Printer Friendly

Systematics, diversity, genetics, and evolution of wild and cultivated potatoes.

Table of Contents


   Genetic Basis of Species Boundaries in Potatoes
   * Self-incompatibility
   * Unilateral Incompatibility
   * Male Sterility
   * 2n Gametes
   * Endosperm Balance Numbers
   * Stylar Barriers
   * Dihaploids
   * Dihaploid-Wild Species Hybrids
   * Tetraploid Genetics
   * Genetic Basis of Species Boundaries in Potatoes

   Wild Potato Taxonomy and Phylogeny
   * History of Taxonomic Treatments of Solarium section Petota
   * Series Treatments in Solarium section Petota
   * Morphological Studies of Species Boundaries Subsequent to Hawkes
   * Molecular Studies of Species Boundaries Subsequent to Hawkes
   * Introgression and Interspecific Hybridization
   * Taxonomic Changes Subsequent to Hawkes (1990)
   * Ingroup and Outgroup Relationships
   * Polyploidy--Occurrence, Taxonomy, Biogeography, Habitats
   * Genome Differentiation in section Petota Identified by Genomic in
     situ Hybridization
   * Polyploidy--DNA Sequence Data
   * Wild Potato Taxonomy: Our New Taxonomy Adopted Here

   Cultivated Potato Taxonomy and Phylogeny
   * Early Classifications of Cultivated Potatoes
   * Cultivated Potato Taxonomy: Our Recent Potato Landrace

   Origin of Cultivated Potatoes
   * Multiple Origin Hypothesis
   * Restricted Origin Hypothesis
   * Recent Studies of Cultivated Potato Species Origins
   * Summary of Cultivated Potato Origins

   From Landraces to Modern Potatoes
   * Domestication Traits
   * Geographic Correlates of Potato Systematics and Diversity
   * Germplasm Collections and Molecular Characterization
   * Germplasm Utilization and Contributions to Cultivar Improvement
   * Breeding Potential of Polyploids
   * Where do we go From Here? New Discoveries From Whole-Genome DNA
     Sequencing Data
Literature Cited


The common potato, Solanum tuberosum L., is grown and consumed worldwide. It is the third most important food crop (FAO, 2013), and has remained so for at least 190 years (Sabine, 1824). Solanum tuberosum is the name traditionally used for landrace (indigenous cultivated) populations grown in lowland Chile and in the high Andes. The name S. tuberosum is also used for the world's potato cultivars grown since the end of the sixteenth century outside of South America. The modem cultivars are the products of extensive breeding between different cultivar groups and wild species. Landrace potatoes and wild potato species, all classified as Solanum section Petota, are widely used for potato improvement. Members of section Petota are broadly distributed in the Americas from the southwestern United States to the Southern Cone of South America (Hawkes & Hjerting, 1969, 1989; Ochoa, 1990a, 1999; Spooner et ah, 2004, 2014). The last comprehensive taxonomic treatment of section Petota was published by John (Jack) Hawkes in 1990; it recognized seven cultivated species and 228 wild species, divided into 21 taxonomic series, including 19 series for tuber-bearing species and two series of non-tuberous species. Here we consider section Petota to include only the tuber-bearing species. Since 1990, intensive field collections from throughout the range of the group, coupled with morphological and molecular studies, have halved the number of species and elucidated new ingroup and outgroup relationships. The recent sequencing of the potato genome (The Potato Genome Sequencing Consortium, 2011) has greatly accelerated investigation of all aspects of potato biology.

The purpose of our review is to provide a historical overview and update since 1990 of the systematics, diversity, genetics, domestication, evolution, and breeding of Solanum section Petota that serves as a reference to aid the next generation of studies in the group. It updates reviews of Spooner and Hijmans (2001) and Spooner and Salas (2006) that were bibliographic summaries of taxonomic changes by many authors. This review is intended to provide our current and thoroughly independent taxonomic decisions regarding the number of species and the interrelationships among species in section Petota. We begin with a presentation of the genetics of the group because this has historically provided key concepts used to form taxonomic decisions and to choose species for breeding programs. Species names serve many purposes, one of which is to link studies across different publications. Because of the large reduction in species adopted here (Table 1), we use both the names in the original publications, followed by our concept of these species in parentheses, for example: S. fendleri (=S. stoloniferum).

Genetic Basis of Species Boundaries in Potatoes


Most diploid tuber-bearing Solanum species are self-incompatible due to a genetically-based gametophytic self-incompatibility system (Pushkamath, 1942; Pandey, 1962). The style produces an S-RNase that inhibits the growth of genetically matching pollen tubes (Luu et al., 2000). The S locus and S-RNase genes have been localized on chromosome 1 (Tanksley & Loaiza-Figueroa, 1985; Rivard et al., 1996). Within wild species populations, self-compatible plants have occasionally been reported (Pushkamath, 1942; Pandey, 1962; Cipar et al., 1964; De Jong et al., 1971; Oldster & Hermsen, 1976; De Jong & Tai, 1977; Flermsen, 1978; Birhman & Hosaka, 2000; Phumichai et al., 2005). In some plants of the wild species S. chacoense, self-compatibility is controlled by the dominant allele of an S-locus inhibitor gene (Sli) (Hosaka & Hanneman, 1998a) that is independent of the S locus. Unlike the S locus, which is expressed in the gametophyte, the Sli gene is expressed in the sporophyte. The Sli gene has been mapped to the distal end of chromosome 12 (Hosaka & Hanneman, 1998b). In addition to Sli in S. chacoense, a diploid S. tuberosum clone (US-W4) expresses a dominant self-incompatibility inhibitor (De Jong et al., 1971). The genetic basis of self-compatibility in US-W4 is not known.

In contrast to diploid potatoes, tetraploid potatoes (both wild and cultivated) are self-compatible. The breakdown of the gametophytic self-incompatibility mechanism in polyploid species is a common phenomenon in angiosperms (Frankel & Galun, 1977; Levin, 1983). In tetraploids, the pistil is still functional in the incompatibility reaction. However, since the pollen is diploid rather than monoploid, it does not elicit an incompatibility response. The molecular basis for the loss of self-incompatibility in polyploids is not understood (Comai, 2005).

When the cultivated potato or its SYi-bearing wild relatives are self-pollinated, a high degree of inbreeding depression is observed in the form of flower bud abortion, lack of flower bud formation, and sterility (De Jong et al., 1971; Birhman & Hosaka, 2000). In addition, selfing causes reductions in vigor and yield, presumably because these traits are controlled largely by heterotic genetic effects (Krantz, 1924, 1929; De Jong et al. 1971; Mendiburu & Peloquin, 1977; Ross, 1986; Golmirzaie et al., 1998). After several generations of self-pollination of the diploid wild species S. chacoense, however, vigorous, fertile clones were produced (Phumichai et al., 2005). In fact, recent SNP analyses of ten wild potato relatives previously used in cultivar development by potato breeders have revealed unexpectedly high levels of homozygosity (Hirsch et al., 2013). Assumptions about abundant heterozygosity within wild populations are called into question as a result of these initial findings. Perhaps wild and cultivated potatoes are actually somewhat tolerant of inbreeding.

Unilateral Incompatibility

Unilateral incompatibility is a phenomenon in which self-compatible species can be crossed as a female, but not as a male, to self-incompatible species (Abdalla & Hermsen, 1972). Pollen tubes fail to penetrate stylar tissue in self-incompatible (female) x self-compatible (male) crosses. Although most diploid potato species are self-incompatible, the Mexican diploid species S. verrucosum is self-compatible. Solanum verrucosum can be crossed as a female, but not as a male, to self-incompatible species (Dinu et al., 2005; Jansky & Hamemik, 2009). The stylar tissue of S. verrucosum does not produce the S-RNases that inhibit pollen tube growth in incompatible crosses (Eijlander, 1998). This is likely why it can be used as a female parent in crosses to species in section Petota (Hermsen & Ramanna, 1976; Jansky & Hamemik, 2009).

It is sometimes possible to find exceptional plants that do not exhibit unilateral incompatibility in self-incompatible x self-compatible interspecific crosses (Pandey, 1962). The identification of such plants allows a breeder to overcome the unilateral incompatibility crossing barrier. For example, exceptional plants ("acceptors") that accept S. verrucosum pollen and produce fertile hybrids have been reported (Eijlander et al, 2000). Apparently some "acceptor" plants will accept pollen of any other plant of S. verrucosum, while others only accept pollen from certain S. verrucosum plants (Hermsen, 1978).

Male Sterility

Male sterility is common in potato cultivars (Krantz, 1924; Howard, 1970). Because the marketable product in potato is not botanical seed, there is no selection pressure for fertility in cultivars. In fact, fruit development may partition resources away from tuber yield, so breeders may inadvertently select against high fertility (Jansky & Thompson, 1990). In addition, recessive sterility alleles can accumulate in tetraploid potato cultivars because they are easily masked by the additional chromosomal sets and are rarely found in a homozygous condition (Krantz, 1924; Lindhout et al., 2011).

Interactions between cytoplasmic and nuclear genes commonly lead to male sterility in potato interspecific hybrids. Cytoplasmic-genetic male sterility has been reported in a number of cultivated x wild potato species hybrids (Grun & Aubertin, 1966; Sanford and Hanneman 1979; Hanneman & Peloquin, 1981; Masuelli & Camadro, 1997; Camadro et al., 1998; Carputo et al., 2000b, 2003b; Jansky & Peloquin, 2006; Caruso et al., 2008; Masuelli et al., 2009; Jansky, 2010; Weber et al., 2012; Larrosa et al., 2012). For example, crosses between Chilotanum group (see "Cultivated Potato Taxonomy and Phylogeny" below for cultivated species group nomenclature) haploids and Andigenum group clones produce male fertile hybrids when the haploids are the male parent, but male sterile hybrids when the haploids are the female parent (Grun et al., 1962; Ross et al., 1964; Carroll, 1975). Nuclear genes that restore fertility to interspecific hybrids have been reported (Iwanaga et al., 1991; Tucci et al., 1996). Cytoplasmic-genetic male sterility provides an isolating mechanism to help maintain the integrity of sympatric species (Hosaka & Sanetomo, 2012, 2013).

2n Gametes

Numerically unreduced (2n) gametes are believed to have been responsible for polyploidization in potato (Den Nijs & Peloquin, 1977a; Camadro & Peloquin, 1980; Camadro et al., 2004). 2n gametes are typically produced by recessive alleles of genes that control meiosis. When homozygous, these mutations interrupt meiosis so that gametes contain the parental (sporophytic) chromosome number rather than half that number. Meiotic mutations occur naturally and frequently in cultivated and wild potatoes (Peloquin et al., 1999; Carputo et al., 2000a). Some meiotic mutations result in the production of 2n eggs (Stelly & Peloquin, 1986b; Wemer & Peloquin, 1991a), while others produce 2n pollen (Quinn et al., 1974; Mok & Peloquin, 1975; Den Nijs & Peloquin, 1977a; Masuelli et al., 1992; Iwanaga & Peloquin, 1982; Watanabe & Peloquin, 1991; Masuelli et al., 1992; Hanneman, 1999; Camadro et al., 2008). Meiotic mutations typically exhibit variable expressivity, so homozygous recessive plants produce both 2n and n gametes (Mok & Peloquin, 1975; Ortiz & Peloquin, 1992; Carputo et al., 2000a, 2003a; Carputo, 2003). However, as discussed below, a cross between a tetraploid and a 2n gamete-producing diploid will produce only tetraploid offspring. The union of an n (2x) gamete from a tetraploid and an n (x) gamete from a diploid will produce a seed with a triploid embryo but inviable endosperm. 2n pollen is easily detected microscopically because diploid (2x) pollen grains are larger than monoploid (lx) pollen grains (Quinn et al., 1974). 2n eggs can also be detected microscopically via a stain clearing technique (Stelly et al., 1984), but this is a laborious procedure and not practical for large-scale screening. Diploid clones that produce 2n eggs can be identified by simply crossing them as females to tetraploids (Erazzu & Camadro, 2006). If seeds are produced, then the diploid parent produces 2n eggs.

The mechanism of 2n gamete production determines the genetic composition of the gametes. Normally, in anthers, the four products of meiosis are separated so that their poles define a tetrahedron and cytokinesis produces four haploid microspores. In contrast, a "parallel spindles" mutation produces two microspores, each with an unreduced (sporophytic) chromosome number. The first division is normal, but in the second division, the spindles are parallel and cytokinesis produces two diploid microspores. Even though the first meiotic division occurs in this mutant, the genetic result of parallel spindles is equivalent to a first division restitution (FDR) genetic mechanism because gametes contain non-sister chromatids from the centromere to the first crossover. Consequently, all loci in this region have the same genetic constitution in the gamete as that of the parent (Park et al., 2007; Peloquin et al., 2008). In the chromosomal region beyond the first crossover, half of the loci that were heterozygous in the parent will remain so in 2n gametes. There is typically only one crossover per chromosome in potato (Yeh et al. 1964; Park et al., 2007). Consequently, FDR 2n gametes transmit approximately 80 % of the diploid parent heterozygosity to tetraploid offspring. They provide a unique and powerful method of transmitting blocks of advantageous dominance (intralocus) and epistatic (interlocus) interactions to polyploid offspring even following meiosis, which usually breaks up such interactions.

While the genetic consequence of 2n pollen formation in potatoes is typically FDR, 2n eggs are formed by a second division restitution (SDR) mechanism (Stelly & Peloquin, 1986a; Wemer & Peloquin, 1990). The SDR gametes contain sister chromatids from the centromere to the first crossover. SDR 2n gametes transmit less than 40 % of heterozygosity to offspring (Peloquin, 1983; Peloquin et al, 2008).

Chromosomal regions near the centromere carry major genes that contribute to yield in potatoes (Buso et al., 1999b). These regions are transmitted intact to the tetraploid level via FDR 2n gametes such as the products of the parallel spindles mutant. An ortholog of the parallel spindles gene (AtPSl) has been isolated and characterized in Arabidopsis (d'Erfurth et al., 2008). The AtPSl protein appears to have a regulatory function and is conserved throughout the plant kingdom. The combination of a common mutation that produces FDR 2n gametes, a high rate of transmission of allelic interactions via FDR 2n gametes, a strong heterotic yield response in potatoes, and the positioning of yield enhancing genes near the centromere provide an advantageous set of circumstances to realize high productivity in the evolution of tetraploid potato.

Endosperm Balance Numbers

Endosperm development is critical for viable seed production in potatoes. Intraspecific, intraploidy crosses in potatoes typically produce viable seeds containing well-developed endosperm. Conversely, in most interploidy crosses, inviable seeds are produced due to endosperm failure (Brink & Cooper, 1947). Flowever, endosperm failure is observed in some intraploidy, interspecific crosses. Conversely, sometimes interploidy, interspecific crosses succeed. The endosperm balance number (EBN) hypothesis proposes that a 2 maternal: 1 paternal ratio of genes, rather than genomes, is necessary for normal endosperm development in potatoes (Johnston et al, 1980). The genetic basis of these endosperm balance factors has yet to be elucidated, although genetic models have been proposed (Ehlenfeldt & Hanneman, 1988a; Camadro & Masuelli, 1995). Genes on more than one chromosome appear to control EBN (Johnston & Hanneman, 1996). Species in section Petota have been assigned endosperm balance numbers (EBN) based on their ability to hybridize with each other (Hanneman, 1994). Viable seeds will be produced from crosses between plants with matching EBN values, as long as other hybridization barriers are absent. Ploidy and EBN combinations in potatoes include 6x (4EBN), 4x (4EBN), 4x (2EBN), 2x (2EBN) and 2x (1EBN).

Hawkes (1988, 1990) proposed that wild potato species arose in Mexico and then spread to South America. Most of the diploid Mexican species are 1EBN (with the exception of S. verrucosum, which is 2EBN), while most of the diploid South American species are 2EBN (Table 1). Hence, the South American diploid 2EBN species may have evolved from the 1EBN Mexican species (Hawkes & Jackson, 1992). Hawkes proposed a reasonable hypothesis that S. verrucosum in Mexico today could be explained by the migration of the diploid 2EBN South American species back north.

Breeders use EBN values to predict crossing success. If two species differ in EBN by a factor of two, then doubling the genome of the species with the lower chromosome number will double its EBN value and increase the probability of hybridization success. Doubling can be achieved via somatic genome duplication (Ross et al., 1967; Johnston & Hanneman, 1982; Sonnino et al., 1988; Carputo et al., 1997, 2000c) or by selecting individuals that produce 2n gametes (Johnston & Hanneman, 1980, 1982; Carputo et al., 1997, 2000c; Hayes & Thill, 2002). Endosperm balance number can be reduced through anther culture or parthenogenesis, as discussed below. It is important to note that, while knowledge of EBN and 2n gamete production often allows for successful cross prediction, there are exceptions. Sometimes intra-EBN crosses fail and, at other times, inter-EBN crosses succeed even without the presence of 2n gametes. Endosperm balance number, therefore, is only one component of a complex system of pre-and post-zygotic interspecific crossing barriers (Masuelli & Camadro, 1997; Chen et al., 2004).

Endosperm balance number and 2n gametes have played a pivotal role in the evolution of both auto- and allopolyploidy in Solarium species (Den Nijs & Peloquin, 1977a, b; Camadro & Peloquin, 1980; Iwanaga & Peloquin, 1982; Carputo et al., 2003a). Because 2n gametes are common in wild Solarium species, they likely contributed to the production of spontaneous tetraploids (Marks, 1966; Quinn et al., 1974; Den Nijs & Peloquin, 1977a; Werner & Peloquin, 1991a). Unilateral sexual polyploidization occurs when hybrids between 4x (4EBN) species and 2n gamete-producing 2x (2EBN) species produce only tetraploid offspring. Triploid seeds are inviable due to endosperm failure, as discussed above. Bilateral sexual polyploidization is also possible when 2n gametes from two diploid species unite to produce tetraploid offspring. In contrast to somatic doubling, sexual polyploidization minimizes the level of inbreeding in a new tetraploid (Den Nijs & Peloquin, 1977a). Disomic polyploid wild potato species are likely the product of bilateral sexual polyploidization (Ortiz & Ehlenfeldt, 1992).

In addition to contributing to recurrent polyploidization, EBN may also serve a valuable function as an isolating mechanism (Ortiz & Ehlenfeldt, 1992). Sympatric species with matching ploidy levels may be sexually incompatible due to differences in EBN values. For example, S. chacoense (2x, 2EBN) and S. commersonnii (2x, 1EBN) have overlapping ranges in Argentina. However, species integrity may be maintained by EBN incompatibility (Ortiz & Ehlenfeldt, 1992). The genes governing EBN have not been identified, so this biological factor governing species integrity is speculative.

Another evolutionary advantage of sexual polyploidization is that it allows for recurrent production of new tetraploids with different combinations of species and clones involved. It is interesting to note that a self-incompatible diploid will produce exclusively tetraploid offspring when self-pollinated, if it produces both 2n pollen and 2n eggs. As described above, self-incompatibility breaks down due to competitive interaction of S-alleles when heteroallelic pollen tubes interact with the style (Mok et al., 1976). The high frequency of the parallel spindles allele for 2n pollen production in potato cultivars supports the idea that tetraploid cultivated potatoes arose via sexual polyploidization (Iwanaga & Peloquin, 1982; Carputo et al., 2003a).

Stylar Barriers

Knowledge of EBN values helps to predict post-zygotic hybridization barriers in potatoes. However, other hybridization barriers are common among species of Solatium section Petota (Camadro et al., 2004; Jansky, 2006). Consequently, while matching EBN values are necessary for successful hybridization, they are not always sufficient. An important and common pre-zygotic hybridization barrier is the inhibition of pollen tube elongation by stylar tissue, ft has been reported in many inter- and intra-EBN crosses (Camadro & Peloquin, 1981; Fritz & Hanneman, 1989; Novy & Hanneman, 1991; Erazzu et al., 1999; Peloquin et al., 1999; Raimondi & Camadro, 2003; Jansky & Hamemik, 2009; Masuelli et al., 2009; Weber et al., 2012). Pollen tube growth may be impeded in the top, middle or bottom of the style (Camadro et al,, 2004). A few seeds are sometimes produced in incompatible crosses, indicating that stylar barriers are incomplete in some cases. A gene-for-gene interaction between stylar tissue and pollen has been proposed (Camadro et al., 2004). This hybridization barrier helps to maintain the identity of sympatric species with identical EBN values. Camadro et ah (2004) argue that genetic cross-incompatibility systems, such as that resulting from pollen-style interactions, are necessary for sympatric species to maintain their integrity.


Haploids are sporophytes with the gametophytic chromosome number. In potatoes, haploids derived from tetraploids are commonly called dihaploids to indicate that they contain two sets of chromosomes. Dihaploids provide a mechanism for direct gene transfer from most of the wild diploid potato relatives and allow breeders to work at the diploid level. Dihaploids can also be used to measure the genetic load in the tetraploids from which they are derived, since they reveal deleterious alleles that were hidden in the tetraploids (De Jong et al. 1971).

Dihaploids are easily produced from female fertile tetraploid clones via parthenogenesis (Hougas et al., 1958; von Wangenheim et al., 1960). Selected diploid S. tuberosum Andigenum group (in literature these clones formerly referred to as cultivar group Phureja) 'pollinator' clones produce diploid offspring when crossed to tetraploids. In these crosses, both sperm cells from the pollinator enter the central cell, allowing normal endosperm to develop. This stimulates the division of the egg cell in the absence of fertilization, resulting in the production of a dihaploid (2x) embryo (Von Wangenheim et al, 1960; Montelongo-Escobedo & Rowe, 1969). Sometimes, functional 2n pollen in the pollinator produces tetraploid offspring. It is important to distinguish between seeds that resulted from fertilization of the egg cell by 2n pollen (which would be tetraploid) and those that were not (and are therefore dihaploid). 'Pollinators' have been selected for homozygosity of a dominant gene that produces a small dark spot on seeds. Seeds expressing the marker are hybrids and are therefore discarded; seeds lacking the marker are retained with the expectation that they are dihaploids (Peloquin & Hougas, 1959; Hermsen & Verdenius, 1973).

Populations of dihaploids provide unique opportunities for the genetic analysis of polygenic traits (Hougas et al., 1958; Kotch et al., 1992). A population of dihaploids from a single heterozygous tetraploid clone represents a random pool of female gametes. Genetic analyses can be carried out on this population without the confounding effects of fertilization. In addition, genetic variability hidden in polyploids can be revealed in populations of dihaploids (Peloquin et al., 1991). As a result of segregation, dihaploids may express traits that were not found in their tetraploid parents. Genetic variation among dihaploids for plant and tuber traits is common and has been widely reported (Peloquin & Hougas, 1960; Matsubayashi, 1979; De Maine, 1984a, b; Rousselle-Bourgeois & Rousselle, 1992; Hutton, 1994). Disease resistance traits are also variable among dihaploids, with some dihaploid clones exhibiting better resistance than their parents. Dihaploids with resistance to late blight, Verticillium wilt, soft rot, common scab, blackleg, potato virus X, and potato cyst nematode have been reported (Hutten et al., 1995b; Carputo et al., 1996; Jansky et al., 2003; Ercolano et al., 2004; Bradshaw et al., 2006b). Dihaploid populations have been used to characterize the genetic basis of total tuber yield, average tuber weight, tuber number, dry matter content, tuber dormancy, vine maturity, and tuber glucose levels (Kotch et al., 1992). Because dihaploids form spontaneously from crosses with certain 'pollinator' clones, they may provide an opportunity to reduce ploidy in natural systems.

Tetraploid potatoes are typically more vigorous and higher-yielding than their dihaploid offspring (Peloquin & Hougas, 1960; De Maine, 1984a; Kotch et al., 1992). The lower vigor and yield in dihaploids is likely due to ploidy reduction and inbreeding depression. The magnitude of this loss at the diploid level varies depending on the tetraploid clone from which the dihaploids were derived (Kotch et al., 1992).

Potato monoploids (lx) can be produced from diploids via anther culture (Veilleux et al., 1985) or pollination (Uijtewaal et al., 1987). While the production of monoploids through anther culture is possible, it can be difficult because it requires the presence of genes for androgenic competence ("tissue culturability"), which are not found in all potato cultivars (Sonnino et al., 1989). A "monoploid sieve" selects against deleterious recessive alleles, allowing only the genotypes with high fitness values to develop into monoploid plants. These monoploids can be somatically doubled to produce homozygous diploids for heterosis breeding (Lightboum & Veilleux, 2007). The first published potato genome sequence was based on the homozygous doubled monoploid DM1-3 (The Potato Genome Consortium, 2011). This has provided the structural framework for the sequencing of heterozygous genomes.

Dihaploid-Wild Species Hybrids

Wild relatives of potatoes are commonly used in breeding programs as sources of genes not found in cultivated potatoes (Leue, 1983; Hermundstad & Peloquin, 1986; Yerk & Peloquin, 1989, 1990; Jansky et al., 1990; Watanabe et al., 1995; Serquen & Peloquin, 1996; Tucci et al., 1996; Oltmans & Novy, 2002; Weber & Jansky, 2012). One strategy to access wild Solanum germplasm is through hybridization with dihaploids of tetraploid potato cultivars. Dihaploid-wild species hybrids allow breeders to capture valuable genes from wild species in an adapted form that can be maintained clonally as tubers.

Yield heterosis often is observed in dihaploid-wild species hybrids (Leue, 1983; Hermundstad & Peloquin, 1986; Santini et al., 2000). The high yield and large tuber size in hybrids allows breeders to determine the contributions of wild species to tuber traits such as dry matter content, dormancy, starch composition, nutritional components, and processing quality (Yerk & Peloquin, 1989, 1990; Jansky et al, 1990; Rousselle-Bourgeois & Rousselle, 1992; Serquen & Peloquin, 1996; Santini et al., 2000; Oltmans & Novy, 2002; Ortega et al., 2005). Many dihaploid-wild species hybrids produce edible tubers with acceptable appearance, even though they contain 50 % wild species germplasm. In addition to variation for tuber traits, dihaploid-wild species hybrids exhibit useful variation for disease resistance and stress tolerance (Carputo et al., 1996, 2000c; Tucci et al., 1996; Jansky & Rouse, 2000; Ortega et al., 2005; Hamemik et al, 2009; Weber & Jansky, 2012).

Tetraploid Genetics

Tetraploid potatoes, 2n=4x=48, contain four sets of chromosomes (4x) in the sporophyte (2n) generation with 48 chromosomes in each somatic cell. Gametes (n) from a cultivar are 2x=24. The cultivated potato is considered to be an autopolyploid and, as such, exhibits tetrasomic inheritance (Howard, 1970; Ross, 1986; Hawkes, 1990). In autotetraploids, three types of gene segregation are possible, depending on the proximity of the gene of interest to the centromere (Little, 1952; Burnham, 1962). If the gene is close to the centromere, then a crossover between that gene and the centromere is unlikely to occur during meiosis and that gene will experience chromosome segregation. That is, the gene segregates with the chromosome on which it resides. Consequently, a triplex (AAAa) genotype will produce 50 % AA and 50 % Aa gametes; no aa gametes are produced. In the other two types of segregation, called random chromatid segregation and maximum equational segregation, the gene is far enough from the centromere that a crossover is likely to occur during meiosis. Consequently, it is possible for the sister chromatids carrying the recessive allele to be transmitted to the same gamete (aa) through a process called double reduction. Four requirements must be met to achieve double reduction: 1) A quadrivalent must form. That is, all four homologous chromosomes must associate with each other through crossing-over at meiosis; 2) Crossing-over must occur between the gene of interest and the centromere; 3) The two pairs of chromosomes that were involved in the crossover must end up at the same pole after the first meiotic division; and 4) Chromatids must separate randomly during the second meiotic division. If these criteria are always met, then maximal equational separation occurs and the frequency of double reduction is 1/6. A less extreme type of segregation occurs when chromatids segregate randomly, resulting in 4/28 or 1/7 gametes carrying sister chromatids. Consequently, random chromatid segregation results in a frequency of double reduction of 1/7. It does not require a crossover between the gene and the centromere in every meiotic cell. Since both types of chromatid segregation produce similar results, very large segregating populations are needed to distinguish between them. With either type of chromatid segregation, all combinations of chromatids must be considered when determining gametic ratios. The gametes produced by a triplex (AAAa) individual would be all pairwise combinations of AAAAAAaa, which would be 15 AA, 12 Aa, and 1 aa, or 27 with the dominant phenotype and one with the recessive phenotype.

Tetrasomic genetic analyses differ from those of diploids in two critical ways. First, the segregation ratio of any gene depends on its location on the chromosome. This is in contrast to diploid segregation ratios, which do not depend on a gene's chromosomal position. Fixed ratios can be predicted from chromosome and random chromatid segregation models, but they represent extremes. In reality, these extremes are rarely attained and ratios fall between them. Exact ratios cannot be predicted because they are determined by crossover events, which differ in every meiotic cell. Second, large samples of segregating populations must be evaluated in order to characterize genetic ratios and to identify clones carrying genes for traits of commercial interest. For example, it is necessary to evaluate at least 1700 plants to distinguish between chromosome segregation and random chromatid segregation when self-pollinating a duplex (AAaa) clone (Little, 1952). In addition, epistatic (interlocus) interactions are magnified in tetrasomic tetraploids, gene dosage effects are often important, and interactions with the environment can be complex. All of these complications result in a loss of resolution at the tetraploid level, so that qualitative traits are difficult to identify. For example, early studies of potato eye depth using tetraploid potato breeding lines were unable to resolve the genetic basis of this trait. However, when a genetic study was carried out at the diploid level, a major gene for eye depth (Eyd) was discovered (Li et al., 2005).

In another study carried out at the tetraploid level, high heritability estimates were found for potato leaf roll virus resistance, indicating that a few major genes are likely to be mainly responsible for resistance. However, it was not possible to identify individual genes and their effects (Brown et al., 1997). In contrast, when inheritance studies were carried out at the diploid level using the wild species S. chacoense, a single dominant resistance gene was identified and parental genotypes were determined based on Mendelian diploid segregation ratios (Brown & Thomas, 1994).

Even highly selected tetraploid potato clones contain undesirable alleles along with desirable ones. The proportion of deleterious alleles in a plant is called the genetic load. Most deleterious alleles are recessive, so they are only expressed when homozygous. Consequently, the genetic load is high in tetraploids where homozygous recessive genotypes are less common than in diploids. These deleterious alleles are hidden by dominant alleles in tetraploid clones and do not typically have a negative effect. However, when tetraploid clones are self-pollinated or crossed to related clones, some of their offspring will be homozygous for deleterious recessive alleles and will exhibit reduced vigor and/or fertility (Krantz, 1929; Phumichai & Hosaka, 2006). These clones are discarded as seedlings in breeding programs. Therefore, one method to measure the genetic load in parents used in breeding programs is to self-pollinate them and measure the proportion of non-vigorous offspring. It may be beneficial to select parents, in part, based on low genetic load.

Gene expression was studied in a lx, 2x, 4x polyploid series created by somatic doubling (Stupar et al., 2007). It is interesting that a linear correlation between gene expression and ploidy was rarely found. That is, the diploid and tetraploid clones exhibited similar gene expression patterns. The diploid plants created by Stuper et al. (2007) were actually more vigorous than the tetraploid ones produced by somatic doubling, and thus not able to exploit the heterozygosity necessary for enhanced fitness in polyploids. The cost to maintain more DNA and larger cells was apparently not compensated by higher vigor. This provides evidence that polyploidy per se is not evolutionarily advantageous in potatoes. Instead, polyploidy must be accompanied by an increase in allelic diversity.

The Genetic Basis of Species Boundaries in Potatoes

Hundreds of successful artificial interspecific hybrids have been reported in the literature (for example, Bukasov, 1933; Bukasov & Kameraz, 1959; Hawkes, 1958; Hawkes & Hjerting, 1969, 1989; Kamaraz, 1971; Ochoa, 1990a, 1999). It is likely, then, that natural hybridization between species is common in the wild as well (Bedonni & Camadro, 2009; Masuelli et al., 2009; Camadro et al., 2012). Even the most universal barrier to interspecific hybridization in potatoes, endosperm balance number (EBN), is not expected to provide an impenetrable barrier between sympatric species. For example, the diploid species S. chacoense and S. commersonii are sympatric, but since the former is 2EBN and the latter is 1EBN, they would not be expected to hybridize. These hybrids have been generated in the lab (Ehlenfeldt & Hanneman, 1988b). Since 2n gametes are common in wild potatoes, they would allow inter-EBN crosses to occur spontaneously. The resulting progeny are typically odd-ploidy. This might be considered a reproductive dead-end, but 2n gametes, asexual reproduction, and perenniality allow gene flow even through triploid and pentaploid plants.

In addition to EBN, unilateral incompatibility may present a barrier to interspecific hybridization in wild potato populations. Sometimes, unilateral incompatibility results when pollen tube growth is inhibited in the style, but the reciprocal cross is successful. This is especially apparent with crosses between self-compatible and self-incompatible species, where hybridization is successful if the self-compatible species is the female, but not when it is the male. Reciprocal cross differences in hybridization success may also be due to cytoplasmic-genetic male sterility. Since wild species populations are typically both male and female fertile, stylar barriers and male sterility likely inhibit, but do not prevent, interspecific hybridization.

The continuous flow of genes across and within ploidy levels in sexually reproducing perennial populations results in a complex aggregation of related genotypes. According to Camadro et al. (2012) "Hybridization and subsequent gene flow and introgression in sympatric populations, within and between ploidy levels, often results in exceedingly complicated patterns of variation." Consequently, the biological concept of a species is difficult to apply to potatoes, as in all plants (Knapp, 2008). Breeders have proposed the concept of crossability groups to aid in utilization of wild and cultivated germplasm (Harlan & De Wet, 1971). Based on EBN and self-compatible/ self-incompatible systems, Fig. 1 proposes five crossability groups in potatoes. First of all, EBN divides the collection of species into three groups. Those groups may be crossed, though, if 2n gametes are present. Consequently, double-headed arrows connect the EBN groups. Within the 1EBN and 2EBN groups, the self-compatible species may be separated from the self-incompatible ones. All 4EBN species are self-compatible. At the 1EBN and 2EBN levels, self-compatible (female) by self-incompatible (male) crosses are typically successful, while reciprocal crosses fail. Hence, single-headed arrows connect these groups. Hybridization within each of the five groups is expected to be successful, although failures have occasionally been reported (see Hawkes, 1958). As discussed above, while hybridization across groups is less likely to be successful than that within groups, no barrier is complete.

Wild Potato Taxonomy and Phylogeny

History of Taxonomic Treatments of Solatium Section Petota

As detailed by Spooner & van den Berg (1992a), section Petota has been the subject of intensive taxonomic work since the description of the cultivated potato, S. tuberosum (Linnaeus, 1753). Different taxonomists applied various taxonomic philosophies and species concepts to the section, but mainly have used morphology to define species. Walpers (1844) accepted only ten species in section Petota. The last attempt to monograph Solarium in its entirety was by Dunal (1852) who included 17 species in section Petota, while Baker (1884) recognized only six species in the section. Bitter (1912-1913), in his monumental work on Solatium, described more than 50 new species, subspecies or varieties of wild potatoes.

The first regional treatment of section Petota was provided by Rydberg (1924), who monographed the Mexican and Central American species and described ten new taxa. Extensive taxonomic investigations were conducted by Nikolai Vavilov's Russian associates Sergei Bukasov and Sergei Juzepczuk, who worked on material gathered on Russian expeditions to Mexico, Central America, and South America in the 1920s and 1930s. They effectively and validly described 30 wild and 18 cultivated species, in addition to publishing a great number of names that were not validly published (Bukasov, 1930, 1933, 1937). This first potato germplasm collection was widely used to study potato cytogenetics by Vladimir Rybin (Rybin, 1929, 1933) and interspecific hybridization by Abram Kameraz (Bukasov and Kameraz, 1959) whose results were used for developing potato taxonomy and phylogeny. Hawkes (1944) treated collections from a series of British expeditions to Mexico and South America in the 1930s and described 52 new species, subspecies, or varieties, of which he accepted only ten in 1990 (Hawkes, 1990). Regional treatments have been provided for North and Central America (Correll, 1952; Spooner et al., 2004); Mexico (Flores Crespo, 1966; Hawkes, 1966; Rodriguez & Vargas, 1994); Peru (Vargas, 1949, 1956; Ochoa, 1962, 1999; Correll, 1967); Bolivia (Hawkes & Hjerting, 1989; Ochoa, 1990a); Argentina, Brazil, Paraguay, and Uruguay (Hawkes & Hjerting, 1969); Bolivia, Argentina, Chile, Paraguay, and Uruguay (Spooner et al., in press), and Chile (Montaldo & Sanz, 1962; Contreras, 1987).

The first modrn comprehensive (from throughout the entire range of the group) taxonomic treatment of section Petota was provided by Hawkes (1956b) who synonymized many species. The treatment of Correll (1962) was similar in its taxonomy, and included extensive specimen citations and excellent illustrations. Other comprehensive treatments have been provided by Hawkes (1956b, 1963, 1990), Bukasov (1978), and Gorbatenko (2006). Since the work by Correll (1962), 176 new taxa have been described, 140 of these by Carlos Ochoa, including 77 new varietal and form names for the Bolivian cultivated species alone (Ochoa, 1988). In total, there are 494 epithets for wild and 626 epithets for cultivated taxa, including names not validly published (Ovchinnikova et al., 2011).

Series Treatments in Solanum Section Petota

As detailed by Hawkes (1989), Bitter (1912) was the first to describe series in section Petota (series Conicibaccata and series Maglia), although he failed to designate series affiliations for many species. Rydberg (1924) divided the Mexican and Central American potatoes into five informal groups (Bulbocastana, Juglandifolia, Oxycarpa, Pinnatisecta, Tuberosa) but failed to designate rank. Hawkes (1944) validated these names as series (except Oxycarpa, which he equated to series Conicibaccata) and described series Cuneoalata. Since that time, 26 additional series have been validly published: series Megistacroloba (Cardenas & Hawkes, 1946); Trifida (Correll, 1950); Cardiophylla, Polyadenia (Correll, 1952); Circaeifolia, Piurana (Hawkes, 1954b); Morelliformia (Hawkes, 1956b); Acaulia, Andigena, Commersoniana, Demissa, Etuberosa, Longipedicellata, Vaviloviana (Bukasov & Kameraz, 1959); Ingifolia (Ochoa, 1962); Clara, Minutifoliola, Tarijensa, Yungasensa (Correll, 1962); Olmosiana (Ochoa, 1965); Lignicaulia (Hawkes, 1989); Bukasoviana, Chomatophylla, Pyriformia, Simpliciora (Gorbatenko, 1989), Simplicissima (Ochoa, 1989b). Gorbatenko (1989) published series Lignicaulia as a later homonym. The date of publication on Gorbatenko (1989) is ambiguous, because the latest date listed on the volume is August 9, after the words (transliterated) "Podpisano v petsat"=signed off for printing. This was not the publication date, however, which was 21 Sep 1989 (letter from Ludmilla Gorbatenko to Jack Hawkes). Hawkes (1989) was released on 29 Aug 1989, giving his Lignicaulia priority at the sectional rank. The following names have been treated as series but were not validly published: Looseriana (Bukasov, 1939); Andreana (Hawkes, 1944); Glabrescentia, Transaequatorialia (Bukasov & Kameraz, 1959); Borealia (Correll, 1962); Alticola, Berthaultiana, Chilotana, Cisaequatorialia, Collina, Subacaulia, and Verrucosa (Bukasov, 1978). Hawkes (1990) and Gorbatenko (1989) recognized 15 and 20 series, respectively, for the South American species, and Hawkes (1990) and Bukasov (1978) recognize 21 and 36 series, respectively, for section Petota. Spooner and van den Berg (1992a) provided a graphic chronological comparison of these varying concepts of series. These series often are not well-defined morphologically, and the affiliations of species to series vary widely among different authors.

Morphological Studies of Species Boundaries Subsequent to Hawkes (1990)

Recent reinvestigations of species boundaries, origins, and phylogeny in sect. Petota have employed extensive field work throughout the range of the group (summarized in Spooner & Salas, 2006) and numerical taxonomic investigations of morphological data gathered from field studies, herbarium specimens, or germplasm grown in field plots (Table 2) (Clausen & Crisci, 1989; Child & Lester, 1991; Lester, 1991; Spooner & van den Berg, 1992b, 2001; van den Berg & Spooner, 1992; Spooner et al., 1993b, 1995b, 2001a, b, 2008a; van den Berg & Groendijk-Wilders, 1993, 1999; Giannattasio & Spooner, 1994a; Miller & Spooner, 1996; van den Berg et al., 1996, 1998; Castillo & Spooner, 1997; Clausen & Spooner, 1998; Kardolus & Bezem, 1998; Kardolus & Groendijk-Wilders, 1998; Kardolus, 1999; Rodriguez & Spooner, 2002; Lara-Cabrera & Spooner, 2005; Alvarez et al., 2008; Ames et al., 2008; Fajardo et al, 2008; Bedonni & Camadro, 2009). These morphological studies documented wide character state variation within species and overlap of character states among closely related species. They were often conducted in parallel with molecular studies of the same accessions, as described below.

Molecular Studies of Species Boundaries Subsequent to Hawkes (1990)

Studies of species boundaries, origins, and phylogeny frequently have involved a wide range of molecular marker and DNA sequence data, sometimes in combination with morphological data (Table 2). These have included isozymes, protein electrophoresis, single- to low-copy nuclear DNA restriction sites, nuclear microsatellites, plastid microsatellites, plastid deletion markers, highly repeated nuclear DNA, mitochondrial DNA RFLPs, DNA sequences from the internal nontranscribed spacer of nuclear ribosomal DNA, and DNA sequences from orthologous nuclear genes, with the polyploid studies (summarized in Polyploidy--DNA Sequence data, below). Similar to the morphological studies (above) these studies frequently documented wide character state variation within species and overlap of character states among closely related species.

Introgression and Interspecific Hybridization

Natural interspecific hybridization has been hypothesized to be a major evolutionary mechanism in section Petota (Ugent, 1970a; Hawkes, 1990). Spooner & van den Berg (1992b) summarized literature proposing 26 potato species (then accepted) to have resulted from hybrid speciation. Five of these were cultivated species and 21 wild species (12 diploid and nine polyploid). Most of these hypotheses have been generated by intermediate morphology, inference from distributional data, artificial reconstruction of the hybrids and comparison with putative natural hybrids, and assessment of reduction of fertility. We here discuss reinvestigations of putative diploid wild species; hypotheses of the cultivated and wild polyploid species are discussed below.

Spooner et al. (1991b) reexamined, with plastid DNA and nuclear ribosomal DNA restriction site data, the hypothesis of Ugent (1970b) that the Peruvian diploid species S. raphanifolium was of recent and ongoing hybrid origin between diploid S. canasense (=S. candolleanum) and S. megistacrolobum (=S. boliviense). Solanum raphanifolium is morphologically intermediate between the putative parents and occurs where the two species overlap in distribution. Solanum raphanifolium, however, was divergent from either putative parent regarding both markers. The putative parents were similar, and no support was provided for the hybrid origin.

Miller and Spooner (1996) reexamined the putative origin of mountain populations of S. chacoense (diploid), hypothesized by Hawkes (1962a) to have arisen from introgression with S. microdontum and lowland populations of S. chacoense. Its hybrid origin was not supported, however, with data from morphology, RAPDs, or nuclear RFLPs.

Clausen and Spooner (1998) reexamined the putative hybrid origin of S. yrechei, hypothesized by Hawkes & Hjerting (1969) and Okada & Hawkes (1978) to be of hybrid origin between S. kurtzianum and S. microdontum. Like S. raphanifolium, S. xrechei occurred at the overlap zone of its two parents. In addition, it had reduced fertility in comparison to natural and artificially constructed hybrids. In contrast to the two studies mentioned above, additive profiles of nRFLPs gave strong support to its hybrid origin.

Additionally, introgression and interspecific hybridization not leading to speciation has been believed to be common in section Petota (Hawkes, 1962a). For example, Hawkes & Hjerting (1969) interpreted 9.5 % of the wild potato specimens they examined for the flora of Argentina, Brazil, Paraguay, and Uruguay to be interspecific hybrids, and Hawkes and Hjerting (1989) and Ochoa (1999) provided extensive lists of natural and artificial interspecific hybrids. Spooner & van den Berg (1992b) and Spooner et al. (2007a) investigated, with morphological and molecular marker data, respectively, hypotheses by Hawkes & Hjerting (1989) that 9.5 % of the natural populations of S. berthaultii and S. tarijense were interspecific hybrids. While both studies showed extremes that could be recognized as variants identified as these two species, there was a near continuum of variation that Spooner et al. (2007a) interpreted as variation in the highly variable species S. berthaultii. This variation included three diploid hybrid species accepted by Hawkes (1990), S. xlitusinum, S. xtrigalense, and S. xzudaniense.

A putative natural hybrid between S. chacoense and S. kurtzianum was described by Brucher (1962) as S. ruiz-lealii, and accepted by Hawkes & Hjerting (1969) and Hawkes (1990). Raimondi et al. (2005) examined the hypothesis of hybridization by phenetic analyses of morphological and molecular data and cytological analyses of interspecific hybrids. They concluded that S. ruiz-lealii is not a recent natural hybrid of S. kurtzianum x S. chacoense but originated by divergence of S. chacoense or by hybridization between S. chacoense and another unnamed taxon. They proposed maintaining the species status of S. ruiz-lealii.

Rabinowitz et al. (1990) tested hypotheses of gene flow between the diploid wild species S. sparsipilum (=S. candolleanum) and the cultivated diploid .S', stenotomum (=S. tuberosum Andigenum group). By use of isozyme markers specific to these populations, they were able to document high levels of gene flow in experimental field plots in the Andes. They used these data to speculate that extensive gene flow occurs among other cultivated and wild species. Similarly, Debener et al. (1991) used phenetic analyses of nuclear RFLPs to support incorporation of wild species germplasm into cultivated species. In addition, Celis et al. (2004) documented, with AFLP markers, the possibility of gene flow from cultivated species to diverse wild species occurring in the Andes. These results would need to be tested in natural situations to see if such hybrid populations would survive in the wild, but if so, they provide support to hypotheses of Ugent (1970a) who proposed that the cultivated species were formed and genetically enriched subsequent to formation by gene flow from the wild species.

Taxonomic Changes Subsequent to Hawkes (1990)

The combined molecular and morphological studies mentioned above and observations of species during collecting expeditions have often failed to support many of the traditionally recognized species of wild potatoes, and form the rationale for our reduction of species (Table 1). This has occurred in almost every group studied. An account of post-1990 taxonomic decisions in section Petota by many workers published in Spooner & Salas (2006) reduced the 235 species of Hawkes (1990) to 190, but our independent taxonomic decisions presented here result in a greatly reduced number of 107 wild and four cultivated species (Table 1).

This is perhaps best illustrated by studies of species boundaries in the wild potato S. brevicaule complex. This complex contains about 20 taxa and has long attracted the attention of biologists because of its similarity to cultivated potatoes (Correll, 1962; Ugent, 1970a; Gran, 1990). Some members of this complex, endemic to central Peru, Bolivia, and northern Argentina, were considered ancestors of the landraces (Ugent, 1970a). The species in the complex share pinnately dissected leaves, round fruits, rotate to rotate-pentagonal corollas, and are largely sexually compatible with each other and with the cultivated potato (Hawkes, 1958; Hawkes & Hjerting, 1969, 1989; Ochoa, 1990a, 1999; van den Berg & Spooner, 1992a). They include diploids, tetraploids, and hexaploids, with traditionally recognized species possessing multiple ploidy levels (S. gourlayi [=S. brevicaule] with diploids and tetraploids; and S. oplocense [=.S. brevicaule] with diploids, tetraploids, and hexaploids). Members of the complex have been so difficult to distinguish from each other that even experienced potato taxonomists Hawkes & Hjerting (1989) and Ochoa (1990a) provided different identifications for identical collection numbers of the Solarium brevicaule complex in fully 38 % of the cases (Spooner & van den Berg, 1992a). Field collections in Peru (Spooner et al., 1999; Salas et al, 2001), Bolivia (Spooner et al., 1994), and Argentina (Spooner & Clausen, 1993); phenetic analyses of morphological data in the Netherlands (van den Berg et al., 1996) the United States (van den Berg et al., 1998) and Peru (Alvarez et al., 2008); single- to low-copy nuclear restriction fragment length polymorphism (nRFLPs) and random amplified fragment length (RAPD) data (Miller & Spooner, 1999); and amplified fragment length polymorphism (AFLP) data (Spooner et al., 2005a) failed to clearly differentiate many wild species in the complex, but defined two geographic subsets: (1) the Peruvian populations, (2) the Bolivian and Argentinean populations. However, even these two groups could only be distinguished by computer-assisted statistical analyses of widely overlapping character states, and not by species-specific characters. We here recognize two morphologically very similar species, S. candolleanum from Peru (and extreme northern Bolivia), and S. brevicaule from Bolivia and Argentina.

There are many other similar examples of difficult species complexes in wild potatoes. For example, an examination by Spooner et al, (in press) of independent identifications of S. megistacrolobum and S. toralapanum by Hawkes & Hjerting (1989) and Ochoa (1990a) showed that they gave different identifications to identical collection numbers of these species 17 % of the time. Further identifications by Spooner et al. (in press) of all specimens from southern South America found that species variation in these species extended even to the long-accepted names S. astleyi, S. boliviense, and S. sanctae-rosae, necessitating their synonymy under the earliest name S. boliviense as shown in Table 1. Combined morphological and molecular studies show similar patterns failing to support traditionally recognized species in formerly recognized Solanum series Conicibaccata (Castillo & Spooner, 1997; Spooner et al., 2001b; Fajardo et al, 2008; Jimenez et al., 2008; Fajardo & Spooner, 2011); series Demissa (Spooner et al, 1995b); series Longipedicellata (Spooner et al, 2001a; van den Berg et al., 2002); and series Piurana (Spooner et al., 1995b; Ames et al., 2008; Ames & Spooner, 2010), as well as in the cultivated species (below).

Ingroup and Outgroup Relationships

Phylogenetic studies in section Petota, including plastid DNA restriction site data (Spooner et al. 1991a, 1993a; Spooner & Sytsma, 1992; Castillo & Spooner, 1997; Rodriguez & Spooner, 1997; Spooner & Castillo, 1997) and nuclear DNA sequencing data (Spooner et al., 2008b; Rodriguez & Spooner, 2009; Rodriguez et al., 2009; Ames & Spooner, 2010; Fajardo & Spooner, 2011; Cai et al., 2012) have greatly changed our understanding of ingroup and outgroup relationships. Solanum section Petota now excludes two non-tuber-bearing series Hawkes (1990) placed in section Petota, now reclassified in near outgroup section Etuberosum (Bukasov & Kameraz) A. Child, section Juglandifolia (Rydberg) A. Child, and section Lycopersicoides A. Child (Peralta) (Contreras & Spooner, 1999; Peralta et al., 2008).

The remaining 19 series of Hawkes (1990) are all tuber-bearing, but we do not recognize series as many of them are not supported by recent studies. Rather, they are divided into four clades (1-4) based on plastid restriction site data or three clades based on nuclear DNA sequencing data (Fig. 2), with both results similar except that the nuclear DNA sequencing data combines species in plastid clades 1+2 (Fig. 2). In addition, many allopolyploid species combine alleles from different clades (Table 1) as outlined in the sections "Polyploidy--DNA Sequence Data," and "Genome Differentiation in section Petota Identified by Genomic in situ Hybridization" (below).

Figures 3, 4, and 5 illustrate species in these three nuclear clades: S. bulbocastanum (clade 1+2; Fig. 3), S. chiquidenum (clade 3; Fig. 4), and S. verrucosum (clade 4; Fig. 5). The diploid species within these clades possess trends in morphological character states, but there are many exceptions. For example, most species in clade 1 +2 (Fig. 3) possess non-shiny leaves, white stellate corollas, and single tubers at the end of stolons; species in clade 3 (Fig. 4) possess shiny leaves, blue to purple (occasionally white) pentagonal corollas and moniliform (arranged like beads on a string) tubers (Fig. 4), and species in clade 4 have non-shiny leaves, variously colored pentagonal to rotate corollas, and single tubers at the end of stolons. The polyploids are mostly allopolyploids as discussed below in "Polyploidy--DNA Sequence Data" and it is more difficult to assign morphological character states to them. For example, some, but not all of the allopolyploids in species formerly assigned to series Conicibaccata (possessing alleles from both clades 3 and 4) possess moniliform tubers (Fajardo et al, 2008), likely inherited from their diploid parents in clade 3 (Table 1).

As a result of these many variants within species and clades, our assignment to groups within section Petota relies mainly on molecular data that divides the diploid species into three nuclear clades, with allopolyploid derivatives that combine genomes of these three clades. Within these three main clades some species are clearly interrelated. Spooner et al. (2004) summarized these relationships as 11 informal "species groups" for the North and Central American species, and Spooner et al., (in press) as six informal species groups for the southern South American species. Three groups (Morelliforme, Conicibaccata, and Acaulia groups) have representatives shared in both North and Central America and in South America.

Polyploidy--Occurrence, Taxonomy, Biogeography, Habitats

All species of the section Petota have the same basic chromosome number x=12. The first indications of the existence of different ploidy levels in the wild potatoes were provided by Salaman (1926), Smith (1927) and Vilmorin and Simonet (1927) for S. chacoense, S. jamesii, S. fendleri {=S. stoloniferum), S. xedinense and S. demissum. Rybin (1929, 1933) first described the polyploid series in wild potatoes (2x, 3x, 4x, 5x, 6x) and a polyploid series in cultivated species (2x, 3x, 4x, 5x). Using the classical taxonomic system of Hawkes (1990), four taxonomic series (Hawkes, 1990) of wild potatoes were wholly or predominantly polyploid: series Acaulia (4x, 6x), Conicibaccata (2x, 4x, 6x), Demissa (6x), and Longipedicellata (4x). Other series of Hawkes (1990) were predominately diploid: Bulbocastana (2x, 3x), Commersoniana (2x, 3x), Maglia (2x, 3x), Pinnatisecta (2x, 3x), Piurana (2x, 4x), and Tuberosa (2x, 3x, 4x, 6x) (Hijmans et al., 2007).

Of the 107 wild potato species we here accept with known chromosome numbers (Table 1), 19 (18%) have multiple cytotypes. Sixty-four (60%) are exclusively diploid, 14 (13%) have diploid and triploid cytotypes, and three (3%) have diploid and polyploid cytotypes exclusive of triploids. Eighteen (17%) are exclusively polyploid at the tetraploid or hexaploid levels; one of these has tetraploid and hexaploid cytotypes, three (3%) are exclusively triploid, one (1%) is exclusively pentaploid, and one (1%) has triploid and tetraploid cytotypes.

Hijmans et al. (2007) analyzed ploidy data with geographic information system (GIS) tools to elucidate the possible relationship of polyploidy to geographical and environmental range expansion in wild potatoes. Through an analysis of 5447 reports of chromosome counts of wild species, they found that the diploids occupy a larger area than the polyploids, but diploid and tetraploid species have similar range sizes, and the two species with by far the largest range sizes are tetraploids. The fraction of the plants that are polyploids is much higher from Mexico to Ecuador than farther south in the center of the sectional range. Compared with diploids, triploids tend to occur in warmer and drier areas, whereas higher-level polyploids tend to occur in relatively cold areas. Diploids are absent from Costa Rica to southern Colombia, the wettest part of the group's range. They concluded that polyploidy played an important role in this group's environmental differentiation and range expansion.

Genome Differentiation in Section Petota Identified by Genomic in situ Hybridization

As noted by Matsubayashi (1991) and Gavrilenko (2007,2011), there is little karyotype variation among potato species. Traditionally, identification of the type of polyploidy (auto- or allopolyploid) is based on the analysis of meiosis of species and interspecific hybrids (Table 3). Multiple cytotypes of predominantly diploid potato species represent autopolyploids, or presumed autopolyploids (Gavrilenko, 2007). Segmental allopolyploidy has been proposed by Matsubayashi (1991) for tetraploid species of Hawkes's (1990) series Acaulia and for wild and cultivated polyploids of series Tuberosa (Hawkes, 1990), including S. tuberosum. Polyploid species of Hawkes's (1990) series Conicibaccata, Demissa, Longipedicellata, and Piurana have been considered as strict allopolyploids based on their regular bivalent pairing (Marks, 1955, 1965; Irikura, 1976; Matsubayashi, 1991). There are frequent contradictions in the hypotheses of the origin and genome composition of allopolyploids (Table 3). Matsubayashi (1991) proposed a five-genome concept that suggested that all diploid potato species comprised one major genomic group 'A' with minor variants designated by superscripts, corresponding to each taxonomic series of Hawkes (1990). Matsubayashi (1991) proposed that allopolyploid species share one common component genome 'A' (or its very similar genomic variants) and differed from each other by their second genome B, C, D or P (Table 3). The diploid North and Central American species S. verrucosum was suggested as the contributor of the 'A' component genome to Mexican allopolyploids based on traditional analysis of chromosome pairing in species and their hybrids (Bains, 1951; Marks, 1955, 1965; Irikura, 1976; Matsubayashi, 1991). However, Matsubayashi (1991) proposed that there are no extant diploid species with the B, C, D and P genomes.

Genomic in situ hybridization (GISH) and fluorescence in situ hybridization (FISH) techniques have been used extensively to investigate polyploid potato species (Pendinen et al., 2008a,b, 2012; Gavrilenko, 2011). These studies support allopolyploid origins of North and Central American tetraploid species S. hjertingii and S. stoloniferum, and Mexican hexaploid species S. hougasii, S. iopetalum and S. schenckii. GISH results support S. verrucosum (or its ancestral species) as an 'A' genome contributor in all North and Central American allopolyploids, confirming the prior hypothesis of classical cytogenetic analysis (Marks, 1965) and DNA sequence data (Spooner et al., 2008b; Rodriguez & Spooner, 2009). GISH supports S. hjertingii and S. stoloniferum to have originated through merging two divergent genomes (A and B) (Pendinen et al., 2008a; Table 3; Fig. 6). Symbol 'B' (rather than [A.sup.pi] [A.sup.pi], as used by Matsubayashi, 1991) has been subsequently adopted to denote the genomes of Mexican diploid species (2n=2x=24, BB) S. cardiophyllum, S. ehrenbergii, and S. jamesii, reflecting their homology to the second component genome B of the allotetraploid Mexican species S. hjertingii and S. stoloniferum (Pendinen et al., 2008a) (Table 3). Genome B may also be homologous to the genome of Mexican diploid species S. bulhocastanum based on similar GISH results (unpublished data).

The genome formula PP (not equivalent to APAP of Matsubayashi, 1991) was proposed by Spooner et al. (2008b) for South American diploid species in clade 3 (Fig. 2), largely consisting of species Hawkes placed in series Piurana (S. andreanum, S. chomatophilum and S. piurae) based on DNA sequence data. GISH data indicated that the 'P' genome diverged from both the A genome of S. verrucosum and the B genome of diploid Mexican species (S. cardiophyllum, S. ehrenbergii, S. jamesii). GISH results suggest that the P genome of diploid South American species of clade 3 and the genomes of S. hjertingii and S. stoloniferum (2n=4x=48, AABB) have homologous segments on only two chromosome pairs of allotetraploids (Pendinen et al., 2008a).

GISH analysis was also used to investigate the genome composition of Mexican hexaploid species of Hawkes's series Demissa, using labeled DNA of diploid species with AA, BB, or PP genomes (Pendinen et al., 2012). The results support S. hougasii, S. iopetalum, and S. schenckii as allopolyploids, suggesting the involvement of A, B, and P genome species as genome contributors (Pendinen et al, 2012). Differences between these allohexaploid species were revealed in the extent of presence of the B and the P genomes (Table 3) that may be the result of genome restructuring subsequent to their formation.

However, the fourth Mexican hexaploid species S. demissum was supported as an autopolyploid A-genome species (Pendinen et al., 2012) (Table 3, Fig. 6). These results suggest that S. demissum may be derived from the related A-genome species or from the same A-genome ancestral species. GISH results support the recent reclassification by Spooner et al. (2004) of the Mexican hexaploid species into the Iopetala group containing S. hougasii, S. iopetalum, and S. schenckii, and the Acaulia group containing S. demissum. A similar autopolyploid nature (A genome polyploids) was revealed by GISH for other members of the Acaulia group South American polyploid species, 5. acaule and S. albicans, of Hawkes's series Acaulia (Pendinen et al., 2012; Table 3).

GISH was unable to differentiate the component genomes in South American tetraploid species of Hawkes's series Conicibaccata, S. colombianum ([A.sup.c][A.sup.c]CC genome according to Matsubayashi, 1991) (Table 3) indicating that the [A.sup.c] and C genomes are closely related (Pendinen et al, 2008b). GISH analysis of S. colombianum also revealed high homology between the genome of S. colombianum and genomes of diploid species of the related species S. violaceimarmoratum and clade 3 species S. andreanum an S. pascoense (Pendinen et al., 2008b).

In summary, GISH supports traditional hypotheses of the allopolyploid origin of Mexican tetraploids and hexaploids (except S. demissum), confirms the genome composition of Mexican tetraploids, contradicts classical hypotheses of genome composition of all Mexican hexaploids as well as the South American hexaploid species S. albicans, supports recent DNA sequence results (see below), and provides new data on parental genome contributors in species of the Iopetala group.

Polyploidy--DNA Sequence Data

Hawkes (1990) proposed that section Petota arose in North and Central America from indigenous but unidentified ancestral species, possessed white stellate corollas, B genomes, and endosperm balance numbers of 1. He speculated that some of the North and Central American 2x (1EBN) species migrated to South America, evolving A genomes, rotate corollas, and EBN numbers of 2 or 4, then followed by a return migration of A genome species back to Mexico and Central America around 3.5 MA, followed by polyploid events leading to species he placed in series Conicibaccata, Demissa, and Longipedicellata with rotate to rotate-pentagonal corollas. Later genome 'B' was identified in Mexican diploid species S. cardiophyllum, S. ehrenbergii, and S. jamesii, (2n=2x=24, genome BB) based on the GISH analysis of Mexican allotetraploids (Pendinen et al. 2008).

DNA sequence data have the potential to infer allopolyploid origins if the parental genomes are divergent and if there has been little change in the homeologs subsequent to hybridization. Spooner et al. (2008b) used DNA sequence data from the GBSS1 (waxy) gene and Rodriguez & Spooner (2009) used DNA sequences from the nitrate reductase gene to study polyploid origins in wild potatoes. Both studies gave similar results. Concordant with prior hypotheses based on classical cytogenetics and GISH data of Pendinen et al. (2008a), S. hjertingii and S. stoloniferum were strongly supported as combining genomes of the B-genome North and Central American

diploids (i.e., any of the diploids in this region exclusive of S. verrucosum), and A-genome species (likely S. verrucosum, the only A-genome species from this region). Also concordant with prior cytogenetic hypotheses, S. albicans and S. demissum (Acaulia group) had all alleles in the A-genome clade containing most South American species. These studies showed new allopolyploid origins, however, of S. hougasii, S. iopetalum, and S. schenckii (Iopetala group) and S. colombianum and S. moscopanum (Conicibaccata group) in that they combined genomes of the Agenome South American species (as expected) but also with genomes of the Pgenome Piurana group (unexpected). Nitrate reductase also showed new alleles in S. schenckii the B-genome North and Central American clade. Fajardo & Spooner (2011) showed these A and P genome allopolyploid origins to be characteristic of a much wider range of species in the Conicibaccata group.

One problem with using orthologous DNA sequences to infer phylogeny of the allopolyploids is the occurrence of PCR recombination and heteroduplex fixation. Rodriguez et al. (2011) optimized an asymmetric single-strand conformation polymorphism technique to isolate allelic variants of highly heterozygous individuals, providing data of greater accuracy, speed, and reduced costs relative to prior procedures using cloning.

Lindqvist-Kreuze et al. (2013) tested the orthology of putative nuclear orthologs by aligning them with a whole genome sequence of potato. They showed that these markers are mostly single- or low-copy by comparison to the potato whole genome sequence (The Potato Genome Sequencing Consortium, 2011) and that there are several breaks in colinearity between the species analyzed. However, they found some nuclear orthologs to be present in multiple copies and these mapped to unexpected locations. Sequence comparisons between species show that some of these markers may be paralogs.

Both the GBSSI and nitrate reductase results were from single genes/regions, but Cai et al. (2012) examined 54 accessions of 11 polyploid species and 34 accessions of 29 diploid species with six nuclear orthologs. The results increased phylogenetic resolution within clades, giving better ideas of diploid progenitors, and showed unexpected complexity of allele sharing within clades. While some polyploid species have little diversity among accessions and concurred with the GBSSI and nitrate reductase results (e.g., S. agrimonifolium, S. colombianum, S. hjertingii, and S. moscopanum), the results gave much better resolution of species-specific progenitors. Seven other polyploid species showed variant patterns of allele distributions suggesting multiple origins and allele loss. Complex three-genome origins were supported for S. hougasii, S. schenckii, and one of the ten examined accessions of S. stoloniferum (the other nine accessions having only two genomes). It was unexpected that six Central American polyploid species (S. demissum, S. hjertingii, S. hougasii, S. iopetalum, S. schenckii, and S. stoloniferum) shared alleles from the South American diploid species S. berthaultii, as well as from the Central American diploid species S. verrucosum. These results, showing genomic complexity of some wild potato polyploids, could be explained by multiple hybrid origins and allele losses, similar to what is found in many allopolyploid groups (Wendel, 2000; Soltis et al., 2009).

Brown et al. (2014) associated cleaved amplified polymorphic site (CAPS) DNA markers and sequence tagged site (STS) DNA markers co-segregating with resistance phenotypes of Columbia root-knot nematode with resistant populations of S. bulbocastanum, artificial hybrids of S. bulbocastanum and S. tuberosum, and plant introductions of S. hougasii and S. stoloniferum. These results support the findings of Cai et al. (2012), showing a B genome (clade 1+2) in some populations of S. hougasii and S. stoloniferum, demonstrating the utility of phylogeny to guide the search for useful allelic variants. Similarly, Sanetomo & Hosaka (2013) documented the exclusive presence of a mitochondrial DNA marker present only in some accessions of North and Central American polyploid members of the Longipedicellata group (tetraploid) and Iopetala group (groups sensu Spooner et al, 2004) and S. demissum (both hexaploid), and in their putative maternal ancestor S. verrucosum (diploid). These results support S. verrucosum as the maternal ancestor of these species, as well as illustrate the genomic complexity these polyploids. They also help to explain the difficulty of delimiting clearly defined species in polyploid potatoes.

Wild Potato Taxonomy: Our New Taxonomy Adopted Here

We summarize above the extensive studies using a variety of morphological, molecular, crossing, and field observation data that have been used to reinvestigate the species boundaries and interrelationships of wild potatoes. Table 1 provides our revised taxonomic decisions relative to Hawkes (1990), recognizing 107 wild species and four cultivated species, and provides hypotheses of interspecific relationships based on the three clade designations. This taxonomy is considerably changed relative to Hawkes (1990) who recognized 228 wild and seven cultivated species divided into 21 taxonomic series (19 tuber-bearing and two non-tuber-bearing).

While these changes since 1990 are extensive, they simply demonstrate that taxonomy of section Petota is inherently complicated by a "perfect storm" of biological factors that hinder the simple partitioning of populations into discrete species. These include the lack of strong biological isolating mechanisms and the resulting interspecific hybridization and introgression, allopolyploidy, a mixture of sexual and asexual reproduction, and recent species divergence (as supported by Sarkinen et al., 2013) (Spooner & van den Berg, 1992a; Spooner, 2009). Recent workers have benefited by the collections of prior workers, personal opportunities to collect germplasm and observe variation in natural settings, access to experimental field plots to grow out and measure problematic groups in replicated field trials, and access to the majority of the type specimens that for a variety of reasons were not shared among previous workers.

The very nature of the complicating biological factors in section Petota makes it difficult to define species. The many problems in the recognition of species in section Petota have been discussed at length from literature reviews (e.g., Masuelli et al., 2009; Spooner, 2009; Camadro et al., 2012), and large scale molecular marker analyses (e.g., Jacobs et al., 2008,2011). We consider our taxonomy (Table 1) to be subject to critique and modification, but to greatly improve the highly splintered and unworkable recognition of the many species recognized by Hawkes (1990). While the interspecific relationships are largely well-supported, our decisions of species boundaries are based primarily on results of morphological and molecular marker analyses (Table 2), combined with a practical ability to distinguish species, following a phylogenetic species concept, i.e., the recognition of an irreducible (basal) cluster of organisms, diagnosably distinct from other such clusters, and within which there is a parental pattern of ancestry and descent (Cracraft 1989).

Clearly, within our more broadly-defined species there exist distinct variants, sometimes distributed in small localized areas, that on first examination appear worthy of taxonomic recognition. Also, there are apparent wider dines of variation as described by Spooner et al. (2004) for S. stoloniferum, where smaller northern plants were previously called S. fendleri, taller southern ones S. stoloniferum, and softly pubescent more southern plants S. papita. The monographic studies leading to our decisions of species boundaries in section Petota (Table 1) rely, on necessity, from an examination of the thousands of living plantings of germplasm accessions in field plots and herbarium specimens. Previous decisions in Spooner's group relied partly on molecular marker analysis to distinguish species, for example, separation of S. megistacrolobum and S. toralapanum (both=5. boliviense; Spooner et al., 1997), or separation of S. astleyi and S. boliviense (both=A boliviense (Giannattasio et al., 1994a,b). While both studies could discriminate these taxon pairs, it was only with the use of many morphological characters that overlapped in range. Field and herbarium studies, however, showed these to be imprecise and impractical, and to extend to yet other similar species (Table 1). We conclude that their maintenance as distinct species, as in many other examples in section Petota, will only perpetuate a taxonomy that is unnatural, unworkable, and continue to perpetuate variant identifications by future worders.
COPYRIGHT 2014 New York Botanical Garden
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:p. 283-326
Author:Spooner, David M.; Ghislain, Marc; Simon, Reinhard; Jansky, Shelley H.; Gavrilenko, Tatjana
Publication:The Botanical Review
Article Type:Report
Date:Dec 1, 2014
Previous Article:Plant conservation challenges in the Bahama Archipelago.
Next Article:Systematics, diversity, genetics, and evolution of wild and cultivated potatoes.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |