We chose G. hirsutum TM-1, G. australe, G. capitis-viridis as materials. G. hirsutum is an cultivated tetraploid cotton with genome [AD]₁ (2n = 4x = 52), while G. australe and G. capitis-viridis are wild diploid cotton species with the genome C and B (2n = 2x = 26). All the wild paternal pollen used for the creation of the trispecific hybrid CHA [G. capitis-viridis ×(G. hirsutum ×G. australe)²] in this work, were kindly supplied by the cotton collection of National Wild Cotton Nursery in Sanya of China.

Tetraploid G. hirsutum TM-1 was first crossed directly with the diploid parents G. australe, creating a triploid hybrid [ADC]. Allohexaploid was further synthesized by colchicine-doubling the sterile intergenomic hybrid G. hirsutum × G. australe. Allopolyploid trispecific CHA hybrid [genome BADC] was finally obtained by allohexaploid crossing with the other wild diploid maternal parent G. capitis-viridis according to the pseudophyletic introgression method [9].

In every growth period of this hybrid, morphology characteristics were observed from its whole plant to leaf style, flower color and performances, pollen grain morphology, its seed set and fertility. Comparisons from different characteristic and performance for this hybrid were also made with its parents.

Immature flower buds in different development stages were collected from three parents and this triple hybrid, fixed between 8.00–11.00 am in freshly prepared Carnoy’s solution (ethanol: acetic acid = 3:1) for 24 h, and then stored at 4°C in 70% EtOH for future studies. Meiotic analysis were carried out on suitable size flower buds. After washing the fixed buds in distilled water, anther squashes were stained on slides in Carbol fuchsin solution and observed under a light microscope. Photographs were taken from freshly prepared slides using an Olympus BX60 microscope with automatic camera. Size and sterility of pollen grains from randomly selected anthers were also studied staining with Carbol fuchsin solution. Interpretations were based on analyses of between 15–50 pollen mother cells from the triple hybrid.

The hybrid identity and its parents was also confirmed by SSR molecular markers. SSR analysis was examined by G. hirsutum, G. australe, allohexaploid, this hybrid and G. capitis-viridis. DNA from each plant was extracted from young leaves using CTAB method and quantified using agarose gel electrophoresis. DNA amplification was carried out in a volume of 10 μL containing 1 μL 10×buffer, 1.6 μL Mgcl₂ (25 mM), 0.5 μL dNTPs (10mM), 2 μL template DNA (50 ng μL^-1), 0.8 μL primers (2.5 μM) (0.4 μL of forward and reverse primer each), 0.1 μL Taq polymerase (5 U/μL, Sangon), 5.6 μL ddH₂O. PCR was programmed with an initial denaturing for 3 minutes at 95°C; followed by 30 cycles of pre-denaturation for 3 minute at 94°C; denaturation for 40 s at 94°C; annealing at 57°C for 45 s; extension for 1 minutes at 72°C, and a final extension for 5 minutes at 72°C in a DNA Mastercycler, Eppendorf, Germany. PCR product (10 μl) was resolved in PAGE by silver staining for the resolving of the bands.

The hybrid plants were perennial, semi-bush with crawl growth characteristic different from any other parents (Fig 1). It has long epidermis hair in whole plant especially on the stem and two side of leaf following with its wild-type parents G. australe (Fig 1). The shape of leaves is heart style with round edge which is intermediate character between its parents and thicker also different from its three parents (Fig 1). It produced many flower buds, and flowered by pink or white petal with unclear or no speck, sometimes pink and white flower at the same day (Fig 2). It had normal stigma and stamen with yellow pollen grains like its upland parents (Fig 2). But most pollen grains were abnormal resulting in its sterility. For several years this hybrid had no seed set. Flower performance and leaf characteristic for this hybrid were also evaluated and compared with its three parents in Table 1 and Fig 3.

The triple hybrid had long epidermis hair on the leaf and stem inheriting from its wild parents. It is highly resistant to insects comparing with other cotton species which always have some damage from cotton worm. In Sanya of China, cotton can grow in whole year because of comfortable hot climate. So cotton is very easy to be harmed by various insects. By four year field investigation from 2010–2014, it was tested that some cultivated cotton species like TM-1, CCRI 12, CCRI 16 were all damaged mostly by various insects like pink worm, cotton bollworm, aphid, thrips, mite, leaf hopper and stinkbug, whereas this triple hybrid performed high insect resistance especially immune to pink worm, cotton bollworm, aphid and mite under no any protection control. The possible reason is that it inherits the resistant gene from its parents G. capitis-viridis and G. australe. We will further confirm its resistant characteristics and identify this resistant gene in next study. The triple hybrid is also high drought resisting because of its long epidermis hair on the leaf and stem.

In addition, the hybrid plants were perennial, semi-bush with crawl growth characteristic different from any other parents. So it can prevent from wind damage suitable for cultivating dwarf and wind resistant species in Hainan of China.

We try to backcross this hybrid with cultivars and hope to recover its fertility incorporating into next cotton breeding program. We will continue to examine the stability of the morphological features and further to check its other excellent agronomic traits of this hybrid in our later study.

In this study, the meiotic behavior in three parents was quite normal in the first division and completely normal in the second division, generating about 100% of normal tetrads. Observations were also made on chromosome morphology and behavior during meiosis for this triple hybrid. After the occurrence of an irregular first cytokinesis, the meiocytes progressed normally to the second division, generating various numbers of unbalanced microspores mainly because anaphase chromosome movements were blocked. As a consequence of this abnormality in chromosome orientation, a few or several micronuclei of various sizes were observed inside the cells in telophase I and II with diads, triads and different polyads (Fig 4). Finally, various abnormal pollen grains were appeared like cracked pollen, big oval-shaped pollen, under-developed pollen and so on (Fig 5). Only a small portion of pollen grains are normal style. These meiotic observations are similar to those previously reported by Rieseberg [10]. It was further proved that this hybrid is actually wide cross product from different three parents and maybe it has a long way to get into harmonious genome interaction from B, A, D and C. But it gives us suggestive thinking that it is possible to produce a novel germplasm recombining three parents characters through the production of trispecific hybrid.

In this paper, the hybrid and its three parents were analyzed using SSR markers from DNA molecular level. A total of 112 selected polymorphism SSR primers were amplified and 98 pairs were screened effective bands in these four accessions. A total of 230 DNA bands were scored from 98 primers for this hybrid, with loci number ranging from one to six per primer pair. By statistic analysis of amplified bands, the genetic proportion coming from three parents in this hybrid is in Table 2. Among the SSR markers in this trispecific hybrid, 47.8% of them were coming from G. hirsutum, 14.3% from G. australe, 7.0% from G. capitis-viridis. Other 30.9% markers were the new and peculiar ones for this hybrid. Different new polymorphics of molecular markers from its parents resulting from difference of sequences might be due to chromosomal rearrangements in this newly formed hybrid [11]. This trispecific hybrid had the characteristic bands of three parents and accordingly had its parent genetic materials from SSR molecule marker analysis. It was also testified that wild genetic material has been transferred into cultivated cotton and this new germplasm can be incorporated into next cotton breeding program. It was also tested that SSR analyses can provide insight into the genome difference in hybrid identification.

Our examinations from morphology and cytogenetic analysis in meiosis following with SSR identification showed that this hybrid was actually wide cross product from three parents G. hirsutum, G. australe and G. capitis-viridis.

Polyploidy is a prominent process in plant evolution. Approximately 70% of angiosperms are thought to have experienced one or more episodes of polyploidy at some point in the past [12]. Among the best studied polyploids are many of the world’s leading crops, including cotton, wheat, oat, soybean, peanut, canola, tobacco, coffee, and banana, each of which evolved by the joining of divergent genomes in a common nucleus [13]. The evolution of the genus Gossypium (cotton) has included a very successful experiment in polyploid formation by hybridization between a maternal Old World ‘‘A” genome taxon resembling G. herbaceum (2n = 2x = 26) and paternal New World ‘‘D” genome taxon resembling G. raimondii [14] or G. gossypioides [15] (both 2n = 2x = 26) [16]. Similarly, the reunion of genomes through hybridization and allopolyploidy is conservatively estimated to account for 2–4% of speciation events in flowering plants and 7% in ferns [17], and most plants appear to be paleopolyploids [18]. Hybridization and introgression may play an important role in evolution and often serves as a repair or replacement strategy, rather than solely as a mechanism for the development and/or acquisition of novel traits [19]. Cells differing only by their ploidy are identical in terms of DNA sequence information and relative gene dosage, and yet are often quite different in terms of physiology, morphology, and behavior [20].

The justification for the use of these characters is the assumption that they have a complex genetic basis of this triple hybrid. Rieseberg [21] indicates that the genetic basis of morphological differences between wild plant species possibly controlled by a single major locus plus modifiers. Vlot et al [22] found a molecular marker locus associated with the major gene controlling pappus part number. Doebley et al [23] also investigated that the genetic control of morphological traits differentiating maize from its wild ancestor fit the general model of a single major locus with large effects plus modifiers. These contrasts indicate polyploidy speciation in plants is accompanied by a diverse array of molecular evolutionary phenomena, which will vary among both genomic constituents and taxa [24], also between some wide cross hybrids and their parents. A better understanding of the number of loci responsible for species differences would greatly assist in assessing the reliability of particular morphological characters, reproductive morphology, frequently employed for phylogenetic and taxonomic inference [21] and hybrids identification.

In most cases, divergent genomes do not coexist peacefully upon first contact and hybrids, if formed at all, are inviable or sterile [25]. Although F₁ sterility or inviability is a common feature of wide interspecific crosses, small introgressions often have indiscernible heterozygous effects while being lethal or sterilizing in the homozygous state [26]. Reproductive barrier formation between newly derived hybrid taxa and their parental species represents a major evolutionary hurdle [10]. Rieseberg [27] studies of the genes causing hybrid incompatibilities is that postzygotic isolation often as a by-product of adaptation. However, natural selection for viability and fertility provides an effective filter that eliminates gametes and individuals carrying maladaptive changes and favors those inheriting advantageous changes [18]. The genomic changes appeared to be largely complete by the third generation following polyploidy formation, and were accompanied by an increase in seed fertility and a reduction of irregular chromosome pairing in meiosis [28].

In this study, we hope to incorporate G. australe and G. capitis-viridis genetic material into upland cotton with some novel traits from the species, such as insect and disease resistance from G. capitis-viridis and low gossypol seed & high-gossypol plant trait introgressed from G. australe. Its other excellent agronomic traits need our further identification.