SHORT COMMUNICATION

High proportion of diploid hybrids produced by interspecific
diploid 3 tetraploid Sorghum hybridization

Stan Cox . Pheonah Nabukalu . Andrew H. Paterson . Wenqian Kong .

Susan Auckland . Lisa Rainville . Sheila Cox . Shuwen Wang

Received: 4 June 2017 / Accepted: 24 October 2017

� Springer Science+Business Media B.V. 2017

Abstract A perennial version of grain sorghum [S.

bicolor (L.) Moench] would create opportunities for

greatly reducing tillage and preventing soil degrada-

tion. Efforts to select for perenniality and grain

production among progeny of hybrids between S.

bicolor (2n = 20) and the weedy tetraploid perennial

S. halepense (L.) Pers. (2n = 40) are complicated in

that F1 hybrids produced by diploid 9 tetraploid

sorghum crosses are usually tetraploid. In 2013, a set

of random pollinations between 19 diploid cytoplas-

mic male-sterile inbred lines and 43 tetraploid peren-

nial plants produced 165 F1 hybrid plants, more than

75% of which had highly atypical plant, panicle, and

seed phenotypes. Phenotypic segregation in F2 popu-

lations derived from atypical hybrids was also anoma-

lous. Examination of mitotic metaphase cells in F1 or

F2 root tips revealed that 129 of the 165 hybrids were

diploid. Parentage of the diploid progenies was

confirmed using simple-sequence repeat analysis.

The mechanism by which diploid hybrids arise from

diploid 9 tetraploid crosses is unknown, but it may

involve either production of monohaploid (n = 10)

pollen by the tetraploid parent or chromosome elim-

ination during early cell divisions following formation

of the triploid zygote. The ability to produce diploid

germplasm segregating for S. bicolor and S. halepense

alleles could have great utility, both for the develop-

ment of perennial sorghum and for the improvement of

conventional grain sorghum.

Keywords Genetic resource � Germplasm

enhancement � Sorghum breeding � Ploidy � Perennial �
Interspecific hybridization � Chromosome

Introduction

Development of perennial grain sorghum (Paterson

et al. 2013; Nabukalu and Cox 2016) would create

opportunities for greatly reducing tillage and other

aspects of annual grain sorghum cultivation that

damage soil structure and lead to erosion. Efforts to

develop perennial sorghum germplasm, underway

since 2002, involve producing hybrids between S.

bicolor and the weedy perennial grass S. halepense

(L.) Pers., selecting for rhizome development and

winter survival in segregating populations, backcross-

ing to S. bicolor, and repeating the cycle (Nabukalu

and Cox 2016).

Electronic supplementary material The online version of
this article (https://doi.org/10.1007/s10722-017-0580-7) con-
tains supplementary material, which is available to authorized
users.

S. Cox (&) � P. Nabukalu � S. Cox � S. Wang

The Land Institute, 2440 E. Water Well Rd., Salina,

KS 67401, USA

e-mail: cox@landinstitute.org

A. H. Paterson � W. Kong � S. Auckland � L. Rainville
Plant Genome Mapping Laboratory, University of

Georgia, 111 Riverbend Road, Rm 228, Athens,

GA 30605, USA

123

Genet Resour Crop Evol

DOI 10.1007/s10722-017-0580-7

https://doi.org/10.1007/s10722-017-0580-7
http://crossmark.crossref.org/dialog/?doi=10.1007/s10722-017-0580-7&domain=pdf
http://crossmark.crossref.org/dialog/?doi=10.1007/s10722-017-0580-7&domain=pdf


S. bicolor is diploid with 2n = 20, while S.

halepense is tetraploid with 2n = 40. Hybridization

can be accomplished by using an induced tetraploid

line of S. bicolor as either the male or the female

parent (Piper and Kulakow 1994) or by fertilizing

male-sterile diploid S. bicolor plants with pollen from

perennial tetraploid plants and obtaining 40-chromo-

some hybrids through the infrequent but consistent

production of 20-chromosome gametes by the male-

sterile parent (Hadley 1958; Nabukalu and Cox 2016).

Either process results in tetraploid populations that

may be used in selection and germplasm development.

In some cases, triploid hybrids occur, but they produce

no seed when self-pollinated or pollinated by S.

bicolor.

Introgressing useful germplasm from S. halepense

via tetraploid hybrids presents several disadvantages: it

limits the scope of the parental gene pool; the approach

to homozygosity with self-pollination is slower than in

diploid sorghum; partial sterility is common; and

genetic analysis is complicated and difficult (Haldane

1930; Gupta 2007). Perennial sorghum germplasm

development will be greatly simplified if alleles for

salient traits such as cold tolerance and rhizome

development can be introgressed from S. halepense

directly into diploid grain sorghum.

In published research on controlled hybridization

between diploid S. bicolor and S. halepense, chromo-

some numbers of hybrids have varied. Hadley (1958)

and Sengupta and Weibel (1971) obtained both

triploid and tetraploid hybrids. When pollinating

thousands of nuclear and cytoplasmic male-sterile

panicles with tetraploid plants during 2002–2012 as

part of our perennial sorghum germplasm research

(Nabukalu and Cox 2016), we obtained hundreds of

germinable tetraploid F1 seeds and only one ger-

minable triploid seed (although we have often

observed shriveled, inviable seeds that we assumed

were triploid).

Dweikat (2005) reported surprising results from

crossing eight nuclear male-sterile (ms3ms3) sorghum

plants with S. halepense. Pollinating an estimated

36,000 florets, he obtained 380 shriveled, inviable

seeds (presumably triploids) and two normally devel-

oped seeds. One of the plants from a normal seed was

found to be diploid. Phenotypic and molecular-marker

analysis confirmed the plant’s parentage.

In 2014, when we grew out 154 F1 plants resulting

from pollinations between male-sterile grain sorghum

inbred lines and perennial tetraploid plants, a large

proportion of them were phenotypically very different

from diploid 9 tetraploid F1 plants we had observed

in our nurseries over the previous decade. In 2016,

after reading for the first time the report by Dweikat

(2005), we formed a hypothesis that the phenotypi-

cally novel F1 plants we had observed in 2014 were

diploids produced by diploid 9 tetraploid pollination,

and we began examining chromosome numbers of

their F2 progeny.

Materials and methods

In May, 2013, 43 tetraploid sorghum plants were

selected from among the self-pollinated progeny of S.

bicolor 9 S. halepense, (S. bicolor 9 S. hale-

pense) 9 S. bicolor, or (S. bicolor 9 S. hale-

pense) 9 S. bicolor2 hybrids and were used to

pollinate inbred lines of grain sorghum carrying A1-

type male-sterile cytoplasm (Suppl. Table 1). The

perennial parents were tetraploid sorghum plants

selected in the course of developing perennial

sorghum germplasm (Suppl. Table 1; Nabukalu and

Cox 2016). Nineteen different inbred lines produced

F1 seed. A total of 154 F1 plants from 55 different

parental combinations (Suppl. Table 1) were grown to

maturity and harvested. In 2016, 11 remnant F1 seeds

from crosses made in 2013 were removed from storage

and germinated, and the seedlings were transplanted in

the field. These F1 seeds had been produced during the

same time period as the 154 hybrids discussed above,

using the same pool of parents.

In 2016–2017, root tips were obtained from F2
progeny of each of the 154 F1 plants grown in 2014, the

11 F1 plants grown in 2016, and plants derived by self-

pollination of themale parents of 2013 crosses. Root tips

were fixed and used for determining mitotic metaphase

chromosome numbers. The technique was based on

procedures reported by Singh (2016) and Sharma and

Sharma (2014), with modifications. Slide preparations

were examined under a compound light microscope

(Axiostar plus; Zeiss, Oberkochem, Germany) with a

magnification of 409. Chromosomes were observed,

counted and photographed using AxioCamERc5 s

camera and Zen 2 (version 10.0) digital imaging

software (Zeiss, Oberkochen, Germany).

Simple-sequence repeat (SSR) genotyping was

used to verify the recorded parentages of a sample of

Genet Resour Crop Evol

123



the diploid hybrids. Young leaf tissue was obtained

from 13 to 26 seedlings belonging to each of nine F2
populations and from all 11 new F1 plants that were

being grown out that year. Leaf tissue was also

sampled from 12 seedlings of each male parent and

one seedling of each female parent involved in

producing the sampled F1 and F2 plants. DNA

extraction from leaf samples and SSR genotyping

were performed as described by Kong et al. (2013).

Results and discussion

Phenotypically, the 165 F1 plants grown out in 2014

and 2016 fell into two distinct groups. A minority

exhibited traits typically seen in tetraploid hybrids

between domesticated sorghum and tetraploid peren-

nial plants. Compared with S. bicolor, these hybrids

displayed more profuse tillering and upper-node

branching, thinner culms, narrower leaves, longer

panicle branches, lower seed-set, and smaller seeds.

As is typical in such hybrids, seed shattering was

delayed or absent. The second group, a majority of the

plants, had phenotypes intermediate between the first

group of hybrids and S. bicolor. They produced fewer

tillers and branches than is typical of tetraploid

hybrids. They had thicker culms and wider leaves,

shorter panicle branches (panicles would have been

classified as semi-compact to semi-lax), higher seed-

set, and larger seed than had been seen in tetraploid

hybrids in previous years. Some displayed early,

complete seed shattering.

Somatic chromosome numbers were determined

for one random F2 plant from each of 127 populations

that we suspected of being diploid on the basis of their

F1 parents’ phenotypes and F2 segregation, finding that

121 were indeed diploid and six were tetraploid.

Fifteen additional plants that we hypothesized to be

tetraploid on the basis of phenotype were examined

cytologically as controls, and all were tetraploid.

Figures 1 and 2 show images of cells at mitotic

metaphase from diploid and tetraploid F2 plants.

Chromosome numbers of F2 plants from an additional

15 populations assumed to be tetraploid on the basis of

phenotypic segregation were not confirmed

cytologically.

Of the 11 F1 plants grown in 2016 from remnant

seed, we found that eight were diploid and three

tetraploid. Adding those counts to those from F2

plants, we concluded that at least 129 of the 165 F1
hybrids produced in 2013 had been diploid and the

remainder had been tetraploid (Suppl. Table 1).

From each of nine of the diploid hybrids confirmed

by counts of 20 chromosomes at mitotic metaphase,

Fig. 1 Nucleus at metaphase in a tetraploid F2 plant from the

cross KS105A (2n = 20) 9 H6-70-8 (2n = 40)

Fig. 2 Nucleus at metaphase in a diploid F2 plant from the

cross KS105A 9 H6-70-8

Genet Resour Crop Evol

123



13–26 F2 progeny were screened with 20 SSRs

previously known to be polymorphic between

BTx623 and Gypsum 9E, one sampling each sorghum

chromosome arm (Suppl. Table 2). All eight diploid

F1 plants were similarly screened (Supp. Table 3).

Among the SSRs, 15 to 18 were informative between

the S. bicolor and (S. bicolorn 9 S. halepense) parents

of the nine F2 progeny arrays. SSR genotypes of F1
plants and F2 families were consistent with their

recorded parentage. Among the nine F2 families, the

portion of loci with S. halepense alleles ranged from

13.9 to 33.2%, with a mean of 25%. No individual

progeny completely lacked S. halepense alleles,

although several had only one. At one locus, Xiabt-

p121, no S. halepense allele was found in any plant;

other loci showed S. halepense alleles in 1.1–56.3% of

individuals, demonstrating that most S. halepense

chromosome arms can be recovered in such lines.

Among F1 plants, the share of loci with a S. halepense

allele derived from the male parent ranged from 8.3 to

75.0%, with an overall presence of 25.9%.

The SSR analysis showed that we can recover in

diploid x tetraploid hybrids most chromosome arms

from the tetraploid parent. Across all loci, the S.

halepense allele was detected in 26% of F1 plants and

25% of F2 plants. Those values are consistent with the

25% expected when the tetraploid male parent used to

pollinate a diploid plant carries a single copy of the S.

halepense allele, and then, as postulated by Dweikat

(2005), half of the male parent’s chromosomes are

eliminated either during pollen formation or after

fertilization. Because the male parents we used were

derived from repeated backcrossing to S. bicolor, it

would not be surprising if many S. halepense alleles

were present as single copies in these parents. More

extensive analysis will be required to resolve ques-

tions surrounding the transmission of alleles from

tetraploid parents to diploid progeny.

The parentage of the diploid F1 plant produced by

Dweikat (2005) contrasts with the parentage of the

diploid hybrids reported herein. His hybrid arose from

a cross between a nuclear male-sterile (ms3ms3) S.

bicolor plant and a S. halepense plant, whereas ours

arose from crosses between cytoplasmic male-sterile

plants and species-backcross tetraploid plants. Dwei-

kat (2005) hypothesized that his diploid hybrid

resulted from either (1) production of a monohaploid

(n = 10) gamete by the S. halepense parent or (2)

elimination of chromosomes in early cell division

following fertilization. With either mechanism, a

central question is whether specific chromosomes

were selectively retained or eliminated.

Further studies will be required to determine which

(if either) of Dweikat’s (2005) proposed mechanisms

is the cause of diploid production by such crosses and

to explain the unprecedented numbers of diploid

hybrids we produced in 2013. More broadly, it would

be useful to identify genetic or environmental factors

that might be manipulated to increase the likelihood

that diploid 9 tetraploid sorghum crosses will pro-

duce diploid hybrids. The diploid S. bicolor 9 S.

halepense germplasm thus produced could have high

utility for the development of perennial sorghum and

sorghum improvement in general.

References

Dweikat I (2005) A diploid, interspecific, fertile hybrid from

cultivated sorghum, Sorghum bicolor, and the common

Johnsongrass weed Sorghum halepense. Mol Breed

16:93–101

Gupta PK (2007) Cytogenetics (2). Rastogi Publications,

Meerut

Hadley HH (1958) Chromosome numbers, fertility and rhizome

expression of hybrids between grain sorghum and john-

songrass. Agron J 50:278–282

Haldane JBS (1930) Theoretical genetics of autopolyploids.

J Genet 22:359–372

KongW, Jin H, Franks CD, KimC, Bandopadhyay R, RanaMK,

Auckland SA, Goff VH, Rainville LK, Burow GB,

Woodfin C, Burke JJ, Paterson AH (2013) Genetic analysis

of recombinant inbred lines for Sorghum bicolor 9 S.

propinquum. G3 3:101–108

Nabukalu P, Cox TS (2016) Response to selection in the initial

stages of a perennial sorghum breeding program. Euphyt-

ica 209:103–111

Paterson AH, Cox TS, Kong W, Navarro M (2013) Viewpoint:

multiple-harvest sorghums toward improved food security.

In: Perennial Crops for Food Security, FAO, Rome,

pp 90–102

Piper JK, Kulakow PA (1994) Seed yield and biomass allocation

in Sorghum bicolor and F1 and backcross generations of S.

bicolor 9 S. halepense hybrids. Can J Bot 72:468–474

Sengupta SP, Weibel DE (1971) Cytological study of hybrids of

Sorghum halepense (L.) Pers. Proc Okla Acad Sci

51:56–60

Sharma AK, Sharma A (2014) Chromosome techniques: theory

and practice, 3rd edn. Butterworth-Heinemann, Oxford

Singh RJ (2016) Plant cytogenetics, 3rd edn. CRC Press, New

York

Genet Resour Crop Evol

123


	High proportion of diploid hybrids produced by interspecific diploid x tetraploid Sorghum hybridization
	Abstract
	Introduction
	Materials and methods
	Results and discussion
	References