Interspecific Hardy Geranium Progenies: Morphological Characterization and Genetic Evaluation

Abstract

Hardy geranium is a popular ornamental plant with a high market demand. As a result, there is a strong need for continuous innovation to improve its ornamental qualities. Interspecific hybridization is a widely used technique for introducing desirable traits and creating new cultivars with added value. To explore the possibilities of creating novelty within the Geranium genus, we conducted 2438 interspecific crosses over three flowering seasons, resulting in the successful harvest of 445 seeds and the final production of 82 seedlings. Using embryo rescue increased the germination rate of a seedling from 0.17% to 1.01%. To validate the paternal genetic contribution in the new seedlings, we used amplified fragment length polymorphism (AFLP) to assess the proportion of uniquely inherited paternal markers in all seedlings. AFLP analysis confirmed that 54 of the 82 seedlings were indeed true hybrids (65.8% hybridization success rate). A morphological evaluation of the flowers and leaves, including flower diameter (mm), style length (mm), number of flowers, flowering period (weeks), and color of both flowers and leaves, provided further confirmation of the F1 hybrid status of the seedlings. The results of our interspecific breeding demonstrated the efficiency of interspecific hybridization in hardy geraniums and its potential to create distinctive and novel cultivars for the ornamental market.

1. Introduction

Novel genetic variation can be created by various means, such as hybridization within or between plant species, mutagenic treatment, or alteration of the ploidy level [1]. The origin of many commercial cultivars can be attributed to the hybridization between botanical varieties, cultivars, and species, whether it occurs intentionally or naturally. Interspecific hybridization, in particular, has been valuable in creating new floriculture crops and cultivars by introducing new desirable traits [2]. Genome hybridization involves merging the complete genomes of two parental species to create a hybrid offspring with a wider gene pool, while amphiploid hybridization is the creation of a hybrid offspring that has two complete sets of chromosomes from each parent, and introgression hybridization involves transferring desirable traits from one species, often a wild relative, to an elite cultivar of another species [3]. However, interspecific hybridization can be largely prevented by the presence of pre- and post-fertilization barriers. To overcome embryo maldevelopment after fertilization, embryo rescue is commonly used [4], enabling the development of immature or mature–lethal hybrid embryos from interspecific and intergeneric crosses that cannot survive in vivo or through traditional plant breeding practices [5,6,7,8,9]. The process involves excising immature or lethal embryos and culturing them in vitro on a specific nutrient medium. Success requires careful attention to species-specific nutrient requirements and an assessment of growth condition parameters [5]. Embryo rescue has been used in many species, including Lycopersicon sp. [10], Brassica napus [11], and Vitis vinifera L. [12] and ornamental plants, such as Chrysanthemum sp. [13,14], Gentiana sp. [15], and Pelargonium sp. [16].

Geranium is a large and diverse genus of flowering plants in the family Geraniaceae, consisting of three subgenera: Erodioidea, Robertium, and Geranium. The largest subgenus, Geranium, is primarily native to the temperate northern regions of Eurasia, extending from the eastern Mediterranean to the western Himalayas. In addition, South America has a remarkable diversity of Geranium species [17,18,19]. The rosette growth habit, with a dense, upright inflorescence emerging from a compact cluster of basal leaves, is common in Geranium. However, some species have an alternative growth pattern with elongated stems and sparse basal leaves [20]. These attractive plants, known for their winter hardiness, extended flowering period, large radially symmetrical flowers, and vibrant colors in both leaves and flowers, have become very popular as ornamentals since the 1990s [19,20].

The first interspecific crosses of hardy geraniums available on the market were between G. macrorrhizum, G. sanguineum, G. himalayense, G. endressii, and G. × oxonianum (G. endressii × G. versicolor). A logistic regression model was developed to predict immature seed production from interspecific hybridization in hardy geraniums (Geranium sp.) using the genetic distance between parental plants [21]. The primary goal of interspecific hybridization in this genus is to use plant diversity to introduce new traits. These traits include extended flowering periods and novel flower and leaf colors and sizes. For example, there is a lack of hardy geraniums with large white flowers that are resistant to powdery mildew. However, while some parental combinations exhibit intriguing traits, certain combinations are incompatible. Overcoming the incompatibility barriers allows breeders to combine desirable traits from different species. The main focus of this article is the evaluation of interspecific seedlings of Geranium using amplified fragment length polymorphism (AFLP) [22], alongside the study of their morphological characteristics to explore the potential for hybridization. Key issues addressed include identifying true hybrids among the seedlings and assessing the extent of morphological differences from their parents. In addition, we provide insight into the importance of embryo rescue in the context of hardy geraniums.

2. Materials and Methods

2.1. Plant Materials

A diverse collection, including species from three different subgenera of Geranium genus, Geranium, Erodioidea, and Robertium (described in [23], see Table S1), was used for crosses in three consecutive years (2020, 2021, and 2022). The collection is located at ILVO in Melle, Belgium, precisely at the coordinates 50°59′31.6″ N, 3°47′07.3″ E. All plant material and genotypes in our collection were verified and confirmed by a plant breeder (Figure S1).

2.2. Male Fertility and Female Receptivity

Pollen fertility was assessed using the fluorescein diacetate (FDA) staining technique. For this, freshly harvested anthers were collected in the morning and placed in a 500 µL solution containing 130 g.L⁻¹ of sucrose. To facilitate pollen release, the anthers were gently stirred in the solution. After removal of the anther debris, FDA was added to the pollen and allowed to incubate for 5 min in the dark. The pollen was then briefly centrifuged at 6000 rpm, the supernatant carefully removed, and the resulting pellet resuspended in 300 µL of sucrose solution. The analysis involved counting 100 pollen grains from each of the three samples collected on three different days, with three replicates for each sample. This resulted in a total of 300 pollen counts (3 samples × 3 replicates × 100 pollen counts). A fluorescence microscope (Leica DMIRB, Wetzlar, Germany) equipped with a filter set suitable for fluorescein (L5) and a magnification of 5× or 10× was used.

Female receptivity for selected cross combinations was monitored by examining in vivo pollen germination and subsequent pollen tube growth by aniline blue staining of pistils. At 48 h after pollination, pistils were harvested and immersed for 24 h in an FAA solution consisting of formaldehyde, acetic acid, and 70% ethanol in a 1:1:18 ratio. After thoroughly rinsing with water, the pistils were transferred to a 6 M NaOH solution for incubation for 16 h. Subsequently, the pistils were immersed in a solution containing 0.033 M K₃PO₄ and 0.1% aniline blue (Acros Organics, Geel, Belgium) for 3 h in the dark. Pollen tubes were visualized and examined using the same fluorescence microscopy and filter set used above. Next to aniline blue staining, a spontaneous seed set in the field was assessed over two consecutive flowering seasons for the whole collection as an alternative method to assess female receptivity.

2.3. Interspecific Crosses

A total of 349, 1155, and 934 interspecific crosses were made in 2020, 2021, and 2022, respectively. These crosses included 96, 236, and 61 different combinations of parental genotypes (described in Tables S2–S4). The emasculation of young flowers was performed for most genotypes, while it was considered unnecessary for the male sterile genotypes (see further in Section 3). A few flowers within each of the genotypes were not emasculated for the purpose of being used as pollinators.

2.4. Embryo Rescue

To identify the optimal medium for rescuing seeds, four different media (designated M₁ to M₄) were tested (

Table 1. Growth media formulations used in hardy geranium embryo rescue.

Medium	Composition
M1	Quoirin and Lepoivre full-strength medium containing sucrose (30 g.L−1), agar (7.5 g.L−1), and activated charcoal (0.05 g.L−1)
M2	M1 supplemented with casein hydrolysate (100 mg.L−1).
M3	M1 supplemented with L-proline (100 mg.L−1).
M4	M1 supplemented with glutamine, asparagine, serine, and arginine (100 mg.L−1 each).

).

A total of 84 open-pollinated seeds were collected, and embryos were dissected from three different genotypes: Geranium ‘Blushing Turtle’ (G80), Geranium ‘Orion’ (G14), and Geranium himalayense ‘Baby Blue’ (G38) at different stages of development, categorized as early stage (A₁), intermediate stage (A₂), and mature stage (A₃) (Figure 1) and assigned to the four different growth media. After an incubation period of approximately five months, the number of germinated seeds and the development of robust seedlings were assessed for each growth medium. From the second crossing season onwards, all embryos were transferred to an in vitro environment, and M₁ was used throughout the final breeding season.

For the in vitro embryo rescue of seeds resulting from interspecific crosses, we determined premature seed harvest days by visually assessing seed characteristics, including color, size, and shrinkage.

2.5. AFLP Analysis

DNA was extracted from 100 mg of young plant leaf samples (of 82 seedlings) stored at −80 °C using the CTAB DNA isolation protocol derived from the method established by Doyle and Doyle [24]. AFLP analysis [25] was performed according to the procedure described in Akbarzadeh et al. (2021) [23]. The commercially available AGG kit from Perkin-Elmer Biosystems (Waltham, MA, USA) was used for fluorescent fragment detection [26,27]. The following primer combinations were used: EcoRI-AAC_ned/MseI-CAG, EcoRI-AAG_hex/MseI-CAG, EcoRI-ACA_fam/MseI-CTT, and EcoRI-AGG_hex/MseI-CAG. The evaluation and scoring of the peaks were carried out in GeneMapper version 5 (Applied Biosystems) based on De Riek et al. (1999). According to Denaeghel et al. (2017), seedlings with more than 25% male unique markers were classified as true hybrids. Those with a number of unique parental markers between 10 and 25% were recorded as partial hybrids. Seedlings with less than 10% unique markers were classified as non-hybrids [28].

2.6. Morphological Analysis

Flower and leaf color was measured using a portable spectrophotometer (Konica Minolta CM-700d, Konica Minolta Sensing, Inc., Osaka, Japan) at anthesis, more specifically at the mature stage of stigma (Figure S2), in three replicates. ΔE*ab (Delta E CIE 1976) was used to evaluate the differences between the flowers and leaves of the parents and the corresponding seedlings. In 1976, the International Commission on Illumination (CIE) introduced the first color difference equation applicable to a known set of CIE L*a*b* coordinates.

$${{\Delta E}^{*}}_{ab}=\sqrt{{\left({L_2}^{*}-{L_1}^{*}\right)}^2+{\left({a_2}^{*}-{a_1}^{*}\right)}^2+{\left({b_2}^{*}-{b_1}^{*}\right)}^2}$$

This equation calculates the color difference between two objects with coordinates (L₁*, a₁*, b₁*) and (L₂*, a₂*, b₂*) [29].

Flower diameter and style length were measured using a conventional ruler on 2-year-old seedlings. These measurements were taken at the mature stigma stage to ensure the accuracy and consistency of the recorded data. Style length was measured from the base of the style to the tip. Flowering time was quantified by recording the number of weeks from the onset of flowering, marked by the appearance of three flowers at anthesis, to the end of the flowering phase. To assess flower abundance, the total number of flowers was counted during the flowering season.

2.7. Statistical Analysis

IBM SPSS Statistics version 29.0.1.0 (171) was used to perform a general linear model analysis followed by a post-hoc test using Tukey, with a significance level (α) set at 0.05 to analyze the embryo rescue data.

3. Results

3.1. Parental Male Fertility and Female Receptivity

To make an accurate crossing table, it is necessary to assess male fertility and female receptibility before hybridization. FDA staining of the pollen revealed a viability between 0% and 79.9% (

Table 2. Evaluation of female receptivity and male fertility (using FDA staining) and the need for emasculation of different Geranium cultivars. Female receptivity is expressed as seed set in open pollination. ‘Yes’ indicates that the genotype is capable of producing fully mature seeds, ‘No’ indicates that the genotype is not capable of producing fully mature seeds, and ‘No^im’ indicates that the genotype is capable of producing immature seeds in open pollination. ‘×’ indicates that emasculation isn’t necessary and ‘✓’ indicates that emasculation is necessary.

Code	Genotype	Female Receptivity Expressed as Seed Set (Open Pollination) *	Male Fertility Expressed as Pollen Viability ± SE	Need for Emasculation
G01	Geranium ‘Ann Thomson’	No	0	×
G02	Geranium ‘Azure Rush’	Noim	24.4 ± 2.0	✓
G03	Geranium ‘Bob’s Blunder’	No	0	×
G04	Geranium ‘Brookside’	Yes	60.9 ± 5.7	✓
G05	Geranium ‘Catherine Deneuve’	Noim	0	×
G06	Geranium ‘Chantilly’	No	24 ± 2.8	✓
G07	Geranium ‘Dragon Heart’	No	0	×
G09	Geranium ‘Galactic’	Yes	19 ± 4.6	✓
G10	Geranium ‘Jolly Jewel Red’	Noim	37.9 ± 20.8	✓
G13	Geranium ‘Mavis Simpson’	No	0	×
G14	Geranium ‘Orion’	Yes	34 ± 2.4	✓
G16	Geranium ‘Rozanne’	Noim	26.2 ± 1.9	✓
G17	Geranium ‘Salome’	No	7.8 ± 2.7	✓
G18	Geranium ‘Sanne’	No	0	×
G19	Geranium ‘Silverwood’	Yes	51.6 ± 0.5	✓
G21	Geranium ‘Sylvia’s Surprise’	Yes	36.4 ± 15.2	✓
G22	Geranium ‘Tanya Rendall’	No	0	×
G24	Geranium ‘Tiny Monster’	Noim	21 ± 8.0	×
G27	Geranium × cantabrigiense ‘Biokovo’	Noim	60.6 ± 0.9	✓
G30	Geranium cinereum ‘Laurence Flatman’	No	61.4 ± 8.6	✓
G35	Geranium endressii	Yes	75.2 ± 11.7	✓
G37	Geranium endressii ‘Trevor Bath’	Yes	39.3 ± 2.1	✓
G38	Geranium himalayense ‘Baby Blue’	Yes	56.5 ± 27.2	✓
G39	Geranium himalayense ‘Derrick Cook’	Yes	20 ± 9.4	✓
G42	Geranium macrorrhizum ‘Czakor’	Noim	- *	✓
G44	Geranium macrorrhizum ‘White Ness’	Noim	45.2 ± 9.6	✓
G45	Geranium maculatum ‘Album’	Yes	72.2 ± 13	✓
G46	Geranium maculatum ‘Elizabeth Ann’	Yes	79.9 ± 8.7	✓
G49	Geranium × oxonianum ‘Katherine Adele’	Yes	49.7 ± 11	✓
G50	Geranium × oxonianum ‘Southcombe Double’	Yes	0	×
G54	Geranium phaeum ‘Angelina’	Yes	54.8 ± 5.9	✓
G57	Geranium pratense ‘Algera Double’	No	0	×
G61	Geranium pratense ‘Purple Ghost’	Yes	10.6 ± 3.1	✓
G62	Geranium psilostemon	No	- *	×
G64	Geranium renardii	Noim	70 ± 3.2	✓
G69	Geranium sanguineum ‘Album’	Yes	46.6 ± 17.6	✓
G71	Geranium sylvaticum ‘Album’	Yes	26.2 ± 7.3	✓
G73	Geranium versicolor	Yes	49.2 ± 14.7	✓
G75	Geranium ‘Bloom Time’	Yes	10.4 ± 1.2	✓
G76	Geranium wallichianum ‘Havana Blue’	Noim	22.1 ± 7.5	✓
G77	Geranium wlassovianum	No	0	×
G80	Geranium ‘Blushing Turtle’	Noim	30.6 ± 8.9	×

* Not measured as it occasionally produces pollen.

). Our evaluation confirmed that Geranium ‘Ann Thomson’ (G01), Geranium ‘Bob’s Blunder’ (G03), Geranium ‘Dragon Heart’ (G07), Geranium ‘Mavis Simpson’ (G13), Geranium ‘Sanne’ (G18), Geranium ‘Tanya Rendall’ (G22), Geranium ‘Algera Double’ (G57), and Geranium wlassovianum (G77) are both male and female sterile. In our collection, pollen production in Geranium psilostemon (G62) could occur but was rare; hence, pollen viability could not be measured (Table 2).

The growth of pollen tubes after pollination is influenced by the parental combination. Figure 2 shows examples of arrested and successful pollen tube growth through the style tissue. Pollen of Geranium ‘Brookside’ (G04) could not germinate in the cross combinations Geranium macrorrhizum ‘White Ness’ (G44) × Geranium ‘Brookside’ (G04) and Geranium ‘Blushing Turtle’ (G80) × Geranium ‘Brookside’ (G04), with pollen tube growth arrested before entering the ovules. On the other hand, pollen tubes of Geranium ‘Brookside’ (G04) and Geranium himalayense ‘Derrick Cook’ (G39) could successfully reach the ovules in both combinations of Geranium ‘Sanne’ (G18) × Geranium ‘Brookside’ (G04) and Geranium ‘Brookside’ (G04) × Geranium himalayense ‘Derrick Cook’ (G39) (Figure 2). As is also shown in Figure 3, if the parental genetic distance (=cJaccard, 1-Jaccard coefficient) between species increased, the number of fully successfully grown pollen tubes decreased. When the cJaccard was below 0.699 [21], pollen tubes in 9 out of 10 combinations successfully reached the ovules. The number of arrested pollen tubes increased as the genetic distance increased between 0.849 and 1 (Figure 3).

3.2. Interspecific Crosses between Geranium Species and Overcoming Post-Fertilization Barriers

In order to optimize the embryo rescue technique for Geranium, we analyzed the responses of the embryo obtained in open pollinated plants in different media. The highest rate of embryo germination was observed in the combination of medium M₁ and intermediate stage embryos (A₂) (

Table 3. The effect of different media and embryo age on the response (absolute numbers/total) of Geranium ‘Blushing Turtle’ (G80), Geranium ‘Orion’ (G14), and Geranium himalayense ‘Baby Blue’ (G38) embryos. (M indicates different media, as detailed in Table 1, and (A) indicates different embryo ages, as shown in Figure 1). Absolute numbers/total were calculated based on observed embryos, including germinated, calloused, and non-reactive embryos. General linear model analysis with Tukey’s post hoc test within a column revealed a significant difference between groups ^a and ^b at α = 0.05.

Medium	Embryo Age	Germinated Embryo (%)	Callused Embryo (%)	Non-Reactive Embryo (%)
M1	A1	2/18 ab	1/29 a	3/37 a
	A2	4/18 a	0 a	1/37 a
	A3	1/18 b	4/29 a	5/37 a
M2	A1	- *	- *	- *
	A2	2/18 ab	3/29 a	6/37 a
	A3	2/18 ab	5/29 a	1/37 a
M3	A1	1/18 ab	4/29 a	4/37 a
	A2	0 b	3/29 a	3/37 a
	A3	0 b	4/29 a	2/37 a
M4	A1	1/18 ab	1/29 a	3/37 a
	A2	2/18 ab	1/29 a	6/37 a
	A3	3/18 ab	3/29 a	3/37 a

* Not available.

).

Conversely, M₂A₃ combinations showed the highest levels of callus formation, which is undesirable in hardy geraniums (Figure 4). Geranium ‘Blushing Turtle’ was the only genotype capable of both direct embryo germination and shoot regeneration from the callus-formed embryo. M₄ showed optimal performance for A₃ embryos close to maturity; however, M₄ is not very effective for embryo rescue. The need for embryo rescue is mainly observed in distant cross combinations (high cJaccard value 0.799 > 1 [21,23]), where embryos fail to reach maturity. In contrast, seeds from close parental combinations (low cJaccard distance) usually reach full maturity.

Table 4. Summary of breeding and seedling production data per growing season (2020, 2021, and 2022). The germination rate was calculated by dividing the actual number of seedlings by the total number of potential seedlings (five for each cross, e.g., 349 × 5 = 1745 potential seedlings, 3/1745 × 100 = 0.17).

Growing Season	2020	2021	2022
Number of all crosses	349	1155	934
Number of different cross combinations	96	236	61
Number of harvested seeds	15	175	255
Number of combinations that produced seeds	5	36	33
Number of combinations that produced seedlings	2	15	13
Number of seedlings	3	32	47
Germination rate (%)	0.17	0.57	1.01

and the Supplementary Tables S2–S5 provide a detailed overview of the interspecific hybridization and seedling production process over the three growing seasons (2020, 2021, 2022), showing the number of crosses made, different cross combinations, seed production, developed seedlings, and germination rate. The average cJaccard coefficient of parental combinations during the breeding year was consistently within a narrow range (0.81, 0.79, and 0.75 correspond to the first, second, and third crossing seasons, respectively). The increase in germination rate from 0.17% in 2020 to 0.57% in 2021 can be explained by the introduction of embryo rescue instead of mature seed germination in 2020 (Table 4). A further increase to 1.01% in 2022 was achieved through parental selection, improved crossing techniques, harvesting time, and the use of the optimized embryo rescue medium.

3.3. Genetical and Morphological Evaluation of F1 Progeny

AFLP analysis revealed that out of 82 seedlings obtained over 3 years, 54 seedlings (65.8%) were identified as true hybrids with more than 25% unique paternal markers. In contrast, a smaller subset of five seedlings were classified as partial hybrids, inheriting between 10% and 25% unique paternal markers. A group of 23 of the seedlings did not meet the criteria for true hybrids as they inherited less than 10% of the unique paternal markers, as shown in Table S5.

Additional data for all seedlings are presented in

Table 5. Hardy geranium seedlings and breeding information are sorted in ascending order by cJaccard. The germination rate (%) is calculated by dividing the number of seedlings by the number of seeds. True hybrid: ♂ unique markers (%) ≥ 25%; partial hybrid: 10% ≤ ♂ unique markers (%) < 25%; non-hybrid: 10% > ♂ unique markers (%).

♀ × ♂	cJaccard	No. of Crosses	No. of Seeds	No. of Seedlings	Germination Rate (%)	True Hybrid	Partial Hybrid	Non-Hybrid
G49 × G35	0.486	22	40	10	25	10	0	0
G50 × G49	0.512	18	7	1	14	1	0	0
G73 × G35	0.528	2	4	1	25	1	0	0
G21 × G75	0.563	10	1	1	100	0	1	0
G14 × G39	0.631	37	38	5	13	5	0	0
G04 × G39	0.641	29	35	4	11	4	0	0
G04 × G09	0.686	33	26	20	77	18	0	2
G09 × G04	0.686	31	10	4	40	2	0	2
G80 × G69	0.689	48	11	2	18	2	0	0
G04 × G61	0.690	4	3	2	67	2	0	0
G09 × G39	0.721	16	33	2	6	2	0	0
G39 × G09	0.721	19	18	4	22	4	0	0
G49 × G62	0.767	5	2	1	50	0	0	1
G39 × G62	0.778	1	2	2	100	0	0	2
G05 × G39	0.795	43	54	1	2	0	1	0
G38 × G62	0.798	2	3	3	100	0	3	0
G35 × G75	0.852	1	3	3	100	3	0	0
G45 × G69	0.866	5	5	3	60	0	0	3
G50 × G21	0.871	5	1	1	100	0	0	1
G45 × G06	0.881	26	5	3	60	0	0	3
G54 × G44	0.918	35	21	9	43	0	0	9

, focusing on the genetic distances between the parental combinations. The cJaccard (a measure ranging from 0 to 1) is provided for each cross combination and is sorted in ascending order. The data suggest that most true hybrids originate from the hybridization of two parents with a cJaccard value lower than 0.800. When the cJaccard value is more than 0.800, it becomes challenging to obtain true hybrids. However, there is an exception, Geranium endressii (G35) × Geranium ‘Bloom Time’ (G75), which has a cJaccard value of 0.852 and resulted in three true hybrids. For the cross between Geranium phaeum ‘Angelina’ (G54) × Geranium macrorrhizum ‘White Ness’ (G44) with a high cJaccard value (0.918), nine out of the twenty-one seeds germinated. However, none of the resulting seedlings proved to be true hybrids.

In

Table 6. Comparison of morphological characteristics between newly flowering true Geranium hybrids and their parental genotypes. The table highlights the morphological characteristics of true hybrids that have entered the flowering stage. Flower diameter and style length in mm and flowering period in weeks. ΔE*ab, a measure of color difference in the CIELAB color space, indicates the similarity (close to 0) or dissimilarity between a seedling and each of its parents. All measurements were performed with three replicates.

	Combinations	Flower Diameter (mm)			Style Length (mm)			Number of Flowers			Flowering Period (Weeks)			ΔE*ab (Flower)		ΔE*ab (Leaf)
Code	♀ × ♂	♀	♂	Progeny	♀	♂	Progeny	♀	♂	Progeny	♀	♂	Progeny	♀	♂	♀	♂
D21	G04 × G39	46	56	39	14	15	11.4	245	53	25	6	3	8	27	38	10	13
D25	G04 × G61	46	35	39	14	12	11	245	na *	85	6	3	14	61	22	8	14
D26	G04 × G61	46	35	39	14	12	11	245	na *	365	6	3	14	52	31	3	19
D34	G14 × G39	57	56	45	-	-	-	-	-	-	-	-	-	4	64	18	17
D59	G49 × G35	32	41	27	10	10	7.3	298	128	490	26	14	16	24	37	2	5
D60	G49 × G35	32	41	30	10	10	8.4	298	128	630	26	14	16	19	32	4	8
D61	G49 × G35	32	41	36	10	10	8.5	298	128	818	26	14	16	43	56	2	4
D62	G49 × G35	32	41	27	10	10	8	298	128	465	26	14	16	44	57	3	8
D80	G73 × G35	26	41	30	8	10	8	na	na	195	11	14	10	21	43	33	25

na *—not available.

, a morphological description of the true hybrid Geranium seedlings that already flowered is given. The D21 seedling resulting from the cross Geranium ‘Brookside’ (G04) × Geranium himalayense ‘Derrick Cook’ (G39) produced progeny with an average flower diameter of 39 mm, which was smaller compared to the 46 mm (G04) and 56 mm (G39) diameters of the parents. Similarly, the style length of the F1 seedling was, on average, 11.4 mm shorter than the parents’ styles. Parental plants produced 245 (G04) and 53 (G39) flowers during the flowering season, while the F1 seedling had an average of 25 flowers. Remarkably, the flowering period of the F1 seedling was significantly longer than that of the parents. An evaluation of flower and leaf color using Delta E CIE 1976 showed that the flower color of the D21 F1 seedling is closer to the maternal plant (ΔE*ab = 27) compared to the paternal plant (ΔE*ab = 38). However, the leaf color was almost the same (Table 6). This is further illustrated in Figure 5, where the seedling with an intermediate flower coloration is shown. D25 and D26 seedlings, both resulting from the cross between Geranium ‘Brookside’ (G04) × Geranium pratense ‘Purple Ghost’ (G61), showed identical flower phenotypes. The main differentiating factor was the difference in flower number, with the D26 F1 seedling having more flowers than the D25 seedling. In addition, the flower color of both F1 seedlings is more similar to the male parent (Table 6).

D59, D60, D61, and D62 F1 seedlings all derived from the same combination (Geranium × oxonianum ‘Katherine Adele’ (G49) × Geranium endressii (G35)) and collectively showed different flower sizes compared to their parents. D59 and D62 produce smaller flowers than their parents. The style length of all these F1 seedlings was shorter than that of the parents, but a notable contrast was observed in the number of flowers, which was considerably higher in the new F1 seedlings (Table 6). In addition, the flowering time of these F1 seedlings is intermediate. Based on the portable spectrophotometer data, the flower colors of D59 and D60 F1 seedlings are more similar to the maternal plant than to the paternal plant. These differences are not observed in D61 and D62 F1 seedlings. The presence of veins in the maternal plant may affect the measurement accuracy. However, the differences between the seedlings are evident, as shown in Figure 6. In particular, the D61 F1 seedling shows a distinct intermediate flower appearance, with veins inherited from the female parent and petal color inherited from the male parent (Figure 6).

4. Discussion

Creating new combinations of desirable traits in commercial cultivars is the challenge of breeders. Interspecific hybridization is an effective way of introducing more diversity and combining desirable traits in plants, but it is time-consuming. It has been found that it is possible to predict hardy geranium seed production for each parental combination using cJaccard (AFLP). This assists breeders in selecting optimal parental combinations and estimating the approximate number of crosses, flowers, and plants required during the breeding program [21,23]. In addition, AFLP assists in pre-flowering F1 selection, especially for progenies that may take two or more years to flower and can speed up breeding programs.

Geranium typically begin their flowering cycle around the second half of May and, in some cases, continue to flower until the first frost in October–November. An assessment of male fertility and female receptivity across different genotypes within the collection, combined with cJaccard distance coefficient, is critical to establishing an effective breeding program. This approach assists in the development of accurate crossing schemes and improves the strategic planning of breeding efforts. Pollen tube growth arrest, a common phenomenon in plant breeding as a pre-zygotic crossing barrier, can be caused by several determinants, including genetic incompatibility, self-incompatibility mechanisms, and environmental stress. In the specific context of parental combinations, such as Geranium macrorrhizum ‘White Ness’ (G44) × Geranium ‘Brookside’ (G04) and Geranium ‘Blushing Turtle’ (G80) × Geranium ‘Brookside’ (G04), it is plausible that pollen tube growth arrest is primarily due to genetic incompatibility since these genotypes are very distantly related according to the cJaccard distance coefficient.

After evaluation of male and female receptivity and performing interspecific crosses, seed germination is crucial. The use of embryo rescue techniques has been confirmed as necessary in the hybridization of Pelargonium [16,30,31], a close relative to the hardy geranium. For Pelargonium, the preferred approach is direct embryo culture using hormone-free media [32,33]. However, there is no information on its efficient use in hardy geraniums. In our observation, opposite to the case of Pelargonium [16], regenerating shoots from the callus derived from embryos is not feasible. All calli-formed embryos of hardy geraniums in our research failed to regenerate into plants, except those of Geranium ‘Blushing Turtle’ (G80). This observation emphasizes the need for direct embryo germination in hardy geraniums. Therefore, it is recommended to use a hormone-free medium without additional callus inducer compounds for embryo rescue in hardy geraniums. Consistent with what we found in hardy geraniums, the influence of proline on inducing callus in rice embryos was remarkable [34]. In contrast to our observation with hardy geraniums, Kamlah et al. (2019) were able to successfully regenerate shoots from callus-forming zygotic embryos in Pelargonium [30].

Over the course of three consecutive flowering seasons, we have refined our knowledge and techniques, resulting in an increase in the probability of obtaining healthy seedlings from 0.17% in 2020 to 1.01% in 2022. Over the course of 3 years, we performed almost 2438 interspecific crosses at the end, resulting in 54 confirmed F1 hybrids. Despite these successes, the limitation of plant regeneration from callus in hardy geraniums has challenged the efficacy of embryo rescue in our initiatives. The difficulty in regenerating plants from callus in Geranium is probably attributed to the production of phenolic compounds by the callus. In order to overcome this significant obstacle in the breeding of hardy geraniums, an in-depth exploration of the intricacies of embryo rescue and the plant regeneration process from callus is essential. It is noteworthy that a significant proportion of these parental combinations required embryo rescue for successful development, highlighting the importance of addressing this challenge for the advancement of our breeding efforts.

Cross-compatibility and successful hybridization leading to seed production in hardy geraniums are primarily influenced by the genetic distance between the parental species. The evaluation of resulting seedlings in this article further supports this by showing that progeny from parents with high genetic distance (cJaccard) often do not represent true hybrids. For example, the parent combination Geranium phaeum ‘Angelina’ (G54) × Geranium macrorrhizum ‘White Ness’ (G44), characterized by a high genetic distance, had a germination rate of 43% (defined as the number of seedlings per all seeds), but none of the resulting seedlings proved to be true hybrids. Since Geranium phaeum ‘Angelina’ was emasculated before hybridization and did not produce self-pollinated seeds under controlled conditions (net bags), it is likely that the seeds were apomictic and stimulated by pollination. This is consistent with findings in Helleborus [35], Corymbia [36], and Kalanchoe [37], where greater genetic distance leads to more difficult crossing and lower hybrid recovery compared to situations with smaller genetic distances.

The hardy geranium F1 hybrids showed morphological variations, such as changes in flower diameter, style length, number of flowers, and flowering time compared to their parents. The ΔE*ab value provided a clear indication of color similarity to either parent. For example, D21, an F1 hybrid of Geranium ‘Brookside’ (G04) and Geranium himalayense ‘Derrick Cook’ (G39), had a flower color similar to that of the maternal parent, whereas D25, an F1 hybrid of Geranium ‘Brookside’ (G04) and Geranium pratense ‘Purple Ghost’ (G61), had a flower color more similar to that of the paternal plant. Among the F1 hybrids of Geranium × oxonianum ‘Katherine Adele’ (G49) and Geranium endressii (G35) (D59, D60, D61, D62), the flower color of D60 was closer to the maternal plant than its siblings. In addition, some F1 hybrids inherited petal veins. These veins play an important role in the beauty of ornamental plants.

Hybridization is one of the oldest breeding techniques and plays a central role in the development of modern plant varieties. In ornamental plant breeding in particular, it is the main driver of genetic diversity.

5. Conclusions

We confirm the indispensability of embryo rescue for the generation of novel trait combinations in hardy geraniums, especially in crosses with genetically distant parents. As shown in the obtained F1 hybrids, hybridization notably altered flower diameter, style length, number of flowers and flowering time across generations. To determine the paternal genetic contribution within the newly developed seedlings, we used AFLP analyses and determined 54 out of 82 seedlings as true hybrids (65.8%). This article highlights the critical role of parental genetic distance in determining the success of interspecific hybridization in hardy geraniums. Seedlings with interesting traits are being prepared for customer trials, which will be carried out in the near future. In addition, our results increase the understanding of the efficacy of hybridization in this genus and its potential to produce unique and innovative cultivars that meet the evolving preferences of the ornamental market.