1839
American Journal of Botany 96(10): 1839–1848. 2009.
Polyploidy has played an important role in plant evolution in
general and has been especially important in crop species. In
addition to the challenges of untangling the often complicated
history of polyploid origins, the study of the origins of some
polyploids involves molecular comparisons of closely related
species, among which there may be little variation in DNA se-
quences. As an example, we here present results of a study
comparing AFLP data of an understudied domesticated crop
with four potential progenitor candidates.
Oxalis tuberosa Molina, commonly known as oca, is primar-
ily cultivated in the central Andes. Like many other “ underuti-
lized ” crops, it plays an important role in the food security of
rural communities ( Pastor, et al., 2008 ). It originated from
within a clade informally known as the “ Oxalis tuberosa alli-
ance, ” a group of several dozen morphologically similar spe-
cies found through the central and northern Andes ( Emshwiller,
2002a ). Cultivated oca was found to be octoploid in most stud-
ies, but the majority of the wild species in the alliance clade
are diploid, with only a few wild polyploids (reviewed in
Emshwiller, 2002b ). Those alliance species for which there are
published chromosome counts share a base chromosome num-
ber rare in Oxalis ( x = 8) (e.g., de Azkue and Mart í nez, 1990 ).
The majority of the alliance species do not form tubers, but
wild, tuber-bearing populations have been found in four geo-
graphic areas from central Peru to northwestern Argentina
( Fig. 1 ) , and these seem to comprise four species (as discussed
later). However, only two of the four taxa have ever been de-
scribed as species (see Discussion section), i.e., O. picchensis R.
Knuth from southern Peru and O. chicligastensis R. Knuth from
northwestern Argentina ( Figs. 1 and 2 ). The other two wild, tu-
ber-bearing Oxalis taxa, which are as yet unnamed, will simply
be referred to in this paper by shorthand designators: LimaW/T
for the wild, tuber-bearing Oxalis populations found in the west-
ern slopes of the Andes in Lima Department, Peru, and BolW/T
for the wild, tuber-bearing Oxalis populations found on the east-
ern side of the Andes in Bolivia ( Figs. 1 and 2 ).
Although generally considered to be a single species, oca
may comprise two molecularly distinct groups. Ethnobotanical
studies (E. Emshwiller, unpublished data) in the district of Pisac
in Cusco Department, Peru, in 1997 found that traditional
Quechua-speaking farmers in that area distinguished two use-
categories of oca: (1) the several sweeter folk cultivars of the
use-category wayk ’ u were usually cooked fresh after a few days
to sweeten in the sun, and (2) the sour khaya use-category, with
only the folk cultivar P ’ osqo, was cultivated separately and
used exclusively for processing into khaya, dried tubers that
can be preserved for years. Preliminary AFLP data using a sin-
gle primer combination found that these two use-categories
formed separate clusters in the results of neighbor-joining anal-
ysis ( Emshwiller, 2006a ), suggesting that their evolutionary

1


Manuscript received 22 October 2008; revision accepted 2 July 2009.
The authors thank C. An é for help with NMDS analysis, M. Wegener for
help with mapping in ArcGIS, K. Elliot for help with fi gures, and the
following people for help with collection of samples of wild, tuber-bearing
taxa from 1995 to 2006: D. Avila, J. G. L ó pez, R. Alberco, C. Gir ó n, M. L.
Ugarte, S. Guam á n, T. Villarroel, J. Almanza, G. Meza, P. Cruz, A. Castelo,
H. Flores, R. Estrada, R. Ortega, C. Arbizu, M. Ram í rez, and A. Andia.
This study was supported by NSF DEB-0732490 (originally DEB-
0426496).

6


Author for correspondence (e-mail: emshwiller@wisc.edu)
doi:10.3732/ajb.0800359
ORIGINS OF DOMESTICATION AND POLYPLOIDY IN OCA ( OXALIS
TUBEROSA ; OXALIDACEAE). 3. AFLP DATA OF OCA AND FOUR
WILD, TUBER-BEARING TAXA 1
Eve Emshwiller, 2,6 Terra Theim, 2 Alfredo Grau, 3 Victor Nina, 4 and
Franz Terrazas 5

2
University of Wisconsin-Madison, Botany Department/Birge Hall, 430 Lincoln Drive, Madison, Wisconsin 53706-1381 USA;

3
Laboratorio de Investigaciones Ecol ó gicas de las Yungas (LIEY), Facultad de Ciencias Naturales, Universidad Nacional de
Tucum á n, CC 34, Yerba Buena 4107, Tucum á n, Argentina; 4 Instituto Nacional de Investigaci ó n Agraria, Estaci ó n Experimental
Andenes, Av. Micaela Bastidas # 310-314, Wanchaq, Cusco, Peru; and 5 Fundaci ó n PROINPA, Av. Meneces s/n Km.4 (El Paso),
Cochabamba, Bolivia
Many crops are polyploids, and it can be challenging to untangle the often complicated history of their origins of domestication
and origins of polyploidy. To complement other studies of the origins of polyploidy of the octoploid tuber crop oca ( Oxalis tube-
rosa ) that used DNA sequence data and phylogenetic methods, we here compared AFLP data for oca with four wild, tuber-bearing
Oxalis taxa found in different regions of the central Andes. Results confi rmed the divergence of two use-categories of cultivated
oca that indigenous farmers use for different purposes, suggesting the possibility that they might have had separate origins of do-
mestication. Despite previous results with nuclear-encoded, chloroplast-expressed glutamine synthetase suggesting that O. pic-
chensis might be a progenitor of oca, AFLP data of this species, as well as different populations of wild, tuber-bearing Oxalis
found in Lima Department, Peru, were relatively divergent from O. tuberosa . Results from all analytical methods suggested that
the unnamed wild, tuber-bearing Oxalis found in Bolivia and O. chicligastensis in NW Argentina are the best candidates as the
genome donors for polyploid O. tuberosa , but the results were somewhat equivocal about which of these two taxa is the more
strongly supported as oca ’ s progenitor.
Key words: AFLP; Andean crops; domestication; neighbor-joining; nonmetric multidimensional scaling; Oxalidaceae; Oxa-
lis tuberosa ; polyploidy.

1840 American Journal of Botany [Vol. 96
Fig. 1. Map of known distributions of the four wild, tuber-bearing Oxalis taxa included in this study, within the range of the Oxalis tuberosa alliance.
More localities are indicated than were sampled in this study. Images of plant habit are at the same scale, and all images of the tubers are at another scale.
Yellow diamonds: localities of the wild, tuber-bearing taxon of Lima Department, Peru (LimaW/T); green squares: O. picchensis ; pink triangles: wild,
tuber-bearing taxon of Bolivia (BolW/T), including specimen EE284; blue triangles: O. chicligastensis ; black circles: other wild species (nontuber-bearing)
in the Oxalis tuberosa alliance clade.

1841October 2009] Emshwiller et al. — AFLP and polyploid origins of OXALIS TUBEROSA
gin of polyploidy of O. tuberosa . Specifi cally, this study uses
analyses of a single AFLP data set to: (1) determine whether the
wild, tuber-bearing Oxalis populations from four geographic
areas are separate taxa and whether the two use-categories of
oca are molecularly distinct groups; (2) assess repeatability
when comparing AFLP data with morphotypes recognized in
folk taxonomy, when using a different, but overlapping, sample
of individual oca plants and an increase in AFLP primer combi-
nations compared to the previous study ( Emshwiller, 2006a );
and (3) assess the evidence that each of the wild taxa might
have been a genome donor of octoploid oca.
MATERIALS AND METHODS
Sampling strategy — AFLP data were generated for 3 – 5 individuals of each
of the four wild, tuber-bearing Oxalis taxa (Table 1), and 57 cultivated O. tu-
berosa from Cusco Department ( Table 2 ). In some cases, the populations of the
wild, tuber-bearing Oxalis sampled are the only ones that are known (e.g.,
LimaW/T). The samples of BolW/T included typical wild plants as well as an
individual, EE284, that was found near an area where cultivated oca was grown,
and which was more robust than the others. The sample of cultivated oca in-
cluded both use-categories: 10 were P ’ osqo, the traditional cultivar used exclu-
sively for khaya, and 47 were other folk cultivars of the wayk ’ u use-category.
To assess repeatability, we resampled 20 of the tubers that had been included in
the previous study, all of them from Pisac District in Calca Province of Cusco
Department, Peru ( Emshwiller, 2006a ). Additional samples were selected from
two nearby areas in Cusco Department ( Table 2 ), which were each less than 50
km straight line distance from the communities that had been collected in 1997,
close enough that they would share some of the same cultivars. Some of these
newer samples appeared similar enough that they probably belong to the same
clonal genotype as those in the prior study, so these samples provided further
tests of congruence of AFLP and morphological data.
Molecular methods — Fluorescent amplifi ed fragment length polymorphism
(AFLP) data were generated for seven primer combinations following Myburg
et al. (2001) with minor modifi cations as follows. DNA isolations used either
DNeasy columns (Qiagen, Carlsbad, California, USA) or CTAB extractions
(Doyle and Doyle, 1990), that were subsequently cleaned using a modifi cation
of the Alexander et al. (2007) protocol after it was determined that the CTAB
extractions did not yield DNA of suffi cient purity for AFLP. Genomic DNA
was digested for 2 h at 37 ° C in a 5- μ L reaction with approximately 84 ng DNA,
0.05 μ L of 100 ng/100 μ L BSA, 5 U of EcoRI, and 5 U of MseI (both New
England BioLabs, Beverly, Massachusetts, USA). Adapters and primers (made
by Integrated DNA Technologies, Coralville, Iowa, USA) used sequences of
Vos, et al. (1995) . Adapters were ligated to the ends of the digested DNA frag-
ments immediately after digestion. The ligation reactions were incubated at
16 ° C for 14 h, and included the following reagents in each 10- μ L volume: 5 μ L
digestion product, 3.6 μ L of double-distilled (dd) H 2 O, 1 μ L of 10 × ligase buf-
fer, 0.19 μ L of 50 μ M double-stranded EcoRI adapter, 0.19 μ L of 50 μ M dou-
ble-stranded MseI adapter, and 40 U of T4 ligase (New England BioLabs).
Ligation products were diluted with ddH 2 O (1 : 5) and then used in the fi rst
round of amplifi cation. These preselective amplifi cations included the follow-
ing reagents in a 25- μ L volume: 2.5 μ L of 10 × buffer, 1.5 μ L of 25 mM MgCl 2 ,
2 μ L of 2.5 mM (each) dNTP, 0.38 μ L of 20 μ M of each primer, which had only
one selective base (C on the MseI primer) or none (EcoRI primer), 1.25 U Taq
polymerase, and 5 μ L diluted ligation product. The thermal cycling protocol
included an initial incubation at 72 ° C for 60 s; 20 cycles of 94 ° C for 50 s, 56 ° C
for 60 s, 72 ° C for 120 s; and a fi nal 72 ° C for 120 s. The resultant product was
diluted in ddH 2 O (1 : 19) to prepare for the fi nal, selective amplifi cation step. The
seven selective primer combinations had two or three selective bases as follows
(indicated by EcoRI selective bases/MseI selective bases): AC/CAC, TC/CTG,
AG/CTG, GC/CGA, AGC/CAT, GC/CTC, ATT/CGA (the fi rst combination is
same as the single AFLP primer pair used in Emshwiller, 2006a ). Selective
amplifi cations included the following in a 25- μ L reaction: 9.25 μ L of ddH 2 O,
2.5 μ L of 10 × buffer, 1.5 μ L of 25 mM MgCl 2 , 3.0 μ L of 2.5 mM (each) dNTP,
0.5 μ L of deionized formamide (Hi-Di, Applied Biosystems, Foster City, Cali-
fornia), 2.5 μ L of 10 μ M MseI primer, 0.5 μ L of 10 μ M EcoRI primer (labeled
with 5 ′ 6-FAM fl uorescent tag), 1.25 U Taq polymerase, and 5 μ L of diluted
preselective amplifi cation product. Thermal cycling protocol included 9 cycles
histories may differ, perhaps in their origins of polyploidy and/
or origins of domestication.
Among the two loci used in previous studies aimed at under-
standing the origins of polyploidy and domestication of oca,
DNA sequence data of chloroplast-expressed (but nuclear-en-
coded) glutamine synthetase (ncpGS) provided more variation
than did the internal transcribed spacer of nuclear ribosomal
DNA (ITS) ( Emshwiller and Doyle, 2002 ). The different se-
quence classes of ncpGS within individual plants of oca and the
Bolivian wild, tuber-bearing Oxalis (BolW/T) were separated
by molecular cloning for use in phylogenetic analyses. Fixed
heterozygosity and separate placement of oca ’ s sequence
classes on the ncpGS gene tree suggested that these classes rep-
resent homeologous loci and that oca is an allopolyploid and
possibly an autoallopolyploid. The ncpGS results identifi ed
both wild, tuber-bearing taxa included in that study, O. pic-
chensis and BolW/T, as progenitor candidates, leading to the
working hypothesis that these two Oxalis taxa may have hy-
bridized to form cultivated oca ( Emshwiller and Doyle, 2002 ;
Emshwiller, 2006b ; Zeder et al., 2006 ).
Two observations suggested that oca ’ s origins might be more
complex than the simple hybridization scenario just presented.
First, the O. picchensis -like ncpGS sequence was absent from
one of the nine individual plants sampled, which might have
had any of several explanations, including the possibilities of
introgression, separate origins of polyploidy, or loss of that se-
quence type after polyploidization, among others (discussed in
Emshwiller and Doyle, 2002 , p. 1054). Second, preliminary
AFLP data of O. picchensis showed few markers that were
shared with all oca samples, contrary to expectations for the
progenitor of a polyploid (E. Emshwiller, unpublished data). In
addition, although the two wild, tuber-bearing taxa sampled in
the prior studies with ITS and ncpGS (i.e., O. picchensis and
BolW/T), were the only such tuber-bearing taxa known at the
time, that is no longer the case. Populations of two different
wild, tuber-bearing Oxalis taxa have since been found in
Tucum á n, Argentina ( O. chicligastensis ), and in Lima Depart-
ment, Peru (LimaW/T). The preliminary AFLP data and the
fi nding of additional tuber-bearing, wild Oxalis populations
both indicated the need to revisit the story of oca ’ s origins, us-
ing a combination of different sources of data.
Polymorphic markers that have been used in evolutionary
studies of Oxalis have included AFLP ( Tosto and Hopp, 2000 ;
Emshwiller, 2006a ) and ISSR ( Malice et al., 2007 ; Pissard et
al., 2006 , 2008a, b) in O. tuberosa , and microsatellites in O.
alpina (Tsyusko et al., 2007). AFLP data have been used in
studies of variation among oca and some wild diploid species
( Tosto and Hopp, 2000 ), in a comparison between AFLP data
and the folk taxonomy of oca traditional cultivars ( Emshwiller,
2006a ), as well as in studies of intraspecifi c variation within
cultivated oca ( Adrianz é n, 2 006; Biondi, 2006; Zorilla, 2006;
Schibli, 2007; K. Vivanco, Universidad Nacional Agraria La
Molina, unpublished manuscript, W. Cruz, Universidad Nacio-
nal Federico Villarreal, unpublished manuscript). An ongoing
international collaborative project is using AFLP data of a much
larger sampling of cultivated oca from throughout the Peruvian
Andes to study the distributions of clonal genotypes of culti-
vated oca and the effect of exchange networks among farmers
on the genetic structure of the crop.
Here we apply AFLP data to compare the four known wild,
tuber-bearing Oxalis taxa in the O. tuberosa alliance with a se-
lection of cultivated oca samples, to clarify whether they are
distinct from each other and whether they had a role in the ori-
¬

1842 American Journal of Botany [Vol. 96
or polymorphic among the sampled oca and wild, tuber-bearing Oxalis
populations (data not shown). We hoped to fi nd peaks diagnostic for each of the
four wild, tuber-bearing Oxalis taxa (i.e., that were fi xed in that particular taxon
and absent from the other three wild taxa), which were also fi xed in one or both
use-categories of oca (our “ gold standard ” ). However, very few gold-standard
peaks were found, and expansion of the criteria to include polymorphic peaks
also yielded equivocal results, so the peak sharing evaluation results will only
be discussed briefl y.
RESULTS
Nonmetric multidimensional scaling — Each of the wild,
tuber-bearing Oxalis populations from different geographic
areas separated from each other in the plot of the fi rst two
of 94 ° C for 50 s, 65 ° C for 60 s (decrease 1 ° C per cycle), 72 ° C for 120 s; fol-
lowed by 20 cycles of 94 ° C for 50 s, 56 ° C for 60 s, 72 ° C for 120 s; and a fi nal
incubation at 72 ° C for 10 min. Amplifi cation products were run on an ABI
3730XL capillary sequencer at the University of Wisconsin Biotechnology
Center with fl uorescently labeled lane standard (500ROX, Applied Biosys-
tems). Chromatograms were visualized using GeneMarker (Softgenetics,
State College, Pennsylvania, USA) and scored for presence/absence of
fragments.
Data analyses — The single binary data set representing presence or absence
of 399 AFLP peaks in each sampled individual was analyzed by (1) nonmetric
multidimensional scaling (NMDS) analyses, based on Jaccard distances, per-
formed using the software package VEGAN for R ( Dixon, 2003 ); and (2)
neighbor-joining analysis (N-J), based on Nei and Li (1979) distances, per-
formed in the program PAUP* version 4.0b10 ( Swofford, 2002 ). We also
evaluated AFLP peaks looking for those that were shared as either fi xed
Fig. 2. Comparative size of tubers of domesticated and wild Oxalis . The cultivated oca tubers shown here were a particularly diverse sample of tubers
grown by one family in Hu á nuco Department, Peru. On the bottom row are tubers of each of the wild, tuber-bearing Oxalis taxa studied here. Because of
our as-yet-limited sampling of some of these taxa, we make no claim that the tubers shown here are typical in size for the taxon.

1843October 2009] Emshwiller et al. — AFLP and polyploid origins of OXALIS TUBEROSA
dimensions in the results of nonmetric multidimensional scal-
ing analysis ( Fig. 3 ). The samples of the populations from
southern and central Peru, O. picchensis and LimaW/T, were
particularly well separated from each other and from any of the
other sampled populations. In comparison with those two taxa,
the samples of O. chicligastensis and those of BolW/T are not
as distantly separated from each other. Nonetheless, the sam-
pled accessions of the Bolivian and Argentinean populations do
not overlap with each other or with cultivated oca under current
sampling. There is also no overlap between the sampled plants
of the two use categories of cultivated oca, confi rming the pre-
vious observation of a distinct separation among the khaya and
the wayk ’ u categories with only one primer combination and a
smaller sample of cultivated oca accessions ( Emshwiller,
2006a ).
Neighbor-joining analysis — Similarly to the NMDS results,
the neighbor-joining results also show that the two Peruvian
wild taxa, O. picchensis and LimaW/T, are the most distantly
separated in their AFLP data among the sample groups included
here ( Fig. 4 ). There is less separation among cultivated oca,
BolW/T, and O. chicligastensis . Among these wild taxa of Bo-
livia and Argentina, most of the sampled plants from Bolivia
formed a cluster close to the cultivated oca accessions, with the
O. chicligastensis samples forming another cluster somewhat
more distantly from oca. However, one accession from Bolivia,
EE284, joined loosely with the O. chicligastensis samples in
this analysis. This particular plant was collected from an area
very close to fi elds of cultivated oca. Because of its proximity
to oca and the observation that it was more robust and had larger
tubers than most wild Oxalis , it was suspected to be either a
hybrid or an escape from cultivation. In the process of adding
AFLP data from more primer combinations, this sample re-
solved differently in different analyses, joining alternatively
with BolW/T, O. chicligastensis , or the P ’ osqo oca samples in
different analyses, but always linked only distantly from any of
these clusters (results not shown). We do not know the reason
for these changes in its position in the results, but one possibil-
ity might be that this plant was indeed of hybrid origin, although
not necessarily a hybrid with cultivated oca. Nonetheless, as
noted before, the Bolivian and Argentinean samples are non-
overlapping with each other or with cultivated oca in the NMDS
results.
Table 1. Voucher information for specimens of wild, tuber-bearing
Oxalis taxa sampled for AFLP data.
Collection no. Taxon Country 1st political div. 2nd political div.
EE259 Oxalis sp. (BolW/T) Bolivia Cochabamba Ayopaya
EE260 Oxalis sp. (BolW/T) Bolivia Cochabamba Ayopaya
EE262 Oxalis sp. (BolW/T) Bolivia Cochabamba Ayopaya
EE284 Oxalis sp. (BolW/T) Bolivia Cochabamba Chapare
EE500 O. picchensis Peru Cusco Cusco
EE531 O. picchensis Peru Cusco Calca
EE1152 O. picchensis Peru Apurimac Cotabambas
EE1001 O. chicligastensis Argentina Tucum á n Chicligasta
EE1002 O. chicligastensis Argentina Tucum á n Chicligasta
EE1003 O. chicligastensis Argentina Tucum á n Chicligasta
EE1004a O. chicligastensis Argentina Tucum á n Chicligasta
EE1004b O. chicligastensis Argentina Tucum á n Chicligasta
EE1110 Oxalis sp. (LimaW/T) Peru Lima Huarochir í
EE1169 Oxalis sp. (LimaW/T) Peru Lima Canta
EE1185 Oxalis sp. (LimaW/T) Peru Lima Canta
Table 2. Collection information for specimens of cultivated oca, Oxalis
tuberosa Molina, sampled for AFLP data.
Collection no. Country Dept. Province District
Use-
category
97: 02 – 13 Peru Cusco Calca Pisac Khaya
97: 11 – 07 Peru Cusco Calca Pisac Khaya
97: 46 – 14 Peru Cusco Calca Pisac Khaya
97: 46 – 15 a Peru Cusco Calca Pisac Khaya a
97: 47 – 13 Peru Cusco Calca Pisac Khaya
05: 26 – 03 – 14 Peru Cusco Urubamba Ollantaytambo Khaya
05: 26 – 08 – 12 Peru Cusco Urubamba Ollantaytambo Khaya
05: 26 – 09 – 03 Peru Cusco Urubamba Ollantaytambo Khaya
05: 26 – 09 – 04 Peru Cusco Urubamba Ollantaytambo Khaya
05: 33 – 03 – 05 Peru Cusco Paucartambo Colquepata Khaya
05: 33 – 03 – 06 Peru Cusco Paucartambo Colquepata Khaya
97: 02 – 05 Peru Cusco Calca Pisac Wayk ’ u
97: 11 – 03 Peru Cusco Calca Pisac Wayk ’ u
97: 11 – 01 Peru Cusco Calca Pisac Wayk ’ u
97: 12 – 05 Peru Cusco Calca Pisac Wayk ’ u
97: 14 – 07 Peru Cusco Calca Pisac Wayk ’ u
97: 19 – 01 Peru Cusco Calca Pisac Wayk ’ u
97: 21 – 06 Peru Cusco Calca Pisac Wayk ’ u
97: 22 – 06 Peru Cusco Calca Pisac Wayk ’ u
97: 31 – 08 Peru Cusco Calca Pisac Wayk ’ u
97: 35 – 04 Peru Cusco Calca Pisac Wayk ’ u
97: 40 – 02 Peru Cusco Calca Pisac Wayk ’ u
97: 46 – 05 Peru Cusco Calca Pisac Wayk ’ u
97: 47 – 06 Peru Cusco Calca Pisac Wayk ’ u
97: 48 – 02 Peru Cusco Calca Pisac Wayk ’ u
97: 50 – 02 Peru Cusco Calca Pisac Wayk ’ u
05: 17 – 06 – 01 Peru Ayacucho Huanta Luricocha-
Huayllay
Wayk ’ u
05: 26 – 01 – 10 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 01 – 09 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 01 – 19 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 02 – 02 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 02 – 12 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 05 – 03 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 05 – 13 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 12 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 24 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 39 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 39 b Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 41 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 41 b Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 47 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 06 – 50 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 07 – 04 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 07 – 46 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 07 – 47 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 08 – 13 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 08 – 42 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 09 – 01 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 09 – 02 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 10 – 20 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 10 – 28 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 10 – 35 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 12 – 30 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 10 – 42 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 10 – 43 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 11 – 07 Peru Cusco Urubamba Ollantaytambo Wayk ’ u
05: 26 – 11 – 14 Peru Cusco Urubamba Ollantaytambo Wayk ’ u

a
This might have been a mixed-up tube. The plant was recorded as
khaya-type in the fi eld and it clustered with other khaya samples in the
previous results of N-J analysis with a single primer combination
(Emshwiller, 2006a). However, in the current N-J results, it grouped with
the wayk ’ u cluster.

b
These were replicate samples.

1844 American Journal of Botany [Vol. 96
DISCUSSION
Comparison of AFLP profi les among domesticated O. tube-
rosa of two use-categories and four wild, tuber-bearing Oxalis
taxa provides evidence regarding the distinctness of each of
these entities. In partial contrast with previous studies using
ncpGS, the AFLP results suggest that the wild taxa that are
most closely related to domesticated oca are those of Bolivia or
possibly Argentina, rather than those of Peru. This study pro-
vides an example of how different sources of data may lead to
different insights.
Insights from AFLP data regarding distinctness and tax-
onomy of wild, tuber-bearing Oxalis taxa — Two of the four
wild, tuber-bearing Oxalis taxa studied here were given spe-
cies-level names, O. picchensis R. Knuth and O. chicligasten-
sis R. Knuth, whereas the populations in Lima Department,
Peru, and those of Bolivia have never been named as distinct
taxa. None of the four taxa were considered distinct species in
the treatment of Oxalis subgenus Oxalis by Lourteig (2000) ,
but our current results have implications for their distinctive-
ness.
Lourteig (2000) considered the name O. picchensis to be a
synonym of O. petrophila R. Knuth, but these taxa differ in
both morphological and molecular data ( Emshwiller, 2002a ,
2006b ). The type specimen of O. picchensis was lost in the de-
struction of the Berlin herbarium, and the populations from
which it was collected were probably extirpated because the
type locality, Cerro Piccho, is now within a built-up area of the
expanding city of Cusco. However, this taxon can still be found
near Cusco city and elsewhere in Cusco and Apurimac depart-
ments. Despite the fact that Lourteig (2000) did not recognize
O. picchensis as a distinct species, she described the new spe-
cies O. apurimacensis Lourteig, for which she did not mention
underground structures. Examination of some of the specimens
listed in the protologue for this species suggested that she in-
cluded material that might actually be O. picchensis (E. Emsh-
willer, personal observations).
The samples of O. chicligastensis used in this study were col-
lected near the type locality in Chicligasta Department in
Tucum á n Province, Argentina. Although Lourteig (2000) con-
sidered this name to be a synonym of O. tuberosa , all the sam-
pled individuals of this taxon separate from all oca samples
tested here in the NMDS and N-J results, suggesting that under
current sampling this taxon appears to be distinct from oca.
The AFLP data presented here also suggest that the two un-
named wild, tuber-bearing Oxalis taxa are also distinct. AFLP
data suggest that the LimaW/T populations are clearly distinct
from O. tuberosa and should probably be considered a new spe-
cies. The molecular separation of the BolW/T populations from
cultivated oca is not as great, but under current sampling they
do not overlap in the NMDS ordination results.
Thus, our current results suggest that each of these four wild,
tuber-bearing Oxalis taxa are separate entities from each other
and from cultivated oca. The separation of O. chicligastensis
and the Bolivian taxon from each other and from cultivated oca
still needs to be further confi rmed, as a single, possibly hybrid,
individual (EE284) violated the separation of all these taxa in
the N-J results. Pending further study of species delimitations,
we provisionally retain both names, O. chicligastensis and O.
picchensis , as distinct species and also suggest that LimaW/T
and BolW/T are probably separate species as well. However,
the geographic sampling of the current study is limited. Samples
In agreement with the NMDS results, the two use-categories
that were recognized by the farming households in the district
of Pisac formed separate groups in distance analyses of AFLP
data. Not only did four P ’ osqo accessions that were resampled
from the Pisac sample of 1997 separate from the wayk ’ u variet-
ies, but additional tubers that appeared morphologically similar
to P ’ osqo tubers, that had been collected from areas not far from
Pisac District, also joined the same cluster with the Pisac P ’ osqo
tubers in the N-J results. One exception (no. 46-15) was prob-
ably a mix-up of DNA tubes, based on its morphology and also
its having joined the P ’ osqo cluster previously ( Emshwiller,
2006a ).
A secondary goal of this study was to test repeatability of
the method by comparing the current results with seven AFLP
primer combinations with previous results using a single
primer combination and somewhat different laboratory meth-
ods ( Emshwiller, 2006a ). In addition to testing whether the
two use-categories would still resolve as distinct clusters as
more data were added, we tested whether samples of each of
several wayk ’ u folk cultivars would cluster together. Repeat-
ability was good. The morphotypes sampled here, from Pisac
and from other provinces in Cusco Department, separated into
similar clusters as had previously been found with a single
AFLP primer combination ( Emshwiller, 2006a ) and were
congruent with the morphotypes distinguished by the local
farmers, with very few exceptions. Samples from Pisac joined
with morphologically similar samples as previously, although
the arrangements within clusters were not necessarily identi-
cal, and they also clustered with morphologically similar tu-
bers from nearby provinces. There were a few cases of
dissimilar tubers joining the same cluster, for which we do not
know whether or not they might involve mix-ups of samples
in the fi eld or laboratory.
Fig. 3. Results of nonmetric multidimensional scaling (NMDS) anal-
ysis of AFLP data from all four wild, tuber-bearing taxa and both use-cat-
egories (wayk ’ u and khaya) of cultivated oca. Yellow diamonds: wild,
tuber-bearing taxon of Lima Department, Peru (LimaW/T); green squares:
O. picchensis ; pink triangles: wild, tuber-bearing taxon of Bolivia
(BolW/T); turquoise-blue triangles: O. chicligastensis ; open red circles:
wayk ’ u oca cultivars; dark blue circles: the khaya oca cultivar P ’ osqo.

1845October 2009] Emshwiller et al. — AFLP and polyploid origins of OXALIS TUBEROSA
Fig. 4. Results of neighbor-joining (N-J) analysis of AFLP data from all four wild, tuber-bearing taxa and both use-categories (wayk ’ u and khaya) of
cultivated oca. Green branches: O. picchensis ; yellow branches: wild, tuber-bearing taxon of Lima Department, Peru (LimaW/T); pink branches: wild,
tuber-bearing taxon of Bolivia (BolW/T); turquoise-blue branches: O. chicligastensis ; red branches: wayk ’ u oca cultivars; darker blue branches: khaya oca
cultivar P ’ osqo. Bootstrap support from 10 000 replicates is shown only on the branches leading to these groups, not within them.

1846 American Journal of Botany [Vol. 96
in Emshwiller, 2002b ), but as of yet, there are no reports of
octoploid wild, tuber-bearing Oxalis . Flow cytometry data indi-
cated that O. picchensis is tetraploid ( Emshwiller, 2002b ), and
O. chicligastensis is probably tetraploid as well, based on a re-
port by Br ü cher (1969) that probably represents that taxon (re-
viewed in Emshwiller, 2006b ). The ploidy level of the Bolivian,
wild, tuber-bearing populations is unknown, but fi xed heterozy-
gosity for ncpGS suggests that they too are probably polyploid
( Emshwiller and Doyle, 2002 ). Ongoing fl ow cytometry studies
at the International Potato Center by Kelly Vivanco (unpub-
lished data) have tentatively found the unnamed populations
from Lima Department to be hexaploid, and some sampled
plants of P ’ osqo oca to be tetraploid, but those observations still
need to be confi rmed.
Expectations for polyploid origins — Although most culti-
vated oca is octoploid, and the wild, tuber-bearing Oxalis stud-
ied here are probably of lower ploidy levels, we do not think
that oca must necessarily be a hybrid between two of the taxa
studied here. There exist several other possibilities. One possi-
bility is that oca may have originated from autopolyploidization
of a wild allotetraploid, possibly one of the taxa studied here.
Another possibility is that one of the wild, tuber-bearing Oxalis
studied here hybridized with another taxon, which would not
itself have to be another tuber-bearing taxon and could even be
a diploid species. Testing all the 2 x species in the alliance with
AFLP was beyond the scope of this paper, but additional tuber-
less, diploid species in the alliance are being tested in work with
single-copy nuclear loci.
Implications of AFLP results for origins of octoploid O. tube-
rosa — AFLP data provide a broader sampling of the genome
than previous studies with ITS and ncpGS data, and the results
with AFLP led to different conclusions than the previous stud-
ies based on ncpGS data ( Emshwiller and Doyle, 2002 ). These
previous results suggested that O. picchensis of southern Peru
might be one of the genome donors of octoploid oca, because
the ncpGS sequence of that taxon was identical to one of the
sequence classes of oca. In contrast, the current results with
AFLP, regardless of analytical method, suggested that O. pic-
chensis does not fi t the pattern expected for a genome donor of
octoploid O. tuberosa . AFLP data of O. picchensis are the least
similar to oca (of either use-category) of the four wild, tuber-
bearing Oxalis . However, the possibility still exists that O. pic-
chensis might have had some role in the evolution of the
octoploid cultigen, perhaps through introgressive hybridization.
The LimaW/T populations are quite divergent from O. picchen-
sis and should probably be described as a new species. They are
likewise divergent enough from cultivated oca that it is unlikely
that they were a progenitor of the crop. Nonetheless, we will
continue to include both these taxa in studies that will use sin-
gle-copy nuclear loci to confi rm the origins of oca.
Compared to the two Peruvian wild taxa, the other two wild,
tuber-bearing Oxalis taxa, found in Bolivia and Argentina, are
more strongly supported by AFLP data as being progenitor can-
didates for octoploid oca, regardless of the analytical methods,
which all concur on this point. Although the latter two wild,
tuber-bearing taxa, BolW/T and O. chicligastensis, are rela-
tively close to each other and to O. tuberosa in both the N-J and
the NMDS results, they are non-overlapping in the ordination
results. In results of both these distance-based analyses, the
samples of the Bolivian populations are closer (more similar) to
cultivated oca than the O. chicligastensis samples. On the other
of LimaW/T are as yet known from only three populations,
and those of O. chicligastensis were collected within a very
limited area. Although the wild, tuber-bearing populations of
Peru and Bolivia appear to be geographically separated from
each other (E. Emshwiller, personal observations), additional
exploration is needed to determine whether those of Bolivia
(BolW/T) and northwestern Argentina ( O. chicligastensis ) are
geographically separated and to clarify whether these taxa are
truly distinct from each other and from cultivated oca.
Distinctness of wayk ’ u and khaya use-categories — Culti-
vated oca samples were selected to permit a re-examination of
some of the samples used in the previous study ( Emshwiller,
2006a ). Also included were oca tubers collected from other
communities in Cusco Department, that were suspected, based
on tuber morphology, of being examples of the same oca clones
as had been included in that study (see Table 2 ). The sampled
oca accessions formed two distinct and non-overlapping groups
in the NMDS results and joined two clusters in the N-J results,
corresponding to the two use-categories that are recognized by
traditional farmers in the district of Pisac in Cusco Department,
Peru. That is, the P ’ osqo cultivar used exclusively for khaya
always separated from the wayk ’ u cultivars in both NMDS and
N-J analyses, with the exception of one possible mix-up of
tubes. This separation reinforces the previous AFLP results of
an overlapping sample ( Emshwiller, 2006a ), suggesting that
these use-categories are molecularly distinct. This molecular
divergence has important implications for the study of the ori-
gins of oca, because it suggests that the use-categories have
some as-yet-unknown difference in their evolutionary histories,
which might possibly represent separate origins of domestica-
tion and/or separate origins of polyploidy.
Quechua and Aymara-speaking farmers in other areas of the
Andes also distinguish “ sweet ” oca folk cultivars from “ sour ”
or “ bitter ” varieties, much as they do with potatoes. However,
we do not yet know whether these categories correspond to the
wayk ’ u and khaya categories from the communities in Cusco
Department that were studied here and that were so divergent
from each other in our results. Ongoing studies will confi rm
whether this separation of use-categories is upheld in a sam-
pling of oca that has been collected from throughout the Peru-
vian Andes. Additional expeditions to collect more samples of
the wild taxa should help to clarify whether the divergence
among the use-categories indicates multiple domestications.
Nonetheless, the fi nding of molecular divergence among these
two use-categories is an example of how ethnobotanical infor-
mation can reveal evolutionary differences that might have
been overlooked otherwise.
With respect to the groupings of the wayk ’ u cultivars in the
N-J results, with the exception of a few accessions that joined
the “ wrong ” cluster, these results were quite congruent with
morphological groups of tubers from these three districts and
also with previous results ( Emshwiller, 2006a ) using a single
primer combination and a different, but overlapping, set of sam-
ples. These results indicate that these seven AFLP combinations
are suffi cient to distinguish clones of oca and that the results are
reliable and reasonably repeatable across different datasets.
Origins of polyploidy in Oxalis tuberosa — State of knowl-
edge of ploidy levels of the sampled taxa — Information on
ploidy levels is important for the interpretation of the current
results, but it is incomplete for the wild, tuber-bearing Oxalis
taxa included here. Most cultivated oca are octoploid (reviewed

1847October 2009] Emshwiller et al. — AFLP and polyploid origins of OXALIS TUBEROSA
populations to improve the geographic representation of these
two taxa, and will also include other sources of data. These fu-
ture studies will assess whether these taxa are indeed separate,
distinct taxa and further test whether they were genome donors
of octoploid oca, and will also investigate whether the two use-
categories of cultivated oca might represent separate origins of
domestication.
LITERATURE CITED
Adrianz é n , P. 2006 . An á lisis de la diversidad gen é tica de ocas cultiva-
das ( Oxalis tuberosa Mol.) de 4 comunidades campesinas del Cusco,
usando AFLP. Thesis for Licenciado en Biolog í a, Universidad
Nacional Agraria La Molina, Lima, Peru.
Alexander , P. J. , G. Rajanikanth , C. D. Bacon , and C. D. Bailey .
2007 . Recovery of plant DNA using a reciprocating saw and silica-
based columns. Molecular Ecology Notes 7 : 5 – 9 .
Allaby , R. G. , and T. A. Brown . 2003 . AFLP data and the origins of
domesticated crops. Genome 46 : 448 – 453 .
Allaby , R. G. , D. Q. Fuller , and T. A. Brown . 2008 . The genetic
expectations of a protracted model for the origins of domesticated
crops. Proceedings of the National Academy of Sciences, USA 105 :
13982 – 13986 .
Biondi , J. 2006 . An á lisis de la diversidad gen é tica de la colecci ó n de
Oxalis tuberosa Molina “ oca ” mantenida en el Centro Internacional de
la Papa, usando marcadores AFLP. Thesis for Licenciado en Biolog í a,
Universidad Ricardo Palma, Lima, Peru.
Br ü cher , H. 1969 . Poliploidia en especies sudamericana de Oxalis. Bolet í n
de la Sociedad Venozolana de Ciencias Naturales. 28 : 145 – 178 .
de Azkue , D. , and A. Mart í nez . 1990 . Chromosome number of the Oxalis
tuberosa alliance (Oxalidaceae). Plant Systematics and Evolution
169 : 25 – 29 .
Dixon , P. 2003 . VEGAN, a package of R functions for community ecol-
ogy. Journal of Vegetation Science 14 : 927 – 930 .
Doyle, J. J., and J. L. Doyle . 1990. Isolation of plant DNA from fresh
tissue. Focus 12: 13 – 15.
Emshwiller , E. 2002a . Biogeography of the Oxalis tuberosa alliance.
Botanical Review 68 : 128 – 152 .
Emshwiller , E. 2002b . Ploidy levels among species in the ‘ Oxalis tu-
berosa alliance ’ as inferred by fl ow cytometry. Annals of Botany 89 :
741 – 753 .
Emshwiller , E. 2006a . Evolution and conservation of clonally propa-
gated crops — Insights from AFLP data and folk taxonomy of the
Andean tuber “ oca, ” Oxalis tuberosa. In T. Motley, N. Zerega, and H.
Cross [eds.], Darwin ’ s harvest: New approaches to the origins, evolu-
tion, and conservation of crops, 308 – 332. Columbia University Press,
New York, New York, USA.
Emshwiller , E. 2006b . Origins of polyploid crops: The example of the
octoploid tuber crop Oxalis tuberosa. In M. A. Zeder, D. Bradley,
E. Emshwiller, and B. D. Smith [eds.], Documenting domestication:
New genetic and archeological paradigms, 153 – 168. University of
California Press, Berkeley, California, USA.
Emshwiller , E. , and J. J. Doyle . 2002 . Origins of domestication and
polyploidy in oca ( Oxalis tuberosa : Oxalidaceae). 2. Chloroplast-
expressed glutamine synthetase data. American Journal of Botany 89 :
1042 – 1056 .
Liu , B. , and J. F. Wendel . 2002 . Non-Mendelian phenomena in allopo-
lyploid genome evolution. Current Genomics 3 : 489 – 505 .
Lourteig , A. 2000 . Oxalis L. subg é nero Monoxalis (Small) Lourteig,
Oxalis y Trifi dus Lourteig. Bradea 7 : 201 – 629 .
Lukens , L. N. , P. A. Quijada , J. Udall , J. C. Pires , M. E. Schranz , and T.
C. Osborn . 2004 . Genome redundancy and plasticity within ancient
and recent Brassica crop species. Biological Journal of the Linnean
Society 82 : 665 – 674 .
Malice , M. , N. Martin , A. Pissard , J. A. Rojas-Beltran , A.
Gandarillas , P. Bertin , and J. P. Baudoin . 2007 . A prelimi-
nary study of the genetic diversity of Bolivian oca ( Oxalis tuberosa
Mol.) varieties maintained in situ and ex situ through the utilization
hand, only O. chicligastensis had any of the “ gold standard ”
diagnostic markers that it shared, as fi xed, with cultivated oca,
in the peak-sharing evaluation (data not shown), so its possible
role in the origin of oca should not be dismissed. Despite the
fact that many peaks were shared, as fi xed, between each of the
wild taxa and oca, these shared, fi xed peaks were usually pres-
ent in more than one wild, tuber-bearing Oxalis taxon (they
were not diagnostic). Only O. chicligastensis had diagnostic
peaks that were absent in all the other wild, tuber-bearing Oxa-
lis taxa: one peak that it shared with all oca samples and another
one that it shared with all wayk ’ u oca samples. Although the
default expectation for polyploids is additivity (e.g., Pires et al.,
2004 ), i.e., that markers that are fi xed in the progenitors will be
fi xed in the polyploid derivative, this default expectation is of-
ten violated in many polyploids studied to date ( Wendel, 2000 ;
Liu and Wendel, 2002 ; Pikaard, 2001 ; Lukens et al., 2004 ), and
several processes may lead to “ loss ” of bands. This process of
loss of molecular markers in polyploids may be the explanation
for the low number of “ gold standard ” peaks observed here. On
the other hand, the pattern may be complicated because the pro-
genitor candidates are closely related, so that fi xed markers are
shared among multiple taxa.
The neighbor-joining results must also be interpreted with
caution, because having a shorter distance between clusters in
the N-J results is not defi nitive by itself to identify crop pro-
genitors. The results indicate that AFLP data of BolW/T are
more similar to those of both use-categories of cultivated oca
than are those of O. chicligastensis , but this should not neces-
sarily be interpreted to mean that the Bolivian populations sam-
pled are necessarily the progenitors of oca. The fact that
neighbor-joining is a distance-based method means that its re-
sults do not necessarily refl ect phylogenetic relationships
among species. When used in studies within species, the poten-
tial for gene fl ow and meiotic recombination among popula-
tions with separate origins can lead to results that appear to
suggest “ monophyletic ” origins when in fact there were multi-
ple origins ( Allaby and Brown, 2003 ; Allaby et al., 2008 ). Poly-
ploidy adds additional complexity and can make the N-J results
even more diffi cult to interpret, especially when such analyses
are undertaken with samples of different ploidy levels in the
same analysis. The representation of N-J results as a tree-like
diagram tempts many to interpret the results as a hierarchical
phylogeny, which is inappropriate if the true history includes
hybrid origins, as is often the case for polyploids. In addition, it
is possible that the dominant nature of AFLP markers would
result in clustering of taxa that have the same ploidy level, re-
gardless of their true relationships, because with dominant
markers only nulliplex (completely homozygous recessive) in-
dividuals will not have a particular marker, whereas one indica-
tor of allopolyploidy is fi xed heterozygosity. Thus, individuals
of several different homeologous genotypes might have the
same AFLP phenotype for a particular marker or group of
markers, because gene dosage is not considered, only the pres-
ence and absence.
The incongruence between the AFLP results presented here
and the previous results with ncpGS ( Emshwiller and Doyle,
2002 ), especially with respect to the role of O. picchensis in the
origins of O. tuberosa , serves to underscore the complexity of
polyploid evolution. Nonetheless, the currently available AFLP
data support the idea that the wild, tuber-bearing Oxalis popula-
tions from Bolivia (BolW/T) and O. chicligastensis of Argen-
tina are the better candidates as genome donors of oca. Future
studies are planned to increase sampling of individuals and

1848 American Journal of Botany
of ISSR molecular markers. Genetic Resources and Crop Evolution
54 : 685 – 690 .
Myburg , A. A. , D. L. Remington , D. M. O ’ Malley , R. R. Sederoff ,
and R. W. Whetten . 2001 . High-throughput AFLP analysis us-
ing infrared dye-labeled primers and an automated DNA sequencer.
BioTechniques 30 : 348 – 352, 354, 356 – 357 .
Nei , M. , and W.-H. Li . 1979 . Mathematical models for studying genetic
variation in terms of restriction endonucleases. Proceedings of the
National Academy of Sciences, USA 76 : 5269 – 5273 .
Pastor , S. , B. Fuentealba , and M. Ruiz . ( 2008 ) Underutilized spe-
cies policies and strategies — Underutilized crops in Peru: Some
conceptual and political considerations. Report commissioned by
the Global Facilitation Unit for Underutilized Species (GFU), Via
dei Tre Denari, 472/a, Maccarese (Fiumicino), Italy. Also at website
http://www.underutilized-species.org/Documents/PUBLICATIONS/
underutilized_species_policies_and_strategies_peru_lr.pdf.
Pikaard , C. S. 2001 . Genomic change and gene silencing in polyploids.
Trends in Genetics 17 : 675 – 677 .
Pires , J. C. , K. Y. Lim , A. Kovarik , R. Matyasek , A. Boyd , A. R. Leitch ,
I. J. Leitch , et al . 2004 . Molecular cytogenetic analysis of recently
evolved Tragopogon (Asteraceae) allopolyploids reveal a karyotype
that is additive of the diploid progenitors. American Journal of Botany
91 : 1022 – 1035 .
Pissard , A. , C. Arbizu , M. Ghislain , A. M. Faux , S. Paulet , and P.
Bertin . 2008a . Congruence between morphological and molecular
markers inferred from the analysis of the intra-morphotype genetic
diversity and the spatial structure of Oxalis tuberosa Mol. Genetica
132 : 71 – 85 .
Pissard , A. , M. Ghislain , and P. Bertin . 2006 . Genetic diversity of the
Andean tuber-bearing species, oca ( Oxalis tuberosa Mol.), investi-
gated by inter-simple sequence repeats. Genome 49 : 8 – 16 .
Pissard , A. , J. A. Rojas-Beltran , A. M. Faux , S. Paulet , and P. Bertin .
2008b . Evidence of intra-varietal genetic variability in the vegeta-
tively propagated crop oca ( Oxalis tuberosa Mol.) in the Andean
traditional farming system. Plant Systematics and Evolution 270 :
59 – 74 .
Schibli , R. 2007 . An á lisis de la diversidad gen é tica de Oxalis tuberosa
Molina “ oca ” de 2 comunidades del Cusco usando marcadores AFLP.
Thesis for Licenciado en Biolog í a, Universidad Nacional Agraria La
Molina, Lima, Peru.
Swofford , D. L. 2002 . PAUP*: Phylogenetic analysis using parsi-
mony (*and other methods), version 4.0b10. Sinauer, Sunderland,
Massachusetts, USA.
Tosto , D. S. , and H. E. Hopp . 2000 . Suitability of AFLP markers for the
study of genomic relationships within the Oxalis tuberosa alliance.
Plant Systematics and Evolution 223 : 201 – 209 .
Tsyusko , O. V. , T. D. Tuberville , M. B. Peters , N. Crawford ,
C. Hagen , S. G. Weller , A. K. Sakai , and T. C. Glenn .
2007 . Microsatellite markers isolated from polyploid wood-
sorrel Oxalis alpina (Oxalidaceae). Molecular Ecology Notes 7 :
1284 – 1286 .
Vos , P. , R. Hogers , M. Bleeker , M. Reijans , T. van de Lee , M. Hornes , A.
Friters , et al . 1995 . AFLP: A new technique for DNA fi ngerprint-
ing . Nucleic Acids Research 23 : 4407 – 4414 .
Wendel , J. F. 2000 . Genome evolution in polyploids. Plant Molecular
Biology 42 : 225 – 249 .
Zeder , M. A. , E. Emshwiller , B. D. Smith , and D. G. Bradley .
2006 . Documenting domestication: The intersection of genetics and
archaeology. Trends in Genetics 22 : 139 – 155 .
Zorrilla , C. 2006 . Diversidad gen é tica de cultivares nativos de papas am-
argas y ocas provenientes de Puno. Thesis for Licenciado en Biolog í a,
Universidad Nacional Agraria La Molina, Lima, Peru.