The morphological diversity of pollen in the genus Gossypium

Abstract

Background

Plant pollen has diverse morphological characteristics that can be consistently passed down from generation to generation. Information on pollen morphology is thus immensely important for plant classification and identification. In the genus Gossypium, however, in-depth research on pollen morphology is lacking, with only few reports on limited cotton species. To evaluate the diversity of pollen in Gossypium, we therefore conducted a comprehensive analysis of the pollen morphology of 33 cotton species and varieties using scanning electron microscopy.

Results

The 33 analyzed cotton samples exhibited common pollen morphological features, including spherical shapes, radial symmetry, echination, panporation, and operculation, while the pollen size, spine shape, spine density and length showed distinctive features. Pollen size varied significantly among species, with diameters ranging from 62.43 μm in G. harknessii to 103.41 μm in G. barbadense. The exine had an echinate sculptural texture, and spines were mostly conical or sharply conical but occasionally rod-like. Spine density varied from 173 in G. incanum to 54 in G. gossypioides, while spine length ranged from 3.53 μm in G. herbaceum to 9.47 μm in G. barbadense. In addition, the 33 cotton species and varieties were grouped at a genetic distance of 3.83 into three clusters. Cluster I comprised five allotetraploid AD-genome cotton species, four D-genome species, and one K-genome species. Cluster II included 13 diploid species from A, B, D, E, and G genomes, whereas Cluster III only consisted one E-genome species G. incanum.

Conclusions

Although pollen characteristics alone are not enough to resolve taxonomic and systematic relationships within the genus Gossypium, our results add to knowledge on palynomorphology and contribute to phenological information on these taxa. Our findings should aid future systematic and phylogenetic studies of the Gossypium genus.

Introduction

Gossypium L. (cotton), the largest and most widely distributed genus in the Gossypieae, comprises more than 50 species (Fryxell 1992), including 46 diploid species (2n = 2x = 26) as well as 5 well-established and 2 newly identified tetraploid species (2n = 4x = 52) (Grover et al. 2014; Wendel and Grover 2015; Wang et al. 2018). The diploid cotton species are believed to have originated from a common ancestor approximately 5 million to 10 million years ago (MYA) and subsequently diversified cytogenetically into eight genome groups (designated A to G and K) that differ in chromosome size but not in number (Wendel and Cronn 2003). Gossypium species vary greatly in morphology, ranging from herbaceous perennials to subshrubs and even to small trees. Moreover, the flowers of Gossypium also differ in size and color. The AD genome group is predominantly characterized by the presence of ivory-white or bright yellow flowers, whereas rosy flowers are prevalent among species belonging to the G, C, and K genome groups of Australia.

Cotton is an economically significant crop worldwide and also serves as a model species for evolutionary studies (Wendel et al. 2012). An extensive amount of molecular phylogenetic research has been conducted on Gossypium (Cronn et al. 2002; Wendel and Cronn 2003; Grover et al. 2008; Xu et al. 2012; Wendel and Grover 2015; Hu et al. 2021; Wang et al. 2022). For example, the phylogenetic relationships of diploid cotton genome groups have been evaluated using 16 nuclear and chloroplast genes, which revealed that cotton genome groups underwent rapid radiation after formation of the genus (Cronn et al. 2002). Grover et al. (2015) has updated previously assumed connections among Gossypium allopolyploids by applying targeted sequence capture of multiple loci in conjunction with both concatenated and Bayesian concordance. The report has discovered that a newly identified allopolyploid species (G. ekmanianum) is closely related to G. hirsutum and has established a more robust phylogeny for allopolyploid Gossypium (Grover et al. 2015). A comparative analysis of 19 Gossypium chloroplast genomes by Chen et al. (2016) has divided the eight diploid genome groups into six clades. The analysis has revealed contrasting evolutionary dynamics in different clades, with parallel genome downsizing in two genome groups and a biased accumulation of insertions in the clade containing cultivated cottons leading to large chloroplast genomes. The rapid global diversification of the Gossypium genus has resulted in some challenging phylogenetic questions.

Pollen morphological characteristics can provide a reliable basis for classifying and identifying the origin and evolution of plants (Rosenfeldt and Galati 2007; Erik 2012; Baser et al. 2016; Mezzonato-Pires et al. 2018; Reunov et al. 2018; Grimsson et al. 2019; Zhang et al. 2021; Umber et al. 2022). For example, pollen grains of Sclerosperma mannii and S. walheri share the same distinct reticulate sculpture, which suggests that these two currently accepted Sclerosperma species are sister taxa of the same intrageneric lineage (Grimsson et al. 2019). Pollen grains from different species have developed unique morphological characteristics during evolutionary while many common morphological features has passed on within families and genera. These morphological features are controlled by genes and not easily be affected by external environment, thus can be used in plant taxonomy.

A previous investigation based on pollen wall stripping and scanning electron microscopy (SEM) has revealed that the pollen grain size of G. barbadense was the largest among four cultivated cotton species (Jia et al. 1988). Another study has used SEM to examine the ultrastructure of upland cotton (G. hirsutum) pollen at various developmental stages and has discovered that the pollen walls are covered with fully grown spiny protuberances and apertures (Liu et al. 1994). Despite these prior investigations of the pollen morphology of some cotton species, no systematic studies have been conducted. In the current study, we therefore performed a comprehensive SEM-based survey of 33 species and varieties of Gossypium. Our findings provide detailed knowledge of cotton palynomorphology and have potential value for elucidating cotton taxonomy and evolutionary history.

Materials and methods

Plant materials

A total of 33 germplasm accessions representing 23 Gossypium species were evaluated in the present study (Table 1). All specimens were planted at the National Wild Cotton Germplasm Resources Nursery (Sanya, Hainan, China), which is supervised by the Institute of Cotton Research, Chinese Academy of Agricultural Sciences (ICR-CAAS), Anyang, Henan, China. Pollen grains wrapped in anthers were collected in the morning from intact, freshly opened flowers. The grains were fixed in 2.5% glutaraldehyde solution (in 0.2 mol·L^–1 phosphate buffer; pH 7.4) for 3 h, washed at least three times with 0.135 mol·L^–1 phosphate buffer (Na₂HPO₄·7H₂O, 30.40 g; NaH₂PO₄·H₂O, 2.98 g; pH 7.4), soaked in 2.5% glutaraldehyde solution, and stored at 4 °C.

Table 1 The information of materials

Scanning electron microscopy (SEM) analysis of cotton pollen

Micro-morphological features of cotton pollen were examined by SEM following the protocol of Lan and Xu (1996). Pollen grains were stepwise dehydrated in 50%, 70%, 90%, and 100% ethanol. Treated pollen grains were then mounted onto the surface of polished aluminum stubs using double-sided adhesive tape. Each stub was sputter coated with a gold layer and taped to the object stage. Observations and micrograph acquisition were carried out under a scanning electron microscope (Hitachi S-350) installed at the Institute of Cotton Research, Chinese Academy of Agricultural Sciences. Biometric measurements were conducted using Image-Pro Plus 6.0 (Media Cybernetics, Silver Spring, Maryland, USA). Measurements were carried out on 20 randomly chosen, mature, non-malformed pollen grains per sample.

Data exploration and statistical analysis

In this study, we measured three quantitative pollen characters of at least 20 pollen grains per collected specimen, including pollen diameter, exine spine length, and exine spine density (the number of spine on the front side of each pollen grain). We also examined the following qualitative characters: pollen shape, spine type, and protuberance at the base of exine spine. Protuberances were classified into three categories: small, medium, and significant. All pollen terminology follows the glossary in Erdtman (1987) and Punt et al. (2007). All statistical analyses were conducted using SPSS18.0 software. Inter-relationships of the six evaluated traits were analyzed based on Pearson correlation coefficients. A euclidean distance matrix of the standardized data was then subjected to cluster analysis by the unweighted pair group method with arithmetic mean as implemented in NTSYS pc version 2.1.

Results

Pollen morphology

The results of SEM analysis of the pollen morphology of 33 cotton specimens are shown in Fig. 1 and Table 2. We found that pollen grains of all examined samples were spherical, radially symmetrical, and panporate. Each pollen grain had more than eight nearly operculate apertures. Aperture number varied widely among cotton taxa, with the largest number present in G. gossypioides (Fig. 1y). In contrast, little variation was observed in pollen shape, which was spheroidal in all Gossypium species.

Fig. 1

SEM micrographs of pollen grains in Gossypium. a G. hirsutum (CRI12), b G. barbadense (CL), c G. barbadense (YM), d G. barbadense (XH7), e G. tomentosum (LZ), f G. tomentosum (JFZ8), g G. tomentosum (01), h G. mustelinum (LZ), i G. mustelinum (16), j G. darwinii (07), k G. herbaceum (ZC1), l G. herbaceum (africanum), m G. arboreum (SXY), n G. arboreum (Rozi), o G. anomalum (LZ), p G. capitis-viridis (LZ), q G. capitis-viridis (01), r G. thurberi (LZ), s G. thurberi (35), t G. armourianum (LZ) u G. harknessii (LZ), v G. davidsonii (LZ), w G. klotzschianum (LZ), x G. raimondii (01), y G. gossypioide s(LZ), z G. schwendimani i(LZ), aa G. stocksii (LZ), ab G. somalense (LZ), ac G. areysianum (LZ), ad G. incanum (04), ae G. bickii (LZ), af G. australe (LZ), ag G. rotundifolium (LZ). Scale bar: 50 μm. Accelerating voltage: 15 kV

Table 2 Pollen morphologies of Gossypium species

Pollen size is normally calculated as the product of polar and equatorial diameters. Because the pollen grains of all examined cotton species were spheroidal, however, polar and equatorial views could not be easily distinguished. We therefore used pollen diameter to represent pollen size. On this basis, the size of pollen grains generally varied widely among cotton species. The largest mean pollen diameter among 33 samples was in G. barbadense, 103.41 μm, and the smallest mean value was in G. harknessii, 62.43 μm (Fig. 2A). In a previous study, the largest pollen grains were those of island cotton (G. barbadense), followed by upland cotton (G. hirsutum), G. arboreum, and G. herbaceum (Jia et al. 1988). We also found that pollen grains of G. barbadense were significantly larger than those of other cotton species. Moreover, we discovered that pollen grains of allotetraploid (AD genome) cotton species, with a mean pollen diameter of 94.7 μm, were much larger than those of diploid cotton species (A–K genomes), which showed mean pollen diameters in the range of 69.66 μm (G genome) to 87.25 μm (K genome) (Fig. 2B). Pearson correlation analysis of cotton genome size (Table 1) and pollen size (Table 2) yielded an r value of 0.648, which indicates a moderately positive correlation between these two characters. Detailed results are listed in Table 3.

Fig. 2

The diameter of pollen in the Gossypium species. A Comparison of pollen diameter among Gossypium species. B Comparison of pollen diameter among different genomes of Gossypium

Table 3 The Pearson correlation test between genome size and pollen size in Gossypium genus

Exine sculpture

We observed a large variation in the shape, number, and length of exine spines within the genus Gossypium. Spine shape was categorized as sharply conical, conical, or rod-like based on the sharp top of spines (Table 2). Accordingly, we uncovered significant differences among the studied species: 15 of sharply conical, 14 of conical and 3 of rod-like. The rod-like type of exine spine was rarely present in G. harknessii or E-genome species G. somalense or G. areysianum. In addition, the size of the protuberance at the base of exine spine, which was classified as small, medium, or obvious, varied among samples (Table 1).

Spine density varied from 54 in G. gossypioides to 173 in G. incanum. Large variations in spine density were observed between cotton species, whereas intraspecific differences were smaller. Exine spine numbers in three varieties of G. tomentosum (AD₃) were 76, 64, and 57, and the number of G. mustelinum (AD₄) were 89 and 87. Notably, the number of spines in G. incanum (173) was higher than that of other E-genome species, such as G. stocksii (118), G. somalense (100), and G. areysianum (106).

Spine lengths in different species ranged from 3.53 μm in G. herbaceum to 9.47 μm in G. barbadense, which was consistent with the measurements of pollen size (Fig. 3A). Similarly, the mean exine spine length of AD-genome species was larger than that of all diploid species except for K-genome member G. rotundifolium (Fig. 3B). Among allotetraploids (AD genome), the spine length of G. hirsutum (AD₁) was similar to that of G. tomentosum (AD₃), whereas the spine length of G. barbadense (AD₂) was nearly the same as that of G. darwinii (AD₄). These results suggested a close relationship of G. hirsutum versus G. tomentosum and of G. barbadense (AD₂) versus G. darwinii (AD₄), consistent with scientifically recognized phylogenetic classifications. A Pearson correlation analysis was performed to examine the relationship between pollen diameter and spine length. The Pearson correlation coefficient was found to be 0.690, indicating a significant correlation between these two variables (Table 4).

Fig. 3

The length of exine spine in the Gossypium species. A Comparison of exine spine length among Gossypium species. B Comparison of exine spine length among different genomes of Gossypium

Table 4 The Pearson correlation test between pollen diameter and spine length in Gossypium genus

Cluster analysis

A cluster analysis was conducted using the morphometric data obtained from 33 samples of 23 Gossypium species (Fig. 4). At a genetic distance of 3.83, samples were separated into three clusters. Cluster I consisted of five allotetraploid AD-genome species (G. hirsutum, G. barbadense, G. tomentosum, G. mustelinum, and G. darwinii) and five diploid species (G. klotzschianum, G. davidsonii, G. raimondii, G. schwendimanii, and G. rotundifolium). Cluster II contained 13 diploid species (G. herbaceum, G. arboreum, G. anomalum, G. capitis-viridis, G. thurberi, G. armourianum, G. harknessii, G. gossypioides, G. stocksii, G. somalense, G. areysianum, G. bickii, and G. australe). Cluster III comprised a single species (G. incanum).

Fig. 4

Dendrogram of Gossypium based on pollen characteristics (A). Three pollen types (right): Type I, big pollen size, long spine length, low spine density and little protuberance (B); Type II, small pollen size, short spine length, low spine density, and obvious protuberance (C); Type III, small pollen size, short spine length, high spine density, and little protuberance (D). Scale bar: 50 μm. Accelerating voltage: 15 kV

According to the cluster analysis, the genus Gossypium was divided into three pollen types (Fig. 4). A key to pollen types and general pollen morphology of the genus Gossypium was provided in Table 2 (mean values were shown). Members of type-I pollen were characterized by the larger pollen size (almost all diameters > 80 μm), the longer exine spine (almost all lengths > 6 μm), exine spine numbers ranging from 57 to 117, the sharply conical exine spine, and the small or medium protuberance at the base of exine spine. Type-II members had the following features: the smaller pollen size (almost all diameters  < 80 μm), the shorter exine spine (almost all lengths < 6 μm), exine spine numbers ranging from 54 to 128, any of the three exine spine shapes, and the medium or obvious protuberance at the base of the exine spine. Finally, pollen characteristics in species of type-III were the smaller pollen size (diameter = 72 μm), the shorter exine spine (4.21 μm), the high exine spine density (173), and the small protuberance at the base of the exine spine.

Discussion

Various species of Gossypium, although distinct morphologically, have many shared features. Pollen grains of all species examined in this study are monad, radially symmetrical, spherical, echinate, panporate, and operculate. These results are consistent with previous studies (Jia et al. 1988; Lan and Xu 1996; Saensouk and Saensouk 2021). For example, Saensouk and Saensouk (2021) has investigated the pollen morphology of subfamily Malvoideae in Thailand and has found that pollen grains of all 19 species (including G. barbadense) are monad, spheroidal, panporate, apolar and radially symmetrical. Pollen shape, which is based on the ratio of polar diameter (P) to equatorial diameter (E), can be classified into five categories, i.e., oblate (P/E < 0.50), suboblate (P/E = 0.5–0.88), spheroidal (P/E = 0.88–1.14), subprolate (P/E = 1.14–2.00), and prolate (P/E > 2.00) (Erdtman 1987). In this study, all investigated species have a spheroidal pollen shape, which is congruent with the results of previous studies (Jia et al. 1988; Lan and Xu 1996). Significant variations in pollen size and sculpture are potentially very useful for species identification and delimitation (Ullah et al. 2018). In this study, we have observed considerable variation in pollen size between different cotton species, but little variation within one species. This pattern suggests that pollen size is a useful taxonomic parameter at the species level.

As is well known, pollen size usually increases with chromosome number. For instance, the pollen grain sizes of tetraploid species of the genus Skimmia tend to be larger compared with those of diploid species, but this is not always true (Fukuda et al. 2008). In the present study, we also observed that the pollen size of allotetraploids (4n = 52) had larger pollen sizes relative to diploid cotton species (2n = 26). The most widespread consequence of polyploidy in plants is resulting increased cell size due to the larger number of gene copies. Consequently, polyploid individuals may exhibit larger organs, such as roots, leaves, tubercles, fruits, flowers, and seeds, compared with the diploid (Sattler et al. 2016). Interestingly, we also observed that pollen of the diploid K-genome species (G. rotundifolium) are larger than other diploid cotton species. The increased pollen size of G. rotundifolium may due to its larger genome, evenif the average genome size of K-genome species (~ 2 570 Mb) is only slightly larger than that of AD-genome allotetraploids (~ 2 400 Mb) (Wendel et al. 2012). In the present study, Pearson correlation analysis uncovered a significant positive correlation (r = 0.65, P < 0.01) between genome size and pollen size in cotton.

As noted in previous studies, the exine ornamentation of pollen is an important diagnostic character for delimiting relationships within tribes, families, and genera in Brassicaceae (Khalik et al. 2002; Erik 2012). In addition, Baser et al. (2016) have reported that exine ornamentation is useful for distinguishing between closely related species in the same genus, such as Pelargonium endlicherianum and P. quercetorum. In the present investigation, we have found that the pollen exine of all cotton species was densely covered with spines with one exception: G. armourianum, whose pollen surface is covered with a thick layer of wax. On the basis of shape, exine spines of cotton pollen grains are separated in two main types: conical or rod-like. The most common exine spine shape in Gossypium is conical (19 species), and a rod-like shape only found in three species. Judging from the SEM photographs, the base of the exine spine commonly bulges to form a tuberculate structure (called as protuberance), but the size of the protuberance varies among the investigated cotton species. Although pollen grains of Gossypium species usually have echinate exine ornamentation, most of them have different spine densities and exine spine lengths. In addition, the results of Pearson correlation analysis indicate that cotton pollen size is highly correlated with exine spine length (r = 0.69, P < 0.001) while weakly correlated with exine spine number (r =  − 0.04, P > 0.05). This suggests that cotton species with larger pollen sizes have longer exine spines. For example, the diploid cotton species G. schwendimanii has larger pollen grains and longer exine spines but fewest number of exine spines. On the contrary, G. incanum has smaller pollen grains and shorter exine spines but with the most exine spines. In regard of G. armourianum, the thick wax coating on its pollen grains is difficult to remove, we are unable to observe and analyze the exine spine during multiple preliminary experiments using pollen from materials of different years. Nevertheless, we believe that G. armourianum has exine spines, as the presence of echinate sculpture is a conserved feature in the Malvaceae family. And G. harknessii, the sister species of G. armourianum has normal exine spines. Further study, however, is necessary to fully reveal the pollen characteristics of G. armourianum.

Cluster analysis based on the six studied pollen traits has grouped all allotetraploid species (AD genome group) into Cluster I, with the wild cotton G. tomentosum (AD₃) showing a closer relationship to the cultivated cotton G. hirsutum (AD₁) compared with other allotetraploid species. Some placements were confusing and unexpected, for example, the nine cotton species from the D-genome group are distributed on different cluster branches, and G. incanum (E₄) is separated from other E-genome species (E₁, E₂, and E₃). Although the cluster dendrogram was not exactly coincident with the molecular phylogenetic tree proposed by Wendel and Cronn (2003), the results of this analysis provide new information, from a palynological perspective, on phylogenetic relationships among 23 analyzed Gossypium species. A sole reliance on pollen characteristics is thus obviously insufficient for understanding taxonomic and evolutionary relationships within the genus Gossypium. Nevertheless, pollen morphological variations are of significance for further taxonomic classification at the species level.

Conclusions

In this study, we acquired SEM images and pollen morphological data for the largest number of cotton species reported to date. We found that pollen size, ornamentation and spine features among species are distinctive and can be used to understand the interspecific relationship. The morphological description from this study not only expands the knowledge in cotton palynomorphology but also sheds light on the palynomorphology based phylogenetic analysis among Gossypium species.