Theor Appl Genet (1997) 95 : 431—441 ( Springer-Verlag 1997
M. Fregene · F. Angel · R. Gomez · F. Rodriguez
P. Chavarriaga · W. Roca · J. Tohme
M. Bonierbale
A molecular genetic map of cassava (Manihot esculenta Crantz)
Received: 5 July 1996 / Accepted: 22 November 1996
Abstract A genetic linkage map of cassava has been
constructed with 132 RFLPs, 30 RAPDs, 3 micro- Introduction
satellites, and 3 isoenzyme markers segregating from
the heterozygous female parent of an intraspecific The genetics of cassava (Manihot esculenta Crantz) are
cross. The F cross was made between ‘TMS 30572’ the least understood of any of the major staple crops
1
and ‘CM 2177-2’, elite cassava cultivars from Nigeria (including rice, maize, wheat, and potatoes) that feed
and Colombia, respectively. The map consists of 20 mankind. This discrepancy is due to the heterozygous
linkage groups spanning 931.6 cM or an estimated nature of the crop, a lack of classical markers, its long
60% of the cassava genome. Average marker density is growing cycle of 9—18 months, its low seed yield per
1 per 7.9 cM. Since the mapping population is an pollination (a maximum of three seeds per cross), and
F cross between heterozygous parents, with unique the limited funding for research on this important crop.
1
alleles segregating from either parent, a second map Cassava is a strongly outcrossing monoecious species
was constructed from the segregation of 107 RFLPs, 50 (mediated by protogyny) and suffers from inbreeding
RAPDs, 1 microsatellite, and 1 isoenzyme marker from depression, making it difficult to develop appropriate
the male parent. Comparison of intervals in the male- stocks for classical genetic studies. Although the crop is
and female-derived maps, bounded by markers hetero- considered to be a segmental allopolyploid (Magoon
zygous in both parents, revealed significantly less et al. 1969) or an allopolyploid (Umannah and Hart-
meiotic recombination in the gametes of the female man 1973), nothing is known about the diploid ances-
than in the male parent. Six pairs of duplicated loci tors of cassava’s 36 somatic chromosomes, nor has
were detected by low-copy genomic and cDNA se- a classical genetic map been developed. Biochemical
quences used as probes. Efforts are underway to satu- marker sudies have recently provided new insights into
rate the cassava map with additional markers, to join the cassava genome. Analysis of ten isoenzyme loci in
the male- and female-derived maps, and to elucidate cassava revealed predominantly disomic inheritance
genome organization in cassava. (Roca et al. 1992; Levefre 1993). Comparison of in-
tralocus and interlocus heterodimeric bands from
Key words Cassava · Molecular markers · isoenzyme analysis of diploid tissue (leaves) and a hap-
Genetic mapping · Polyploidy loid mixture (pollen grains) made it possible to differen-
tiate between heterozygosity (which is characteristic of
cassava) and duplicated loci, providing evidence of
Communicated by J. W. Snape locus duplication in at least two of the isoenzyme loci
M. Fregene ( ) · F. Angel · F. Rodriguez studied (Levefre and Charrier 1993).
W. Roca · J. Tohme Native to the New World tropics and a member of
Biotechnology Research Unit, CIAT, Cali, Colombia family Euphorbiaceae, cassava and some 90 other spe-
M. Bonierbale cies make up the genus Manihot. Cassava is the only
Cassava Program, CIAT, Cali, Colombia widely cultivated member, being valued for its starchy
R. Gomez tuberous roots, which provides food or an export com-
Centro Internacional de Fisica (CIF), Bogota, Colombia modity in nearly all tropical countries of the world.
P. Chavarriaga Evidence of the allopolyploid origin of cassava
Department of Botany, University of Georgia at Athens, (2n"36) relies heavily on its possession of two sets
Athens, GA, USA of dissimilar nucleolar organizing regions, on the
TAG 018
432
repetition of chromosome types (Magoon et al. 1969; in a set of cultivars from different cassava growing
Umannah and Hartman 1973), and on the basic chro- regions (Angel et al. 1993). Restriction fragment length
mosome numbers of other genera in the Euphor- polymorphism (RFLP) between the most genetically
biaceae, which range from 6 to 11 (Perry 1943). No dissimilar pair was around 40%, indicating that intras-
evidence of tetrasomic inheritance or of wild Manihot pecific crosses are suitable for mapping in cassava.
relatives with chromosome numbers of 2n"18 has A molecular genetic map of cassava has the promise of
been found, which supports the allopolyploid theory in helping to identify molecular markers linked to traits of
cassava. On the contrary, chromosome pairing in inter- interest and to apply these in cassava improvement.
specific hybrids between cassava and 10 Manihot Marker-enhanced selection aids the identification of
species, all possessing 2n"36 chromosomes, shows recombinant individuals by permitting a more exact
comparatively normal meiosis, with 18 bivalents at assessment of genotypes bearing introgressed genes of
prophase and occasionally 2—4 univalents or one tet- interest, with minimum flanking DNA, than is possible
ravalent, suggesting a high percentage of genome sim- through phenotype or progeny testing alone (Young
ilarity between cassava and its wild relatives (Hahn et and Tanksley 1989). In addition, markers are a power-
al. 1990; Bai et al. 1992). Normal chromosome pairing ful tool for studying the genetics of complex traits.
at meiosis occurs in hybrids of Manihot species that are
morphologically very different and belong to separate
primary phylogenetic lineages according to cpDNA Materials and methods
studies (Fregene et al. 1994). This combined with the
lack of a strong interspecies boundaries in the genus
Plant materials
has led to the theory that a polyploidization event
occurred before species differentiation (Lefevre and The cassava mapping population is comprised of 90 F plants from
Charrier 1993; Bryne 1984). 1
an intraspecific cross between ‘TMS 30572’ (the female parent), an
Flow cytometry measurements of nuclear DNA in elite cassava cultivar developed at the International Institute of
cassava have revealed a diploid DNA content of Tropical Agriculture (IITA), Nigeria, and ‘CM 2177-2’ (the male
1.67 pg per cell nucleus (Awoleye et al. 1994). This value parent), a successful cassava cultivar resulting from breeding at the
Centro Internacional de Agricultura Tropical (CIAT) in Colombia.
corresponds to 772 mega-base pairs in the haploid The ‘TMS 30572’]‘CM 2177-2’ cross (CM 7857) was chosen from
genome and puts cassava’s genome size at the lower a preliminary evaluation of three intraspecific crosses using RFLP
end of the range for higher plants (Bennet et al. 1992). and random amplified polymorphic DNA (RAPD) markers to de-
The relatively small size of this genome favours the termine the percentage of single-dose markers (Gomez et al. 1995).
development of a saturated genetic map and molecular Apart from being heterozygous, due to the fairly large number of
diverse cassava accessions in their pedigrees, the parental accessions
tags, which may contribute to an understanding of the have the following advantages:
inheritance of many important quantitative traits
despite the heterozygous nature of cassava. 1) ‘TMS 30572’ is highly tolerant to the African cassava mosaic
The literature describes genetic approaches to map- disease (ACMD),
ping polyploid genomes with molecular markers (Ritter 2) ‘CM 2177-2’ shows high photosynthetic rates, good cooking
quality, and tolerance to the cassava mealy bug,
et al. 1991; Wu et al. 1992; Al Janabi et al. 1993; Da 3) both are tolerant to cassava bacterial blight (CBB).
Silva et al. 1994). These approaches attempt to simplify
the determination of allelism by analyzing a special Resistance to ACMD is thought to have been introgressed from
class of markers known as single-dose restriction frag- M. glaziovii into lines leading to ‘TMS 30572’. While a subset of 90
ments (SDRFs) (Wu et al. 1992). SDRFs are DNA individuals was used to develop the first framework map, a total of
150 individuals from the same cross is available for higher resolution
markers that are present in one parent and absent in and quantitative trait mapping. Clonal propagation from stem cut-
the other and segregate in a 1 : 1 ratio (absence : pres- tings in the greenhouse and the field provided leaf tissue for molecu-
ence) in the progeny. They represent the segregation lar analysis and stocks for the production of roots for isoenzyme
equivalent of an allele at a heterozygous locus in a di- analysis.
ploid or an allopolyploid genome or a simplex allele in
an autopolyploid. Linkage analysis using SDRFs in an
F population requires the presence of a number of DNA isolation and restriction enzyme analysis
1
unique segregating polymorphisms (heterozygosity)
and normal meiosis in either or both parents. It results Genomic DNA was isolated from fresh, young tissue of cassava
in two separate linkage maps based on male and female leaves according to Dellaporta et al. (1983). The total DNA obtained
was dissolved gently overnight at 4°C and quantified by fluorimetry
sources of markers. (TKO 100 Hoefer). Large restriction enzyme digestions to produce
Our objectives were to construct a molecular genetic uniform Southern blots were performed with EcoRI, EcoRV, HaeIII,
linkage map of the cassava genome and elucidate HindIII, and PstI according to the manufacturer’s instructions
genome organization using an intraspecific Manihot (Amersham PLC). Ten micrograms of digested cassava genomic
esculenta F cross designed to segregate for several DNA was loaded onto 0.9% agarose gels and electrophoresed in
1 TBE buffer for 16—18 h. Separated DNA fragments were transferred
important traits. As a first step, several restriction by the alkaline method of Southern blotting to Hybond N` nylon
enzyme-by-genomic probe combinations were assessed membranes (Amersham), as recommended by the manufacturer.
TAG 018
433
Parental survey filters contained DNA of the two parents digested RAPD analysis
with the five above-mentioned restriction enzymes. Progeny filters
containing restricted DNA from the 90 F plants, including DNA
1 A slightly modified version of the protocol described by Yu and
from both parents in the first 2 lanes, were replicated, four to eight Pauls (1992) was used to obtain RAPD data (Gomez et al. 1995).
times per restriction enzyme. A total of 740 10-nucleotide oligos of random sequence (Operon
Technologies, Alameda, Calif.) served as primers for the amplifica-
tion of parental DNA and a subset of the progeny in order to detect
polymorphisms. Amplification reactions were carried out in a
Genomic DNA, cDNA libraries, and Southern hybridization 12.5-ll volume containing 25 ng of cassava genomic DNA, 10 mM
TRIS-HCl (pH 9.0), 50 mM KCl, 0.01% Triton X-100, 2.5 mM
Angel et al. (1993) have described the construction of nuclear MgCl , 0.2 mM of each dNTP (New England Biolabs), 0.8 lM,
genomic libraries, totaling 200 clones, from cassava DNA sequences 2
of primer and 1 U ¹aq DNA polymerase (Perkin Elmer-Cetus
generated with HindIII, XbaI, EcoRI, and PstI. Two other PstI Corporation).
genomic libraries of about 1,500 clones each were also constructed
from cassava nuclear DNA according to standard procedures (Sam-
brook et al. 1989; Vayda et al. 1991). The copy number of DNA
sequences was estimated by Southern hybridization of total cassava Microsatellite analysis
DNA, labelled with Horseradish peroxidase, according to the ‘‘ECL
direct’’ protocol of Amersham PLC, with dot blots of whole plasmid Cassava nuclear DNA was digested to completion with EcoRI and
preparations. Genomic clones that produced a strong signal after XhoI restriction enzymes (New England Biolabs) according to the
two 0.5] SSC, 0.4% SDS washes at 55°C for 10 min, detection, and manufacturer’s instructions; 0.3- to 0.8-kb fragments were isolated
autoradiography were judged to be repetitive sequences and were from a 0.5% low-melting-point agarose gel using the phenol chloro-
left out of the parental survey. form purification procedure (Sambrook et al. 1989). Purified frag-
For the cDNA libraries, total RNA was extracted from leaf ments were cloned in lambda ZAP II (Stratagene), and 8]105
tissue using the method of Bothwell et al. (1990); polyadenylated recombinants clones were obtained, as determined from the IPTG,
RNA was purified away from the total RNA preparation using X-gal screening procedure. The library was then screened with
Dyna beads oligo (dT)25 (Dynal inc), and a cDNA library was (GA)15, (GT)15, (AT)15, (TTAG)8, and (TCT)10 oligonucleotides.
constructed from messenger RNA in lambda ZAP II vector, follow- Positive clones were sequenced on an automated sequencer (Applied
ing the protocol of the suppliers (Stratagene). Approximately 17,500 Biosystems) using T3 and T7 primers. Primers were designed for
recombinant clones were obtained in the primary library. A total of microsatellites containing no less than ten perfect or imperfect re-
about 200 cDNA clones were isolated; they ranged in size between peats using the PRIMER 0.5 software. PCR reactions to search for
0.6 and 1.3 kb, as determined by polymerase chain reaction (PCR) microsatellite polymorphism between the parents and to score
amplification of inserts using T3 and T7 primers (Stratagene). For microsatellite loci in the progeny were carrried out in a 100-ll
both cDNA and genomic clones, probes were prepared for Southern volume containing 0.1—0.7 ng/ll genomic DNA, 0.2 lM of each
hybridization by PCR amplification using the appropriate primers. primer in 10 mM TRIS-HCl, 50 mM KCl, 1.5 mM MgCl , 0.01%
T3 and T7 primers were used for cDNA clones in lambda ZAP II 2
gelatin, 200 lM each of dNTPs, and 2.5 U of Ampli-¹aq poly-
and genomic clones in pBluescript, and M13 forward and reverse merase (Perkin Elmer-Cetus). One denaturation cycle was per-
primers (New England Biolabs) for clones in PUC 18. A PCR formed at 94°C for 5 min prior to 30 cycles of denaturation at 94°C
thermal profile of 1 min at 94°C, 34 cycles of 30 s at 94°C, 1 min at for 1 min, annealing at 45°C or 56°C for 1 min, extension at 72°C
55°C, and 2 min at 72°C, with a final extension time of 10 min at 1 min, and a final extension at 72°C for 5 min. PCR-amplified
72°C in a Perkin Elmer-Cetus thermo-cycler, was used in insert products were visualized on 5% metaphor agarose gels stained with
amplification. EtBr, or labelled directly by incorporation of a[32P]dATP during
Southern hybridization of parental surveys of the cDNA and PCR, and run on 6% polyacrylamide gels.
genomic libraries was carried out using the ECL direct nucleic acid
hybridization and detection kit of Amersham PLC. A batch of 30
survey filters (6]4 cm) was routinely hybridized in small plastic
boxes (6]6 cm) with 10 ml of ECL direct hybridization buffer Isoenzyme analysis
and 100—200 ng of labelled probe for 16—18 h; this was followed
by two medium stringency washes of 0.5] SSC, 0.4% SDS for Nine isoenzyme systems (i.e., alpha and beta-esterase, acid phos-
10 min at 55°C, then chemiluminescent detection, and autoradiogra- phatase (acp), diaphorase, peroxidase, glutamate oxaloacetate
phy for 1—2 h. Filters were used more than 20 times. This provided transaminase (got), shikimate dehydrogenase (skdh), and malate
a fast and relatively inexpensive way to screen large numbers of dehydrogenase) were examined in the parents of the mapping popu-
RFLP clones; about 150 were handled in a 5-day week. Southern lation. Beta-esterase yielded a single-dose fragment segregating from
hybridization of progeny filters, with a greater emphasis on clarity the male gametes, while shikimate dehydrogenase, diaphorase,
of autoradiograms to permit unambiguous scoring of RFLP data, glutamate oxaloacetate, and acid phosphatase had unique alleles
was done with a[32P]dNTP-labelled probes. Purified PCR segregating from the female gametes. Protocols for the isoenzyme
products of DNA inserts (using Sephadex G-50 columns) from assay have been described elsewhere (Ocampo et al. 1992).
both genomic and cDNA libraries were labelled with a[32P]dATP
by the random primer method of Feinberg and Volgestein (1983)
to a specific activity of about 1—8.0]10 cpm and employed
as probes in Southern hybridization using the Church and Gilbert Data analysis and map construction
(1984) hybridization buffer at 65°C. Prehybridization and hybridiza-
tion were for 4 and 18 h, respectively. Post-hybridization Monogenic segregation ratios of present, absent classes for each
washes were at 65°C with 2] SSC, 0.1% SDS for 30 min and 0.5] marker were tested for goodness of fit to the expectation of 1 : 1 to
SSC, 0.1% SDS for 20 min, followed by autoradiography for identify single-dose markers by chi-square analysis. Two separate
2—4 days at !80°C with one or two intensifying screens. Filters mapping data sets were obtained from the segregation of single-
were reprobed. The previous probe was stripped off by incubating dose markers in the F mapping population. One data set was
1
the filters in 0.5% SDS solution (initial temperature 100°C) at 65°C from single-dose markers segregating in the gametes of the female
for 30 min, up to 12 times without any loss of the hybridization parent, and the other was from single-dose markers segregating in
signal. the gametes of the male parent. Markers with segregation ratios
TAG 018
434
significantly different from the expected ratio of 1 : 1 were tested for Table 1 Percentage polymorphism (unique allele) found with re-
other possible ratios — such as 3 : 1, which is expected for the spect to male and female parent with RFLP, RAPD, microsatellite,
segregation of double-dose markers on homoeologous chromo- and isoenzyme loci
somes of an allo- or autopolyploid (Wu et al. 1992). Double-dose
markers represent the duplex (double simplex) condition at a hetero- RFLP % polymorphism Average number of
zygous locus. In order to compensate for the random assignment of detected fragments detected
‘‘1’’ or ‘‘0’’ to alternate alleles at a locus and to detect linkage in by probe
repulsion, we duplicated the data matrices, and markers of the
second half were recoded by inverting the scores before linkage Female Male Female Male
analysis. This resulted in a mirror image of each linkage group,
which was later discarded. EcoRI 10.0 12.0 1.5 1.5
The unduplicated mapping data sets consisted of 195 markers for EcoRV 7.2 7.7 1.5 1.5
the female and 203 for the male parent. The test for linkages was HaeIII 5.2 4.0 1.3 1.3
done using the computer package MAPMAKER 2.0 running on HindIII 8.9 1.1 1.4 1.4
a Macintosh Centris 650 and Mapmaker 3.0 Unix version on PstI 5.7 6.1 1.2 1.2
a SPARC workstation (Lander et al. 1987). A LOD score of 4.0 and Total RFLP 37 40.8 1.4 1.4
recombination fraction of 0.30 served as the threshold for declaring RAPD 40.0 38.0 — —
linkage. Map units (in centiMorgans, cM) were derived using the Microsatellite 83.0 58.0 — —
Kosambi function (Kosambi 1944). Maximum likelihood orders of Isoenzyme 50 12.5 — —
markers were verified by the ‘‘ripple’’ function, and markers were
said to belong to the framework map if the LOD value, as calculated
by the ‘‘ripple’’ command, was *2.0. Markers that could not be
placed with LOD*2.0 were added to the map in the most likely
interval between framework markers. four libraries generated with non-methylation-sensitive
Once linkage groups were drawn, they were checked for markers enzymes (HindIII, XbaI, BamHI, and EcoRI). About
linked in repulsion to distinguish between random chromosome 900 low-copy sequences from two PstI libraries, 100
assortment, as in autopolyploids, and preferential pairing, as in
diploids or allopolyploids. Only pairs of adjacent loci with one genomic clones from the four smaller libraries, and 75
shared allele and one parent-specific allele, for which the ‘presence’ cDNA clones have so far been screened for detecting
class had not been assigned at random, were considered adequate for polymorphism between the parents of the mapping
this comparison. Pairs of loci for which ‘presence’ was linked with population with five restriction enzymes, EcoRI,
‘absence’ of segregating alleles were counted as being linked in
repulsion. EcoRV, HaeIII, HindIII, and PstI. Of the low-copy
Recombination rates in the gametes of the male and female genomic clones 41% and 37% detected unique seg-
parents were compared by a t-test (P(0.01) of 10 map intervals regating alleles in the gametes of the male and female
bounded by markers for which both parents were heterozygous and parents, respectively, with at least one restriction en-
had one allele in common, termed allelic bridges (Ritter et al. 1990). zyme. Of the cDNA clones, 20% and 26% revealed
similar markers in gametes of the male and female
parents, respectively, with at least one restriction en-
Estimation of genome size zyme. The percentage of RAPD markers with a unique
allele in the male and female parents has been described
A simple and useful method of estimating genome length, G, from
linkage data of organisms that undergo normal meiosis has been elsewhere (Gomez et al. 1995). Twelve microsatellite
described (Hulbert et al. 1988). The method estimates G based on the loci, ranging in length from 12 to 21 di- or tri-nucleo-
probability that a randomly chosen pair of loci will lie within xcM of tide repeats were screened for their ability to detect
each other is approximately 2x/G; where x is assumed to be small polymorphisms between the parents; 10 microsatellites
compared to the mean genetic length of the chromosome. G is or 83%, were heterozygous in the female parent and 7,
mathematically determined from linkage data by solving the
equation: or 58%, in the male parent. Nine isoenzyme loci
detected four unique alleles segregating in the female
G"MX/K (roughly 50%) but only one (13%) in those of the male
Where M"number of informative meioses, X"an interval in cM parent. Percentage polymorphism of RAPD markers,
at some minimum LOD score, K"actual number of pairs of with unique alleles, in the male and female parents, and
markers observed that border the interval x or less. percentage polymorphism of similar RFLP markers,
for the same cross, showed no significant difference.
Table 1 gives a breakdown of polymorphism found in
Results both parents for the 1075 clones screened. Greater
levels of polymorphism were found with EcoRI and
Library characterization and parental survey HindIII; HaeIII was the least successful (Table 1).
About 2,700 clones, or 90% of the total from the two
PstI libraries of 1,500 clones each, were judged to be Segregation analysis
low-copy sequences based on dot blot hybridization of
whole recombinant plasmid with total cassava genomic Three hundred and seventy-two genomic and cDNA
DNA as probe. One hundred low-copy clones (50%) clones were scored in the F mapping population, yield-
1
were obtained from 200 genomic clones derived from ing segregation data for 158 single-dose RFLPs for the
TAG 018
435
Fig. 1 Southern hybridization showing segregation of heterozygous
unique alleles from both parents (allelic bridges). F Female parent,
M male parent
female and 145 for the male. Five additional poly-
morphic RFLP markers segregated in the mapping
population with a ratio of 3 : 1 (3 from the female and
2 from the male; these were not included in the linkage
analysis). Eighteen markers, polymorphic in the par-
ents, with more than one restriction fragment, did not
segregate in the F mapping population (10 from the
1
female and 8 from the male). Another 95 of the poly- Fig. 2 Number of DNA low-copy sequences detecting 1, 2, 3, 4, and
morphic markers were either pseudo F markers,
2 5 fragments with EcoRI nad HindIII restriction enzymes
monomorphic between the parents but heterozygous
and segregating in the F population (12 markers, ex-
1
cluded from linkage analysis), or were difficult to score only one or two fragments, which was to be expected
after Southern hybridization and are being reanalysed. for unique loci in the homozygous or heterozygous
About a quarter of the polymorphic RAPD markers state. About 100 single- and low-copy sequences detec-
segregating as single-dose markers were chosen for ted more than two fragments. This was expected for
linkage analysis based on several factors, including unique loci in an allo- or autopolyploid with at least
consistency of banding pattern after two or three ream- one of the two homologous groups having two alleles
plifications, clarity of gels, and number of amplified at duplicated loci or was due to the presence of an
fragments, with fewer fragments being more acceptable. internal site for the restriction enzyme used in at least
All microsatellite and isozyme markers polymorphic one of the alleles at a marker locus. Duplication was
in the male or female parent segregated as single- confirmed, by linkage analysis, for six of the sequences
dose markers and were scored in the F mapping
1 presenting more than 2 alleles. These markers detected
population. nonallelic segregating fragments in the male or female
Thirty genomic clones detected a unique segregating gametes and were mapped to different linkage groups.
fragment in each parent and a common allele in both Three of the duplicated loci had 1 locus, each segregat-
parents and were mapped to similar positions on the ing in the gametes of the male and the female parent,
male/female derived linkage group. Such allelic bridges and were mapped to linkage groups identified as
(Ritter et al. 1991) are crucial for identifying the analog- nonanalogous in the male- and female-derived frame-
ous linkage groups in the male- and female-derived work maps. The duplicated loci, GY25a and GY42a
maps, as they detect the same locus on both parental on linkage group D and GY 101a on linkage group
chromosomes, except when they represent duplicated F, are shown on the female-derived map described
sequences. Figure 1 shows an example of segregation of below. Three other duplicated loci segregated in
a marker heterozygous in both parents with a shared the gametes of the male parent (not shown). The rest
allele, or allelic bridge, used to reconcile linkage groups of the RFLP loci having three or more fragments
drawn on the independent segregation of markers in consisted of : loci having 1 mapped locus and
male and female gametes. additional monomorphic fragments; loci with two
nonallelic segregating fragments (unmapped); or loci
with no heterozygous fragments. Efforts continue to
Sequence duplication in the cassava genome map secondary (or primary) loci, as shown in Fig. 3,
using additional restriction enzymes. The addition of
We assessed the number of fragments detected by 1075 microsatellites to the map may also assist here, as
single- and low-copy DNA sequences with the two preliminary evaluation indicates that they detect higher
most polymorphic restriction enzymes, EcoRI and levels of allelic diversity/heterozygosity than genomic
HindIII (Fig. 2). The majority of sequences detected or cDNA clones.
TAG 018
436
from the gametes of the male parent scored in the
F mapping populations, defined 24 linkage groups
1
(not shown) with a total distance of 1, 220 cM. A similar
estimate of genome size, conducted as in Hulbert et al.
(1988), based on mapping data from the male gametes
suggested that the length of the cassava genome is 2,
010 cM. The most salient difference observed between
the male- and female-derived framework maps is the
greater genetic distances on the male-derived map be-
Fig. 3 Southern hybridization of clone GY92 with HindIII digests tween markers common to both parents (allelic
of parental line and mapping population. F Female parent, M male bridges); see Fig. 5. Intervals were observed to be larger
parent. Two faint fragments can be observed to be present and
polymorphic in the male but absent in the female parent in the male-derived map than in the female-derived
map in eight instances, and in only two instances, did
the female-derived map display larger genetic distances
The genetic linkage map (linkage groups J and K). A paired t-test of the ten
intervals showed significantly (P(0.01) greater distan-
One hundred and fifty-eight RFLP, 30 RAPD, 3 micro- ces in the male-derived map, suggesting a reduced re-
satellite, and 4 isoenzyme single-dose markers, seg- combination rate in gametes of the female parent. This
regating in the gametes of the female parent of our observation was reinforced by the greater overall
F mapping population, were tested for linkage using length of the male-derived famework map and its larger
1
the MAPMAKER computer package. One hundred estimate of genome size compared to the female. The
and thirty-two RFLP, 30 RAPD, 3 microsatellite, and mean interval length between adjacent allelic bridges in
3 isoenzyme loci were found to define 20 linkage groups the female-derived map was 38% less than in the male-
spanning 931.6 cM, with an average marker density of derived map. Though distances between adjacent allelic
1 marker every 8 cM (Fig. 4); 26 RFLP markers and bridges may not be representative of all parts of the
1 isoenzyme marker remained unlinked. Linkage cassava genome, the similarity between means of inter-
groups are named alphabetically until they can be val differences and the overall difference in length of the
correlated to earlier named chromosome karyotypes two maps provided a reasonable basis for comparing
(Magoon et al. 1969). The most densely populated recombination frequency between the male- and fe-
linkage group (D) spanned 51.2 cM, with 26 markers, male-derived parents. We are currently working with
while the least populated group (I), also the longest the computer package JOIN MAP 2.0 (Stam et al.
group, had 8 markers spanning 80.6 cM. This wide 1993) to develop a consensus map based on male- and
range of marker density indicated differing degrees of female-derived framework maps.
saturation of linkage groups with markers.
In Fig. 4, 139 (84%) attached to linkage groups by
horizontal bars make up the LOD 2.0 framework map Chromosome pairing in cassava
of cassava, the remaining 29 markers (in parenthesis)
are placed in the most likely intervals between frame- To test the hypothesis of preferential pairing against
work markers (LOD*2.0). The existence of 27 unlin- the alternative of random chromosome assortment, we
ked markers suggested that some regions of the cassava examined the framework map for markers linked in the
genome have not yet been mapped. Based on the segre- repulsion phase compared to those linked in the coupl-
gation data reported here, we estimated the length of ing phase. The expected ratio of single-dose markers
the cassava genome to be 1, 610 cM. With a total map linked in repulsion to markers linked in coupling is 1 : 1
length of 932 cM, this implied that the present frame- for disomic inheritance, while no markers are expected
work map covers roughly 60% of the cassava genome. to be linked in repulsion for autosomic inheritance (Wu
Regions with distorted segregation ratios could not be et al. 1992). A total of 40 RFLP markers (30%) were
identified in the cassava genetic map, since only those found to be linked in repulsion. Only cases with one
polymorphic restriction fragments segregating in a 1 : 1 allele in common between the parents and the alternate
ratio (SDRF) were included in the linkage analysis. The allele linked in repulsion were considered in this calcu-
non-inclusion of markers with distorted monogenic lation (Fig. 6). Linkage groups B and H presented a 1 : 1
ratios may have excluded certain genomic regions from ratio of 6 markers linked in repulsion to 6 markers
any map based on these criteria. linked in coupling, as expected for disomy. For all other
linkage groups there appeared to be some degree of
random pairing, based on the number of markers
Comparison of the male and female genetic maps linked in the repulsion phase. Linkage groups A, C, D,
I, J, M, N, and T have 2 markers linked in the repulsion
One hundred and seven RFLP, 50 RAPD, 1 microsatel- phase as against 14, 8, 26, 8, 8, 5, 6, and 3 markers
lite, and 1 isoenzyme single-dose markers, segregating in the groups, respectively. Groups E, K, and P have
TAG 018
437
Fig. 4 Molecular genetic map of cassava based on segregation of 4 markers linked in repulsion as against 7, 6 and
RFLP (CD½ cDNA, G½ genomic), microsatellite (GA), isozyme 9 markers in the linkage groups, respectively. Group
(acp, skdh, and got), and RAPD markers (prefix A-AP through Z,
Operon primer nomenclature) in gametes of the female parent L has 6 of its 8 markers linked in the repulsion phase.
(TMS30575), in a F cross (with CM2177-2 as male parent) of 90 No markers were found linked in repulsion phase in
1
individuals. Markers with the suffix ‘‘a’’ represent duplicated loci. linkage groups F, G, O, Q, R, and S, which agrees with
Markers adjacent to horizontal lines belong to the framework the expected behaviour of random assortment under
(LOD'2.0) map; those following on the same line cosegregate, and
remaining markers (in parenthesis) are placed in the most probable autopolyploidy. The sizes of linkage groups G, O, Q,
interval. Map distances, shown on the left, are indicated in Kosambi and R, however, suggested that a sizeable part of the
map units chromosomes remains to be mapped.
TAG 018
438
TAG 018
439
30%, is significantly higher than this. It is not clear
at this stage if the higher number of markers found
in repulsion than expected from the karyology of
cassava (Magoon et al. 1969) supports the predomi-
nance of disomic inheritance as suggested by pairing
behaviour (Bai 1992) and inheritance of isoenzyme loci
(Hussain et al. 1987; Roca et al. 1992; Lefevre and
Charrier 1993).
More confounding is the fact that only 6 duplicated
loci were detected with the segregation of single-dose
markers derived from 36 cDNA, and over 200 genomic
sequences. In well-known allopolyploids such as maize,
Fig. 6A,B Examples of the category of markers considered in iden- wheat, and cotton, blocks of duplicated loci have been
tifying linkage in repulsion, with one parent heterozygous (A) clearly identified by RFLP mapping (Helentjaris et al.
or both parents heterozygous (B). In both examples, unique alleles
(a and b) from the female parent segregated in repulsion in 1988; Devos et al. 1993; Reinisch et al. 1994). Our
the F progeny results, in contrast, have revealed only a few randomly
1 distributed duplicated loci, less than 5% of the total
number of markers, a number corresponding roughly
to that reported in many diploids (Causse et al. 1994;
Discussion Tanksley et al. 1995). There is no doubt that the cassava
genome contains duplicated regions, but it remains to
The cassava genetic linkage map reported here spans be determined whether the duplicated loci represent
931.6 cM, has 20 linkage groups, and is estimated to vestiges of an ancient allopolyploid or random genomic
cover 60% of the cassava genome. If a polyploid origin duplication events in a diploid or diploidized genome.
for the cassava genome is assumed, the expected num- We are currently placing more cDNA markers on the
ber of linkage groups in cassava would depend on cassava map in an attempt to resolve the enigma of
chromosome assortment. Eighteen groups are expected genome duplication in cassava, a putative allopoly-
where preferential pairing is exclusive, up to 36 where ploid. cDNA sequences are known to be relatively
there is mostly random pairing, and a number between more conserved than genomic clones (Helentjaris et al.
18 and 36 where there is a mixture of preferential and 1988), and are expected to be very valuable additions to
random pairing. Thirty percent of all RFLP markers the cassava map.
on the cassava genetic map were found to be linked in The clusterings of markers on some groups, com-
the repulsion phase. This figure is less than the 50% pared to others, separated by intervals larger than
expected for linkage of single-dose markers in allo- 15 cM, suggests that recombination is not occurring
polyploids or diploids (Wu et al. 1992). On the other uniformly across the cassava genome. Nonrandomness
hand, it differs markedly from the complete absence of of recombination frequency has been reported for sev-
linkages in the repulsion phase expected for autopoly- eral other crops (Bonierbale et al. 1988; Causse et al.
ploids. The finding that fewer than 50% of the markers 1994; Lagercrantz and Lydiate 1995).
were linked in the repulsion phase and of entire groups Allelic bridges (markers that are heterozygous in the
not presenting markers linked in repulsion implies that gametes of both the male and female parents and share
a significant amount of random pairing occurs in cas- a common allele) have been useful in reconciling the
sava. The karyology of the 18 haploid chromosomes of separate maps drawn on segregation from gametes of
cassava reveals six identical pairs and three different the male and female parents. Specifically, these markers
pairs of homologous chromosomes (Magoon et al. help compare rates of recombination in the gametes of
1969). Assuming random assortment between homo- the male and female parents, a task of biological inter-
eologous chromosomes of the six pairs of identical est and possible practical use. Genetic distances be-
chromosomes, a reduction of 66.7% would be expected tween pairs of analogous markers were found to be
of all markers linked in repulsion (16.7% as against larger in the male-derived map than in the female in
50%). The percentage of markers linked in the repul- eight out of ten intervals bordered by allelic bridges,
sion phase in the genetic map of cassava reported here, suggesting a reduced recombination in gametes seg-
regating from the female parent. Earlier reports in
Drosophila (Baker et al. 1976), humans (Bowden et al.
1989), and tomatoes (de Vincente and Tanksley 1991)
$&&&&&&&&&&&&&&&&&&&&&&&&& found a significant genome-wide reduction in recombi-
Fig. 5 Genetic maps of single-dose markers segregating in the fe- nation in gametes segregating from the male parent.
male and male gametes. The intervals shown are those bounded by
markers heterozygous in both parents, and distances are as cal- Enhanced male recombination frequencies were, how-
culated by MAPMAKER’s ‘‘pairwise’’ comand. Alphabetical names ever, found to be associated with terminal chromo-
of linkage groups are according to the female map somal regions in Brassica, while enhanced female
TAG 018
440
recombinations were adjacent to putative centromere Awoleye F, Van Duren M, Dolezel J, Novak FJ (1994) Nuclear
positions (Lagercrantz and Lydiate 1994). We cannot DNA content and in vitro induced somatic polyploidization
determine at this point whether the reduced rates of (Manihot esculenta Crantz) cassava breeding. Euphytica 76 :
195—202
recombination observed in the gametes of the female Bai KV (1992) Cytogenetics of Manihot species and interspecific
parent of this cross are due to its genetically diverse hybrids. In: Roca W, Thro AM (eds) Proc 1st Int Sci Meet
ancestry (‘TMS 30572’ is understood to contain intro- Cassava Biotechnol Network. CIAT, Cali, Colombia, pp 51—55
gressed genes from wild Manihot species), or to poten- Baker BS, Carpenter ATC, Esposito RE, Sandler L (1976) The
tial genetic sex differences in cassava. It is known, genetic control of meiosis. Annu Rev Genet 10 : 53—134
Bennet MD, Smith JB, Heslop-Harrison JS (1992) Nuclear DNA
however, that increased recombinations, such as those amounts in Angiosperms. Proc R Soc London Ser B 216 :
found in the male parent, can serve as a tool to minim- 179—199
ize linkage drag during backcrossing schemes (Young Bonierbale MW, Plaisted RL, Tanksley SD (1988) RFLP maps
and Tanksley 1989). based on a common set of clones reveal modes of chromosomal
To our knowledge, the cassava genetic map is the evolution in potato and tomato. Genetics 120 : 1095—1103
Bothwell AL, Yancopoulis GD, Alt FW (1990) Methods for cloning
first of its kind to be constructed for the family Euphor- and analysis of eukaryotic genes. Jones and Bartlett Publ, Boston,
biaceae. It could prove useful for comparative mapping Mass., pp 8—19
in other economically important genera of the family, Bowden DW, Gravius TC, Green P, Falls K, Wuster-Hill D, Noll
such as Hevea (rubber), or the markers could be applied W, Muller-Kahle H, Donis-Keller H (1989) A genetic linkage map
of 32 loci on human chromosome 10. Genomics 5 : 718—726
to minor related species, for which molecular cloning Bryne D (1984) Breeding cassava. Plant Breed Rev 2 : 73—134
may not be justified. Furthermore, the map provides an Causse MA, Fulton TM, Cho YG, Ahn SN, Chunwongse J, Wu K,
important tool for cassava improvement, based on the Xiao J, Yu Z, Ronald PC, Harrington SE, Second G, McCouch
association of molecular markers with useful genes. SR, Tanksley SD (1994) Saturated molecular map of the rice
Cassava is a staple throughout the tropics, and impor- genome based on an interspecific backcross population. Genetics
138 : 1251—1274
tant agronomic and culinary trait as well as adaptation Church GM, Gilbert W (1984) Genomic sequencing. Proc Natl Acad
to pests and diseases vary in cassava culitvars from one Sci USA 81 : 1991—1995
ecological zone to another. Although intra- and even Da Silva JAG, Sorrels ME, Burnquist WL, Tanksley SD (1993)
interspecies crosses are easy to make in cassava, the RFLP linkage map and genome analysis of Saccharum sponta-
introgression of traits from one cassava genepool to neum. Genome 36 : 782—791
Dellaporta SL, Wood J, Hicks JR (1983) A plant DNA miniprepara-
another is complicated by difficulties in understanding tion: version II. Plant Mol Biol Rep 1 : 19—21
the inheritance of complex traits in highly heterozygous de Vincente MC, Tanksley SD (1991) Genome-wide reduction in
genetic backgrounds. The map can serve as a tool recombination of backcross progeny derived from male versus
for efficiently transferring traits between gene pools. female gametes in an interspecific cross of tomato. Theor Appl
Future perspectives for the genetic map of cassava Genet 83 : 173—178
Devos KM, Millan T, Gale MD (1993) Comparative RFLP maps of
includes the addition of more molecular markers, pre- the homeologous chromosomes of wheat, rye, and barley. Theor
ferably microsatellite markers and cDNA clones, and Appl Genet 85 : 784—792
joining the male/female-derived maps. Microsatellite Feinberg AP, Vogelstein B (1981) A technique for radiolabelling
markers showed a higher level of polymorphism than DNA restriction endonuclease fragments to high specific activity.
Anal Biochem 137 : 266—267
any other marker in the intraspecific cross used in Fregene MA, Vargas J, Ikea J, Angel F, Tohme J, Asiedu RA,
generating the cassava genetic map, while cDNAs will Akoroda MO, Roca WM (1994) Variability of chloroplast
be most useful in defining the genome structure. DNA and nuclear ribosomal DNA in cassava (Manihot
esculenta Crantz) and its wild relatives. Theor Appl Genet
89 : 719—727
Acknowledgements We would like to thank Dr. C. Iglesias for Gomez R, Angel F, Bonierbale M, Rodriguez F, Tohme J, Roca
generating the mapping population, Maria Mercedes Maya and W (1994) Selection of heterozygous parents and single dose
Matt ‘‘Mateo’’ Sayre for excellent technical support, Cesear Ocampo markers for genetic mapping in cassava. In: Roca W, Thro
for the isoenzyme analysis, and Nathan Russell for editing. Thanks AM (eds) Proc 2nd CBN Meeting. CIAT, Cali, Colombia, pp
also to Dr. S. Bebe for helpful comments. This work was supported 113—124
by grant RF 91060 No. 2 of the Rockefeller Foundation’s biotech- Hahn S, Bai KV, Asiedu RA (1990) Tetraploids, triploids, and 2n
nology programme. MF was supported by a post-doc fellowship, pollen from diploid interspecies crosses with Cassava. Theor Appl
also from the Rockefeller foundation and RG by COLCIENCIAS Genet 79 : 433—439
project No. 228-07-003-92. Helentjaris T, Weber D, Wright S (1988) Identification of the
genome locations of duplicate nucleotide sequences in maize by
anlysis of restriction fragments length polymorphisms. Genetics
References 118 : 353—363
Hulbert SH, Ilott TW, Legg EJ, Lincoln SE, Lander ES, Michelmore
RW (1988) Genetic analysis of the fungus, Bremia lactucae, using
Al Janabi SM, Honeycutt RJ, McClelland M, Sobral BWS (1993) restriction fragments length polymorphisms. Genetics 120 :
A genetic map of Saccharum spontaneum L. ‘‘SES 208’’. Genetics 947—958
134 : 1249—1260 Hussain A, Ramirez H, Roca W, Bushuk W (1987) Identification of
Angel F, Arias D, Tohme J, Iglesias C, Roca W (1993) Towards the cassava (Manihot esculenta Crantz) by electrophoretic patterns of
construction of a molecular map of cassava (Manihot esculenta esterase isozymes. Seed Sci Technol 15 : 19—22
Crantz): comparison of restriction enzymes and probe source in Kosambi DD (1944) The estimation of map distances from recombi-
detecting RFLPs. J Biotechnol 31 : 103—113 nation values. Ann Eugen 12 : 172—175
TAG 018
441
Lagercrantz UFL, Lydiate DJ (1995) RFLP mapping in Brassica Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning,
nigra indicates differing recombination rates in male and female a laboratory manual. Cold Spring Harbor Laboratory Press, New
meioses. Genome 38 : 255—264 York
Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln Stam P (1993) Construction of integrated genetic linkage maps
SE, Newburg L (1987) MAPMAKER: an interactive computer by means of a new computer package: JOIN MAP. Plant
package for constructing primary genetic linkage maps of experi- J 3 : 739—744
mental and natural populations. Genomics 1 : 174—181 Tanksley SD, Ganal MW, Prince PJ, deVincente MC, Bonierbale
Lefevre F, Charrier A (1993) Heredity of seventeen isozyme loci in MW, Broun P, Fulton MW, Giovannoni JJ, Grandillo S, Martin
cassava (Manihot esculenta Crantz). Euphytica 66 : 171—178 GB, Messeguer R, Miller JC, Miller L, Paterson AH, Pineda O,
Magoon ML, Krishnan R, Bai KV (1969) Morphology of the pachy- Roder MS, Wing RA, Wu W, Young ND (1995) High density
tene chromosomes and meiosis in Manihot esculenta crantz. Cytol molecular linkage maps of the tomato and potato genomes.
Tokyo 34 : 612—626 Genetics 132 : 1141—1160
Ocampo C, Hershey C, Iglesias C, Iwanaga M (1992) Esterase Umanah EE, Hartmann RW (1973) Chromosome numbers and
isozyme fingerprinting of the cassava germplasm collection held at karyotypes of some Manihot species. Am Soc Hortic Sci:
CIAT. In: Roca W, Thro AM (eds) Proc 1st Int Sci Meet Cassava 98 : 272—274
Biotechnol Network. CIAT, Cali, Colombia, pp 81—89 Vayda ME, Freeling M (1986) Insertion of the Mu1 transposable
Perry BA (1943) Chromosome number and phylogenetic relation- element into the first intron of maize Adh1 interferes with tran-
ship in the Euphobiacea. Am J Bot Club 30 : 527—543 scription elongation but does not distrupt chromatin structure.
Reinisch AJ, Dong J, Brubaker CL, Stelly DM, Wendel JF, Paterson Plant Mol Biol 6 : 441—454
AH (1994) A detailed RFLP map of cotton Gossypium hir- Wu KK, Burnquist W, Sorrels ME, Tew TL, Moore PH, Tanksley
sutum]Gossypium barbadense: chromosome organization and SD (1992) The detection and estimation of linkage in polyploids
evolution in a disomic polyploid genome. Genetics 138 : 829—847 using single dose restriction fragments. Theor Appl Genet
Ritter E, Debener T, Barone A, Salamini F, Gebhardt C (1991) RFLP 81 : 471—476
mapping on potato chromosomes of two genes controlling extreme Young ND, Tanksley SD (1989) RFLP analysis of the size of
resistance to potato virus X (PVX). Mol Gen Genet 227 : 81—85 chromosomal segments retained around the Tm-2 locus of
Roca WM, Henry G, Angel F, Sarria R (1992) Biotechnology re- tomato during backcross breeding. Theor Appl Genet
search applied to cassava improvement at the International 77 : 353—354
Center of Tropical Agriculture(CIAT). AgBiotechnol News Inf Yu K, Pauls KP (1992) Optimization of the PCR program for
4 : 303N—308N RAPD analysis. Nucelic Acids Res 20 : 2606
.