Project Plant
Resume:
Genetic diversity evolution through participatory maize
breeding in Portugal
Maria Carlota Vaz Patto
Pedro Manuel Moreira
Nuno Almeida Zlatko Satovic
Silas Pego
Received: 31 October 2006 / Accepted: 6 June 2007 / Published online: 5 July 2007
Springer Science+Business Media B.V. 2007
Abstract Natural, and in particular, arti cial (hu- that the variation among selection cycles represented
man) selection may pose a danger to the existing crop only 7% of the total molecular variation. However,
genetic diversity. Nevertheless, on-farm breeding the number of private alleles varied among the
systems seem to achieve phenotypic improvements selection cycles, being the highest detected at the
even though preserving variability. Using SSR beginning of the selection project. These ndings
markers, we analysed several selection cycles, over demonstrate that an allele ow took place during the
a 20 years period, of a Portuguese on-farm partici- on-farm selection process of Pigarro but the level of
patory maize OPV- Pigarro breeding project. No genetic diversity was not signi cantly in uenced.
signi cant differences in allelic richness (Nar), Since interesting phenotypic improvements were also
observed heterozygosity (HO), expected heterozygos- achieved, on-farm breeding projects, like this one,
ity (or gene diversity; HE) or inbreeding coef cient should be valued as a way to preserve unique
(f) were detected among the selection cycles. 58 out Portuguese maize landraces in risk of disappearing.
of 107 alleles were common to all the selection cycles
Keywords Genetic diversity Maize On-farm
studied. The analysis of molecular variance showed
Participatory breeding SSR Zea mays L.
M. C. Vaz Patto N. Almeida
Instituto de Tecnologia Qu mica e Biologica, Plant Cell
Introduction
Biotechnology Lab, Universidade Nova de Lisboa, Apt.
127, Oeiras 2781-901, Portugal
e-mail: ******@****.***.** Maize was introduced in Portugal during the XVI
century and spread rapidly throughout the country.
P. M. Moreira
The establishment and further expansion of this new
Departamento de Fitotecnia, Escola Superior Agraria de
crop during the XVII and XVIII centuries, in a
Coimbra, Bencanta, Coimbra 3040-316, Portugal
polycrop system (maize + beans + forage), lead to an
Z. Satovic
agricultural revolution, enhancing the rural popula-
Faculty of Agriculture, Department of Seed Science and
tion s standard of living. Numerous landraces (open
Technology, University of Zagreb, Svetosimunska 25,
pollinated varieties, OPV) have been developed
Zagreb 10000, Croatia
during the centuries of cultivation, adapted to speci c
S. Pego
regional growing conditions as well as farmer s
Estacao Agronomica Nacional (EAN), Instituto Nacional
needs. However, after World War II, Portugal was
de Investigacao Agraria, Av. Republica, Oeiras 2784-505,
one of the rst European countries to introduce
Portugal
123
284 Euphytica (2008) 161:283 291
production still plays an important economic and
American hybrids which initially were not well
social role in Central and Northern Portuguese rural
accepted by the Portuguese farmers due to several
communities. This bread making ability seems to
handicaps as late maturity or kernel type, not tted
depend on a range of particular traits not found in the
for food. Subsequently, due to the big accomplish-
available commercial hybrid varieties, and this is
ments of the maize hybrid development in Portugal,
probably why traditional maize landraces have not,
with several national breeding stations releasing
in these regions, been totally replaced by hybrid
adapted hybrid varieties, maize landraces were pro-
varieties.
gressively replaced. Nevertheless, since the late 70 s
One regional maize OPV was selected based on
there has been a growing concern that numerous
the farmers needs and introduced in the PPB project,
Portuguese maize landraces may have been lost
at that time with support of CIMMYT. The selected
forever. In 1975, Portugal took the rst initiative of
OPV was named by the farmer as Pigarro .
collection of maize germplasm, and later on, with
Pigarro is a white int type, with FAO 300 maturity
FAO support, a national collection program was
rating, high level of fasciation, bread making ability
implemented and a national germplasm bank was
and adapted to polycropping (with beans). Since then,
established in the city of Braga. However, genetic
a farmer s selection criterium based on mass selec-
diversity was still being lost on the farmers elds.
tion methodology was applied to Pigarro, on the
It is said that participatory plant breeding (PPB)
farmer s eld, by the farmer himself, but in close
may encourage farmers to continue growing landrac-
collaboration with the breeder (Silas Pego). The
es by enhancing their current use value (Smale et al.
breeder, on the other hand, worked side by side with
2003). PPB, with the involvement of farmers, uses
the farmer, using recurrent selection methodologies
mostly material generated from crosses among local
with careful respect for the local traditional agricul-
landraces, leading to a dynamic form of in situ
ture, accepting low input and intercropping charac-
genetic conservation and genetic enhancement. On
teristics, and favouring diversity (as the basis for pest
the other hand, PPB programs meet the needs of low-
and disease tolerance) and quality as priorities.
input, small-scale farmers who are often overlooked
In the case of the farmer s selection, a two parental
by conventional crop breeders. The returns from PPB,
control mass selection was applied to Pigarro . In the
compared to conventional breeding, are generally
eld, before pollen shedding, male owers were
higher because it costs less and the bene ts to farmers
detasseled, and the weakest, diseased and pest
are realised earlier (Virk et al. 2003; Witcomb et al.
susceptible plants were removed. After that, and just
2003).
before harvesting, plants were again selected based
Taking all this into account, Silas Pego led, in
on the ear size, root and stalk quality, pest and disease
1984, a detailed survey on farmer s maize elds at
tolerance and proli cacy. Finally, at the storing
Vale do Sousa Region (Sousa Valley Region) in
facilities, after harvesting, selection was focused on
the Northwest of Portugal. The collected materials
the ear length and the kernel row number, avoiding
were the starting point of a PPB project, with
damage to ears. Signi cant agronomic improvements
simultaneous on-farm breeding and on-farm conser-
have been achieved with this approach (Pego and
vation objectives (VASO- Vale do Sousa - project).
Antunes 1997).
The project was focused on solving the problem of
The seeds of each selection cycle (in a total of 20
small farmers with scarce land resources, due to high
mass selection cycles) have been put in cold storage.
demographic density; with polycroping production
Concern has been expressed that genetic diversity
systems, quality being the rst priority over quantity.
might be reduced by natural and arti cial (human)
The Sousa valley is a traditional maize cultivation
selection. The main commercial maize hybrids that
region with polycropping systems for human uses
have substituted the traditional OPVs worldwide,
(bread production), which is very fertile, with good
involve a restricted number of key inbred lines, thus
water availability and local germplasm adapted to
limiting the available genetic diversity. However, it is
local conditions during centuries of cultivation. This
known that traditional farmers, when selecting their
PPB project concerned mainly int-type OPV land-
landraces seed, have been successful in preserving
races with technological ability for the production of
variability as a way to guarantee production under
the traditional maize bread called broa . Broa
123
Euphytica (2008) 161:283 291 285
any circumstances (Pego and Antunes 1997). Only biotech) by comparison with internal and external
recently an objective assessment of genetic diversity size standards. Size estimates were rounded up or
has become possible with the introduction of molec- down using the criteria de ned by Matsuoka et al.
ular marker technologies. SSR markers have proven (2002) as described in Vaz Patto et al. (2004). To
their ef ciency as genetic markers to assess genetic reduce variance in the estimate of fragment sizes
diversity evolution in numerous plant species between runs, control samples (B73) were run
(Khlestkina et al. 2004; Struss and Plieske 1998; Le repetitively on all the gels corresponding to the same
Clerc et al. 2005; Maccaferi et al. 2003). SSR locus.
To study the genetic diversity evolution during The 16 SSR markers used in the study were chosen
traditional maize landraces development, we analy- from MaizeDB based on their repeat unit and bin
sed different selection cycles of the on-farm partic- location. The SSR primers were scattered throughout
ipatory maize breeding project (VASO project), in the maize genome and represented various repeat
progress since 1984, in the Portuguese Sousa Valley classes (Table 1). Primer sequences are available
region, with simple sequence repeat (SSR) markers. from the MaizeDB (www.maizegdb.org).
Data analysis
Materials and methods
The GDA software (Lewis and Zaykin 2001) was
Plant materials used for calculating allele frequencies and estimating
the average number of alleles (Na), number of private
PPB using mass selection was carried out within the alleles (Npa), the observed and expected heterozyg-
VASO project each year and OPVs seed was stored osities (HO, HE) and xation index (f) in each
from each selection cycle. selection cycle. FSTAT v. 2.9.3.2 programme pack-
From each of three different Pigarro on-farm age (Goudet 1995, 2002) was used for estimating the
mass selection cycles (1984, 1993 and 2004), allelic richness Nar as the measure of the number of
approximately 30 individuals were randomly selected alleles per locus independent of sample size. The
from seed. In total, 89 individuals were analysed estimates of Nar, HO, HE and f in each selection cycle
using SSRs markers. In addition, B73, an US inbred were compared using the Kruskal-Wallis test in SAS
line, was used as a control. software (SAS Institute 1999).
GENEPOP v. 3.4 (Raymond and Rousset 1995)
SSR ngerprints was used to test genotypic frequencies for confor-
mance to Hardy-Weinberg (HW) expectations, to test
DNA was isolated from 2-week old seedlings, the loci for linkage disequilibrium and to estimate the
employing a modi ed CTAB procedure (Saghai- signi cance of genic differentiation between selec-
Maroof et al. 1984). tion cycle pairs. All signi cance tests were based on a
SSR marker technique was performed as described Markov chain method (Guo and Thompson 1992;
by Vaz Patto et al. (2004). Fragment analysis was Raymond and Rousset 1995) using 10,000 de-mem-
carried out using an automated laser uorescence orization steps, 100 batches and 5,000 iterations per
(ALFexpress II) sequencer (Amersham Biosciences). batch. Sequential Bonferroni adjustments (Holm
For this, 0.5 ll of each ampli cation reaction was 1979; Rice 1989) were applied to correct for the
mixed with 3 ll of formamide loading buffer, 3 ll of effect of multiple tests using SAS Release 8.02 (SAS
TE (pH 8.0) and 0.3 ll of each of two internal sizers Institute 1999).
labelled uorescently (Cy5) with sizes anking the The distribution of gene diversity was conducted
ampli ed fragments. After denaturation at 948C for according to the model proposed by Nei (1973), in
3 min, and cooling down on ice, the 8 ll samples which the total genetic diversity mean (HT) is
were loaded onto a standard sequencing gel (Repro partitioned in two components: the gene diversity
gel, Amersham Biosciences). Fragment sizes were mean within selection cycles (HS), and between
determined using the computer program ALFwin cycles (DST). The proportion of total gene diversity
Fragment analyser v. 1.00 (Amersham Pharmacia between cycles (GST), or genetic differentiation, was
123
286 Euphytica (2008) 161:283 291
Table 1 Repeat motifs, size ranges and number of alleles for 16 SSR loci used in 89 maize plants from three selection cycles
(SC1984, SC1993, SC2004)
Locus Repeat motif Bin location Size range No. of alleles
SC1984 SC1993 SC2004 Total
umc1013 GA 1.08 129 167 5 5 4 5
umc1823 TG 2.02 85-165-**-**-*-**
umc1635 GAAGG 2.05 119 144 3 3 4 4
umc1907 AT 3.05 109-***-**-*-*-**
umc1528 TGCG 3.07 148 164 3 3 3 3
bmc2323 AG 5.04 144-***-**-*-*-**
umc1524 GGACTG 5.06 128 158 4 4 3 4
umc1143 AAAAT 6.00 73 83 3 3 3 3
umc1229 AG 6.01 218-***-**-** 9 14
umc1066 GCCAGA 7.01 138 150 2 2 3 3
umc1483 ACG 8.01 154 160 3 3 3 3
umc1858 TA 8.04 117 159 6 4 6 7
umc1279 CCT 9.00 91 100 4 2 3 4
umc1120 GGCAT 9.04 92 112 3 3 4 4
umc2067 CATG 10.03 143 155 3 3 3 3
umc2021 TGG 10.07 112 136 5 6 6 6
Total 92 79 80 107
Average 5.75 4.94 5.00 6.69
calculated as GST = DST/HT. The partition of total selection cycles, while 21 were private alleles.
genetic diversity was performed separately for SC1984 Average frequencies of private alleles (4.45%) and
vs. SC1993, SC1993 vs. SC2004, and SC1984 vs. SC2004 alleles found in two out of three selection cycles
using FSTAT. (6.45%) were considerably lower than the average
The proportion-of-shared-alleles distance (Bow- frequencies of common alleles (24.97%). As
cock et al. 1994) between pairs of individuals was expected, the highest number of private alleles (12)
calculated using MICROSAT (Minch et al. 1997) and was detected in SC1984.
the distance matrix was subjected to the analysis of The average number of alleles per selection cycle
molecular variance (AMOVA; Excof er et al. 1992) was the highest in SC1984 (5.750) and the lowest in
using ARLEQUIN version 2.000 (Schneider et al. SC1993 (4.938), but the gene diversity (HE) had even
2000). The signi cance of /-statistics was obtained slightly increased from 0.599 in SC1984 to 0.612 in
non-parametrically after 106 permutations. SC2004. However, no signi cant differences were
observed among the three selection cycles in any of
the analysed parameters including Nar, observed
heterozygosity (HO), expected heterozygosity (or gene
Results
diversity; HE) and inbreeding coef cient (f) (Table 2).
After accounting for multiple comparisons, only
A total of 107 alleles were detected within 89
three loci (umc1823, umc1907, umc1229) were
individuals across 16 SSR markers (Table 1). The
signi cantly out of Hardy-Weinberg equilibrium
number of alleles/locus ranging from 3 (umc1528,
(P
umc1143, umc1066, umc1483 and umc2067) to 17
of homozyogotes. Additionally, locus umc1524
(umc1907), with a mean value of 6.69 alleles/locus.
showed signi cant heterozygotes excess in SC2004
Out of 107 alleles, 58 were common to all the
(Table 3).
selection cycles, 28 were detected in two out of three
123
Euphytica (2008) 161:283 291 287
Table 2 Genetic variability estimates for three selection cycles
Selection cycle n Na Nar Npa HO HE f
SC1984 30 5.750 3.861 12 0.498 0.599 0.170
SC1993 29 4.938 3.611 3 0.588 0.606 0.031
SC2004 30 5.000 3.640 6 0.529 0.612 0.140
Mean 5.229 3.896 0.538 0.606 0.114
P* 0.938 0.238 0.996 0.310
* P-value of Kruskal-Wallis test among selection cycles. n: number of individuals, Na: average number of alleles, Nar: allelic
richness, Npa: number of private alleles, HO: observed heterozygosity, HE: gene diversity or expected heterozygosity, f: inbreeding
coef cient
Table 4 P-values for test of null hypothesis that assumes
Table 3 Inbreeding coef cients per locus and selection cycle
identical allelic distribution across cycles
and signi cant deviations from Hardy-Weinberg equilibrium
Locus SC1984/ SC1993/ SC1984/
Locus SC1984 SC1993 SC2004
SC1993 SC2004 SC2004
umc1013 0.461 0.281 0.387
bmc2323 1.000* 0.036 0.004
0.319
umc1635 0.239 0.165
umc1013 1.000 0.000 0.095
umc1823 0.638** 0.177* 0.475**
umc1066 0.091 1.000 1.000
0.162 0.193
umc1528 0.058
umc1120 1.000 0.687 0.022
umc1907 0.497** 0.344** 0.537**
umc1143 1.000 0.672 0.907
0.664 0.365 0.843**
umc1524
umc1229 0.000 0.278 0.000
bmc2323 0.110 0.025 0.005
umc1279 1.000 1.000 1.000
0.425
umc1143 0.053 0.182
umc1483 1.000 1.000 1.000
umc1229 0.311* 0.385** 0.350*
umc1524 1.000 0.890 1.000
0.033 0.055
umc1066 0.123
umc1528 0.493 1.000 1.000
0.082
umc1858 0.081 0.041
umc1635 1.000 1.000 1.000
0.290
umc1483 0.223 0.106
umc1823 0.000 0.000 0.000
0.111
umc1279 0.147 0.101
umc1858 0.011 0.000 0.000
0.048
umc1120 0.232 0.388
umc1907 1.000 0.116 0.004
0.287 0.230
umc2067 0.025
umc2021 1.000 1.000 1.000
umc2021 0.219 0.127 0.150
umc2067 1.000 1.000 1.000
Signi cant deviations from Hardy-Weinberg equilibrium after No. of 3 4 6
sequential Bonferroni corrections: ** corresponds to signi cant
signi cance at the 1% nominal level, and * signi cance at tests
the 5% nominal level; no marking depicts non-signi cant
* P-value as obtained after sequential Bonferroni corrections
values
Among a total of 360 tests for linkage disequilib- considerable, but also the cycles SC1984 and SC2004
rium between pairs of loci, 30 were signi cant at did not differ greatly in allelic frequencies.
P
Contact this candidate