (Gompper 1995, 1997, Gompper et al. 1997, 1998, Valenzuela & Ceballos 2000, Hass 2002). This means that female capacity to migrate is low and the high levels of philopatry may lead to pronounced fine-scale genetic structuring (Ruiz-García 1998, 1999). Although there are some reports that males have moved more than 20 km between years (Lanning 1976), our spatial genetic results suggest that the three coati taxa have limited capacity for dispersion and, therefore, their behaviours are strongly similar because they are not very differentiated taxa, which is more indirect evidence to include Nasuella within Nasua.

The karyotype of N. olivacea

This is the first time that the karyotype of N. olivacea (one male, and one female) is reported. Although our banding pattern for N. olivacea was not comparable with the banding patterns obtained for the two species of Nasua, the chromosome morphology is comparable. As mentioned, the chromosome morphology of the karyotype of N. olivacea was un-differentiable from that reported by Wurster & Benirschke (1968) for N. nasua. It is composed of 28 metacentric, submetacentric, and subtelocentric autosomic chromosomes, eight acrocentric autosomic chromosomes, one submetacentric X chromosome, and one subtelocentric Y chromosome. In fact, the chromosome morphology of N. narica, although highly similar to that of the other coatis, showed some minor differences to that described for N. olivacea and N. nasua. It has one additional pair of metacentric and submetacentric autosomic chromosomes and one less pair of the acrocentric autosomic chromosomes, as well as a different acrocentric or small submetacentric Y chromosome.

All other Procyonidae genera also have 2n = 38. However, these karyotypes show some differences to the karyotype of the coatis. For example, Bassaricyon gabii has an autosomal complement of 28 meta- and submetacentric chromosomes and eight acrocentric chromosomes. One pair of small acrocentric chromosomes in this species has satellites on its short arms. The X chromosome is a medium-sized submetacentric chromosome, similar to that found in Nasua and Nasuella, but the Y chromosome is a small subacrocentric chromosome different to that of Nasua and Nasuella (Wurster & Bernirsche 1967, 1968). The North American raccoon (P. lotor) has 30 metacentric, submetacentric or subtelocentric chromosomes and six acrocentric or telocentric chromosomes (different numbers than the coatis). The X chromosome is submetacentric and the Y chromosome is submetacentric or subtelocentric. A pair of small subtelocentrics (pair 14) possesses a distinctive satellite on each short arm, similar to the E1 pair of domestic cat (Benirschke et al. 1966, Hsu & Arrighi 1966, Todd et al. 1966), which is not present in the coatis. Finally, the karyotype of Bassariscus astutus (ring-tailed cat) has a FN = 68, which is different from the FN = 72 of Nasua and Nasuella. The autosome chromosomes were 36 submetacentrics and subtelocentrics. Additionally, the karyotype of this species includes a large submetacentric X chromosome and a small acrocentric Y, which is not present in the coatis (Hsu & Arrighi 1966, Wurster-Hill & Gray 1975). Although the karyotype of the Procyonidae is conservative, the differences observed among the different genera and the close similarity between N. nasua and N. olivacea support one unique coati genus, Nasua, rather than two well-differentiated genera. One of the Nasua species (N. nasua) was more related to Nasuella than to the other species of Nasua (N. narica).

Taking into consideration the mitochondrial and the karyotype results presented here, it seems clear that all coatis belong to a unique genus: Nasua. Nuclear genes, immunological, reproductive, and ethological studies should be conducted to further investigate the status of Nasuella as a “true” genus.