Systematics, diversity, genetics, and evolution of wild and cultivated potatoes.
Early Classifications of Cultivated Potatoes
Landraces refer to indigenous cultivated crops. There are perhaps 3000 landraces of potato still grown by indigenous farmers in South America. Linnaeus (1753) recognized a single cultivated potato species, S. tuberosum. Dunal (1852) also recognized this single species, but with a separate variety that is now recognized as the wild potato S. chacoense (Ovchinnikova et al., 2011). De Candolle (1886) was the first to name the Chilean landraces as a distinct taxon (S. tuberosum var. chiloense A.DC. [=5. tuberosum Chilotanum group]). Here we use both formal Linnean nomenclature and non-Linnean group nomenclature. Table 4 lists a comparison of taxonomic treatments of cultivated potatoes at the Linnean ranks of series, species, and subspecies, and at the non-Linnean rank of groups and subgroups as discussed below.
Sergei Juzepczuk, Sergei Bukasov. The Russian taxonomists Juzepczuk & Bukasov (1929) were the next to describe the diversity of landrace potatoes. They expanded the concept of cultivated potato species, based on examination of gennplasm collections and their observations in expeditions to South America by Bukasov (Colombia in 1926), Juzepczuk (Peru, Bolivia, Chile, 1927-1928), and Nicolai Vavilov (Ecuador, Peru, Bolivia, Argentina, Chile, Brazil, 1932-1933) (Juzepczuk & Bukasov, 1929; Bukasov, 1933; Juzepczuk, 1937). Most of their taxonomic descriptions were made from living plantings of germplasm collections at the experimental stations of the All-Union Institute of Plant Industry, now the Vavilov Institute of Plant Industry (VIR), Russia. The extensive diversity of these collections led Juzepczuk & Bukasov (1929) to at first consider S. tuberosum as a 'collective species' (5. tuberosum sensu lato). They further subdivided S. tuberosum into 13 species (named using the Linnaeus's binomial system) and defined S. tuberosum in a narrow sense (sensu stricto) as restricted to native Chilean landraces (Juzepczuk & Bukasov, 1929; Bukasov, 1933).
The taxonomic treatment of Juzepczuk & Bukasov (1929) was based mainly on a morphological species concept, but they indicated that these species were supported by distinctive ploidy levels and ecogeographical criteria. Rybin (1929, 1933), Bukasov (1933, 1937, 1960, 1971, 1978) and Lekhnovich (1971) published karyological, geographical, ecological, physiological, biochemical and anatomical information about the landrace species they accepted, using a complex approach that was novel for section Petota. Rybin (1929, 1933) first determined that landrace potatoes exist in a polyploid series from diploid (2n=2x=24), triploid (2n=3x=36), tetraploid (2n=4x= 48), to pentaploid (2n=5x=60), and proposed to use ploidy levels to distinguish species. In many cases, ploidy levels were needed to discriminate morphologically similar cultivated species (e.g., triploid S. chaucha from Andean diploid and tetraploid landraces). However, even this criterion was not absolute for taxonomic recognition, as Lekhnovich (1971) and Bukasov (1978) later indicated the existence of autotriploid forms (cytotypes) in diploid S. goniocalyx and S. stenotomum.
Ecogeography was an important criterion in the taxonomic system of Juzepczuk & Bukasov (1929) and Bukasov (1930, 1933, 1938, 1978). Potato landraces were originally restricted to South America, distributed from western Venezuela to northern Argentina, and also in south-central Chile with a disjunction of 560 km due to the Atacama Desert separating the upland Andean and lowland Chilean potatoes. Based on an informal ecogeographic classification of the cultivated species (Bukasov, 1938) some (but not all) named landraces possessed distinct geographic or/and ecological characters. For example, S. tuberosum sensu stricto was endemic to the lowlands of south-central Chile and to neighboring islands growing at or near sea level, and able to produce tubers under the long days of central coastal Chile. Most of the cultivated diploids and triploids were relatively narrow endemics. However, these characters often relied on minor and overlapping morphological characters, and some of the cultivated species had the same ploidy level or/and occupied common habitats.
The 13 cultivated species (Juzepczuk & Bukasov, 1929) were later modified to 14 (Bukasov, 1933), then 18 (Bukasov, 1937), then 21 (Lekhnovich, 1971), and finally to 17 (Bukasov, 1978), 16 of these from the Andes of western Venezuela to northern Argentina, and one from south-central Chile (S. tuberosum s. str. in system of Juzepczuk & Bukasov (1929) and Bukasov (1933) or S. chilotanum in the latest treatment of Bukasov (1978) (Table 4). In addition, Russian taxonomists recognized hundreds of intraspecific taxa at various taxonomic ranks (subspecies, convarieties, varieties and forms) in order to characterize the tremendous variation (Bukasov, 1933; Lekhnovich, 1971). Many of these names were not validly published, most commonly because a Latin diagnosis was not provided (Lechnovich, 1971) or a type specimen was not designated, as reviewed in Ovchinnikova et al. (2011).
Bukasov (1933, 1938, 1939) attempted to classify these species into natural informal ecogeographical subgroups, and in his latest treatment (Bukasov, 1978) divided them into the three Linnean series Andigena Bukasov, Chilotana Bukasov, and Subacaulia Bukasov (Table 4). An ecogeographical pattern is evident in this classification, with 1) series Subacaulia composed of upland natural hybrids from the altiplano of Bolivia and Peru, 2) series Andigena in a wide range of latitudes and altitudes but generally in lower elevations than members of Subacaulia, and 3) series Chilotana of lowland central Chilean landraces (Bukasov, 1978). Physiological characters reflecting the ecological conditions also were used. For example, Chilean landraces are able to produce tubers under the long days of central coastal Chile, whereas the Andean landraces form tubers under the shorter day length (Bukasov, 1933). Upland landraces of series Subacaulia are frost resistant (Bukasov, 1933, 1938) with bitter-tasting tubers that need special processing to remove high level of glycoalkaloids (Juzepczuk & Bukasov, 1929; Bukasov, 1933). Landraces of series Subacaulia can be readily distinguished from members of other series by the distinctive morphological character of high pedicel articulation (Juzepczuk & Bukasov, 1929).
John G. [Jack] Hawkes
The English taxonomist Jack Hawkes (1944) originally recognized 18 cultivated species with many formally named varieties and forms, informally grouped into geographic regions. This classification was very similar to the 13-species system of Juzepczuk & Bukasov (1929) but with the addition of five new species described by Hawkes. Hawkes (1944) subsequently named many varieties and forms, grouped into geographic regions similar to Bukasov (1933). Before he set out on his first collecting expedition to South America in 1938, Hawkes visited the All-Union Institute of Plant Industry in Leningrad, where he met Russian scientists who had been describing and documenting potato diversity using their system. Hawkes developed reservations about this system, however, saying: "when I later described and classified my own collections of potatoes I followed Vavilov in establishing far too complex a system. Much later I had to simplify this drastically" (Hawkes, 2004). Hawkes's subsequent classifications (Hawkes 1956a, b, 1963, 1990) greatly reduced his landrace taxa, converging on seven species and seven subspecies (Hawkes, 1990; Table 4). Much of Hawkes's (and colleagues) research was devoted to the biology (Hawkes, 1949), crossability (Jackson et al., 1978), ecology (Hawkes, 1954a), taxonomy (Hawkes 1956a, b), chemotaxonomy (Schmiediche et al., 1980; Huaman et al., 1983; Cribb & Hawkes, 1986) cytology (Hawkes, 1958), history (Hawkes & Francisco-Ortega, 1992, 1993), breeding value (Hawkes, 1958), classification theory (Hawkes, 1986), artificial resynthesis of putative hybrids (Hawkes, 1962b; Astley & Hawkes, 1979; Schmiediche et al., 1982; Cribb & Hawkes, 1986), and ethnobotany (Hawkes, 1947; Jackson et al., 1980) of cultivated potatoes.
Carlos M. Ochoa
Carlos Ochoa spent his entire career working on the collection, systematics, and breeding of potato, first at the Universidad Nacional Agraria in La Molina, Peru and later at the International Potato Center (http://agro.biodiver.se/2008/12/ carlos-ochoa/). Most of his research was on the systematics of wild potatoes, where he used data from morphology of herbarium specimens and living plants grown in greenhouses, ploidy levels, and crossability to delimit taxa. His early monograph on the wild potatoes of Peru (Ochoa, 1962) was followed by much more complete treatments of the wild and cultivated potatoes of Bolivia (Ochoa, 1990a) and wild potatoes of Peru (Ochoa, 1999), in which he summarized his many new wild potato species from South America. His only taxonomic treatment of cultivated landraces is in his monograph of the wild and cultivated potatoes of Bolivia (Ochoa, 1990a), summarizing 77 new varietal and form names for the Bolivian cultivated landraces. This intraspecific classification was a complex inter-nested series of subspecies, varieties and forms, similar to that used by the Russian taxonomists. Ochoa never completed a planned treatment of the Peruvian cultivated potatoes. In total (including landrace taxa he mentioned as accepted in his Peruvian treatment; Ochoa, 1999), he recognized eight species and three subspecies of cultivated potatoes, similar to Hawkes (1990) (Table 4).
Cesar Vargas was a lecturer at the National University of Cusco, Peru. He described new wild Peruvian potato species and treated the cultivated potatoes of Peru (Vargas, 1949, 1956) using the species names proposed by Juzepczuk & Bukasov (see above) and Hawkes (1944). He recognized 14 cultivated species (7 diploid, 5 triploid, 1 tetraploid [as S. andigenum], and 1 cultivated pentaploid species).
Alfonso Castronovo, Ludmila Kostina, Andres Contreras and Ingrid Castro (Chilean Potatoes)
The Chilean taxonomist Alfonso Castronovo (1949) collected and provided cultivar names for 113 landrace potatoes in Chile as "papas Chilotas". The Russian taxonomist Ludmila Kostina (1978) provided a treatment of 360 landrace potatoes of Chile (as S. chilotanum Hawkes), classified into about 50 "varietal types". The Chilean agronomists Andres Contreras & Ingrid Castro (2008) described and provided photographs of many of the 289 accessions of S. tuberosum subsp. tuberosum (-S. tuberosum Chilotanum group) in Chile maintained at the Universidad Austral de Chile in Valdivia.
The Russian taxonomist Vladimir Lekhnovich (1971) described hundreds of Andean and Chilean landraces using different taxonomic ranks (subspecies, convarieties, varieties and forms).
All of the taxonomic treatments above classified the group of landraces as distinct Linnaean taxa (e.g., species, subspecies, varieties; the current Linnaean taxonomic code is the International code of nomenclature for algae, fungi, and plants, ICN; McNeill et al., 2012). The English taxonomist John Dodds (1962), in contrast, treated the landraces under the International Code of Nomenclature of Cultivated Plants (ICNCP; the latest version is Brickell et al., 2009) using the group nomenclature. "Cultivar-groups"(the current terminology) are taxonomic categories used by the ICNCP to associate cultivated plants with traits that are of use to agriculturists (Spooner et al., 2003). Dodds suggested that there was poor morphological support for most cultivated species, and recognized only S. xcurtilobum, S. xjuzepczukii, and S. tuberosum, with five "groups" present in the latter (Table 4). The cultivar-group classification of Dodds (1962) was based on comparative morphology, reproductive biology, cytological and genetic data, and cultural practices. He contended that the morphological characters used by Hawkes (1956a) to separate cultivated species exaggerated the consistency of qualitative and quantitative characters. He showed that Andean farmers grow landraces of all ploidy levels together in the same field and that these can all potentially hybridize. He showed no genetic differentiation of the cultivated diploids existed (Dodds & Paxman, 1962) and contended that his classification was conservative in that it "provides a genetically reasonable classification that disturbs the established usage of words [taxonomic names] as little as possible" (Dodds, 1962, p. 530).
Later data supported Dodds's (1962) hypothesis of poor morphological separation of the cultivated species and suggested that they form a genetically diverse assemblage of genotypes of multiple and complex hybrid origins. Some "escaped" and persistent cultivated tetraploids in the Andes ("Araq" potatoes) and some putative "wild" species may be revertants from cultivation (Spooner et al. 1999; De Haan et al., 2012). Biological factors support gene flow among wild and cultivated potatoes. For example, Watanabe & Peloquin (1989, 1991) showed both diploid and unreduced gametes to be common in the South American wild and cultivated species, allowing gene transfer among different ploidy levels. Huaman (1975) showed evidence of natural crosses between the diploid wild species S. megistacrolobum (=S. boliviense) and the diploid cultivated species S. stenotomum (=S. tuberosum Andigenum group). Open pollinated hybrid fruits were found in all experimental plots containing 10, 25, 50, and 90 % of S. megistacrolobum plants within isolated plots of S. stenotomum grown in Huancayo, Peru. Rabinowitz et al. (1990) documented high levels of natural gene flow between the diploid wild taxon S. sparsipilum (=S. brevicaule) and S. stenotomum.
Cultivated Potato Taxonomy: Our Recent Potato Landrace Classification
Huaman & Spooner (2002) examined morphological support for the classification of potato landraces, using 267 accessions of representatives of all seven species and most subspecies as outlined in Hawkes (1990) (Table 4). The results showed some phenetic support for S. ajanhuiri, S. chaucha (=S. tuberosum Andigenum group) S. curtilobum, S. juzepczukii, and S. tuberosum subsp. tuberosum (=S.. tuberosum Chilotanum group) but little support for the other taxa. However, most of this morphological support relied on a suite of characters, all of which are shared with other taxa (polythetic support).
Spooner et al. (2007b) examined 742 accessions of the same cultivated taxa and eight closely related wild species progenitors with 50 nuclear microsatellites and a plastid DNA deletion marker that distinguishes most lowland Chilean from upland Andean landraces (Hosaka et al., 1988). The results highlighted a tendency to separate three groups: 1) putative diploids, 2) putative tetraploids, and 3) the hybrid cultivated species S. ajanhuiri (diploid), S. juzepczukii (triploid), and S', curtilobum (pentaploid). However, there are many exceptions to grouping by ploidy. Strong statistical support occurred only for the species S. ajanhuiri, S. curtilobum, and S. juzepczukii. In combination with the morphological results of Huaman & Spooner (2002) and an examination of the identification history of these collections, Spooner et al. (2007b) classifed the cultivated potatoes into four species: 1) S. tuberosum, with two cultivar groups (the Andigenum group of upland Andean genotypes containing diploids, triploids [except triploid S. juzepczukii]), and tetraploids, and the Chilotanum group of lowland tetraploid Chilean landraces), 2) S. ajanhuiri (diploid), 3) S. juzepczukii (triploid), and 4) S. curtilobum (pentaploid). Gavrilenko et al. (2010) used phenetic analysis of morphological data from an experimental field in the Saint Petersburg Region of Russia and 19 nuclear microsatellites, to study 238 landraces of all cultivated species from the VIR germplasm collection. This study had similar results of Huaman & Spooner (2002) for the morphological data and Spooner et al. (2007b) for the nuclear microsatellite data. The main difference between these two studies was that the VIR study failed to distinguish S. ajanhuiri (five accessions) from the majority of the other landraces.
Gavrilenko et al. (2013) studied 237 accessions of all of Hawkes's (1990) cultivated species and 155 accessions of closely related wild species using 15 plastid microsatellites. All 15 loci were polymorphic and identified a total of 127 haplotypes. As is typical for most cultivated plants, large decreases in genetic diversity were revealed in landraces in comparison with wild ancestral species. Phylogenetic analysis revealed two distinct groups: 1) the majority of accessions of the Solanum tuberosum Andigenum group and the majority of accessions of northern members of the wild progenitor S. brevicaule complex, 2) most of the wild species accessions and almost exclusively hybrid landraces which have introgressed plastid genomes from the other wild gene pools. Lack of clustering of traditionally recognized cultivated species (e.g., Hawkes, 1990) supported the revised four-species classification of cultivated potatoes of Spooner et al. (2007b) and Ovchinnikova et al. (2011; Table 4).
This new classification of cultivated potatoes was incomplete, however, because it failed to account for the many taxonomic names, many published in the Russian literature and not readily available to a non-Russian audience. Ovchinnikova et al. (2011) compiled all 602 basionyms of cultivated taxa, located their type specimens, designated lectotypes when possible, and placed these names (including names not validly published) in synonymy with this new classification.
In summary, landrace potatoes are grown throughout mid to high (about 3000-3500 m) elevations in the Andes from western Venezuela to northern Argentina, and then in lowland south-central Chile, concentrated in the Chonos Archipelago. The widely used classification of Hawkes (1990) divided cultivated potatoes into seven species and seven subspecies, but Bukasov (1978) and Lechnovich (1971) recognized 17 and 21 species respectively, and Ochoa (1990a, 1999) recognized nine species and 141 intraspecific taxa for the Bolivian cultivated species alone. Like the S. brevicaule complex, the S. tuberosum Andigenum group is characterized by ploidy variation and contains diploids, triploids, and tetraploids. Investigation of species boundaries in this group used data from morphological phenetics from a field plots in Peru (Huaman & Spooner, 2002) and the Saint Petersburg Region, Russia (Gavrilenko et al, 2010), nuclear microsatellites (Raker & Spooner, 2002; Ghislain et al., 2006; Spooner et al., 2007b; Gavrilenko et al., 2010), DNA sequence data of nuclear orthologs (Rodriguez et al., 2010), plastid microsatellites (Sukhotu et al., 2004, 2005, 2006; Gavrilenko et al., 2013) and plastid DNA deletion data (Hosaka, 2003; Sukhotu et al., 2004; Ames & Spooner, 2008; Gavrilenko et al., 2013). These results supported a classification of the cultivated potatoes into four species: (1) S. tuberosum, with two cultivar groups (the Andigenum group of upland Andean genotypes containing diploids, triploids, and tetraploids and the Chilotanum group of lowland tetraploid Chilean landraces), (2) S. ajanhuiri (diploid), (3) S. juzepczukii (triploid), and (4) S. curtilobum (pentaploid) (Table 4).
Origin of Cultivated Potatoes
Two classes of hypotheses have long competed concerning the origin(s) of cultivated potatoes: (1) a multiple origin hypothesis developed by Russian scientists, and (2) a restricted origin hypothesis developed by English scientists.
Multiple Origin Hypotheses
Russian scientists (Juzepczuk & Bukasov, 1929; Bukasov, 1933; Vavilov, 1935, 1939; Juzepczuk 1937; Bukasov, 1938, 1939) first developed the multiple origin hypotheses. They followed Vavilov's (1926, 1928) idea that the center of origin of crop plants corresponded to geographic areas(s) containing the greatest diversity of cultivated species and their wild relatives, and Vavilov's ideas concerning the role of weedy wild crop relatives in crop domestication (Vavilov, 1962,1965,1989). Juzepczuk & Bukasov (1929) postulated that the greatest diversity in potato landraces was concentrated in two different centers corresponding to independent domestication events: 1) the Peruvian and Bolivian plateau, and 2) southern Chile, in the region of Chiloe Island and the adjoining islands.
Their observations were based on expeditions by Bukasov to Mexico, Guatemala, and Colombia from 1925 to 1926, and by Juzepczuk to Peru, Bolivia, and Chile from 1927 to 1928, supplemented by examination of living collections at the experimental stations of All-Union Institute of Plant Industry, Leningrad (now the Vavilov Institute of Plant Industry in Saint Petersburg, Russia (Juzepczuk & Bukasov, 1929; Bukasov, 1930, 1933; Juzepczuk, 1937). These studies documented extensive polymorphism in landrace morphology (Juzepczuk & Bukasov, 1929; Bukasov, 1930, 1933), chromosome numbers (Rybin, 1929, 1933), and physiological characters of frost tolerance, photoperiodic response, earliness, and dormancy (Bukasov, 1932, 1933; Razumov, 1931), reflecting adaptations to diverse ecogeographic conditions. As amplified below, these scientists proposed potato landraces to have two main separate origins, derived from different wild species in different geographic areas.
Solanum Tuberosum Andigenum Group
Andean Diploid Landraces. Juzepczuk and Bukasov (1929) hypothesized that Andean landraces evolved from wild species endemic to the Peruvian and Bolivian plateau, often growing in indigenous resident's fields, with ongoing hybridization after domestication. They proposed the Peruvian diploid wild species S. multiinterruptum and the Bolivian diploid wild species S. sparsipilum (=S. brevicaule) as wild species progenitors. Bukasov (1966, 1968, 1970, 1978) extended the list of putative wild species progenitors and postulated that each diploid cultivated species had an independent origin from separate diploid wild species. He suggested that the current distribution of cultivated species reflected their geographic origins. In agreement with Hawkes (1958), Bukasov (1966, 1978) indicated that the wild species S. canasense (=S. candolleanum) and S. leptophyes (=S. brevicaule) both from Peru may have been involved in the origin of the polymorphic cultivated diploid S. stenotomum (=S. tuberosum Andigenum group). Bukasov (1966) agreed with Cardenas (1950) that the Peruvian wild species S. candolleanum was involved in the origin of S. phureja (=S. tuberosum Andigenum group) and suggested that S. phureja was a result of crosses between the Peruvian wild species S. candolleanum and S. leptophyes (=S. brevicaule).
Bukasov (1966, 1978) proposed independent endemic origins of diploid landraces from Ecuador and Columbia (S. canarense, S. kesselbrenneri, S. rybinii [all =S. tuberosum Andigenum group], Table 4). The following northern Andean wild species were suggested as likely progenitors: S. flahaultii, S. paucijugum, S. regularifolium (=S. andreanum), and S. solisii (=S. andreanum); however, this suggestion was not supported by further studies (see below). Ugent (1970a) also proposed multiple origin hypotheses for the cultivated species, followed by continued hybridization with the wild species.
Andean Triploid Landraces (Exclusive of S. juzepczukii). Bukasov (1939, 1966, 1978) proposed that natural crosses of diploid and tetraploid landraces in various ecogeographic regions produced the cultivated triploids S. chocclo, S. mamilliferum, and S. tenuifilamentum (Table 4; all now classified as S. tuberosum Andigenum group). Bukasov (1939) postulated that S. chaucha (=S. tuberosum Andigenum group) a hybrid triploid species lacking tuber dormancy, formed from a cross between S. phureja (2x) and S. andigenum (4x) (both S. tuberosum Andigenum group). Lekhnovich (1971) and Bukasov (1978) later recognized S. chaucha as an autotriploid of S. phureja.
Andean Tetraploid Landraces. Bukasov (1939) suggested that Andean tetraploid landraces are of multiple origins arising through meiotic polyploidization (fusion of unreduced gametes of different cultivated diploids). Later, Bukasov (1966, 1978) proposed origins of Andean tetraploid landraces through interspecific hybridization between cultivated diploids and various diploid wild species with subsequent polyploidization of these interspecific hybrids.
Bitter Potatoes: S. juzepczukii (3x) and S. curtilobum (5x). Bukasov (1978) classified S. juzepczukii and S. curtilobum in series Subacaulia. They share frost resistance encountered at the high altitudes of Peru and Bolivia. The morphological similarity of S. juzepczukii and S. curtilobum to the sympatric wild species S. acaule was earlier noted by Juzepczuk & Bukasov (1929). Juzepczuk (1937) later proposed that S. juzepczukii and S. curtilobum were of hybrid origin involving species in series Acaulia (S', acaule, S. punae, S. depexum Juz. [all =S. acaule]; all tetraploids). Bukasov (1939) proposed that S. juzepczukii was derived from natural crosses between an unknown cultivated diploid and the wild tetraploid species S. acaule, and that S. curtilobum was derived from natural crosses between S. juzepczukii and Andean cultivated tetraploids (S. andigenum [=S. tuberosum Andigenum group]), matching their placement into series Acaulia together with S. acaule (Bukasov, 1955, 1966; Lekhnovich, 1971). Bukasov (1939) indicated that S. juzepczukii is similar to experimental interspecific hybrids produced in different combinations including S. acaule and cultivated diploids that were then recognized as S. canarense, S. gonicalyx, and S. rybinii (=S. tuberosum Andigenum group). Further evidence as to the hybrid origins of S. juzepczukii and S. curtilobum was provided by morphological and cytological data generated by experimental resynthesis (Hawkes, 1962b; Schmiediche et al., 1982). All later taxonomists recognized S. juzepczukii and S. curtilobum at the species rank (Table 4), and agreed with the hybrid origin hypothesis involving S. acaule.
Solanum Tuberosum Chilotanum Group
Chilean Landraces. Following Darwin (1845), de Candolle (1912), and Bitter (1913), Juzepczuk & Bukasov believed that Chilean landraces evolved in the lowland region of southern Chile and adjoining islands independently from upland Andean potatoes (Juzepczuk & Bukasov, 1929; Juzepczuk, 1937; Bukasov, 1933, 1939, 1978), and as a result classified them into series Chilotana and series Andigena respectively (Table 4). Russian taxonomists hypothesized that Chilean landraces evolved from the wild Chilean tetraploid species S.fonckii (a nomen nudum from a herbarium annotation made by R.A. Philippi in SGO), S. leptostigma, and S. molinae. Hawkes (1956a) suggested that all these taxa represent naturalized escapes from cultivation and treated them as S. tuberosum subsp. tuberosum (=S. tuberosum Chilotanum Group).
Ugent et al. (1987) proposed a Chilean origin of tetraploid Chilean landraces but from another ancestor, the wild species S. maglia, today known from coastal Chile and a single valley in Argentina, but with all locations 1000 km north of Chiloe Island where S. tuberosum Chilotanum group landraces are today grown. This hypothesis was based mainly on starch grain analysis from the fossil tuber skins found in archaeological sites of south-central Chile compared to starch grains from extant S. maglia and Chilean landraces.
Restricted Origin Hypothesis
The restricted origin hypothesis was developed by Salaman (1946), Hawkes (1956a, 1990, 1999), and Simmonds (1964, 1995) who proposed that potato domestication took place in South America somewhere between Colombia and Bolivia from diploid wild species, followed by polyploidization. They then suggested a subsequent expansion of those short-day adapted landraces into new ecological conditions north to Colombia and Venezuela and south to coastal Chile.
Solanum Tuberosum Andigenum Group
Diploid Andean Landraces. Hawkes (1990) considered S. stenotomum (-diploid, S', tuberosum Andigenum group) to be the most primitive diploid cultivated species. Hawkes (1958) proposed the origin of S. stenotomum from wild ancestors related to the present day wild species S. canasense (=S. candolleanum) S. leptophyes (=S. brevicaule) and S. soukupii (=S. candolleanum) but later narrowed this to just S. leptophyes (Hawkes, 1994). He considered S. phureja to be selected from S. stenotomum (both =S. tuberosum Andigenum group) for quick maturity and lack of tuber dormancy (Hawkes & Hjerting, 1989).
According to Ugent (1970a), the cultivated diploids originated from a group of morphologically similar wild species distributed from central Peru to northern Argentina: S. abbottianum, S. brevicaule, S. bukasovii, S. canasense, S. leptophyes, S. liriunianum, S. multidissectum, S. multiinterruptum, S. ochoae, S. souhupii, S. spegazzinii, S. vidaurrei. Ugent (1970a) grouped all these 'microspecies' into the 'Solarium brevicaule complex' and proposed continuing hybridization of cultivated species with yet another wild species outside the complex (S. acaule, S. megistacrolobum [=S. boliviense], S. raphanifolium) that continued to enrich the cultivated gene pool. Bracher (1975) hypothesized the wild Argentinian species S. vernei (not a member of S. brevicaule complex) as the ancestor of cultivated diploids; but this was not supported by recent molecular data (below).
Andean Triploid Landraces, Exclusive of S. juzepczukii. Hawkes (1963) synonymized all triploid cultivated species recognized by Bukasov (1978) with S. chaucha (=5. tuberosum Andigenum group) except S. juzepczukii (Table 4) and suggested that this cultivated triploid originated from natural crosses between the cultivated tetraploid S. tuberosum subsp. andigenum and cultivated diploid species S. stenotomum (both =S. tuberosum Andigenum group).
Tetraploid Andean Landraces. Hawkes (1956a) proposed two scenarios for the origin of tetraploid Andean landraces, both in the region of southern Peru and northern Bolivia. The first was from somatic chromosome doubling of widely distributed diploid landrace S. stenotomum, the second from natural crosses of S. stenotomum with wild diploid S. sparsipilum (=S. brevicaule). Ugent (1970a) reported the existence of natural interspecific hybridization between S. stenotomum and S. sparsipilum. Cribb & Hawkes (1986) synthesized this interspecific combination and analyzed its morphology and tuber proteins; their results did not contradict a hypothesis of the hybrid origin of subsp. andigenum. Rabinowitz et al. (1990) demonstrated, with isozyme markers, high levels of interspecific hybridization between S. sparsipilum and S. stenotomum in experimental plots in the Andes. Matsubayashi (1991) hypothesized that tetraploid Andean landraces originated from crosses of the two diploid cultivated species S. phureja and S. stenotomum (both =S. tuberosum Andigenum group) followed by chromosome doubling.
Cultivated Bitter Species, S. ajanhuiri, S. curtilobum, and S. juzepczukii. All taxonomists recognized S. juzepczukii and S. curtilobum at the species rank (Table 4) and agreed that they were hybrids involving the tetraploid wild species S. acaule. Hawkes (1962b) and Schmiediche et al. (1982) synthesized artificial triploids which were morphologically similar to the natural species S. juzepczukii in crosses between S. acaule (maternal parent) and cultivated diploid species. Schmiediche et al. (1982) could not resynthesize pentaploid hybrids in crosses of S. juzepczukii (maternal parent) with tetraploids of subsp. andigenum, although Hawkes (1962b) reported success in such crosses.
Ugent (1970a) suggested that the gene pool of cultivated diploids was enriched by natural hybridization with the wild diploid species S. megistacrolobum (=S. boliviense) and S. raphanifolium but did not mention cultivated diploid S. ajanhuiri. Huaman et al. (1982) resynthesized S. ajanhuiri with crosses of diploid cultivated S. stenotomum (=S. tuberosum Andigenum Group) as the maternal parent and wild diploid S. megistacrolobum (=5. boliviense) as the male. Reciprocal combinations with S. megistacrolobum as the female parent produced only a few seeds with very poor germination (Huaman et al., 1982). Johns & Keen (1986) presented field data supporting a hybrid origin of S. ajanhuiri from S. stenotomum and S. megistacrolobum.
Solanum Tuberosum Chilotanum Group
Chilean Landraces. The main contradiction between the multiple and restricted origin hypotheses concerns the origin and taxonomic status of tetraploid Chilean landraces. As outlined above, Russian taxonomists proposed independent origins of tetraploid Chilean and Andean landraces in separate regions from separate indigenous ancestors, and hence treated them as different taxa: 1) S. tuberosum s. stricto (5. tuberosum var. chilotanum Bukasov & Lechn., later as S. chilotanum [both =S. tuberosum Chilotanum group]) (Bukasov, 1978)], and 2) S. andigenum Juz. & Bukasov, respectively.
Based on the restricted origin hypothesis, Chilean landraces initially were of Andean origin and then introduced into Chile after the potato was already domesticated in the central Andes. Thus, Andean tetraploids somehow appeared in Chile and evolved to the Chilean type including the ability to produce tubers under the long day conditions of Chile (Salaman, 1946; Hawkes, 1944, 1956a, 1990, 1999; Simmonds, 1964). This restricted origin hypothesis is based on two facts: 1) presently in the region of southern and central Chile there have been no diploid cultivated potatoes (except scattered reports) or of wild species from which S. tuberosum could have originated (Hawkes, 1944, 1956a), and 2) long-term selection experiments were reported to have adapted Andean tetraploids to a Chilotanum-like form (referred to as Neo-Tuberosum), especially regarding long day length adaptation (Salaman, 1946; Simmonds, 1966, 1969). This artificial selection to create Neo-Tuberosum was subsequently used by many authors as the model for the evolution of Andigenum germplasm to Chilotanum germplasm in south-central Chile. Using these two ideas, Hawkes (1956a) grouped all tetraploid potatoes (Chilean and Andean) under S. tuberosum L., considering this species as a complex of polymorphic forms that initially evolved in the Andes of southern Peru and northern Bolivia and subsequently spread north to Venezuela and Colombia and south to coastal Chile (5. tuberosum subsp. tuberosum [=5. tuberosum Chilotanum group]). Dodds (1962) and Briicher (1998) concurred with this idea. Gran (1990) proposed a modified scenario where the Chilean landraces originated from hybridizations of Andean landraces with a wild species, possibly S. chacoense, or an unidentified wild species. However, the restricted origin hypothesis lacks supporting evidence of movement from the Andes to southern Chile.
Recent Studies of Cultivated Potato Species Origins
Solanum tuberosum Chilotanum Group. The S. tuberosum Chilotanum and Andigenum groups can be separated by morphology, nuclear-cytoplasmic interactions, and day length responses (Grun, 1990). Data from nuclear and plastid microsatellites and morphology show that these groups often intergrade (Huaman & Spooner, 2002; Raker & Spooner, 2002; Spooner, et al. 2007b; Gavrilenko et al. 2010, 2013). One of the arguments of the restricted origin hypothesis (proposing an Andean origin of Chilean landraces) was based on the results of Simmonds (1966) who worked with "Neo-Tuberosum" clones. Neo-Tuberosum refers to cultivated potato adapted to long-day tuberization and a syndrome of related morphological and physiological traits developed by intercrossing and selection of short-day adapted potatoes of the Andigenum group. The putative rapid selection of Neo-Tuberosum suggested that this process could occur naturally to produce Chilotanum group germplasm.
Ghislain et al. (2009b) demonstrated with nuclear microsatellites, however, that Neo-Tuberosum germplasm is related to the Chilotanum group, not the Andigenum group. They interpreted this unexpected result to be caused by strong rapid selection against the original Andigenum clones after unintended hybridization with Chilotanum group germplasm that occurred in nearby experimental fields. This result questioned a separate hypothesis that the European potato was derived from the Andigenum group (Salaman, 1937), and supported an earlier hypothesis that the European potato was derived from the Chilotanum group (Juzepczuk & Bukasov, 1929).
As was mentioned above, Grun (1990) proposed a modification of the restricted origin hypothesis and suggested that Chilean landraces originated from crosses between tetraploid Andigenum group and an unidentified wild species. Hosaka's group proposed an evolutionary pathway where hybrids of the wild species S. tarijense (=S. berthaultii', as a maternal ancestor) and tetraploid Andigenum group species were transferred to the southern regions to Chile. This view was based on plastid RFLPs, dividing cultivated species into five main "types:" A, C, S, T, W (Hosaka et al., 1984; Hosaka, 1986, 1995; Hosaka & Hanneman, 1988) or according to new nomenclature based on multiplex PCR and using an additional mitochondrial marker: A, M, P, T, W types (Sanetomo & Hosaka, 2011; Hosaka & Sanetomo 2012) (Table 5).
None of these plastid types were species-specific (Table 5). Thus, the T-type plastid DNA was predominantly found in the Chilotanum group (88 %), and rarely (about 1 %) in tetraploid Andigenum group, mainly from Argentina and southern Bolivia (Hosaka, 2002; Hosaka and Sanetomo, 2009). Similar frequencies of the T-type plastid DNA in tetraploid landraces was detected by Spooner et al. (2007b) and by Gavrilenko et al. (2013). Further screening of 566 accessions of 35 wild species including putative wild ancestors of the Chilotanum group revealed T-type plastid DNA in about 18 % of S. berthaultii (including S. tarijense) and S. neorossii (Table 5) but not any other examined wild species (Hosaka, 2002, 2003). The T-type plastid DNA could be distinguished by the presence of a 241 bp deletion in the ndhC/trnV intergenic spacer region (Kawagoe & Kikuta, 1991); this deletion is easily detected by PCR marker HI (Hosaka 2002). Gavrilenko et al. (2013) demonstrated with plastid microsatellites that the T-type plastid DNA is distinct not only by the presence of the 241 bp deletion, but also by the combination of many plastid simple sequence repeat SSR (microsatellite) alleles which were not detected in the other haplotypes of cultivated species. All accessions with the 241 bp deletion (representatives of the T-type plastid DNA) share the same plastid SSR haplotype TIT which was detected only in Chilean landraces and in a few S. berthaultii accessions (Table 5, Fig. 7). Other representatives of S. berthaultii without the 241 bp deletion having different plastid SSR haplotypes were close to haplotype III in the plastid SSR tree (Gavrilenko et al. 2013). All the results of molecular studies mentioned above support S. berthaultii sensu lato from southern Bolivia to northern Argentina as the maternal ancestor (cytoplasm donor) of Chilean landraces with the T-type of plastid DNA.
The hypothesis of Ugent et al. (1987) about the origin of the Chilotanum group from the wild species S. maglia found support from nuclear microsatellites (Spooner et al. 2012), which grouped S. maglia with the Chilotanum group. Rodriguez et al. (2010) using DNA sequence data of the waxy gene, also found that two of three examined accessions of the Chilotanum group had alleles grouping with S. maglia and in a clade containing the Andigenum group and related wild species, supporting S. maglia as a hybrid contributor to the Chilotanum group. However, the results were ambiguous because the two S. maglia accessions lack the 241 bp plastid deletion that is shared by most accessions of the Chilotanum group. Only one of 34 accessions of Chilean landraces had the same plastid SSR haplotype ('IT) as all three examined accessions of S. maglia and as the majority of tetraploid Andean landraces in the plastid microsatellite study of Gavrilenko et al. (2013), whereas 88 % of Chilean landraces had another plastid SSR haplotype ('III') and all shared a 241 bp deletion in the ndhC/trnV region (Table 5). Two accessions of S. maglia analyzed for plastid DNA type (Hosaka, 1986) also showed haplotypes common with majority of tetraploid Andean cultivated potatoes. Based on results of Spooner et al. (2012) and the results of plastid DNA studies, Gavrilenko et al. (2013) proposed S. maglia as a possible paternal contributor to Chilean tetraploid landraces. In conclusion, the origin of the Chilotanum group remains unresolved.
Rodriguez et al. (2010) studied the hybrid origins of S. ajanhuiri from the Andigenum group diploids x S. boliviense, S. juzepczukii from the Andigenum group diploids x S. acaule, and S. curtilobum from the Andigenum group tetraploids x S. juzepczukii. For the tetraploid Cultivar groups of 5. tuberosum, hybrid origins are suggested entirely within much more closely related species, except for two of three examined accessions of the Chilotanum group that appear to have alleles from the wild species S. maglia. Two hybrid origins proposed by others received no support, that is, the crop/weed species S. sucrense (from Andigenum group tetraploids and S. oplocense), and S. vernei as a wild species progenitor of the Andigenum group.
Solatium tuberosum Andigenum Group. The first plastid DNA RFLP studies supported the multiple origin hypotheses for diploid landraces of the Andigenum group (Table 5). Hosaka (1995) detected four "types" of plastid genomes (W, A, S, C) within S. stenotomum that he interpreted to support multiple origins of the diploid Andigenum group from different closely related wild species. However, further analyses of additional landrace and wild species accessions revealed restricted plastid DNA polymorphism within diploids of the Andigenum group. It showed a predominance of two major plastid DNA types 'S' (74 %) and 'A' (24 %) in the diploid Andigenum group, (Table 5); these two types were also found in many representatives of wild species progenitors in the S. brevicaule complex (Sukhotu and Hosaka, 2006; Sukhotu et al. 2004, 2005, 2006; Hosaka & Sanetomo, 2009) (Table 5). These results suggested that diploid landraces either 1) had dual origins from two different wild species or 2) had introgression with haplotype A (Sukhotu & Hosaka, 2006; Sukhotu et al. 2006).
Gavrilenko et al. (2013) used another set of plastid SSR markers and detected only two haplotypes among 100 diploid landraces (S. phureja and S. stenotomum [both =S. tuberosum Andigenum group]), 83 % had predominant haplotype 'I' and 13 % had haplotype 'II' (Fig. 7). Plastid diversity in the ancestral S. brevicaule complex exhibited much higher diversity, with 93 wild species accessions having 69 haplotypes; most of them were unique. The predominant haplotype T of diploid landraces was found also in two accessions (2 %) of representatives both northern and southern members of the S. brevicaule complex (Gavrilenko et al., 2013).
These data, together with the results of Sukhotu et al. (2006) and Sukhotu & Hosaka 2006), support the domestication of Andigenum group diploids from members of the S. brevicaule complex having the predominant in cultivated diploids 'S' type plastid DNA (or plastid SSR haplotype 'I'); which could be easily distinguished by the presence of the allele NTCP6 127 described earlier by Hosaka (2003) as having a 48 bp deletion in rps16/trnQ region of plastid DNA.
Results of plastid DNA studies of Hosaka's group (Hosaka, 2003; Sukhotu et al. 2004, 2005,2006; Sukhotu & Hosaka 2006) and Gavrilenko et al. (2013) in general are in agreement, although the results are obtained with different germplasm collections and different plastid DNA markers. Species differentiation based on plastid DNA studies demonstrate that all accessions of the diploid and triploid Andigenum groups were grouped together with representatives of wild species accessions of'S. brevicaule complex'--mostly of northern members (S. bukasovii, S. canasense, S. multidissectum [all =S. candolleanum]), but also with a few representatives of the southern members of 'S. brevicaule complex' and with a few accessions of wild species from other gene pools (as S. boliviense, S. maglia, S. tarijense [=S. berthaultii]). These results could reflect subsequent hybridization of landrace and wild species as proposed by Ugent (1970a).
Within the landraces, the Andigenum group tetraploids have the highest level of plastid DNA polymorphism (Hosaka, 1995; Sukhotu et al. 2004, 2005; Sukhotu & Hosaka, 2006; Gavrilenko et al., 2013) (Table 5, Fig. 7). This supports earlier hypotheses that they arose both from cultivated diploids by sexual polyploidization and from hybridization with wild species (as maternal parents). Some of the plastid microsatellite haplotypes specific only to Andean tetraploid landraces were detected in the samples from the southern Andes and were absent farther north (Gavrilenko et al, 2013). In contrast, some haplotypes detected in the northern Andes were not found in tetraploid landraces from the southern Andes, supporting possible independent introgression events with representatives of different wild species (Gavrilenko et al., 2013).
AFLP analysis (Spooner et al., 2005a) supported a hypothesis of a single origin of Andean landraces from the northern members of the Solanum brevicaule complex indigenous to southern Peru and northern Bolivia. These northern members of the S. brevicaule complex are here combined into the single highly polymorphic species S. candolleanum sensu lato (Table 1). Thus, disagreement between two hypotheses of origin of diploid members of the Andigenum group (multiple origins vs. a single origin) is simply a result of differing taxonomic circumscriptions of wild species belonging to members of the northern S. brevicaule complex.
One of the arguments of a multiple origin of the Solanum tuberosum
Andigenum group was based on the putative distinct ecogeographical habitats of landraces that exist in the polyploid series forming this group. In addition, ploidy level has been a major character helping to classify cultivated potatoes under previous taxonomic systems. Spooner et al. (2010) examined associations of environments to ploidy levels (2x, 3x, 4x, 5x) of all landrace populations in South America using a database of 2048 georeferenced accessions examined with random-Forest library (Liaw and Wiener, 2002) (in R; R Development Core Team, 2010). Except for the Chilotanum group and extreme northern and southern range extensions of the Andigenum group, it was impossible to find distinct habitats for the ploidy variants of the S. tuberosum Andigenum group.
Solarium ajanhuiri, S. curtilobum, and S. juzepczukii. Solarium ajanhuiri, S. curtilobum, and S. juzepczukii have long been proposed to be of hybrid origin from members of the Andigenum group and the wild species S. acaule and S. boliviense sensu lato (Juzepczuk & Bukasov, 1929; Juzepczuk, 1937; Bukasov, 1939; Hawkes, 1944, 1958). Nuclear DNA sequence data (Rodriguez et al., 2010) have supported their origins by showing additivity of alleles from their proposed parents; plastid SSRs (Gavrilenko et al., 2013) have supported the wild species parents as the maternal ancestors for S. juzepczukii and S. ajanhuiri. However, there are rare exceptions to classically proposed hybrid origins as seen with plastid microsatellite data, suggesting possible multiple origins of S. ajanhuiri and S. juzepczukii from reciprocal crosses (Gavrilenko et al., 2013) (Table 5, Fig. 7).
Contradiction between nuclear SSR and plastid SSR results in relation to S. curtilobum supports an alternative lineage in its maternal origin related to the Andigenum group (Gavrilenko et al., 2013), and not to S. juzepczukii as was proposed before. All ten examined accessions of S. curtilobum have the plastid SSR haplotype 'F common with members of the Andigenum group, grouping S. curtilobum separately from S. acaule and S. juzepczukii (Table 5, Fig. 7). Accordingly, different scenarios for the origin of the pentaploid cultivated species S. curtilobum were proposed, such as Andigenum group tetraploids x S. juzepczukii (unreduced gametes) and Andigenum group triploids (unreduced gametes) x S. acaule (Gavrilenko et al., 2013). These assumptions correlate with observations of Hawkes (1962b) that mixed fields of representatives of the cultivated species S. curtilobum, S. juzepczukii, and members of Andigenum group frequently co-occur with the wild species S. acaule.
Summary of Cultivated Potato Origins
Two hypotheses have been advanced for the origin of cultivated potatoes, 1) a multiple origin hypothesis, and 2) a restricted origin hypothesis. Much of the disagreement between these two hypotheses stems from the taxonomic circumscription of the putative progenitor species, mainly proposed to be members of the taxonomically difficult S. brevicaule complex. However, there remain gaps in archaeological and genome sequence data that need to be filled in to delineate these origins. For example, we have no knowledge of possible human transport of domesticated potatoes from the Andes to southern Chile. There may have been a (now extinct) widespread tuber-bearing wild species progenitor in southern Chile. Molecular data from accessions currently residing in genebanks may not truly represent the genetics of the original accessions. Finally, many wild species are represented by very few accessions, which may introduce a bias in these studies.
Recent investigations gave support to some points of both hypotheses of potato domestication and proposed the following scenarios. The Andigenum group originated in a single domestication event from diploid wild species in the S. brevicaule complex in southern Peru and immediately adjacent northern Bolivia (Spooner et al., 2005a). The ancestral diploid population(s) and domesticated diploids probably had the same plastid DNA haplotype that is predominant in the present day diploid landraces (plastid DNA type 'S' or plastid SSR haplotype 'I'). The origin of the triploid and tetraploid forms of the Andigenum group could have multiple origins both through meiotic polyploidization events (unreduced gametes) of diploid landraces and through interspecific hybridization.
Later differentiation of the Andigenum group likely involved nuclear and organellar introgression from wild species other than members of the S. brevicaule complex, allowing cultivated species to spread to broader ecological conditions and wider geographical areas, with yet more rounds of hybridization (Gavrilenko et al., 2013). Thus, the origin of the frost resistant species S. juzepczukii, S. curtilobum, S. ajanhuiri involved S. acaule and S. boliviense in the highland Andes of southern Peru and Bolivia. Introgression from S. berthaultii (as a maternal ancestor) and possibly S. maglia led to the formation Chilean landraces in lowland southern coastal Chile. In conclusion, the question of the origin of the cultivated potato is not fully resolved and may need additional data from a wider sample of in situ collections and further investigations in genomics and archeology.
From Landraces to Modern Potatoes
Landrace potatoes are today widely distributed from western Venezuela to northern Argentina, with another group of landraces in coastal Chile (Spooner et al., 2010). Fossil evidence (as preserved tubers) document potatoes in various sites along the dry coast of Peru as early as 8000 BC (Engel, 1970; Ugent et al., 1982; Ugent & Peterson, 1988) and in south-central Chile at the Monte Verde archaeological site (as potato skins) at 11,500 BC. The primary domestication of potatoes in the Andean uplands likely occurred around Lake Titicaca at the boundary of Peru and Bolivia (Hawkes, 1944; Ugent, 1970a; Spooner et al., 2005a). Potatoes were first observed in South America by outsiders by Spanish explorers in 1536 in the tropical lowlands of the Magdalena River Valley in present-day Colombia (Castellanos, 1886) .
Potato germplasm has a huge reservoir of genetic and morphological diversity. This implies that predictions based on one or a few clones are not representative of the potato crop. This has implications for properly assessing the potential of potatoes for food security under climate change; predictions of global yields, adaptation to climate, or influence of climate change are only true for that particular clone or set of clones. For example, it is possible to make crosses in northern Ireland and select progeny in the Negev desert resistant to heat stress up to 40[degrees] Celsius as well as resistant to Verticillium wilt and early blight (Susnoschi et al., 1987). In summary, the potential of potatoes as a crop may have a much better range of adaptability to climate change than previously predicted through modeling (Hijmans, 2003; Schafleitner et al., 2011).
An obvious and major domestication trait in potatoes is the shortening of stolons and a corresponding increase in tuber size. Wild potato species typically produce small tubers on the ends of stolons which may be a meter or more in length. This is an adaptive feature that allows for the production of asexual propagules over a large area. Domesticated potatoes, in contrast, require short stolons for commercial production systems and large tuber size for high marketable yield.
Members of the Solanaceae produce glycoalkaloids, which are toxic and can cause DNA damage when consumed (Korpan et al., 2004). Glycoalkaloids are found in both leaves and tubers, where they impart a bitter taste (Camire et al., 2009). Wild potato species contain varying levels of a wide array of glycoalkaloids (Friedman, 2006). However, the cultivated potato contains low glycoalkaloid levels, typically only solanine and chaconine, suggesting these bitter compounds were selected against during domestication (Johns & Alonso, 1990).
Cultivated potatoes likely originated in a broad band of the equatorial regions of South America, where photoperiod remains near 12 h throughout the year. However, potatoes grown in the major production areas in Europe, North America and Asia must be able to tuberize under the long photoperiods of temperate zone summers. One of the most important traits required for the adaptation of South American potatoes to Europe was the ability to tuberize under a long photoperiod. Genetic models for the tuberization response to photoperiod have been proposed. In diploid cultivated x wild species hybrids, tuber production under a 14-h photoperiod appears to be dominant over that for the inability to tuberize (Hermundstad & Peloquin, 1985; Jacobsen & Jansky, 1989; Yerk, 1989; Jansky et al., 2004; Kittipadakul et al., 2012). Wild species do not produce tubers when grown under the photoperiods of the summer production season at temperate latitudes. Cultivated potatoes segregate for this trait and, when crossed to wild species, produce some hybrid offspring that tuberize under long days (Hermundstad & Peloquin, 1985; Kittipadakul et al. 2012).
The physiological basis for tuberization under long photoperiods involves biochemical and molecular signals that link photoperiod perception in leaves to changes in cellular growth patterns in stolons (Rodriguez-Falcon et al., 2006; Sarkar, 2008, 2010). The tuberization stimulus is perceived in above-ground stems and transmitted to underground stolons (Gregory, 1956). Some of the essential players in this long-distance signaling pathway have been identified and include phytochrome B (Batutis & Ewing, 1982; Hannapel et al., 2004), phloem transmissible StBe15 mRNA (Banerjee et al., 2006; Hannapel, 2010), miR172 microRNA (Martin et al., 2009), gibberellins (Krauss & Marschner, 1982; Carrera et al., 1999; Martinez-Garcia et al., 2002), POTH1 (Chen et al., 2003), StSP6A (Navarro et al., 2011), CO (Rodriguez-Falcon et al., 2006; Navarro et al., 2011), sucrose (Chincinska et al., 2008) and temperature (Krauss & Marschner, 1982). Gibberellins, cytokinins, and jasmonate-like compounds are important in regulating tuberization that is activated in the stolon apex (Hannapel et al., 2004). In temperate zone cultivars, short photoperiods, cool temperatures, and low levels of available nitrogen promote early tuberization (Ewing & Wareing, 1978; Krauss, 1985; Sarkar, 2008). Recently, a major effect quantitative trait locus for plant maturity and tuber initiation was found to be controlled by a transcription factor that acts as a mediator between the circadian clock and StSP6A (Kloosterman et al., 2013). Consequently, even though the pathway for the perception of photoperiod and the response to it is complex, it appears that a major regulatory factor controls tuberization under long days. We can speculate that it would have been easy to select for this simply inherited, dominant genetic system during potato domestication and adaptation to worldwide production systems. In fact, it would be self-selecting. In segregating populations, any genotypes that did not tuberize would not have been maintained.
Geographic Correlates of Potato Systematics and Diversity
The 107 wild potato species can be found between 38[degrees]N to 41[degrees]S, between 0 and 5000 m altitude, within habitats from -1[degrees]C to 26[degrees]C annual average temperature, and with mean annual rainfall from less than 100 mm (S. x neoweberbaueri) to more than 3700 mm (S. acaule, S. boliviense). Most wild potato species can be found between 35[degrees]N and 35[degrees]S, between 1500 m and 4000 m altitude, 7.5 to 20[degrees]C mean annual temperature, and 250 to 1250 mm annual rainfall (Hawkes, 1994; Hijmans & Spooner, 2001). These ranges of ecological parameters of wild potato species are paralleled by high morphological polymorphism within and among the species (Spooner et al., 2004), making it difficult to delineate species.
Geographic information systems (GIS) have played a major role in improving the accuracy of accession locations since the late 1990's (Hijmans et al., 1999) and spatial methods are increasingly applied to the analysis of the ecology of crop wild relatives, including potatoes. While earlier analyses were largely descriptive (Hijmans & Spooner, 2001), later ones frequently used specialized Bayesian approaches (e.g. Simon et al., 2010). Though spatial methods have made significant contributions there are still some caveats to consider: a) the temporal and spatial resolution of the underlying datasets may be insufficient (e.g. some databases for current climate often refer to a climate scenario before climate change based on average data from the 1960's to 1990's, and spatial data are mostly based on interpolations that may not be representative of micro-climates) and b) the bias introduced by the selection of standard sets of variables, i.e., some elements of environment may not be included, like soil data and ground-water level. Notwithstanding, spatial or GIS analytical tools have made many contributions to the analysis of the distribution and ecology of potatoes, including improved distribution maps (Hijmans et al., 1999, 2001), genebank management (Hijmans et al., 2000; Jansky et al., 2013), the compatibility of climatic niches (Simon et al., 2010, 2011; Spooner et al., 2010), the confirmation of the role of ploidy in range expansion (Hijmans et al., 2007), or in the analysis of species-level traits, such as disease resistance (Spooner et al., 2009).
In some species, such as S.jamesii, S. stoloniferum, and S. sucrense (=5. brevicaule) there are no associations of genetic variation and ecogeographical variation (del Rio et al., 2001; del Rio & Bamberg, 2002; Bamberg & del Rio, 2008). However, an association of geography and genetic variation was found within S. verrucosum (del Rio & Bamberg, 2004). This was not explained by introgression with other nearby species (Bamberg & del Rio, 2008). A variety of confounding factors that influence the association have been identified including a) sampling bias due to ease of access near roads (Hijmans et al., 2000); b) differential dispersion (Bamberg et al., 2010; Bamberg & del Rio, 2011); c) mis-identification of wild materials after introduction into genebanks (del Rio & Bamberg, 2003; Bamberg et al., 2009); and d) considerable genetic change over time at sampling locations (del Rio et al., 1997b; Bamberg & del Rio, 2003).
The fate of germplasm in situ and the efficiency of its collection, preservation, and evaluation in the genebank are all influenced by the interaction of population structure and population sampling techniques. When multiple generations of seed propagation in the genebank were compared with RAPDs by del Rio et al. (1997a), those generations showed small, insignificant differentiation compared to the distinction of different populations.
Bamberg & del Rio (2004) showed that RAPD markers segregated in accord with reputed population structures of four model potato species. Bamberg & del Rio (2006) also devised experiments to test for techniques that could promote genetic shifts during the genebank's propagation of populations. They tested germplasm expected to be most vulnerable to genetic drift from selecting the most vigorous seedlings for parents and unbalanced maternal seed bulks (Bamberg & del Rio, 2009), and found that neither introduces opportunity for large genetic changes. This is because more robust seedlings typically result from random early sprouting, and even when parents have a high degree of heterozygosity, few loci have low frequency alleles. Furthermore, few loci fail to fix both alleles in at least one population in the genebank (Bamberg & del Rio, 2003), so even if an individual population would lose an allele in the next generation, no alleles would be lost overall from the genebank.
Hijmans & Spooner (2001) mapped 6073 locality records of wild potatoes using the 196 species then recognized in Spooner and Hijmans (2001). For comparative purposes in this review, we re-assessed the effect of both the revised potato taxonomy (Table 1) and increased numbers of observation records. We qualitatively contrasted these latest data with those reported in Hijmans and Spooner (2001) to look for general trends. In comparison to the data underlying Hijmans and Spooner (2001) our database contains 13,117 records. Of those, 11,485 are georeferenced (87.5 %) and used to compile Table 6, Fig. 8 (corresponding to Table 1 and Fig. 6 in Spooner & Hijmans, 2001).
The overall number of observation records more than doubled (from 6073 to 13,115) while the number of species was reduced from 199 to 107 (Table 6). Consequently, the average observations per species quadrupled (from 30 to 125). The biggest increase in observations came from Argentina (from 1688 to 4547). In terms of number of species, Peru remains the country with the most species (reduced from 93 to 51) and also remains the country with most rare species (from 42 to 13). Rare species are defined following Hijmans (2001) as species with overall observations less than five. However, Peru's share of rare species increased from 58 to 65 %. Peru is followed in terms of number of species by Mexico, Argentina, and Bolivia as in Hijmans and Spooner (2001). Mexico as before occupies a middle range between Peru and Argentina/Bolivia (from 36 to 27). Despite the absolute reduction, Mexico represents a much bigger share of the species genepool (7.6 % more; from 18.1 to 25.2 %). Yet the biggest increase in relative terms is Honduras that has three species as opposed to only one reported in 2001 (from 0.5 to 2.8 % of the species genepool).
Species richness can be represented by spatial explicit counts of wild potato species per half degree grid cell (Fig. 8). The resulting map shows an overall similar pattern of hot spots to the one published in Hijmans and Spooner (2001). The main differences are in the relative changes in the number of species at the hot spots: the most species-rich areas are now distributed from central to northern Peru, whereas the relative number of hot spots in Bolivia and Argentina decreased overall. The hot spot in central Mexico remains roughly the same with perhaps a more even distribution of species richness. Peru still contains almost half of total number of species (48 %) and has an increased share of rare species. Mexico's importance as a secondary center of species richness increased while Bolivia's importance in the share of the wild potato genepool decreased. As mentioned above, these hot spot patterns and changes probably bear little relevance for trait screening purposes but may be of interest for national or regional management of biodiversity.
Conclusions about genetic changes due to one genebank's techniques may not apply to all genebanks. For example, del Rio et al. (2006) compared the germplasm samples in genebanks in the USA and Peru which had been propagated from the same original source population. Differences were usually small and insignificant, but illustrated that evaluation data generated in one genebank may not be completely transferable to the reputed identical germplasm in another genebank.
A number of studies have investigated the relationship between disease and pest resistance diversity with geography, taxonomy, ploidy, and breeding system (Jansky, et al. 2006, 2008, 2009; Spooner et al., 2009; Cai et al., 2011; Chung et al., 2011; Khiutti et al., 2012; Uribe et al., 2013; Perez et al., 2014; Limantseva et al., 2014). All of these studies found high intra-accession and intraspecific variation of disease and pest resistance, but none of them found conclusive association with geography alone. Hence, the current way to find genetic variation for such traits is still through broad screening studies of large populations until a more efficient strategy using taxonomic, molecular, or biogeographic parameters can predict discovery of such useful traits.
Jackson (1990) proposed use of Nikolai Vavilov's (1922) law of homologous variation for predictivity. Basically, Vavilov proposed that knowledge of traits in one species can be used to predict the presence of similar traits in related species. However, as can be inferred from above, this is not a generally promising strategy in wild potatoes, in contrast to other crops such as wheat (Endresen, 2010).
Germplasm Collections and Molecular Characterization
Wild and cultivated potatoes have been collected extensively and intensively throughout their entire range and are maintained by national and international genebanks. These include the International Potato Center (CIP), The Centre for Genetic Resources and Wageningen University, The Netherlands (CGN), the Institute of Plant Genetics and Crop Plant Research Gatersleben, Genebank External Branch North, Germany (IPK), the Vavilov Institute of Plant Industry (VIR), the United States Potato Genebank (National Research Support Project-6, NRSP-6), and national programs in countries where potatoes were collected. Huaman et al. (2000a) constructed a database of germplasm collections of major genebanks that show collections held in common; this is available by writing to the US genebank (NRSP-6). The ready availability of these genetic resources and associated laboratories, greenhouses, and field stations, has proven invaluable for the study of the genetics and taxonomy of section Petota.
The study and management of ex situ germplasm collections and the study of in situ genetic diversity required new technologies not available to plant taxonomists before the 1990s. The advent of DNA markers led to the development of rapid and reliable methods to uniquely identify potato samples. This can be difficult because they are generally closely related. Common methods use morphological and physiological traits to compare cultivated varieties (Reid et al., 2011), but there are many inherent problems with these techniques, leading to a search for molecular methods for discrimination (Morell et al. 1995; Cooke, 1999). DNA fingerprinting refers to a DNA-based assay to uniquely identify individual accessions. Many fingerprinting markers have been applied, including AFLPs, isozymes, ISAPs, ISSRs, RAPDs, nuclear microsatellites, and nuclear RFLPs (Table 2; Gebhardt et al., 1989; Douches & Ludlam, 1991; Douches et al., 1991; Powell et al., 1991; Gorg et al., 1992; Mori et al., 1993; Hosaka et al., 1994; Cisneros & Quiros, 1995; Kawchuk et al., 1996; Oganisyan et al., 1996; Provan et al., 1996a; Sosinski & Douches, 1996; Ford and Taylor 1997; Milboume et al., 1997; Schneider & Douches, 1997; Kim et al., 1998; Prevost & Wilkerson, 1999; McGregor et al., 2000; Isenegger et al., 2001; Bornet et al., 2002; Coombs et al., 2004; Braun & Wenzel, 2005; Hale et al., 2005; Barandalla et al. 2006; Mathias et al., 2007; Reid & Kerr 2007; Fu et al., 2009; Gavrilenko et al., 2010; Karaagac et al., 2010; Reid et al., 2011; Seibt et al., 2012; Karaagac et al., 2014). When used in combination with easily scored morphological traits of market classes of tubers (e.g., Schneider & Douches, 1997), many of these markers serve as effective discriminators of most cultivated varieties. Currently, the hypervariability and well-screened database of nuclear microsatellites (e.g., Ghislain et al, 2009a) make these ideal markers for fingerprinting applications. However, the continuous drop of sequencing cost is leading to the development of sequence-based markers.
Germplasm Utilization and Contributions to Cultivar Improvement
Even though potato breeders have been experimenting with the introduction of wild relatives into their programs for 150 years, the genetic diversity within and among major cultivars remains low (Mendoza & Haynes, 1974a; Wang et al., 2011). While the introgression of a few specific genes from wild species has had a significant impact on cultivar development, only a handful of species have been used extensively (Bradshaw et al., 2006a). These include S. acaule, S. chacoense, S. demissum, S. spegazzinii (=S. brevicaule), S. stoloniferum, and S. vernei, mainly as sources of major genes for resistance to late blight, viruses, and potato cyst nematodes. Although not developed as a processing cultivar, 'Lenape' containing genes from S. chacoense, is in the pedigree of many modem chip cultivars and is credited with contributing to major advances in breeding for chip quality in the late twentieth century (Love et al., 1998). 'Lenape' was removed from the market, however, due to excessive levels of glycoalkaloids in its tubers, likely coming from S. chacoense (Zitnak & Johnston, 1970). This example illustrates the need for germplasm enhancement programs to carry out comprehensive evaluations of their products to avoid the inclusion of undesirable properties in eventual varieties.
To broaden the genetic base of the common potato gene pool and to combine different resistance genes introgressed from wild potato species, various methods have been used, including ploidy manipulations and bridge crosses, embryo rescue, hormone treatments, reciprocal crosses and protoplast fusion (Jansky, 2006). The range of sexual hybridization has been broadened using biotechnological methods that allowed the use of new species that have never been used before in breeding programs such as species outside of section Petota (S. etuberosum, S. palustre, S. nigrum), and species in section Petota but in a clade distantly related to cultivated potatoes (5. tarnii, S. cardiophyllum, S. bulbocastanum (review in Gavrilenko, 2007).
Direct genetic transfer of genes from wild potatoes into existing widely-adopted varieties is another tool available to breeders. The current development of late blight resistant varieties using genes from S. bulbocastanum is an example of the shortcut breeders can use instead of the 45 years it took from the original bridge crosses between cultivated and wild species to introduce useful traits (Haverkort et al., 2009). However, public resistance to genetically modified organisms is still delaying full exploitation of such direct transfer.
Breeding tetraploid potatoes is a challenge due to the heterozygous nature of parents used in breeding. Selfing produces severe inbreeding depression in potatoes and has impeded the elimination of unfavorable alleles and the fixation of alleles responsible for important traits after 150 years of potato breeding. Wild potatoes can be a genetic source of a self-compatible breeding system that can revolutionize potato breeding through the development of hybrid potatoes from diploid inbred lines (Lindhout et al., 2011).
Breeding Potential of Polyploids
Fifty years ago, a sexual polyploidization breeding strategy was proposed to increase yield in tetraploid potatoes by maximizing heterozygosity (Chase, 1963b). Since then, a number of studies have revealed the contributions of inter- and intra-allelic interactions to yield in potatoes (Rowe, 1967b; Mendoza & Haynes, 1974b; Mendiburu & Peloquin, 1977; Peloquin, 1983; Carputo et al., 2000a). Because an autotetraploid can theoretically carry four alleles per locus, the number of combinations within a gene and in epistatic interactions among genes is much higher than can be achieved in diploids. However, recent genomic studies have revealed that tri-allelic and tetra-allelic loci are rare in potato cultivars (Hirsch et al., 2013). In addition, evidence for a heterosis threshold has been published (Sanford & Hanneman, 1982). Three-way hybrids were never superior to two-way hybrids for vigor or yield. Rather than overall heterozygosity, the presence of certain alleles may be more important for high yield (Bonierbale et al., 1993). In addition, genes that contribute to yield are predominantly located near centromeres, where recombination is limited (Buso et al., 1999b). This likely limits advances through conventional breeding methods. So, despite optimism that yield gains from maximizing heterozygosity would be realized via new tetraploid breeding strategies, yield has remained steady for the past century (Douches et al., 1996; Jansky et al., 2013). However, improvements in market traits, such as processing quality and disease resistance have been realized. Consequently, a major concept in potato breeding, that intra-locus interactions in general are required in competitive cultivars, must be reevaluated.
Polyploidy offers a strategy to maximize hybrid vigor (Chen, 2010) and is a major mechanism in potato evolution. Two options are available to bring diploids to the tetraploid level. First, they can be somatically doubled through chemical means such as colchicine (Ross et al., 1967) or through tissue culture (Sonnino et al., 1988). However, somatically doubled diploids exhibit slower growth rates, later maturity, reduced vigor, lower yield, and lower fertility than their diploid counterparts. In one study, yield of the diploid clones was nearly twice that of the tetraploids, mainly due to high tuber number (Rowe, 1967a). In another study, when diploids were crossed with each other and then their somatically doubled counterparts were intercrossed, the highest yielding clones included 27 tetraploid and 13 diploid individuals (Rowe, 1967b).
An alternative method to double chromosome number is through sexual polyploidization using 2n gametes (Chase, 1963a). Unilateral sexual polyploidization results from polyploidization of one parent, while the other parent is already at the polyploid level (4x female x2x male, or 2x female X4x male to produce 4x offspring). Bilateral sexual polyploidization results from polyploidization of both parents (2x*2x to produce 4x offspring). Triploid offspring are not produced from these crosses due to endosperm failure. Sexual polyploidization can produce three types of heterozygotes (simplex-Aaaa, duplex-AAaa, and triplex-AAAa) and up to four alleles per locus. Complex combinations of triallelic (AiA2A3A3) and tetraallelic (AiA2A3A4) loci also can be produced. In addition, sexual polyploidization produces a wide array of complex epistatic (interlocus) interactions.
Initial studies of sexual polyploidization in potatoes focused on tuber yield and quality. Yield heterosis is common following unilateral sexual polyploidization in which the tetraploid female parent is typically a potato cultivar or advanced breeding selection and the diploid male parent is a dihaploid x wild species hybrid or a cultivated diploid x dihaploid hybrid (De Jong et al., 1981; Bani-Aameur et al, 1991; De Jong& Tai, 1991; Buso et al., 1999a; Alberino et al., 2004). Unlike somatic doubling, sexual polyploidization transmits a large proportion of heterozygous loci and epistatic interactions to the tetraploid offspring (Peloquin et al., 2008). This allelic diversity likely buffers against environmental variability, leading to yield stability (Darmo & Peloquin, 1990; Ortiz et al., 1991). Sexual polyploidization has been used to transfer many additional traits to tetraploid offspring, including abiotic stress tolerance (Sterrett et al., 2003), processing quality (De Jong & Tai, 1991; Hutten et al., 1996; Hayes & Thill, 2002; Jansky et al., 2011) and disease resistance (DeMaine et al. 1986; Herriott et al., 1990; Watanabe et al., 1992; Carputo et al., 2000b; Capo et al., 2002; Frost et al., 2006).
Bilateral sexual polyploidization provides an alternative sexual polyploidization mechanism. In this scenario, both parents are diploid and produce 2n gametes. The potential advantage of bilateral sexual polyploidization is that highly heterotic offspring can be produced by hybridization between diverse diploid parents. The disadvantage is that both parents must produce 2n gametes. Tetraploid progeny from bilateral sexual polyploidization are highly heterotic and typically out-yield their diploid full-sibs (Mendiburu & Peloquin, 1977; Sanford & Hanneman, 1982; Hutten et al., 1995a) and even tetraploid commercial cultivars (Werner & Peloquin, 1991b). The yield gains from bilateral sexual polyploidization are typically higher than those from unilateral sexual polyploidization, presumably due to the contributions of heterozygosity from both parents (Werner & Peloquin, 1991b).
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||p. 326-356|
|Author:||Spooner, David M.; Ghislain, Marc; Simon, Reinhard; Jansky, Shelley H.; Gavrilenko, Tatjana|
|Publication:||The Botanical Review|
|Date:||Dec 1, 2014|
|Previous Article:||Systematics, diversity, genetics, and evolution of wild and cultivated potatoes.|
|Next Article:||Systematics, diversity, genetics, and evolution of wild and cultivated potatoes.|