Characteristics of Phenotypic Variation of Malus Pollen at Infrageneric Scale

Abstract

Pollen carries extensive genetic information, which may provide clues regarding the kinship of Malus, whose genetic relationships are complex. In this study, the phenotypic variation of pollen from 107 Malus taxa was investigated using combined methods of intraspecific/interspecific uniformity testing, cluster analysis, and Pearson correlation analysis. The family aggregation distributions in Malus sections, species, and cultivars were analyzed to infer their pedigree relationships. The results showed that (1) compared with pollen size and morphology, aberrant pollen rate and ornamentation were highly interspecifically differentiated, but ornamentation was also intraspecifically unstable, especially perforation densities ($\overline{c.v.}$ > 15%). (2) The pollen alteration direction from the original to the evolutionary population of Malus was large to small, with elliptic to rectangular morphologies, large and compact to small and sparse ridges, and low to high perforation densities. However, there was no significant change in pollen size. (3) The 107 studied taxa were divided into four groups. Malus species were relatively clustered in the same section, while homologous cultivars showed evidence of family aggregation distribution characteristics (92.60% of cultivars were clustered with their parents). (4) M. baccata and M. pumilar var. neidzwetzkyana were high-frequency parents, participating in 38.7% and 20.7% of cross-breeding, respectively. Overall, this study provides a reference for identifying Malus’ pedigree relationship.

1. Introduction

Pollen morphology has developed during long-term evolution and exhibits species-specific characteristics [1,2]. Therefore, pollen traits are often used in investigations on plant distribution, genetic evolution, or systematic taxonomy [3,4,5]. Previous studies mainly used morphological comparisons, cluster analyses, or principal component analyses of pollen phenotypic parameters [6,7,8] and then deduced evolutionary relationships based on similarity or clustering genetic distance. Alternatively, some researchers mapped morphological structures onto a molecular phylogenetic tree to analyze the evolutionary trends of phenotypes [9,10]. However, most researchers only use descriptive methods to illustrate these results, without statistical analyses to support their conclusions. The systematization of germplasm materials is an essential basis for the statistical analysis of phylogenetic data. Previous studies mainly analyzed the genetic relationships among species, genera, or families [11,12,13,14]. However, there is value in exploring the relationship between parent and progeny and taxonomic sections below the level of the genus or even the species.

Malus spp. are of high ornamental value, and there is a rich diversity of Malus germplasm globally [15]. More than 30 species have been recorded [16,17], and nearly one thousand Malus taxa have been bred in the past 200 years, but only over 200 cultivars can be found in the nurseries [15,18]. Unfortunately, the genetic backgrounds and inter-/intra-species relationships of most cultivars are still unclear [6]. So far, researchers have identified ancestors for 65 cultivars, which provides a systematic basis for the statistical analysis of the phylogenetic relationship of the genus [19,20].

Based on pollen phenotypes of 107 Malus taxa (23 species and 84 cultivars), we combined several statistical analysis methods to examine Malus plant classification and pedigree relationships at two levels: classification units (sections, species, and cultivars) and parent–progeny populations. The objectives are (1) to clarify the characteristic of phenotypic variation of pollen from the original to the evolutionary Malus population; (2) to explore the family aggregation distribution characteristics of Malus classification units and parent–progeny populations; and (3) to investigate how the frequency of hybridization with important parents affects the affinity of Malus cultivars.

2. Results

2.1. General Characteristics of Malus Taxa Pollen

All SEM micrographs of pollen grains of the Malus taxa and corresponding data are shown in Figure S1 [21] and Table S1, respectively. The aberrant pollen rates exhibited considerable variation, ranging from 0% to 95.60% (Figure 1a). A total of 13.08% of the taxa had an abnormal pollen rate exceeding 40%. Nevertheless, most taxa (75.70%) displayed abnormal pollen rates below 20.00%. The morphological characteristics of normal pollen grains are summarized below.

The average length of the polar axis (P) of the Malus taxa was 44.95 μm (Figure 1b). The smallest mean P was found for pollen of M. honanensis (29.44 μm), while the largest mean P was found for ‘Brandywine’ (52.35 μm). The mean equator diameter (E) was 23.65 μm, ranging from 20.95 μm in ‘Van Eseltine’ to 29.64 μm in ‘Pink Princess’. The area of the equatorial view with two colpi (S) ranged from 560.25 to 1108.89 μm². The P/E ratio ranged from 1.22 to 2.21, the P/E′ ratio ranged from 1.60 to 2.56, and the E′/E₀ ratio ranged from 0.76 to 0.88 (Figure 1c). The ridge width (RW) of the striate exine ornamentation ranged from 0.13 to 0.25 μm, and the furrow width (FW) of the striate ranged from 0.04 to 0.28 μm (Figure 1d). A total of 35 taxa had no perforations (32.71%), while 72 taxa had perforations, with perforation densities (PDs) ranging from 0.11 per μm² in M. prunifolia to 11.18 per μm² in ‘Van Eseltine’.

In summary, according to the pollen size classification created by Erdtman [22], most pollen was medium-sized (25.1–50 μm; 89.7%), and a few taxa were relatively large (10.3%) (Figure 1a). All pollen grains were bilaterally symmetrical, with perprolate, prolate, or sub-rectangular shapes in the equatorial view (P/E = 1.22–2.21) and tricolpate with three germinal furrows in the polar view. Exine ornamentation was striate–perforate; some were striate or smooth (‘Regal’) without perforations.

2.2. Intraspecifc Uniformity Test and Interspecific Distinctness Analysis of Pollen Morphology Traits

Compared with pollen size and morphology, ornamentation was highly interspecifically differentiated and intraspecifically unstable (Figure 1d,e). Except for PD and FW of the striate, the remaining seven traits had significant intraspecific uniformity ($\overline{c.v.}$ ≤ 15%) (Figure 1d). As for the interspecific distinctness of pollen traits, all traits differed significantly among the taxa (p < 0.0001) (Figure 1e). However, pollen extine ornamentation (PD, FW, and RW) showed a higher degree of distinctness (c.v. ≥ 15%). Because the PD and FW vary significantly within the taxa, the RW of striate ornamentation was more suitable as the main index for identification.

2.3. Cluster Analysis Based on Pollen’s Phenotypic Traits

A cluster analysis of Malus taxa based on six independent variables (P, E, E′/E, RW, FW, and PD) was conducted (Figure 2). The 107 taxa were divided into four groups: A, B, C, and D. The pollen size (P, E, and S) and morphological parameters (P/E, P/E′, and E′/E) of the A group were significantly smaller than those of the other groups (p < 0.0001,

Table 1. The pollen phenotypic traits of Malus taxa based on cluster analysis.

Pollen Phenotypic Traits	Group/Subgroup
	A	B	C	D	D1	D2	D3
P (μm)	36.26 ± 4.21 d	48.42 ± 4.36 a	46.54 ± 2.69 abc	45.29 ± 3.01 bc	46.18 ± 3.58 abc	47.14 ± 2.26 ab	43.90 ± 2.54 c
E (μm)	24.24 ± 1.02 bc	27.42 ± 0.82 a	22.95 ± 1.33 de	23.39 ± 1.11 de	24.86 ± 0.76 b	23.72 ± 0.85 cd	22.74 ± 0.79 e
E′ (μm)	19.25 ± 1.10 c	22.41 ± 0.67 a	19.85 ± 1.20 bc	19.78 ± 0.90 bc	20.49 ± 0.76 b	20.14 ± 0.83 b	19.35 ± 0.75 c
S (μm2)	690.05 ± 74.49 e	1 043.46 ± 71.34 a	840.68 ± 77.43 bcd	834.03 ± 83.22 cd	902.58 ± 76.27 b	879.84 ± 64.92 bc	785.28 ± 63.12 d
P/E	1.51 ± 0.21 d	1.78 ± 0.20 c	2.04 ± 0.13 a	1.94 ± 0.11 ab	1.86 ± 0.15 bc	1.99 ± 0.08 a	1.94 ± 0.10 ab
P/E′	1.90 ± 0.24 c	2.17 ± 0.19 b	2.36 ± 0.15 a	2.30 ± 0.13 a	2.27 ± 0.19 ab	2.35 ± 0.09 a	2.28 ± 0.12 ab
E′/E	0.80 ± 0.02 d	0.82 ± 0.03 c	0.87 ± 0.00 a	0.85 ± 0.02 b	0.82 ± 0.01 c	0.85 ± 0.01 b	0.85 ± 0.01 b
RW (μm)	0.19 ± 0.04 a	0.20 ± 0.02 a	0.17 ± 0.03 a	0.19 ± 0.03 a	0.18 ± 0.06 a	0.17 ± 0.02 a	0.20 ± 0.03 a
FW (μm)	0.10 ± 0.05 c	0.12 ± 0.04 bc	0.20 ± 0.04 a	0.12 ± 0.04 bc	0.10 ± 0.05 c	0.15 ± 0.03 b	0.12 ± 0.03 bc
PD (No./μm2)	3.66 ± 3.49 b	1.81 ± 1.92 c	9.23 ± 1.41 a	1.87 ± 2.23 c	0.42 ± 0.64 c	4.72 ± 1.53 b	0.57 ± 0.81 c
Taxa quantity (No. and %)	7 and 6.54	6 and 5.61	6 and 5.61	88 and 82.24	14 and 15.91	28 and 31.82	46 and 52.27

The different lowercase letters indicate significant differences (p < 0.05) in each pollen morphology trait among different groups or subgroups.

). The pollen size of the B group was significantly greater than that of the other groups (p < 0.0001). The PD and morphological parameters (E′/E) of the C group were significantly greater than those of the other groups (p < 0.0001). There was no apparent trend for the pollen phenotypic parameters in the D group. The germplasm distribution was uneven among the four groups, with 82.2% of the taxa in the D group. The D group was divided into three subgroups (D₁, D₂, and D₃). The number of taxa in the three D group subgroups was also uneven (D₁: 15.9%, D₂: 31.8%, and D₃: 52.3%), and the pollen size of the D₁ and D₂ subgroups was significantly higher than that of the D₃ subgroup (p < 0.0001). The pollen morphology parameters of the D₂ and D₃ subgroups were significantly greater than those in the D₁ subgroup (p < 0.0001). The D₂ subgroup had significantly greater PD than the other two subgroups (p < 0.0001).

2.4. The Characteristics of Family Aggregation Distribution

2.4.1. The Family Aggregation Distribution between Species and Their Sections

According to the Malus taxonomy system proposed by Rehder [16] and Li [17], the genus Malus can be classified into six sections (Docyniopsis, Chloromeles, Sorbomalus, Baccatus, Malus, and Eriolobus). The 23 species used in this study represent all sections except for Eriolobus: I, Sect. Docyniopsis (1 species); II, Sect. Chloromeles (3 species); III, Sect. Sorbomalus (6 species); IV, Sect. Baccatus (3 species) and Sect. Malus (10 species) (Figure 2). The species distribution within the same section was relatively concentrated (Figure 2 and Figure 3a). None of the species we studied were classified in the C group. All species in Sect. Docyniopsis and Baccatus were distributed in the D group. Two-thirds of the species in Sect. Chloromeles and Sorbomalus were distributed in the D group, and 90% of species belonging to Sect. Malus were distributed in the D group. The total percentage of the family aggregation distribution reached 82.6% in the same group and 52.2% in the same subgroups (Figure 3a).

According to the order from original to evolved in the classic taxonomy system [17], the five sections of Malus species were valued as follows: Sect. Docyniopsis (1) → Sect. Chloromeles (2) → Sect. Sorbomalus (3) → Sect. Baccatus (4) → Sect. Malus (5). There was a negative but insignificant correlation between the five sections and pollen size (r < −0.3, p > 0.05) (Figure 3b).

2.4.2. The Family Aggregation Distribution between Cultivars and Their Parents (Species)

In this study, we could trace the origin of the parents of 28 cultivars (

Table 2. The parent traceability and identification of family aggregation distribution characteristics of Malus cultivars.

No.	Progeny	Breeding Routes	References	Class Group in Figure 2	Is the Family- Aggregation Distributed?		Is the Perforation Density of the Progeny Higher than that of one of Its Parents?
					A, B, C, D	A, B, C, D1, D2, D3	Unknown/Untested Taxa Are Considered with Lower Perforation Densities.	Unknown/Untested Taxa Are Excluded.
1	M. ‘Adams’	M. baccata × unknown → M. ‘Adams’	[19]	D2 × unknown → D2	Yes	Yes	Yes	Yes
2	M. ‘Almey’	M. baccata × M. pumila var. neidzwetzkyana → M. ‘Almey’	[19]	D2 × D3 → D2	Yes	Yes	Yes	Yes
3	M. ‘Brandywine’		[18,19]		No	No	Yes	No
4	M. ‘Cardinal’		[18,19]		Yes	No	Yes	No
5	M. ‘Dolgo’		[18]		Yes	Yes	Yes	Yes
6	M. ‘Eleyi’		[19]		Yes	Yes	Yes	No
7	M. ‘Flame’	M. pumila × unknown → M. ‘Flame’	[18]	D3 × unknown → D3	Yes	Yes	Yes	No
8	M. ‘Golden Hornet’	M. mandshurica × M. sieboldii → M. ‘Golden Hornet’	[19]	D3 × untest→ D3	Yes	Yes	Yes	Yes
9	M. ‘Hillier’	M. floribunda × M. prunifolia → M. ‘Hillier’	[23]	D2 × D1 → D3	Yes	No	Yes	Yes
10	M. ‘Hopa’	M. baccata × M. pumila var. neidzwetzkyana → M. ‘Hopa’	[18]	D3 × D3→ D3	Yes	Yes	Yes	Yes
11	M. ‘Klehm’s Improved Bechtel’	M. ioensis × unknown → M. ‘Klehm’s Improved Bechtel’	[17]	D1 × unknown → D2	Yes	No	Yes	Yes
12	M. ‘Lisa’	M. ioensis × unknown → M. ‘Lisa’	[18]	D1 × unknown → D2	Yes	No	Yes	Yes
13	M. ‘Liset’		[18]		Yes	Yes	Yes	No
14	M. ‘Makamik’	M. pumila var. neidzwetzkyana × Unknown → M. ‘Makamik’	[18]	D3 × unknown → D3	Yes	Yes	Yes	No
15	M. ‘Prairie Rose’	M. ioensis × unknown → M. ‘Prairie Rose’	[18]	D1 × unknown → D2	Yes	No	Yes	Yes
16	M. ‘Professor Sprenger’	M. mandshurica × M. sieboldii → M. ‘Professor Sprenger’	[23]	D3 × untest → D3	Yes	Yes	Yes	Yes
17	M. ‘Profusion’		[19]		Yes	Yes	Yes	No
18	M. ‘Purple prince’		[18]		Yes	Yes	Yes	Yes
19	M. ‘Radiant’		[19]		Yes	Yes	Yes	No
20	M. ‘Red Jade’		[19]		Yes	No	Yes	No
21	M. ‘Red Splendor’		[19]		Yes	Yes	Yes	Yes
22	M. ‘Robinson’	M. baccata × M. pumila → M. ‘Robinson’	[19]	D2 × D3 → D3	Yes	Yes	No	No
23	M. ‘Royalty’		[19]		Yes	Yes	Yes	No
24	M. ‘Rudolph’	M. baccata × unknown→ M. ‘Rudolph’	[18,19]	D2 × unknown →D3	Yes	No	Yes	No
25	M. ‘Sparkler’		[18]		Yes	Yes	Yes	No
26	M. ‘Spring Snow’		[19]		Yes	Yes	Yes	Yes
27	‘Van Eseltine’		[19]		No	No	Yes	Yes
28	M. zumi ‘Calocarpa’	M. mandshurica × M. sieboldii → M. zumi ‘Calocarpa’	[17]	D3 × untest →D3	Yes	Yes	Yes	Yes
The probability of family aggregation distribution (the percentage of the progeny and its parents belonging to the same group, %)					92.86	67.86	/	/
The percentage of the progeny whose perforation density is higher than that of its parents (%)					/	/	96.43	53.57

, Figure 2). Cultivars in the A group had no traceable parents, whereas cultivars in the B and C groups each had one traceable parent. There were 26 traceable cultivars in the D group, and all exhibited family aggregation. Overall, according to the statistical analysis of the four cluster groups (A, B, C, and D), the family aggregation distribution of the cultivars was as high as 92.9%. According to our statistical analysis of the six cluster groups/subgroups, the percentage of parent tendency distribution was as high as 67.9%.

Furthermore, E′/E, FW, and PD showed an increasing trend from species to cultivar populations. The positive correlation between populations E′/E and FW reached a significant level (p = 0.0015, p = 0.039) (Figure 3b). The percentage of the progeny whose perforation density is higher than that of its parents is 53.57%. If the perforation density of the unknown/untested taxa is assumed to be lower, this percentage can be up to 96.43%.

2.4.3. Hybridization Frequency of Species

The hybridization route of 65 cultivars is recorded in the literature, involving 18 Malus species [17,18,23,24]. The hybridization frequency of the 18 species can be classified into high-, medium-, and low-frequency parents (Figure 4a). M. baccata and M. pumila var. neidzwetzkyana are high-frequency parents, with hybridization frequencies of 38.7% and 20.7%, respectively. We found that our hybridization frequencies for 12 species in this study were highly correlated with what has been recorded in the literature (R² = 0.89) (Figure 4b).

3. Discussion

3.1. High Aberrant Pollen Rates May Be an Effective Indicator for Apomixis or Multiploid Cultivar Selection

The occurrence of aberrant pollen is regular in various angiosperm families [25,26]. These aberrant pollen grains can be distinguished by features such as shape and size, aperture number and arrangement, and ornamentation type, different from the typical pollen morphology of the species. Halbritter et al. [27] pointed out some apomictic Malus species that only produce irregularly shaped pollen due to asexual reproduction. In this study, we found that 10.28% of taxa had abnormal pollen proportions exceeding 40%. Of these taxa with high abnormal pollen rates, some, including M. mandshurica, ‘Mary Potter’, ‘Regal’, and ‘Firebird’, have been proven to be apomixis and multiploid [28], with a low pollen germination rate [29]. The multiploid plants in the Malus genus fail to produce viable pollen during meiosis due to unequal chromosome allocation, which may be the reason for the aberrant pollen [30]. If the apomixis characteristics of Malus taxa with aberrant pollen are confirmed, the grafting breeding method for these cultivars can be replaced by seed propagation. Moreover, if confirmed, the high aberrant pollen rate can serve as an effective indicator for apomixis or multiploid cultivar selection.

3.2. Variation Tendency of Malus Pollen Morphology

Plants evolve at both macro and micro levels. Intra-genus (inter-species and inter-cultivar) evolution is also considered as microevolution, which reflects the evolutionary process of plant populations within a short term. Inter-genus (or above: inter-family, inter-ordo) evolution can also be regarded as macroevolution, which reflects the origin and the phylogenetic process of plant populations within a geologic age [24]. Different research approaches are required because macroevolution differs from microevolution in terms of the research object type and the time scale. Paleobiology and comparative morphology are often used to investigate macroevolution, whereas genetics, ecology, and systematics are mainly used to investigate microevolution [31].

Current studies have shown that the evolutionary patterns of pollen are observed at the macroevolutionary level. By using both fossil and sample pollen and based on morphological phylogenetics, Walker et al. [13,32,33] studied the evolutionary trend (at macro- and multi-levels) of the pollen morphology of inter-family or inter-generic taxonomic groups and concluded that the evolution directions were as follows: pollen size: large → small; pollen shape: boat-shaped → globose; exine ornamentation: smooth → foveolate and ditch-shaped → clavate, drumstick-shaped and echinate → crisped, reticulum and striate; perforation: nonexistence →existence/small → large; ridge interval: short → long. However, for taxonomic groups of infrageneric germplasms (microlevel), trends in evolution are either highly defined [34,35] or undetectable [4,36].

The present study investigated nine traits of pollen size, shape, and exine ornamentation of Malus taxa. We compared the changes in pollen traits from Malus species to cultivars and from the original section to the evolutionary section (microlevel taxonomic groups) via correlation analysis. We determined the following trends: big pollen → small ones (insignificant); elliptic morphologies → rectangular ones (E′/E, significant); large and compact ridges → small and sparse ones (RW, insignificant; FW, significant); no perforations or low perforation densities → high perforation densities (PD, insignificant) (Figure 5). Our conclusion agrees well with the viewpoint of Walker et al. [13,33,34]. Moreover, FW and PD had low intraspecific uniformity ($\overline{c.v.}$ > 15%), which means that the distribution of perforations and stripes on the exine was uneven. RW was more suitable as the main index for identification ($\overline{c.v.}$ < 15%, c.v. > 15%). E′/E, as the new morphological index, had high intraspecific uniformity and directivity of evolution, which is better for comparing the degree of originality or evolution of Malus taxa.

Notably, we observed a higher density of perforations in the D₂ cluster, which comprised the evolutionary sections, including Sect. Baccatus and Sect. Sorbomalus and their progeny. These evolutionary groups were characterized by many perforations, such as M. baccata, M. rockii, M. floribunda, and M. fusca (Table S1). The number of perforations in their progeny may be influenced by the characteristics of their parental species. Furthermore, we calculated the percentage of the progeny exhibiting an upward trend in perforation density across generations. We discovered that over half (53.57%) of the progeny displayed perforation densities exceeding those of their parents (Table 2). This trend of increasing variation in perforation density, from lower to higher, is unequivocally evident in macroevolution (tectate–imperforate → tectate–perforate → semitectate → intectate) [32,37]. The adaptive significance of this evolutionary trend in perforation was related to the evolution of an infratectal–intercolumellar storage area for sporophytic incompatibility proteins [32]. This protein may be stored within the perforations or reticulum [32]. However, it is unclear if the increased perforation density in hybrid offspring indicates a more pronounced self-incompatibility in subsequent generations. This hypothesis warrants further investigation for validation.

3.3. Genetic Relationships of Malus Taxa Based on Phenotypic Traits of Pollen

Malus has various germplasm and complex genetic relationships [17]. Many studies have explored the genetic relationships between Malus species and cultivars using molecular markers [38,39], isozymes [40], and palynology [41]. However, these studies had small sample sizes and did not analyze the entire genus. Previous studies also did not analyze the genetic relationship of Malus using statistical methods.

The classic taxonomic system of Malus is based on plant morphological characteristics [16,17]. Here, using the cluster analysis of phenotypic traits in Malus pollen, we found that the distribution of Malus species from the same section was relatively concentrated within groups. The family aggregation distribution percentage was 92.9%, which indicated that the classification system based on pollen phenotypic traits was consistent with the traditional classification system based on phenotypic traits.

There were 18 species involved in crossing, and M. baccata and M. pumilar var. neidzwetzkyana were determined to be high-frequency parents (38.7% and 20.7%, respectively). M. baccata is widely distributed in China and has large white flowers [42]. M. pumilar var. neidzwetzkyana is the genetic source of red flower cultivars. There is no red color germplasm in the original species except for M. pumilar var. neidzwetzkyana, which has great significance in color improvement breeding [43]. In this study, the species belong to the D₂ and D₃ subgroups, respectively. These two subgroups not only have a higher percentage of Malus cultivars but also have a high family aggregation distribution. This is in accordance with Fiala [18] and Jefferson [19]. Meanwhile, this also verified why the phenotypic diversity of existing crabapple cultivars is not rich enough. There are some excellent germplasms, such as M. ioensis, M. hupehensis, M. halliana, and M. micromalus. If these germplasms can be frequently used as parents to carry out crossbreeding, their offspring will likely be able to fill vacancies in existing cultivars (few early blooms, few orange fruits, and a faint aroma).

Future research could focus on establishing connections between phenotypic variations and taxonomic specificities, functionality, fruit yield, and the phenotypes of flowers, leaves, and fruits.

4. Materials and Methods

4.1. Materials

In all, 107 Malus taxa (including 23 species and 84 cultivars) were used (

Table 3. A list of the 107 taxa collected from the national repository of the Malus taxa.

Species	Cultivars	Cultivars	Cultivars
Malus angustifolia II	M. ‘Abundance’	M. ‘Hopa’	M. ‘Red Sentinel’
M. baccata IV	M. ‘Adams’	M. ‘Indian Magic’	M. ‘Red Splendor’
M. domestica var. binzi V	M. ‘Almey’	M. ‘Indian Summer’	M. ‘Regal’
M. floribunda V	M. ‘Ballet’	M. ‘Kelsey’	M. ‘Robinson’
M. fusca III	M. ‘Brandywine’	M. ‘King Arthur’	M. ‘Roger’s Selection’
M. honanensis III	M. ‘Butterball’	M. ‘Klehm’s Improve Bechtel’	M. ‘Royal Beauty’
M. ioensis II	M. ‘Cardinal’	M. ‘Lancelot’	M. ‘Royal Gem’
M. mandshurica IV	M. ‘Centurion’	M. ‘Lisa’	M. ‘Royal Raindrop’
M. micromalus V	M. ‘Cinderella’	M. ‘Liset’	M. ‘Royalty’
M. ombrophila III	M. ‘Cloudsea’	M. ‘Lollipop’	M. ‘Rudolph’
M. platycarpa II	M. ‘Coralburst’	M. ‘Louisa Contort’	M. ‘Rum’
M. prunifolia V	M. ‘Darwin’	M. ‘Makamik’	M. ‘Show Time’
M. pumila V	M. ‘David’	M. ‘Mary Potter’	M. ‘Snowdrift
M. Pumila var. neidzwetzkyana V	M. ‘Dolgo’	M. ‘May’s Delight’	M. ‘Sparkler’
M. robusta V	M. ‘Donald Wyman’	M. ‘Molten Lava’	M. ‘Spring Glory’
M. rockii IV	M. ‘Eleyi’	M. ‘Perfect Purple’	M. ‘Spring Sensation’
M. sieversii V	M. ‘Everest’	M. ‘Pink Princess’	M. ‘Spring Snow’
M. spectabilis V	M. ‘Fairytail Gold’	M. ‘Pink Spires’	M. ‘Sugar Tyme’
M. sylvestris V	M. ‘Firebird’	M. ‘Prairie Rose’	M. ‘Sweet Sugar Tyme’
M. toringoides III	M. ‘Flame’	M. ‘Prairifire’	M. ‘Thunderchild’
M. tschonoskii I	M. ‘Furong’	M. ‘Professor Sprenger’	M. ‘Tina’
M. turkmenorum V	M. ‘Golden Hornet’	M. ‘Profusion’	M. ‘Vans Eseltine’
M. yunnanensis III	M. ‘Golden Raindrop’	M. ‘Purple Gems’	M. ‘Velvet Pillar’
	M. ‘Gorgeous’	M. ‘Purple Prince’	M. ‘Weeping Madonna’
	M. ‘Guard’	M. purpurei ‘Neville Copeman’	M. ‘White Cascade’
	M. halliana ‘Pink Double’	M. ‘Radiant’	M. ‘Winter Gold’
	M. ‘Harvest Gold’	M. ‘Red Baron’	M. ‘Winter Red
	M. ‘Hillier’	M. ‘Red Jade’	M. zumi ‘Calocarpa’

The superscript numbering from I to V indicates the evolutionary sequence of sections and series, with I being the most ancient and V being the most advanced. I: Sect. Docyniopsis, II: Sect. Chloromeles, III: Sect. Sorbomalus, IV: Sect. Baccatus, V: Sect. Malus [16,17]. Cultivars in bold font indicate that their parentage is known, according to Zhou et al. [20].

). Pollen was collected from the national repository of Malus germplasm (lat. 32°42′ N, long. 119°55′ E, Yangzhou City, Jiangsu Province, China). The tree ages were between 5 and 8 years. We collected 30 flowers at the large bud stage, wrapped them in litmus paper, laid them in a cooler at 4 °C, and transported them back to the laboratory within 8 h. Anthers were peeled from the flowers and air dried at room temperature for one week, after which the pollen sacs were split open. The dried pollen was collected for further observation.

4.2. Pollen Observations

The dried pollen grains were spread on a glass slide with a double-sided adhesive tape and were sprayed with gold for 120 s at an electric current of 16 mA using a magnetron ion-sputtering device (E-1010, Hitachi Ltd., Tokyo, Japan). Then, a field-emission scanning electron microscope (SEM; S-4800, Hitachi Ltd., Japan) was used to observe pollen traits. The sample support was kept at room temperature, and the acceleration voltage was 15 kV. Representative pollen grains were photographed (2.50 K) and presented in the book [21].

Image J 1.54 software [44] was used to measure the abnormal pollen rate (AP), pollen size, morphology, and ornamentation according to original SEM micrographs of pollen grains (Figure 6) [21]. Pollen grains that differ in shape, size, number arrangement of apertures, and ornamentation type from the typical pollen type of the species are considered as abnormal pollen. The percentage of total pollen count that consists of abnormal pollen is defined as the abnormal pollen rate. The pollen size indicators measured were the length of the polar axis (P), equatorial diameter (E), the diameter at the equatorial plane halfway between the equator and pole (E′), and the area of equatorial view with two colpi (S). The pollen morphology indicators measured were the ratio between the length of the polar axis and equatorial diameter (P/E), the ratio of the length of the polar axis and diameter at the equatorial plane halfway between the equator and pole (P/E′), and the ratio between the diameter at the equatorial plane halfway between the equator and the pole and the equatorial diameter (E′/E). The pollen ornamentation indicators measured were the ridge width (RW), furrow width (FW), and perforation density (PD). Thirty replicates were used for each indicator.

4.3. Data Analysis

Family aggregation means that the progeny and their parents were gathered via cluster analysis, which can provide a reference for their genetic relationship.

The intraspecific uniformity test for quantitative traits is expressed by the mean coefficient of variation ($\overline{c.v.}$). If $\overline{c.v.}$ ≤ 15%, then the trait has met the uniformity requirements. The interspecific distinctness analysis of quantitative traits was expressed by the coefficient of variation (c.v.) of the mean value of each trait in all taxa. If c.v. ≥ 15%, the differentiation degree of this trait is high among all the taxa.

$$\overline{c.v.}={\displaystyle \frac{1}{n}}{\sum }_{i=1}^n{(S}_i/\overline{X_i})$$

(1)

$$c.v.=S^{\prime }/{\overline{X}}^{\prime }\times 100\%$$

(2)

Here, n denotes the number of taxa; S_i and ${\overline{X}}_{\mathit{i}}$ denote the rank for the standard deviation of the observed values and the mean observed values for each trait in each taxon’s repetitions, respectively. S′ and $\overline{\mathrm{X}}$′ denote the standard deviation and the average of observed mean values of each trait in all taxa, respectively.

OriginPro 9.8.0.200 (OriginLab Corp., Northampton, MA, USA) and Adobe Illustrator CC 23.0.2 (Adobe Inc., San Jose, CA, USA) were used for processing data and plotting graphs. ANOVA was conducted using SPSS 26.0 (IBM Corp., Armonk, NY, USA).

5. Conclusions

In this study, we explored the characteristics of the phenotypic variation of Malus pollen at the infrageneric scale. Overall, (1) an aberrant pollen rate and ornamentation were highly interspecifically differentiated, but ornamentation was also intraspecifically unstable, especially for perforation densities. (2) The pollen alteration direction from the original to the evolutionary population of Malus was from large to small, from elliptic to rectangular morphologies, from large and compact to small and sparse ridges, and from low to high perforation densities. (3) The distribution of family aggregation represents a notable characteristic. In summary, this study provides a reference for identifying Malus’ pedigree relationships. These findings improve our understanding of pollen phenotypic variation’s taxonomic significance at the genus level.