Solanum candolleanum Berthault

The Solanum brevicaule complex has long attracted the attention of biologists because of its close relationships and breeding value to cultivated potatoes, and the perplexing taxonomic problems associated with its circumscription (Correll 1962; Grun 1990; Ugent 1970). As described below, we here recognize two species in this complex, one from the “northern” part (as S. candolleanum in central and southern Peru, barely into northwestern Bolivia), and the other from the “southern” part (S. brevicaule from northern Bolivia to central Argentina). Some members of this complex, endemic to central Peru, Bolivia, and northern Argentina, were considered ancestors of the landraces of the cultivated potato (Ugent, 1970). The species of the complex share pinnately dissected leaves, round fruits, rotate to rotate-pentagonal corollas, and are largely sexually compatible with each other and with cultivated potato (Hawkes, 1958; Hawkes & Hjerting 1969, 1989; Ochoa 1990, 1999). Solanum candolleanum is exclusively diploid, and S. brevicaule includes diploids, tetraploids, and hexaploids, with some species recognized by Hawkes (1990) possessing multiple ploidy levels (S. gourlayi with diploids and tetraploids; and S. oplocense with diploids, tetraploids, and hexaploids). Members of the complex are so similar that even experienced potato taxonomists Hawkes & Hjerting (1989) and Ochoa (1990) provided different identifications for identical collection numbers of the S. brevicaule complex in fully 38% of the cases (Spooner et al., 1994).

Field studies in Argentina (Spooner & Clausen 1993), Bolivia (Spooner et al. 1994), and Peru (Spooner et al. 1999; Salas et al. 2001), phenetic analyses of morphological data from common garden studies in the United States (van den Berg et al. 1998) and Peru (Alvarez et al. 2008), single- to low-copy nuclear restriction fragment length polymorphism and random amplified fragment length (RAPD) data (Miller & Spooner 1999), and amplified fragment length polymorphism (AFLP) data (Spooner et al. 2005) failed to clearly differentiate many wild species in the complex, but defined two geographic subsets: (1) a “northern” group composed of the Peruvian populations (to including S. achacacense and S. candolleanum from extreme western Bolivia near the Peruvian border), (2) a “southern” group of Bolivian and Argentinean populations. In all data sets, however, even these two groups could be distinguished only by computer-assisted analysis of widely overlapping character states, and not by species-specific diagnostic characters. The AFLP data provided the best species-specific support, with all accessions of S. candolleanum of the northern S. brevicaule complex forming a clade. Multiple accessions of S. avilesii, S. hoopesii, S. incamayoense, S. spegazzinii, S. ugentii, and S. vidaurrei also formed clades, but support for each was low, with bootstrap values below 50%, and was not concordant with the morphological results that could not reliably distinguish the species.

An exception was S. spegazzinii, the southern-most representative of the southern S. brevicaule complex that showed all eight examined accessions forming a monophyletic group with 74% bootstrap value using AFLPs (Spooner et al. 2005). Solanum spegazzinii has never been questioned as a valid species and can sometimes be differentiated from other members of the southern members of the S. brevicaule complex by the presence of a larger number of interjected leaflets (5-15) and generally by five subequal pairs of lateral leaflets, but these are very inconstant features. Furthermore, S. spegazzinii was indistinguishable by morphological? phenetic studies (van den Berg et al. 1998).

An opposite problem was S. oplocense, that was somewhat distinct using morphological data from plants grown in the United States (van den Berg et al. 1998), but not from plants grown in Peru (Alvarez et al. 2008) nor with AFLP data (van den Berg et al. 1998). Morphologically many populations can be defined by finely denticulate leaf margins, glaucous leaves with frequently purple leaf nerves, and a long exserted style. But these features are widely scattered through the complex and the maintenance of this species would result in unstable identifications.

Brücher (1964) considered S. setulosistylum as identical with S. chacoense and not a hybrid of S. chacoense with S. leptophyes or S. spegazzinii, as stated by Hawkes (1956). Brücher (1964) also stated that S. setulosistylum was not related to S. puberulofructum as assumed by Correll (1962). Hawkes and Hjerting (1969) considered S. setulosistylum and S. puberulofructum to be hybrids between S. chacoense Bitter and S. spegazzinii. Brücher stated that it was not possible to consider S. sleumeri and S. puberulofructum as different species. We have analyzed many collections identified as these species in the literature or on herbarium sheets and consider that they are synonymous with our broad definition of S. brevicaule. One of the collections of S. setulosistylum cited by Correll (Brücher 534) has a pubescent berry, a character thought to distinguish S. puberulofructum.

Populations identified as Solanum spegazzinii are extremely variable. Ispizúa (1994) studied specimens from much of the range of the group and documented three somewhat intergrading morphotypes based on multivariate analysis of 42 morphological characters: 1) populations from Tafí del Valle in Tucumán Province and in central Salta Province near Rosario de Lerma and Chicoana, 2) populations along the length of Quebrada Calchaquí from eastern Catamarca Province near Santa Maria north to Cachi and La Poma in central Salta Province, 3) populations in south-central Catamarca Province around Belén. Specimens from area 1 are characterized by relatively narrow leaflets. Specimens from areas 2 and 3 have wider leaflets and are similar morphologically except for the dense pubescence of leaves and fruits that characterize specimens from area 2. The type of the densely pubescent S. puberulofructum [that we synonymize here], however, is found in area 3 where no pubescent collections were found.

Erazzú et al. (1999) made crosses within and among different accessions of S. spegazzinii from these three areas, and assessed reproductive success by seed set and observation of pollen tube inhibition in stigmas and styles. They found no crossing barriers within accessions from area 1, but sometimes no but also intermediate to high crossing barriers within areas 2 and 3 and crosses among accessions of the three areas. They attributed these crossing barriers to interspecific hybridization with other sympatric species such as S. acaule, S. chacoense, S. berthaultii [as S. tarijense], S. sanctae-rosae, and S. vernei, and suggest that these species are maintained as distinct by eventual selection against maladapted hybrid combinations.

We entered into our monographic studies influenced by all of these results, and the AFLP data led us to consider recognizing S. candolleanum as a northern Bolivian representative of the complex, and S. avilesii, S. hoopesii, S. incamayoense, S. spegazzinii, S. ugentii, and S. vidaurrei as southern representatives. However, like the morphological results of living representatives in common gardens (van den Berg et al. 1998; Alvarez et al. 2008), we found it impossible to identify many collections of these southern “species” with herbarium specimen data. The only way we could provide species-level identifications was to use determinations of prior taxonomists (which differed considerably, see above), or inferences from distributional data. We conclude that populations of the complex are sometimes partly differentiated into phenetically distinguishable units, but that clear boundaries among taxa are confounded by recent and only partial species divergence, introgression, hybridization, and allopolyploidy (Spooner et al., 2008; Rodríguez & Spooner, 2009). We also conclude that molecular marker support for some taxa (as seen in the case of S. spegazzinii) signals divergence through geographical isolation, but that this is not yet accompanied by traits needed for practical species recognition such as morphological divergence or breeding barriers.

Despite our extensive synonymy based on the impracticality of morphological data to distinguish these species, and of low statistical support with the AFLP data, we recognize the northern (S. candolleanum) and southern (S. brevicaule) components of the S. brevicaule complex as distinct species, despite their extreme similarity, because all morphological and molecular data sets support them. In a practical sense they can be distinguished largely by geography. Our key is our best attempt to separate them based on insights from the replicated field trials (van den Berg et al. 1998; Alvarez et al. 2008) and herbarium specimens, but this will often fail.

The extensive synonymy we propose for S. candolleanum and S. brevicaule may appear extreme to those accustomed to these long-accepted species names, but was at least partly expected by us based on our prior studies. We suspect that our synonymy will be questioned by those using molecular marker data, but we suggest that such workers wishing to resurrect our synonymized names do so only with morphological data needed to support the identification of these species in realistic and practical ways. Without such data, we envision the perpetuation of alternative taxonomies, variant identifications by different workers, and the maintenance of unstable names and taxonomic confusion in these important but taxonomically confusing species related to the cultivated potato.

Mandon 397, the type collection of Solanum mandonii A.DC., is a mixed gathering of S. candolleanum and S. tuberosum. Ochoa (1999) discusses the identities of the various duplicates in detail, and has effectively lectotypified this species with the duplicate in P [P00324973], the other duplicate of Mandon 397 at P [P00384634] is a specimen of S. tuberosum, according both to Correll (on annotation slip) and Ochoa (1999).