Since its reintroduction to the Americas during the time of Spanish colonization (Luís et al., 2006; Jimenez et al., 2012), the horse has played a significant role in the history and culture of Mexico (LeCompte, 1985; Griffith and Fernández, 1988; Nájera-Ramírez, 1994; Palomar, 2004; Pineda and Díaz, 2012). Mexico has one of the largest horse populations in the world (da Mota and Regitano, 2012) with a substantial number of mixed-breed horses used for agricultural, farming, transportation, tourism and patrolling (i.e. mounted police) purposes (Loroño-Pino et al., 2003; Velázquez-Beltrán et al., 2011; Cantú-Martínez et al., 2012; Sánchez-Casanova et al., 2014). However, a significant number of horses in Mexico are purebred, mainly promoted by four official horse breeding associations that manage the studbooks for breeds such as the Pura Raza Española, Lusitano, Azteca, Westphalian, Holsteiner, Dutch and Belgian warmbloods, Selle Français, Hanoverian, Thoroughbred, Quarter Horse and Arabian (CONARGEN, 2017; Domínguez-Viveros et al., 2014). Purebred horses in Mexico are used for several equestrian events, including horse racing, Olympic disciplines and traditional equestrian competitions known as charreadas (Palomar, 2004; Cantú-Martínez et al., 2012; Pineda and Díaz, 2012; Rodríguez-Sánchez et al., 2015). Additionally, a barely documented, but broadly extended use for these purebred animals is the training for horse dancing competitions during traditional regional festivities. Another relevant horse breed in Mexico is the Creole (Criollo) horse, which also plays a significant role in farming and Mexican equestrian traditions (Domínguez-Sánchez et al., 2015). The Creole breed is derived from feral horse populations introduced during the first Spanish expeditions to the Americas (Domínguez-Sánchez et al., 2015), similar to other Creole breeds in South America (Kelly et al., 2002; Mirol et al., 2002; Jimenez et al., 2012).

The central region of Mexico has experienced a sizeable introduction of purebred horses. However, a potential problem with these introductions is the maintenance of breed integrity. Conservation of breed identity and adequate effective population size is pivotal for sound breeding management in any species, and genetic markers such as short tandem repeat (STR) loci (i.e. microsatellites) have been extensively used to examine genetic diversity and breed characterization in horse populations worldwide (Aberle et al., 2004; Marletta et al., 2006; Druml et al., 2007; Zuccaro et al., 2008; Felicetti et al., 2010; Chauhan et al., 2011; Ling et al., 2011; Takasu et al., 2012; Fornal et al., 2013; Berber et al., 2014). In this study, we used 17 microsatellite DNA markers to provide genetic diversity and effective population size of four horse breeds of economic and cultural importance in Mexico.

The microsatellite panel was highly informative, with PIC ranging from 0.518 to 0.860 (Table 1), very high individual combined exclusion probabilities in the four evaluated populations (EP > 0.999) and individual EP ranging from 0.766 to 0.971 for HMS1 and ASB23 loci, respectively. The mean number of alleles per locus was 9.41.

The four breeds showed heterozygosity values around 0.800 and lower, with the highest HE values observed in the CRL and AZT populations (Table 2). All HO values were higher than their corresponding HE values and the higher values were estimated in THB. Although values for AR were similar between the four breeds, the highest and the lowest values were found in the CRL and THB groups, respectively (Table 2). Values of FIS were negative and near zero, directly related to heterozygosity estimates. The estimated Ne values were low and varied from 1.0 to 14.4 for the THB and CRL breeds, respectively, by LD method. Whereas, the higher Ne values from the CM method was estimated for QHR. Both methods provided low but different estimations of Ne for analysed breed groups; however, in general, lower effective population estimates were observed in THB (Table 2).

Genetic assignment analysis showed no possible pattern of genetic flow between assessed breeds since all individuals were assigned to their reference-compared populations. However, a resemblance between breeds was suggested by the population structure analysis; the Evanno's test indicated that the most informative number of subpopulations were K = 2 (Fig. 1).

Documentation in the specialized literature on the genetic diversity and structure of current horse breeds in Mexico is scarce. However, it is well known that some breeds were introduced for different recreational purposes. It is important to highlight that the population analysed in the present study has a significant relevance for breeding practices as it is composed of stallions and young male breeding candidates that will be used for breed dissemination locally and regionally by natural breeding and/or with the use of assisted reproduction (e.g. artificial insemination).

In some genetic diversity studies with microsatellites the number of horses analysed per breed ranged from 17 to 24 (Aberle et al., 2004; Solis et al., 2005; Marletta et al., 2006; Burócziová and Říha, 2009; Leroy et al., 2009; Chauhan et al., 2011; Ling et al., 2011; van de Goor et al., 2011; Prystupa et al., 2012; Khanshour et al., 2013; Berber et al., 2014), which is similar to the number of animals used per breed in the present study.

Consistent with the total average number of alleles observed, the allelic richness estimations by breed were low, promoted perhaps by the reduced sample size or possible founder events (Greenbaum et al., 2014). Since AR is considered a useful indicator of population size reduction and past bottleneck events (Nei et al., 1975), the implication for our sampled breeds could be more important as no information on the initial allelic composition available could suggest a deep reduction in original diversity. Is important to observe that a decrease in the allelic richness could lead to a reduction in the population's potential to adapt to future environmental changes, hence the importance for conservation management (Greenbaum et al., 2014).

Heterozygosity was slightly higher than that reported for Thoroughbred horses (Cunningham, et al., 2001; Glowatzki-Mullis, et al., 2006; van de Goor et al., 2011) and Quarter Horses (Luis et al., 2007) in previous studies with the same or similar loci and could be due in part to the random origin of the samples (different ranches). Similarly, lower heterozygosity values were reported in Argentine, Brazilian, Chilean and Venezuelan CRL horses (Cothran et al., 2011).

The negative FIS values found in the four breeds indicate high intra-breed variability and lower heterozygosity. Petersen et al. (2013), using whole-genome single-nucleotide polymorphism data, suggested that horse breeds with low within-breed diversity, such as Thoroughbred and Quarter Horse, may be related to population bottlenecks, intense selective pressure or closed conditions due to their long breed histories. For the Thoroughbred breed, they reported a high mean individual inbreeding coefficient (larger than 10 %) (Petersen et al., 2013). In the present study, high heterozygosis showed no current inbreeding risk. Nevertheless, the two methods used for Ne estimation consistently indicated very low effective populations. However, the LD method was suggested to have estimation bias when used on small samples (England et al., 2006). Recently, improvement in contemporary effective population size estimation methods (e.g. NeEstimator v.2.0) suggests highly reliable estimates of Ne under migration (LD method) and no migration (MC method) assumptions, even with small samples (Gilbert and Whitlock, 2015). Perhaps further and wider sampling would confirm the present results.

In general, the present results confirmed the usefulness of microsatellites in the assessment of genetic diversity and integrity of different horse breeds around the world (Aberle et al., 2004; Solis et al., 2005; Marletta et al., 2006; Luis et al., 2007; Burócziová and Říha, 2009; Leroy et al., 2009; Chauhan et al., 2011; Ling et al., 2011; van de Goor et al., 2011; Conant et al., 2012; Prystupa et al., 2012; Khanshour et al., 2013; Berber et al., 2014). Genetic characterization helps to assure breed integrity and to assign individuals to defined populations (Glowatzki-Mullis et al., 2006). In the present assessment, considering the population structure results showing the two most likely conforming populations, and the partially mixed CRL population could suggest some degree of genetic flow and resemblance that could be explained by the crossbred strategies used in the creation of the CRL breed, leading to a breed that is not well defined.

Therefore, both records on breeding management and genetic monitoring by genetic markers may be necessary to avoid undesirable consequences of effective population size reduction and breed dilution. Together with genetic information in classical horse breeding and recent achievements in population and conservation genetics (Achmann, et al., 2004; Hasler et al., 2011; Janova et al., 2013), the results of the present study have implications for better development of management strategies for local populations such as the breeds assessed here. As such, a comprehensive approach would include a periodical assessment of known local populations to ensure the acceptable effective population size and breed integrity documentation of genetic flow and reproductive management, as most of the introduced animals in the studied area are used as breeders, being relatively expensive and highly valued animals.

In conclusion, using an informative microsatellite panel, we verified diversity indicators of four introduced horse populations and confirmed small effective population sizes as a risk for future breed integrity if not considered during breeding management. Attention to these indicators will avoid undesirable consequences in the breed integrity of these popular animals.