The forensic community often debates which data sets are valid for genetic comparisons and for determining the significance of a DNA match. Human populations and races and breeds of domesticated animals can be clearly genetically defined with molecular and DNA markers, suggesting substructure. However, genetic markers that have common frequencies in all populations can alleviate this concern for finding the appropriate data set to determine match probabilities. Mitochondrial DNA is expected to have less discriminatory power than nuclear markers due to its maternal inheritance and lack of recombination [34
]. Although mtDNA has a high mutation rate, regions with too much genetic variation can lead to heteroplasmy that complicates forensic interpretations [35
], especially in tissue types, such as hair, which have a high mitotic index [37
]. Thus, a mtDNA region with a balance of variation, high for discrimination of an individual, but low to minimize heteroplasmy, needs to be defined in each species. The region analyzed in this study excludes the two highly repetitive regions in the mtDNA that show extensive heteroplasmy [2
] and this study and our own recent work suggests that the region has low heteroplasmy [26
] (unpublished data Huang et al., submitted)
Assessment of human mtDNA mitotypes has focused on hyper-variable regions I and II. These regions demonstrate a high degree of genetic variability coupled with low match probabilities. Likewise in cats, a 402 bp segment of the cat mtDNA CR has shown to have strong discriminatory power and low heteroplasmy [2
]. To date, statistical support for the cat mtDNA CR as a forensic tool has been limited to the work of Halverson and Basten [2
]. This study has extended the initial investigations of the same mtDNA region with additional USA populations, 25 worldwide populations, and 364 cats representing 26 USA breeds.
The 402 bp sequenced region from worldwide sampling of nearly 1,400 domestic cats reveals 12 universal mitotypes that have a frequency of >1% of the combined population set. However, most of the 12 mitotypes can be collapsed into two major mitotypes, mitotypes A and D, as B, C, F, G, J, K and L have only three or fewer mutations in difference. Mitotypes E, H, and I remain distinct, as do the rare outlier mitotypes OL1 – 3, resulting in 8 distinct cat mitotypes throughout the world. Mitotype K appears to have derived in the USA from mitotype A, while mitotype J is limited to the Eastern Mediterranean, derived from mitotype D. An interesting comparison would be evaluating the CR of the five major mtDNA types identified in the domestication study of the cat to the five major types identified in this study [23
To effectively define the universal mitotypes within a geographic location, excessive sequencing efforts may not be required. For example, the Egyptian and the Northern California, data sets were able to identify eleven or more of the twelve universal types with sample sizes of 130 and 133, respectively. The 100 samples of the New York data set failed to identify mitotype F, a mitotype found in all other reasonably sampled United States populations, however, ten major mitotypes were found. Similarly, the Southern California population had 99 representative samples and identified nine major mitotypes. Estimates of the required sample sizes for human populations are similar. Interestingly, the most common western (USA and Europe) mitotype B, is not found in several Middle Eastern countries but is present in India, China and Southeast Asian. Thus, when evaluating population structure in both forensic and evolution studies, the extent of historical and recent migration patterns needs to be considered. In addition, some missing mitotypes may be due to a sampling bias if broad areas for a given region are not considered.
The proportion of unique mitotypes within a population was generally less than 10%, except for the two European populations of Germany and Italy, where over 50% of cats had unique mitotypes. The two regions also had less than 25 sampled individuals. Thus, the “unique” mitotypes may actually represent universal mitotypes or subtypes that have not yet had sufficient sampling to identify additional samples with the same type. The within European pairwise differences was significantly higher (8.56 ± 4.0) than any other within or between group comparison except the Europe – East Asia comparison. This data supports previous STR studies that suggest European and Southeast Asia cats are fairly distinct [27
]. Hence, a more extensive survey of European mtDNA domestic cat structuring seems warranted.
Purebred cats were undifferentiated with respect to mitotypes. None formed monophyletic groups nor were they restricted to a single mitotype although a subset of Maine Coon cats do possess a distinguishing mitotype (16820C in the C mitogroup). Some were representative of their hypothetical country of origin Korat (Thailand), Birman (Burma) and Siamese (Thailand), Turkish Van (Turkey), while others were not, such as the Turkish Angora (Turkey). Overall, the mtDNA correlates breeds to their geographical origins, but the mitotypes are not sufficiently distinct to define a breed or population.
The presented data set includes mtDNA CR data from 1,394 individual cats. The mtDNA region examined by Japanese researchers [25
] was avoided due to concerns regarding inclusion of a repetitive sequence at the 3’ area that is polymorphic but difficult to sequence in forensic samples [2
]. However, comparisons between the current study and the Japanese study yielded similar results. The 1,394 samples in this study have a genetic diversity value of 0.8813 ± 0.0046 compared to 0.8767 ± 0.0277 for the Japanese feline sample set. The random match probability was 12.0% compared to 14.1% for the Japanese samples. The similarity of both the genetic diversity values and the random match probabilities suggests that the regions are comparable in forensic applications and both regions could be useful if sufficient DNA is available for analysis.
The cat mtDNA CR examined in this study also has comparable power to a similar study in the dog[39
]. Persians, which represent >50% of fancy breed cats in the USA, have a diversity value of 0.8456 with an RMP of 20.4%. Increasing sample size may reduce the random match probabilities. In the small sample set from a human Japanese population (N = 50), genetic diversity of 0.973 and random match probability of 4.6% [40
] was determined. Diversity values increased minimally, ranging from 0.996 to 0.998 with increasing sample sizes. Correspondingly, random match probabilities decreased from 2.2 to 0.6% [41
]. Across the cat breeds, heterozygosity approaches 0.880 with a random match probability of 13%. Since unique alleles are in low frequency in any given cat population, increasing sample numbers will not necessarily result in increased diversity values or decreased RMPs. More likely, heterozygosity values will decrease as common allele numbers increase with efforts to identify unique mitotypes within a single population. Caution should be noted with cat populations however, as they may not disperse as readily as other species, and parent offspring relationships may be common in a given area, thus, we reinforce that mtDNA data be used primarily as an exclusionary tool.
Differences between human, canine, and feline genetic diversities and random match probabilities are a reflection of the historical age of each group. Human mtDNA mitotypes have been diverging for greater than 150,000 years [42
] with Eurasians diverging from a single population 60,000 – 80,000 years ago [43
]. Although domestic dog progenitors may have diverged from their wolf ancestors as long as 100,000 years ago [44
], dog domestication and subsequent mitochondrial mitotype proliferation has been more recent. Mitochondrial data suggest all dogs originated from 3 females in China roughly 15,000 years before present [45
], and the accumulation of mutations has accelerated with a relaxation of selective constraints has been suggested as well [46
]. Also, extensive selection for breeds has occurred within the last 200 years and currently there are an estimated 400+ breeds recognized by various worldwide kennel clubs [47
Cat domestication is more recent than for most animal breeds. Archeological evidence suggests cats were in direct contact with humans as early as 9,500 years ago [48
]. From a pragmatic perspective, this is the logical consequence of the domestication of grains occurring in the same time interval [49
]. The archeological evidence is supported by mtDNA studies that suggest five mitochondrial lineages were present roughly 9,000 years before present and from these five lineages all contemporary mitotypes have been derived [23
]. However, as domestication is subjective, especially with respect to cats, the intentional breeding of cats may not have occurred until as late as the 19th
century B.C., during the 12th
Egyptian dynasty [51
]. Breed proliferation has occurred within the last 100 years with most breeds originating within the last 50 years. Moreover, several cat “breeds” are merely coat color or length variants of the different breeds, fostering the exchange of mitochondrial mitotypes across breeds. The relatively short time-span of domestic cat expansion and corresponding mitochondrial mitotype divergence addresses the issue of decreased genetic diversity values compared to humans and dogs. However, it also may prove advantageous as the extensive sampling required to resolve sub-structuring in human populations may prove unnecessary in the domestic cat. The robust sampling of cats from USA indicate little sub-structuring between the populations, all populations having the most common mitotypes and similar frequencies of unique mitotypes. The mean pairwise differences within random bred cat (5.55 ± 2.7) groups were similar to the level found between groups (5.6 ± 2.7). However, the European and Southeast Asian populations appear distinct and the high level of unique mitotypes in the European sampling suggests that more extensive non-USA data sets for cat mtDNA may be warranted.