Biodivers Conserv (2007) 16:28472865

DOI 10.1007/s10531-006-9025-x

ORIGINAL PAPER

AFLPs and morphological diversity of

Phaseolus lunatus L. in Cuban home gardens:
approaches to recovering the lost ex situ collection

Leonor Castineiras Felix Alberto Guzman

Myriam Cristina Duque Tomas Shagarodsky
Raul Cristobal M. Carmen de Vicente

Received: 11 March 2005 / Accepted: 8 March 2006 / Published online: 11 May 2007

Springer Science+Business Media B.V. 2007

Abstract The genetic diversity of 76 accessions of lima bean (Phaseolus lunatus

L.), collected mostly from home gardens, was assessed with AFLPs and seed
descriptors to evaluate the potential for recovering a lost ex situ collection in
Cuba. The sample contained 60 accessions collected from 25 home gardens in the
three main geographical regions of Cuba and represented the three cultivated
types found on the island. Four more accessions were part of the former ex situ
collection and the remaining 12 accessions were selected from the world bean
collection held at the International Center for Tropical Agriculture. Some mor-
phological measurements discriminated among cultivated types. The analysis of 62
polymorphic bands obtained with two AFLP primer combinations indicated that
the three cultivated bean types were comparable in terms of molecular diversity
and that no pattern of variation was associated with geographical distribution.
However, a multiple correspondence analysis with the same molecular data de-
tected different genetic groups. Three of these groups included all the cultivated
accessions collected from home gardens, but could not be explained by the seed
descriptors. The results therefore suggest that a scientifically sound collecting
strategy to recover the former Cuban ex situ gene bank should consider com-
bining geographical, morphological, and molecular data. The findings also suggest
that any proposed methodologies should be considered before developing a
conservation strategy based on an ex situ or combined ex situ and in situ
approaches.

L. Castineiras T. Shagarodsky R. Cristobal

Instituto de Investigaciones Fundamentales en Agricultura Tropical Alejandro de Humboldt
(INIFAT), Calle 1 esq. 2, Santiago de las Vegas, Ciudad de la Habana C.P 17200, Cuba

F. A. Guzman M. C. de Vicente (&)

Regional Office for the Americas International Plant Genetic Resources Institute (IPGRI),
c/o CIAT, A.A. 6713 Cali, Colombia
e-mail: c.devicente@cgiar.org

M. C. Duque
Centro Internacional de Agricultura Tropical (CIAT), A.A. 6713 Cali, Colombia
123
2848 Biodivers Conserv (2007) 16:28472865

Keywords Genetic structure In situ conservation Lima bean

Molecular diversity Multiple correspondence analysis Phaseolus lunatus

Introduction

Studies were carried out on the variation of lima bean (Phaseolus lunatus L.) con-
served in the gene bank of the Instituto de Investigaciones Fundamentales en
Agricultura Tropical Alejandro von Humboldt (INIFAT), Santiago de las Vegas,
La Habana, Cuba. The Institute held accessions of the three cultivated groups de-
scribed within the species (Sieva, Papa, and Big Lima), some cultivated accessions
with intermediate traits (Sieva/Papa and Sieva/Big Lima), and some wild material
(Esquivel et al. 1990; Castineiras et al. 1991). A recent regeneration of the ex situ
collection failed for various reasons (Castineiras et al. 2001) and, as a result, the
persistence of cultivated forms of P. lunatus in Cuba depends solely on their con-
tinued cultivation and conservation in situ in home gardens.
A recent global project on the role of home gardens in the conservation and use
of plant genetic resources (Watson and Eyzaguirre 2002) found P. lunatus growing in
all 39 home gardens sampled in the countrys three main geographical zones: wes-
tern, central, and eastern. This observation confirmed that Cuban home gardens are
indeed reservoirs of lima bean diversity (Castineiras et al. 2002). The project eval-
uated the genetic diversity of different crops in Cuban home gardens, using mor-
phological characters (Castineiras et al. 2002; Fundora et al. 2004; Shagarodsky
et al. 2004). Morphological measurements have been widely used to study crop
diversity, including that of P. lunatus (Vargas et al. 2003).
However, in our study, while the use of morphological descriptors was considered
necessary and useful, they were also deemed insufficient for a thorough and dis-
tinctive evaluation of existing genetic diversity. Consequently, an analysis, using
DNA marker technology, was conducted to supplement the information
already available and to add value to the research (Hoogendijk and Williams 2002;
de Vicente 2002).
Markers such as AFLPs (amplified fragment length polymorphisms) were effec-
tive for measuring the genetic diversity of numerous species (e.g. soybean, Maughan
et al. 1996; Phaseolus vulgaris, Tohme et al. 1996; sunflower, Hongtrakul et al. 1997;
velvet bean, Capo-chichi et al. 2001; and papaya, Kim et al. 2002). They were also
the methodology of choice in a recent diversity study of Capsicum spp. in home
gardens in Guatemala (Guzman et al. 2005). In fact, results obtained from the
Capsicum study provided an important basis for successfully implementing a com-
plementary conservation scheme that included ex situ and in situ activities.
This paper describes the results of studies of the genetic variation of P. lunatus
accessions found in Cuban home gardens, using both AFLPs and morphological seed
descriptors. Research objectives were to assess the consistency of data obtained with
both types of markers and their usefulness in helping to define complementary
conservation strategies. The specific goal was to restore the Cuban ex situ P. lunatus
collection, using the genetic variation discovered in home gardens as a starting point.
In this instance, the comparison of morphological and molecular studies is a means
of determining the best methodology for thoroughly examining diversity and its use
as a decision-making tool in the important venture of conserving crop diversity to
ensure its long-term availability. The outcomes of this research would be of interest
123
Biodivers Conserv (2007) 16:28472865 2849

not only for the current situation in Cuba, but also to countries where conservation
of genetic resources in home gardens is either the only possible or the most sus-
tainable scheme in this endeavour.

Materials and methods

Plant materials

A sample of 64 P. lunatus accessions was used, where (i) 59 accessions were collected
from 25 home gardens in the western (Pinar del Ro), central (Cienfuegos), and
eastern (Guantanamo) regions of Cuba (Fig. 1). These represented 32 accessions of
the Sieva type, 12 of the Papa type, and the remaining 15 of the intermediate type,
Sieva/Papa. (ii) One wild accession came from a Cienfuegos garden; and (iii) four
accessions (one wild and three cultivated forms) left over from INIFATs ex situ
gene bank collection (Table 1). In addition, the study included 10 cultivated
P. lunatus accessions (Table 1) and two wild accessions from Peru. These additional
accessions were selected from the world bean collection kept at the International
Center for Tropical Agriculture (CIAT, its Spanish acronym) to estimate the
molecular similarity of the Cuban accessions with others found in adjacent regions
(Mesoamerica) and with the Peruvian gene pool. In all, 76 accessions were
examined.
An accession from Cuban home gardens was formed by pooling seeds collected
from two to five plants of similar morphological type. Seeds of the Cuban ex situ
collection were harvested from as many plants with as similar morphological traits as
possible in a given location. This group of seeds was then identified as a single
accession. The Cuban accession identified as wild, in fact, refers to a genotype
collected from a feral or weedy population.

Morphological characterization

Nine seed descriptors (seven quantitative and two qualitative) were used to char-
acterize the 64 Cuban accessions, following the IBPGR (1982) guidelines and the
recommendations of Castineiras et al. (1991). The descriptors used were average

Fig. 1 Map of Cuba indicating the collecting sites for home gardens accessions. Departments
coloured in grey: 1. Pinar del Ro, 2. Cienfuegos, 3. Guantanamo

123
 Table 1 Origin and morphological characterization of the accessions
2850

Accession
Origin Number Seed colour 100SW Length Width Girth Ratio Ratio Cultivated type MCA group
of testa (g) (L) (W) (G) L/W G/W

123
colours Primary Secondary (cm) (cm) (cm)

OC-1-1l PR 1 Black 47.69 1.5 1.1 0.6 1.36 0.57 Sieva A

OC-1-2 PR 1 Red 59.96 1.6 1.1 0.5 1.33 0.41 Sieva C
OC-1-3a PR 2 Grey Red 56.25 1.4 0.9 0.6 1.55 0.66 Sieva/Papa A
OC-1-4 PR 1 Red 40.00 1.4 1.0 0.4 1.40 0.40 Sieva A
OC-2-1k PR 1 White 49.55 1.2 0.8 0.4 1.50 0.50 Sieva B
OC-8-1 PR 1 Red 53.75 1.4 0.9 0.4 1.50 0.41 Sieva A
OC-8-2a PR 1 White 42.00 1.4 0.9 0.8 1.50 0.88 Papa A
OC-10-1 PR 1 Red 36.07 1.1 0.8 0.5 1.37 0.62 Sieva/Papa A
OC-12-1k PR 1 Maroon 40.00 1.6 1.3 0.4 1.34 0.33 Sieva B
OC-12-2 PR 1 Dark brown 40.00 1.4 1.0 0.4 1.40 0.40 Sieva C
OC-12-3 PR 1 White 43.00 1.5 1.0 0.5 1.50 0.50 Sieva A
OC-13-1 PR 1 Red 49.50 1.3 0.8 0.8 1.62 0.89 Papa C
C-1-1l CF 2 Grey Black 43.22 1.4 1.0 0.6 1.40 0.60 Sieva/Papa A
C-1-2 CF 1 Red 46.67 1.5 0.8 0.3 1.87 0.37 Sieva C
C-1-3l CF 2 White Brown 43.33 1.3 0.7 0.4 1.72 0.43 Sieva A
C-1-4l CF 1 Red 48.44 1.3 0.9 0.5 1.44 0.55 Sieva A
C-1-5b CF 1 Red 42.00 1.2 0.8 0.7 1.50 0.87 Papa A
C-1-6 CF 2 Grey Light brown 41.90 1.2 0.8 0.7 1.33 0.77 Papa C
C-2-1b CF 1 Grey 58.94 1.5 1.1 0.6 1.36 0.54 Sieva A
C-2-2 CF 2 Grey Red 33.92 1.0 0.8 0.6 1.25 0.75 Papa C
C-5-1c CF 1 White 52.82 1.5 1.0 0.6 1.53 0.54 Sieva A
C-6-1 CF 2 Grey Maroon 41.28 1.3 1.0 0.6 1.30 0.60 Sieva/Papa C
C-6-3k CF 2 Red Black 47.00 1.4 1.0 0.6 1.40 0.60 Sieva/Papa B
C-6-4 CF 2 Red Brown 50.05 1.3 0.8 0.5 1.59 0.55 Sieva A
C-6-5 CF 1 White 41.58 0.9 0.7 0.5 1.28 0.71 Sieva/Papa A
C-6-6l CF 2 White Red 50.00 1.2 0.7 0.4 1.20 0.57 Sieva A
C-7-1 CF 1 Brown yellowish 8.50 0.8 0.7 0.4 1.14 0.57 Wild G
C-7-2 CF 2 Grey Brown 34.29 1.2 0.8 0.7 1.50 0.87 Papa C
C-7-3 CF 1 White 39.27 1.1 0.8 0.4 1.38 0.50 Sieva B
C-8-1d CF 1 White 32.00 1.2 0.8 0.6 1.50 0.75 Papa A
Biodivers Conserv (2007) 16:28472865
 Table 1 continued

Accession
Origin Number Seed colour 100SW Length Width Girth Ratio Ratio Cultivated type MCA group
of testa (g) (L) (W) (G) L/W G/W
colours Primary Secondary (cm) (cm) (cm)

C-9-1 CF 1 Dark brown 32.31 1.2 0.9 0.6 1.20 0.66 Sieva/Papa C
C-10-1d CF 1 Black 50.87 1.4 0.9 0.7 1.55 0.77 Papa A
C-10-2c CF 2 White Black 61.00 1.3 1.0 0.6 1.30 0.60 Sieva/Papa A
C-10-3 CF 2 Grey Light brown 52.00 1.3 0.9 0.6 1.44 0.66 Sieva/Papa A
C-10-4 CF 1 White 47.79 1.3 0.9 0.7 1.44 0.77 Papa A
C-11-1 CF 1 Red 47.27 1.4 1.1 0.6 1.41 0.54 Sieva C
C-11-2 CF 2 White Black 40.70 1.3 0.9 0.6 1.44 0.66 Sieva/Papa A
O-2-1 GT 1 White 54.00 1.4 1.1 0.5 1.27 0.45 Sieva B
O-2-2 GT 2 Grey Maroon 48.67 1.3 1.0 0.6 1.30 0.60 Sieva/Papa C
Biodivers Conserv (2007) 16:28472865

O-2-3 GT 1 White 39.33 1.4 1.0 0.5 1.40 0.50 Sieva A

O-3-1e GT 1 Red 50.00 1.4 1.0 0.5 1.40 0.50 Sieva C
O-4-3 GT 1 White 36.00 1.2 0.9 0.6 1.33 0.66 Sieva/Papa C
O-5-1 GT 1 Red 44.58 1.1 0.9 0.7 1.22 0.77 Papa C
O-5-2f GT 1 Red 34.29 1.2 0.8 0.7 1.50 0.87 Papa C
O-6-1 GT 1 White 40.70 1.3 0.9 0.6 1.44 0.66 Sieva/Papa C
O-7-1 GT 1 Red 40.22 1.4 1.0 0.5 1.40 0.50 Sieva C
O-7-2 GT 2 Red Bright red 38.50 1.4 1.0 0.5 1.40 0.50 Sieva B
O-7-3g GT 1 Red 43.33 1.3 0.9 0.5 1.44 0.55 Sieva C
O-7-4g GT 1 Grey 35.00 1.2 0.9 0.5 1.33 0.55 Sieva C
O-7-5h GT 1 White 48.67 1.2 1.0 0.6 1.20 0.60 Sieva/Papa C
O-7-6 GT 2 Grey Purple 43.00 1.3 1.0 0.3 1.30 0.30 Sieva A
O-7-7f GT 1 Dark brown 39.00 1.4 0.9 0.4 1.55 0.44 Sieva C
O-7-8h GT 1 White 51.33 1.4 1.0 0.5 1.40 0.50 Sieva C
O-7-9 GT 1 White 36.40 1.2 0.9 0.5 1.33 0.55 Sieva A
O-9-1 GT 2 Grey Maroon 53.33 1.2 0.9 0.8 1.33 0.88 Papa C
O-9-2 GT 1 White 42.00 1.3 0.9 0.5 1.44 0.55 Sieva C
O-9-3 GT 1 Red 61.33 1.5 1.1 0.5 1.36 0.45 Sieva C
O-10-1 GT 2 Red Maroon 58.00 1.5 1.0 0.5 1.50 0.50 Sieva C
O-12-1 GT 2 Red Maroon 56.00 1.5 1.1 0.5 1.36 0.45 Sieva/Papa C
O-14-1e GT 1 Red 43.03 1.2 1.0 0.5 1.20 0.50 Sieva C

123
2851
 2852

123
Table 1 continued

Accession
Origin Number Seed colour 100SW Length Width Girth Ratio Ratio Cultivated type MCA group
of testa (g) (L) (W) (G) L/W G/W
colours Primary Secondary (cm) (cm) (cm)

P-1353 SC 2 Red Maroon 40.00 1.4 0.9 0.5 1.55 0.55 Sieva C
P-1836 PR 1 Red 48.00 1.5 1.1 0.5 1.36 0.45 Sieva C
P-2266 CF 2 Brown yellowish Brown 8.20 0.4 0.3 0.2 1.33 0.66 Wild H
P-2775 VC 2 Red Maroon 42.60 1.3 0.9 0.5 1.44 0.60 Sieva/Papa C
G25265 Slv 1 Black 52.51 D
G25700 Pan 2 Pinkish Black 42.80 A
G25914 Per 3 Grey Other 14.00 Wild F
G25915 Per 3 Grey Other 27.20 Wild F
G25992 Gtm 1 Black 36.60 E
G26170i Mex 2 Grey Brown 42.00 A
G26171i Mex 2 Grey Brown 34.70 A
G26173 Mex 2 Grey Purple 38.10 A
G26174 Mex 2 Other Red 21.20 D
G26287 Mex 1 Pinkish 27.70 D
G26307j Cri 1 White 46.60 D
G26308j Hon 3 Other Red 51.40 D

Accessions starting by an O or a C refer to those collected in home gardens, by a P to those belonging to the Cuban ex situ collection and by a G to those from the
worldwide ex situ collection at CIAT. 100SW: weight of 100 seeds (g); PR: Pinar del Rio (West); CF: Cienfuegos (Center); VC: Villa Clara (Center); SC: Santiago
de Cuba (East); GT: Guantanamo (East); Slv: Salvador; Pan: Panama; Per: Peru; Gtm: Guatemala; Mex: Mexico; Cri: Costa Rica; Hon: Honduras


Accessions with the same superindex are those that shared the band profile; accessions without a superindex had a unique band profile
Biodivers Conserv (2007) 16:28472865
Biodivers Conserv (2007) 16:28472865 2853

100-seed weight (100SW), length (cm), width (cm), girth (cm), length-to-width ratio
(L/W), girth-to-width ratio (G/W), number of seed-coat colours, primary colour
(PC), and secondary colour (SC) (Table 1). Seed measurements were taken for all
plants growing in each home garden (25 plants) during the cropping cycles of 2001
and 2002.

Molecular characterization

For the plants collected from the 25 home gardens (point i above), seeds were
harvested separately from all plants forming an accession. Three seeds per plant
were then sown in a greenhouse. Twelve days after planting, young leaves were
taken from each seedling. Leaf tissues of 15 seedlings from each accession were
mixed and lyophilized, following the protocol described by Murray and Thompson
(1980). The lyophilized tissue was kept at )80C until DNA was extracted.
DNA was extracted, starting from either the mixture of lyophilized tissue (64
Cuban accessions) or the young leaves of one individual for each additional acces-
sion (i.e. of the 12 from the world collection), using tissue pieces measuring 125 mm2
and following the extraction protocol described by Vallejos et al. (1992). DNA
quality was checked in an agarose gel, diluted to a final concentration of 25 ng/ll,
and stored at )20C.
AFLP assays (Vos et al. 1995) were performed, using the Analysis System I and
the AFLP Starter Primer kit (Gibco-BRL, Life Technologies), with EcoRI/MseI
primers. DNA digestions, adapters ligation, and preselective and selective amplifi-
cations of the restricted fragments were performed, following the instructions pro-
vided by the kit manufacturers.
Five randomly chosen DNA samples were used to check the performance of
19 primer combinations. EcoRI + ACA/MseI + CAC and EcoRI + ACA/MseI + -
CAT primer pairs were selected, according to the number of amplified fragments,
degree of polymorphism, and band resolution. All amplifications were run in a PTC-
100 programmable thermal cycler (MJ Research, Inc., Waltham, MA).
Electrophoresis was performed in 6% (w/v) denaturing polyacrylamide gels
containing 5 M urea and TBE 0.5 buffer. Six microlitres of buffer [95% (v/v)
formamide; 0.20 mM EDTA pH 8.0; 0.05% (w/v) bromophenol blue; and xylene
cyanol FF] were added to each amplification product and 5ll of each sample were
loaded into the gel. In each gel, a reference sample was included to help compare
bands in different gels. Furthermore, to test the reproducibility of the experiments,
some randomly selected accessions were also run in different gels. The electro-
phoresis was run for about 2 h at 110 W and 50C. The resulting separation of bands
was visualized with silver staining.

Data analysis

Morphological

Morphological data were analyzed to quantify the diversity of home garden beans
according to their geographical origin (western, central, or eastern Cuba) and their
cultivated type. A multivariate analysis of variance (MANOVA) was performed to
test for the existence of multivariate differences between regions and cultivated
types with respect to the morphological measurements (Table 2). In addition, to
123
2854 Biodivers Conserv (2007) 16:28472865

better understand the relationships between morphological variables and their

expression in cultivated types, several combined variables were also considered, that
is, shape (G/W) and size (L/W), in addition to the 100-seed weight conventionally
defined.
With these new variables, another MANOVA was performed, both against region
and cultivated type (Table 2). Where significant differences (Wilks lambda) were
found among sources of variation (region, cultivated type, and their interactions),
the mean separation was calculated, using the RyanEinotGabrielWelsch Multi-
ple Range Test method (REGW Q test; Proc. GLM of SAS v.9.1.3).

Molecular

Bands that were clearly observed and showed polymorphism in the sample were
visually scored and resulting data entered in a spreadsheet as a binary matrix
(1 = present, 0 = absent).
A multiple correspondence analysis (MCA; Proc. Corresp. SAS v.8.2) was carried
out to develop an exploratory approximation, visualize data in Euclidean space, and
identify some type of natural clustering of accessions based on their molecular band
profile and its relationship with regional or cultivated type classifications. The MCA
showed the distribution of individuals as reflecting their chi-squared distance based
on their band profiles. This Euclidean distance takes weights for each individual and
for each band, depending on the frequency with which they score the present state
(i.e. the appearance of a particular band in the gel). Clustering of individuals was
carried out with the unweighted pair-group method with arithmetic mean (UPGMA
method; Proc. Cluster SAS v.8.2) on coordinates from the MCA.
Neis (1987) calculations were used to estimate the total genetic diversity
(HT = HS + DST). Because AFLP bands do not represent codominant loci, Neis
coefficient cannot be interpreted as a heterozygosis parameter but, instead, as an
approximate estimator of heterogeneity. Genetic variation within each group
(groups generated by the MCA, or regions or cultivated types) was expressed as Hi
and the average genetic diversity within groups as HS. The diversity between groups
(DST) was calculated as the difference between HT and HS. DST/HT is the coefficient
of genetic differentiation (GST) and measures the relative amount of total variation
caused by the difference between groups.

Results

Morphological variation in the Cuban accessions

The predominant primary colours observed in the 64 Cuban accessions were red
(38%), white (31%), and grey (19%) (Table 1). Forty-two accessions had only one
seed coat colour, with white (38%) or red (40%) being the most commonly found.
Twenty-two accessions had two seed coat colours and their prevalent secondary
colours were maroon (32%), brown (18%), black (18%), and red (13%).
Table 3 shows the ranges for length, width, girth, and average 100-seed weight in
the Cuban accessions. The average ratios ( their standard deviations) for each
cultivated type can be estimated from Table 3 as follows: (i) L/W: Sieva

123
 Table 2 Significance obtained for region, cultivated type and their interaction as a result of the multivariate analysis with morphological measurements

Source of variation Model with the original variables Model with combined variables and 100SW
Length Width Girth Weight of Wilks L/W G/W Weight of Wilks
(L) (W) (G) 100 seed k 100 seed k
Biodivers Conserv (2007) 16:28472865

(100SW) (100SW)

Region ns ns ns ns ns 0.0409 ns ns ns
Cultivated type 0.0051 0.0213 <0.0001 ns <0.0001 ns <0.0001 ns <0.0001
Region Cultivated type ns ns ns ns ns ns ns
F-statistics 0.0153 0.00093 <0.0001 0.7734 0.0674 <0.0001 0.07734

ns = non-significant

123
2855
 2856

123
Table 3 Values of four seed descriptors in the three group categories analyzed

n 100SW Length Width Girth

m M A SD m M A SD m M A SD m M A SD

Cultivated type
Sieva 32 35.00 61.33 46.41 7.21 1.10 1.60 1.38 0.12 0.70 1.30 0.96 0.13 0.30 0.60 0.47 0.08
Papa 12 32.00 53.33 42.21 7.31 1.00 1.40 1.23 0.11 0.80 0.90 0.84 0.05 0.60 0.80 0.71 0.07
Sieva/Papa 15 32.31 61.00 45.43 8.32 0.90 1.50 1.27 0.14 0.70 1.10 0.93 0.10 0.50 0.60 0.58 0.04
Origin
West 12 36.07 59.96 46.48 7.49 1.10 1.60 1.40 0.15 0.80 1.30 0.97 0.15 0.40 0.80 0.53 0.15
Center 24 32.00 61.00 44.94 7.65 0.90 1.50 1.28 0.15 0.70 1.10 0.88 0.12 0.30 0.70 0.57 0.11
East 23 34.29 61.33 45.07 7.80 1.10 1.50 1.32 0.12 0.80 1.10 0.97 0.08 0.30 0.80 0.53 0.10
MCA group
Aa 25 32.00 61.00 46.09 7.33 0.90 1.50 1.32 0.14 0.70 1.10 0.90 0.11 0.30 0.80 0.55 0.12
B 6 38.50 54.00 44.72 6.41 1.10 1.60 1.35 0.18 0.80 1.30 1.00 0.19 0.40 0.60 0.47 0.08
Cb 28 32.31 61.33 44.74 8.17 1.00 1.60 1.31 0.14 0.80 1.10 0.94 0.10 0.30 0.80 0.56 0.12
D 5 21.20 52.51 39.88 14.45
E 1 36.60
F 2 14.00 27.20 20.60 9.33
G 1 8.50 0.8 0.7 0.4
H 1 8.20 0.4 0.3 0.2

n: number of accessions per group; 100SW: average weight of 100 seeds (g); m: minimum value; M: maximum value; A: average; SD: standard deviation; MCA:
Multiple correspondence analysis
a
If non-Cuban accessions included, then 100SW (A) = 45.17
b
If ex situ accessions included then 100SW (A) = 44.62 and length (A) = 1.32
Biodivers Conserv (2007) 16:28472865
Biodivers Conserv (2007) 16:28472865 2857

Table 4 Genetic diversity values, number of polymorphic bands and unique bands of cultivated
accessions collected in home gardens shown by cultivated type, geographical origin and MCA groups

n Hi (SD) HS (SD) GST Polymorphic Unique

bands bands

Cultivated type
Sieva 32 0.118 (0.217) 23 3
Papa 12 0.113 (0.228) 17 0
Sieva/Papa 15 0.126 (0.233) 22 2
0.119 (0.004) <0.001
Origin
West 12 0.121 (0.232) 19 1
Center 24 0.112 (0.232) 21 2
East 23 0.100 (0.197) 20 3
0.109 (0.008) 0.080
MCA group
A 25 0.097 (0.191) 20 3
B 6 0.032 (0.107) 6 2
C 28 0.097 (0.184) 21 1
0.090 (0.020) 0.238

n: total number of accessions per sample; Hi: within sample diversity; SD: standard deviation; HS:
average genetic diversity among samples; GST: coefficient of genetic differentiation. HT = 0.119
(SD = 0.217) (total genetic diversity)

(1.43 0.13), Papa (1.44 0.13), and Sieva/Papa (1.35 0.09); and (ii) G/W: Sieva
(0.48 0.07), Papa (0.83 0.08), and Sieva/Papa (0.62 0.06). If the calculations
were done per region, the average ratios ( their standard deviations) were (i) L/W:
western (1.45 0.09), central (1.43 0.16), and eastern (1.37 0.09); and (ii) G/W:
western (0.55 0.19), central (0.63 0.13), and eastern (0.56 0.14).
Table 2 shows that, in the model with the original variables (L, W, G, and
100SW), cultivated type is the only significant source of variation for L, W, and
G. Region is not significant, nor is its interaction with cultivated type. The Sieva
beans are significantly longer than those of the other two cultivated types (mean
values and mean separation with EMS = 0.016: Sieva 1.38 a; Sieva/Papa 1.27 b; Papa
1.23 b). Papa is significantly narrower than Sieva and Sieva/Papa (mean values and
mean separation with EMS = 0.011: Sieva 0.96 a; Sieva/Papa 0.93 a; Papa 0.84 b).
The three cultivated types differ significantly in terms of girth: Papa is the bean type
with largest girth, and Sieva/Papa is an intermediate type between Sieva and Papa
(mean values and mean separation with EMS = 0.004: Sieva 0.47 c; Sieva/Papa 0.58
b; Papa 0.71 a).
In the model with combined variables, G/W and L/W were used (G/W is an
indicator of shape and L/W an indicator of size). In this case, region was moderately
significant as a result of a univariate test with L/W, although the multivariate test
showed region as an overall nonsignificant source of variation (Table 2). The most
important significance was obtained with cultivated type as a source of variation,
particularly when tested against the combined variable G/W, that is, shape. The
mean separation test indicated that Papa is the roundest bean, Sieva is the flattest,
and Sieva/Papa again an intermediate type (mean values and mean separation with
EMS = 0.004: Sieva 0.48 c; Sieva/Papa 0.62 b; Papa 0.82 a).
Hence, the most discriminatory criteria and, consequently, those that define the
cultivated types are girth (G) in the simple variable model and shape (G/W) in the
combined variable model.
123
2858 Biodivers Conserv (2007) 16:28472865

In general, analysis of morphological measurements fit the classification of the

cultivated bean types, Sieva, Sieva/Papa, and Papa, indicating that morphological
data may be used to effectively discriminate among cultivated types. However, no
association was found between morphological data and geographical origin.

AFLP polymorphism

Twenty-seven and 35 polymorphic bands were generated, respectively, with the

primer combinations EcoRI + ACA/MseI + CAC and EcoRI + ACA/MseI + CAT.
The scored fragment sizes ranged from 450 to 110 bp. The AFLP profiles were
highly reproducible in replicated assays of the same individual.
The observation of bands in the gels showed: (i) 46 unique profiles, (ii) 11 profiles
shared by two accessions, (iii) a common profile for a group of three accessions, and
(iv) a profile shared by five accessions. In total, 59 different band profiles were
detected for the 76 accessions analyzed. This indicates the presence of a significant
level of variation in the samples and that the selected primer combinations had good
discriminatory power.

Molecular diversity of the Cuban accessions

The molecular profiles of the accessions grouped by cultivated type and origin
showed unique bands (Table 4), except for the Papa type, which shared all bands. If
a group of accessions had unique bands, they represented at least 5% of the poly-
morphic ones.
As a group, accessions from the central region presented 42 out of the 62 poly-
morphic bands detected. Eleven of the 13 unique bands specific to this region were
present in the wild accession collected from a home garden in Cienfuegos Province.
If this wild accession was removed from the sample, the number of polymorphic

Fig. 2 Spatial distribution of groups of cultivated home gardens accessions obtained with the three
main axes of variation after multiple correspondence analysis
123
Biodivers Conserv (2007) 16:28472865 2859

bands (21) and unique bands (two) contained in the central region was then com-
parable with the other regions (Table 4). The comparison of genetic diversity in
home garden accessions according to their cultivated type showed that the amount
of genetic diversity (Hi) within types was almost identical to the total diversity (HT),
which means that diversity among types is very low and, therefore, that the coeffi-
cient of differentiation is also low (GST<0.001) (Table 4).
Given that the sample contained one wild accession and that this accession
accounted for several unique bands, the diversity analysis based on geographical
region was done on the total sample from home gardens both with and without the
wild accession. The values of total genetic diversity with the wild accession
(HT = 0.130, SD = 0.219), the diversity within the central region (Hi = 0.141,
SD = 0.224), the average diversity among regions (HS = 0.122, SD = 0.018), and the
coefficient of genetic differentiation (GST = 0.069) were not significantly different
when compared with those calculated, excluding the wild accession (Table 4). Dif-
ferences among cultivated accessions from these three samples (western, central, and
eastern) accounted for 8% of the total diversity. Therefore, the analysis of genetic
diversity based on the geographical origin of the accessions showed that diversity
within regions was similar.

Multiple correspondence analysis

The entire sample

The clustering analysis applied to the coordinates generated by the multiple corre-
spondence analysis (MCA) and based on the R2 value indicated that eight groups
(AH, Table 1) explained 97% of the total variation present in the sample.
All the cultivated Cuban accessions were included in groups A, B, and C. Group
A included 25 Cuban accessions, three accessions from Mexico (G26170, G26171,
and G26173), and one accession from Panama. Group B was composed of six Cuban
accessions. The three cultivated accessions from the Cuban ex situ collection clus-
tered in group C, together with 28 other accessions from several Cuban home gar-
dens. Group D was formed by two Mexican accessions (G26174 and G26287), one
from Costa Rica, one from Honduras, and one from El Salvador. Group E contained
only one accession from Guatemala (G25992). The two wild accessions from Peru
formed group F. The wild accession collected from a Cuban home garden and the
other wild accession from the Cuban ex situ collection formed groups G and H,
respectively.

Table 5 Group composition (A, B and C) according to the cultivated type and origin of the Cuban
accessions collected in home gardens

Group n Cultivated Origin

type
S P S/P W C E

A 25 13 5 7 7 15 3
B 6 5 0 1 2 2 2
C 28 14 7 7 3 7 18

n: number of accessions per group; S: Sieva; P: Papa; SP: Sieva/Papa; W: West; C: Center; E: East

123
2860 Biodivers Conserv (2007) 16:28472865

A genetic diversity analysis was performed on the eight groups (data not shown)
and the resulting coefficient of genetic differentiation (GST) indicated that the dif-
ferences among groups accounted for 42% of the total variation of the sample
(HT = 0.163, HS = 0.099). In general, the eight groups separated the wild from
cultivated accessions.

Cuban cultivated accessions

The three genetic groups that contained the cultivated accessions collected from
home gardens were resolved by the third dimension of the multiple correspondence
analysis (Fig. 2). The second dimension distinguished between groups A and C.
Comparing the band profiles of accessions collected from home gardens (groups
A, B, and C) permitted detection of unique bands in each group, none of which was
present in all the accessions in a given group (Table 4). The cultivated accessions of
the old Cuban ex situ collection contributed three polymorphic bands, out of which
one was unique to the group (C). Similarly, the number of total polymorphic and
unique bands present in group A increased because of the presence of Mesoamer-
ican accessions, which contributed seven polymorphic bands (three unique bands). It
is noteworthy that only four of the Mesoamerican accessions added substantial new
polymorphism to group A. This clustering of Mesoamerican accessions (three
Mexican and one Panamanian) in group A indicates a degree of relationship among
cultivated types from Cuba and Mesoamerica, especially Mexico. These results
agree with previous studies by Lioi et al. (1991), who found the Mesoamerican seed
protein profile present in accessions type Sieva and Papa from the now lost ex situ
Cuban collection.
The rest of the Mesoamerican accessions (six) that formed groups D and E were
significantly distant from groups A, B, and C (data not shown). This discrimination
between Mesoamerican and Cuban cultivated accessions might have resulted from
the geographical isolation of the Cuban populations, coupled with autogamy.
Groups A and C were formed by the largest number of home gardens accessions
and showed the highest diversity values (Hi) (Table 4). Group B, formed by six
accessions, showed the lowest diversity. The value of genetic differentiation (GST)
indicated that the differences among these three groups accounted for 24% of the
variation among the home gardens accessions. If the analysis was performed after
including the four accessions from Mesoamerica (in group A) and the three acces-
sions from the Cuban ex situ collection (in group C), similar values of diversity were
obtained; the contribution of these accessions to the diversity of these three groups is
therefore negligible (HT = 0.125, HS = 0.100, GST = 0.204).
Groups A and C included accessions from the three cultivated types, while group
B contained accessions from the Sieva type (83%) and from the Sieva/Papa type
(17%) only. Each genetic group contained accessions collected from home gardens
in all three Cuban regions (Table 5).
For all 76 accessions, certain seed colours were particular to some groups (Table 1).
Primary colours black (7%) and pinkish (10%) were particular to a number of
accessions in group A and dark brown was found only in 10% of the accessions in
group C. Maroon as a primary colour was found exclusively in an accession of group B.
Morphological data [length, width, girth, 100-seed weight (100SW), and the ratios
L/W and G/W] were analyzed against group composition (A, B, and C) as they
assembled all samples from Cuban home gardens to identify possible associations
123
Biodivers Conserv (2007) 16:28472865 2861

between morphological data and groups formed with the molecular data. The
multivariate analysis of variance was not significant for any of the morphological
traits neither for their relationships (data not shown). Even so, observation of the
spatial distribution of groups A, B, and C, resulting from the MCA (Fig. 2), suggests
that the main factor dividing them in the third dimension is the shape (G/W) gra-
dient. Our sample did not allow us to substantiate this visual observation, and
confirmation would be possible only with either an increased number of molecular
markers and/or a larger sample.
Neither morphological nor molecular information of P. lunatus accessions col-
lected from Cuban home gardens showed any association with their collection site
in distinct geographical regions. Morphological data allowed distinction among
cultivated types, while molecular data separated accessions in groups that did not
correspond to the cultivated type. This means, therefore, that the molecular infor-
mation revealed a different type of diversity than that contained in the cultivated
type. Moreover, because morphological diversity also allows a useful type of clas-
sification, it may be stated that both types of information add to each other, and
should be considered as complementary for a thorough assessment of genetic
diversity of P. lunatus.

Discussion

Molecular and morphological polymorphism

The two AFLP primer combinations produced 62 polymorphic bands and 59 band
profiles that allowed differentiating 78% of the accessions analyzed; in addition, they
identified a few unique bands. Despite the discriminatory power of the primers
selected, a few Cuban accessions were not distinguishable, even though they
belonged to different cultivated types and had diverse morphological traits (colour
and seed measurements). Because the primer pairs used differed only in the third
selective nucleotide of the MseI primer, it can be speculated that the use of other
combinations or an increased number of assays with more primer combinations
could lead to higher molecular differentiation of accessions which at present have
identical band profiles.
Two of the 13 repeated molecular profiles were shared by pairs of Mesoamerican
accessions (G26170 and G26171; G26307 and G26308). Nine profiles were shared by
Cuban accessions collected from home gardens within the same region and two
profiles were common to accessions collected in different but adjacent geographical
regions (OC-1-1, C-1-1, C-1-3, C-1-4 and C-6-6; OC-2-1, OC-12-1 and C-6-3).
Phaseolus lunatus is an autogamous species. However, in highly humid tropical
regions such as Cuba, the proportion of allogamy may reach 100% of natural crosses
(Allard 1967). It could be that natural gene flow among genotypes in the same
geographical region is the main reason for the low molecular differentiation of
accessions despite their morphological differences.
Moreover, in Cuba, lima beans are not included in existing commercialization
networks, that is, seed exchange happens through informal farmer systems, which
promotes wide dissemination of seeds, with similar consequences to those of natural
crossing, but also between geographically distant accessions. This has been certainly
aided in recent times by the migration of farmers across the country. These situations
123
2862 Biodivers Conserv (2007) 16:28472865

help the flow of genetic material between regions and may explain the difficulty
encountered in molecularly distinguishing some of the accessions collected from
different home gardens. In fact, AFLP polymorphism reflects changes in random
neutral loci and, as such, those changes are not necessarily correlated with specific
phenotypes.
Morphological descriptorsgirth and G/W in particulardifferentiate cultivated
types, possibly resulting from intense selection pressure for these characteristics in
cultivated beans. In contrast, as a whole, AFLP bands did not single out bean types.
They do, however, reflect objective differences in the DNA sequence and, as a
consequence, allow a different and perhaps more detailed calculation of genetic
variation. Even so, any application of a molecular marker technology will always be
limited by the number of markers (amount of polymorphism) obtained and as such
the calculation of variation will be an estimate of the total variation, depending on
the number of markers analyzed.

Genetic diversity of P. lunatus in home gardens

Home gardens represent a unique reservoir of agrobiodiversity. They store species

and landraces for long periods of time and accept new genotypes, depending on the
requirements and preferences of the families managing them.
The Cuban component of the global home gardens project (Watson and Eyzag-
uirre 2002), which included the work presented here, proposed that the three main
geographical regions be considered as minimum in situ conservation units of genetic
resources in home gardens (Castineiras et al. 2002). Moreover, these regions were
selected to be part of the national strategy for in situ conservation and management
of natural biodiversity. The choice of these regions was made according to both
biological and social considerations.
This study suggests that, together with conventional ecogeographic criteria,
morphological and molecular information help define minimum conservation units
of genetic diversity. The comparison of the total genetic diversity of the Cuban
accessions as a whole (HT = 0.119) with the diversity explained by the distribution of
accessions into regions (western, central, and eastern) (GST = 8%) shows that the
factor region is not sufficient to define a minimum unit of conservation. However,
although collecting missions carried out in Cuba have identified the presence of wild
lima beans in the central region (Cienfuegos Province), no guarantee exists that the
genetic diversity in this region is, for that reason, particularly high or unique. Only
molecular analysis can determine the presence of low frequency genetic variants. For
example, the wild accession collected in Cienfuegos showed 11 unique bands, which
confirmed the assumption that exceptional variation exists in the central part of the
country.

Restoring the ex situ collection

The definition of eight basic groups based on multiple correspondence analysis

(molecular data only) shows that the sample had a structure. Furthermore, the fact
that three genetic groups include all the cultivated accessions found in home gardens
and that these groups do not correspond to morphological or ecogeographic clusters
suggests that a sound strategy to restore the ex situ collection of P. lunatus should
not rely either on morphological characters or on the geographical location of
123
Biodivers Conserv (2007) 16:28472865 2863

genotypes, but should also take into account the information offered by the three
genetic groups. The comparison of cultivated types does not suggest the existence of
a population structure (GST< 1%), and the value of the coefficient of genetic dif-
ferentiation (GST) for groups A, B, and C is three times the value obtained from
comparing regions, that is, the molecular analysis better differentiates genetic
groups.
Particular features of the genetic structure of the sample of Cuban cultivated
lima bean accessions are (i) the presence of three genetic groups; (ii) a tendency
for central and western accessions to come together in group A, and eastern
accessions in group C; and (iii) all the accessions in group B shared the characters
of the Sieva cultivated type (overall, flat seeds) and were equally distributed in the
three geographical regions (two each) (Table 5). In addition, accessions collected
in the same home garden appeared in the three genetic groups (e.g. accessions
OC-12, C-6, and O-7) (Table 1). This observation may point to the fact that,
despite close proximity, natural crossing of genotypes does not always exist.
Phenological studies of P. lunatus have not been carried out in Cuba, but some
populations in the same home garden have been observed to have unsynchronized
flowering, which restricts mating among populations in the same location and
favours genetic differentiation. Also important to note is that, while P. lunatus is
less frequent in the west, western accessions were present in the three genetic
groups, thus supporting the significance of that region for future collecting
missions.
This work confirms the importance of home gardens to the endeavour of restoring
the lost ex situ collection of P. lunatus. However, it also suggests that the selection of
germplasm to restore the collection be done, not only on the location of those home
gardens nor only on the morphotypes to which accessions belong, but also on rep-
resentative subsets of the genetic groupings detected through the multiple corre-
spondence analysis of molecular data. Hence, a more appropriate description of
these genetic groups should be established by conducting further research (adding
more markers and/or increasing the sample size) to uncover the relationships
between these groups and criteria other than those involving cultivated type and
origin. This statement is further supported by the observation that groups A and C,
which are divided by the multiple correspondence analysis, do not differ in their
composition according to cultivated type but do in their composition according to
geographical origin (Table 5).
The overall results of this work emphasize the relevance of using appropriate
information to define conservation strategies for P. lunatus. The combination of
morphological evidence with the use of a molecular marker technology and proper
statistical analysis to understand the genetic diversity of varieties held in home
gardens may provide novel and unexpected information to support a rigorous def-
inition of strategies for complementary conservation.

Acknowledgements This research was a spin-off from the global project on genetic resources in
home gardens carried out by the International Plant Genetic Resources Institute (IPGRI) and
entitled Contribution of Home Gardens to the in situ Conservation of Plant Genetic Resources in
Farming Systems, which was partially funded by the German Federal Ministry for Economic
Cooperation and Development (BMZ) through GTZ (Deutsche Gesellschaft fur Technische
Zusammenarbeit). The authors also wish to express their appreciation to Maritza Garca and Fidel
Hernandez (Estacion Ecologica Sierra del Rosario) and to Celerina Giraudy (Union de Servicios
Ambientales de Guantanamo), members of the Cuban component of the global Projects research

123
2864 Biodivers Conserv (2007) 16:28472865

team for their help in collecting samples and morphological characterization; to Joe Tohme (CIAT)
for allowing use of laboratory facilities at the CIAT Biotechnology Unit; to Luigi Guarino, Xavier
Scheldeman, Jose Luis Chavez, and V. Ramanatha Rao for their comments that helped improve the
manuscript.

References

Allard RW (1967) Principios de la mejora genetica de las plantas. Edicion Revolucionaria, La

Habana, Cuba
Capo-chichi LJA, Weaver DB, Morton CM (2001) AFLP assessment of genetic variability among
velvetbean (Mucuna sp.) accessions. Theor Appl Genet 103:11801188
Castineiras L, Esquivel M, Rivero N, Marino A (1991) Variabilidad en semillas de P. lunatus L.
colectadas en Cuba. Revista del Jardn Botanico Nacional 12:109114
Castineiras L, Shagarodsky T, Fuentes V, Fundora Z, Fernandez L, Moreno V, Barrios O, Sanchez
P, Walon L, Perez MF, Puldon G (2001) El frijol caballero (Phaseolus lunatus L.) un cultivo
marginal y en peligro de erosion genetica en Cuba. Revista del Jardn Botanico Nacional
22(1):133138
Castineiras L, Fundora-Mayor Z, Shagarodsky T, Moreno V, Barrios O, Fernandez L, Cristobal R
(2002) Contribution of home gardens to in situ conservation of plant genetic resourcesCuban
component. In: Watson JW, Eyzaguirre PB (eds) Proceedings of the second international home
gardens workshop: contribution of home gardens to in situ conservation of plant genetic
resources in farming systems. Witzenhausen, Federal Republic of Germany, 1719 July 2001,
IPGRI, Rome Italy, pp 4256
de Vicente MC (2002) Molecular techniques to facilitate prioritization of plant genetic resources
conservation and further research. AgBiotechNet 4(ABN092):15
Esquivel M, Castineiras L, Lioi L, Hammer K (1990) Origin, classification, variation and distribution
of lima bean (P. lunatus L.) in the light of Cuban materials. Euphytica 49:8997
Fundora Z, Shagarodsky T, Castineiras L (2004) Sampling methods for genetic diversity study in
home gardens in Cuba. In: Eyzaguirre P, Linares O (eds) Home gardens and agrobiodiversiy.
Smithsonian Books, Washington, pp 5657
Guzman FA, Ayala H, Azurdia C, Duque MC, de Vicente MC (2005) AFLP assessment of genetic
diversity of Capsicum genetic resources in Guatemala: home gardens as an option for conser-
vation. Crop Sci 45:363370
Hongtrakul V, Gordan MH, Knapp SJ (1997) Amplified fragment length polymorphism as a tool for
DNA ingerprinting sunflower germplasm: genetic diversity among oilseed inbred lines. Theor
Appl Genet 95:400407
Hoogendijk M, Williams DE (2002) Characterizing the genetic diversity of home gardens crops:
some examples from the Americas. In: Watson JW, Eyzaguirre PB (eds), Proceedings of the
second international home gardens workshop: contribution of home gardens to in situ conser-
vation of plant genetic resources in farming systems. Witzenhausen, Federal Republic of
Germany, 1719 July 2001, IPGRI, Rome, Italy, pp. 3440
IBPGR (1982) Descriptors of Phaseolus lunatus. International Board for Plant Genetic Resources,
Rome, Italy
Kim MS, Moore PH, Zee F, Fitch MMM, Steiger DL, Manshardt RM, Paull RE, Drew RA, Sekioka
T, Ming R (2002) Genetic diversity of Carica papaya as revealed by AFLP markers. Genome
45(3):503512
Lioi L, Esquivel M, Castineiras L, Hammer K (1991) Lima bean (Phaseolus lunatus L.) landraces
from Cuba: electrophoresis analysis of seed storage proteins. Biol Zent bl 110:7679
Maughan PJ, Saghai Maroof MA, Buss GR, Huestis GM (1996) Amplified fragment length poly-
morphism (AFLP) in soybean: species diversity, inheritance and near-isogenic lines. Theor Appl
Genet 93:392401
Murray MG, Thompson WF (1980) Rapid isolation of high molecular weight plant DNA. Nucleic
Acids Res 8:43214325
Nei M (1987) Molecular evolutionary genetics. Columbia University Press, New York, USA
Shagarodsky T, Castineiras L, Fuentes V, Cristobal R (2004) Characterization in situ of the vari-
ability of sapote or mamey in Cuban home gardens. In: Eyzaguirre P, Linares O (eds) Home
gardens and agrobiodiversity. Smithsonian Books, Washington, pp 266281
Tohme J, Gonzalez DO, Beebe S, Duque MC (1996) AFLP analysis of gene pools of a wild bean
core collection. Crop Sci 36:13751384
123
Biodivers Conserv (2007) 16:28472865 2865

Vallejos CE, Sakiyama NS, Chase CD (1992) A molecular marker-based linkage map of Phaseolus
vulgaris L. Genetics 131(3):733740
Vargas EM, Castro E, Macaya G, Rocha OJ (2003) Variacion del tamano de frutos y semillas en 38
poblaciones silvestres de Phaseolus lunatus (Fabaceae) del Valle Central de Costa Rica. Revista
de Biologa Tropical 51(3):707724
Vos P, Hogers R, Bleeker M, Reijans M, van de Lee T, Hornes M, Frijters A, Pot J, Peleman J,
Kuiper M, Zabeau M (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acids Res
23:44074414
Watson JW, Eyzaguirre PB (eds) (2002) Proceedings of the second international home gardens
workshop: contribution of home gardens to in situ conservation of plant genetic resources in
farming systems, 1719 July 2001, Witzenhausen, Federal Republic of Germany. International
Plant Genetic Resources Institute, Rome

123