Seasonal and Simultaneous Cleistogamy in Rostrate Violets (Viola, subsect. Rostratae, Violaceae)

The special mixed reproductive system, i.e., the ability of an individual plant to develop both open, chasmogamous (CH) flowers adapted to cross-pollination and closed, cleistogamous (CL) flowers with obligate self-pollinating, is a common phenomenon in Viola L. In most sections of Northern Hemisphere violets, cleistogamy is seasonal, and CH and CL flowers develop sequentially in the season. Non-seasonal cleistogamy (simultaneous) is a rare phenomenon in rostrate violets. In the current study, we focused on modification of the CH/CL mating system in V. caspia by environmental conditions, resulting in a gradual switch from temporal cleistogamy, occurring in nature, to simultaneous cleistogamy under greenhouse conditions. V. reichenbachiana with seasonal cleistogamy was the control for V. caspia with the labile seasonal/simultaneous cleistogamy system. In simultaneous cleistogamy, the CH and CL flowers, fruits and seeds developed on an individual plant at the same time on the same branch. The typical difference between CH and CL flowers’ pistils is a straight style ending with a head-like stigma in CH and a curved style in CL adapted to self-pollination. This trait persists in the fruit and seed stages, allowing for easy recognition of fruit of CL and CH flowers in simultaneous cleistogamy. Floral meristems of CH flowers of V. reichenbachiana developed on the rhizome at the end of the growing season under short-day conditions and remained dormant until the following season. The CL floral meristems formed under long-day conditions on elongating lateral branches in the upper leaf axils. The daily temperature influenced the variable CH/CL ratio of V. caspia in nature and greenhouse conditions. Regulation of the CL/CH flower ratio by modifying environmental factors is important for basic research on genetic/epigenetic regulation of cleistogamy and for practical use to produce genetically stable lines of economically important species via CL seeds.


Introduction
Cleistogamy is a special sexual breeding system in which permanently closed, selfpollinated flowers are formed. This phenomenon has intrigued botanists since the 19th century. The term was introduced by Kuhn [1], who observed bud-like flowers that never opened but produced fruits and named them as cleistogamous (CL) flowers. Later, Darwin [2] added that in a CL species, the CL flowers may be the only flower type produced by a plant or may accompany open, insect-pollinated chasmogamous (CH) flowers developing on the same plant. The transition between CH and CL flowers could be gradual, and intermediate flowers with a reduced corolla size and curved pistil style that are facultatively autogamous, termed 'semi-cleistogamous' (SEMCL), are produced by numerous Viola species [3,4]. This special mixed breeding system was found in a variety of plant taxa. It is widespread in angiosperms, occurring in ca. 50 families of monocots and dicots, and may have evolved 34 to 41 times [3][4][5]. Cleistogamy might be beneficial in unfavorable environmental and resource-poor conditions where the abundance of pollinators is drastically reduced, and self-pollinated CL flowers assure reproductive success [5][6][7][8][9][10]. This breeding system is plastic, and the frequency of CL and CH flowers on an individual plant and the timing of flowering may vary depending on environmental conditions as adaptation to heterogenous habitats [11].
The interest in cleistogamy research is mostly focused on the morphology of CH and CL flowers, factors inducing CL flowering and the genetic and molecular basis of the development of both flower types. Different heterochronic processes are involved in the formation of small, unopened CL flowers in Viola odorata (sect. Viola), which appear to end development prematurely relative to CH [12][13][14]. The CH flowers of V. odorata develop in response to short days and the CL flowers in response to long days [15 and references cited therein]. Induction of CL flower development is important in crops (rye, barley, soya) and medicinal and ornamental plants to produce genetically uniform selected lines through self-pollination. The physiology, genetic/epigenetic regulation and molecular mechanisms of cleistogamy still need extensive investigation, although several papers have been published over the last decades [15][16][17][18][19][20][21][22][23].
Cleistogamy is relatively common in Viola, the largest genus in the family Violaceae, comprising ca. 600 species distributed mainly in the Northern Hemisphere [24,25], which has been used as a good model for studying cleistogamy [5,11,19,21,22,[26][27][28]. Most North Hemisphere Viola sections are characterized by seasonal cleistogamy in which CH and CL flowers do not develop simultaneously on an individual plant, but temporal production of CH and CL flowers is strongly influenced by environmental factors such as light quantity, canopy cover, photoperiod, temperature, soil pH and moisture [11].
Both species studied herein, V. reichenbachiana Jordan ex Bor. and V. caspia (Rupr.) Freyn, are caulescent perennials belonging to the palaeotetraploid sect. Viola subsect. Rostratae (Kupffer) W. Becker [29,30]. Viola reichenbachiana is distributed in western Eurasia (from central Spain and the British Isles east to Estonia and Greece, northern Turkey, the Caucasus and northwestern Iran) [31]. Viola caspia, an important ornamental and medicinal plant in northern and northwestern Iranian folklore, is distributed from Turkey in Europe and Crimea eastwards to northern Iran [30,32,33]. The two species are sympatric in northern Iran and in other regions; they have an overlapping ecology and share one genome (2n = 20). In this study, V. reichenbachiana with seasonal cleistogamy was included as the biological background of the flowering process in Rostratae violets vs. V. caspia with the labile seasonal/simultaneous cleistogamy system.
The main focus of the research was to test the influence of variable experimental conditions on the flowering timing of the CH/CL mating system in V. caspia with the implementation of additional research on constant floral traits allowing to distinguish CH and CL flowers and fruits at different stages of development in non-temporal cleistogamy.

The Location of CH and CL Flower Meristems in Viola reichenbachiana
At the end of the growing season (starting from November/December) under shortday conditions, the meristems/buds of CH flowers were observed on the rhizome (Figure 1a-c). Dormant buds survived the winter and developed into CH flowers the following season (Figure 1d). phloem and adaxial xylem were separated by parenchyma tissue.
The CL floral meristems were formed on elongated lateral branches in the upper leaf axils under long-day conditions in the angles between the lateral stem, the petiole and the peduncle of CH flowers (Figure 1h-m).   Meristems of CL flowers developed at CH flowering. To determine the location of CL meristems, the stem, petiole and CH peduncle anatomy was analyzed. All organs had characteristic wings; the vascular system was organized as a collateral bundle. In petioles, abaxial xylem and adaxial phloem formed an arched bundle (Figure 1e). In a peduncle, four vascular bundles (Figure 1f), and in stem, several bundles ( Figure 1g) with abaxial phloem and adaxial xylem were separated by parenchyma tissue.
The CL floral meristems were formed on elongated lateral branches in the upper leaf axils under long-day conditions in the angles between the lateral stem, the petiole and the peduncle of CH flowers (Figure 1h-m).

Differences in the Floral and Fruit Morphology between CH and CL Flowers
Deep violet CH flowers with a spur longer than the sepal appendages of V. reichenbachiana (Figure 2a After pollination, the enlarged ovary developed into a fruit (capsule) filled with seeds. At this stage, the erect style of the pistil ending with the stigma was still visible as a characteristic feature, allowing to distinguish the CH flower from the CL one with a curved style (Figure 3e

Greenhouse Conditions Switched Seasonal Cleistogamy to Simultaneous in Viola caspia
Viola reichenbachiana seasonally developed CH and CL flowers.  Table S1). After March, the number of CH flowers decreased and CL flowers increased, with the highest CH/CL ratio (18) in March and the (e-g) CL bud-like flower (e), enlarged ovary, curved style of pistil (arrow), pollen grains on stigma and ovary (arrowheads) (f), five stamens with reduced anthers (arrows) ending orange appendages (arrowheads) (g), enlarged ovaries (h,i); note differences in style of pistil shape, short and curved in CL (arrowhead in (i)), straight in CH (arrow in (i)). (j) Enlarged ovules in CL fruit, (k) seeds in CL fruit.  Table S1). After March, the number of CH flowers decreased and CL flowers increased, with the highest CH/CL ratio (18) in March and the lowest in May (0.4) (Table S1). These two habitats differed significantly in terms of mean daily temperature but much less so in the case of humidity (Table S3). The changing temperature throughout the year in the LCIV conditions (mean daily temp. ranged from 0.42 to 25.83 ˚C; Table S3) had an impact on the CH flower production, and the negative correlation was confirmed (r = −0.62466, 0.02 < p < 0.05). Such a correlation was not observed in the greenhouse conditions. In turn, in this temperature-stable conditions (mean daily temp. ranged from 20.4 to 24.4 ˚C; Table S3), the number of CL flowers was positively correlated with the number of CH flowers (r = 0.8883, p ≤ 0.001).

Discussion
Changed environmental variables in this experimental study, modified the mixed CL/CH reproductive system. The greenhouse conditions, differing from natural conditions, led not only to a change in the flowering time of CH and CL flowers and the CH/CL ratio in V. caspia, but also to a gradual switch from seasonal cleistogamy (occurring in nature) to simultaneous CH and CL flower production. There were conspicuous differences in temperature between the Live Collection of Iranian Violets (LCIV) and greenhouse conditions. In LCIV, the daily temperature varied throughout the year (0.42 ˚C in March, 25.83 ˚C in September); in the greenhouse, it was more stable and did not drop below 20 ˚C. The daily temperature influenced the variable CH/CL ratio of V. caspia in LCIV and greenhouse conditions. The photoperiod, one of the main factors inducing CH/CL flowering in nature, was the same in the LCIV and greenhouse conditions (artificial light was not supplied in the greenhouse), thus not influencing the differences in the CH/CL ratio in V. caspia.
Our results add to the research on the impact of environmental variables on the development of the two flower morphs in cleistogamous species. Following the studies of Viola pubescens which clearly indicated that, in nature, seasonally variable environmental factors can drastically modify the mixed breeding system of this species [11], we showed that the experimental manipulation of environmental variables could preferentially induce each bud type. These results open the possibility of experimental modification of the In greenhouse conditions, CH and CL flowering duration was extended. CH flowering lasted six months (January-June) and CL flowering lasted twelve months (January-December). Abundant CL and CH flowering was observed in March ( Figure 5). Both flower morphs developed on a single plant simultaneously during the period from January until June. After this period, CH flowering stopped, whereas CL flower production continued until December. During the whole season, the CH/CL ratio was evidently lower in greenhouse conditions than in the Live Collection of Iranian Violets, with the highest in January (1.77) and the lowest in June (0.28) (Table S2).
These two habitats differed significantly in terms of mean daily temperature but much less so in the case of humidity (Table S3). The changing temperature throughout the year in the LCIV conditions (mean daily temp. ranged from 0.42 to 25.83 • C; Table S3) had an impact on the CH flower production, and the negative correlation was confirmed (r = −0.62466, 0.02 < p < 0.05). Such a correlation was not observed in the greenhouse conditions. In turn, in this temperature-stable conditions (mean daily temp. ranged from 20.4 to 24.4 • C; Table S3), the number of CL flowers was positively correlated with the number of CH flowers (r = 0.8883, p ≤ 0.001).

Discussion
Changed environmental variables in this experimental study, modified the mixed CL/CH reproductive system. The greenhouse conditions, differing from natural conditions, led not only to a change in the flowering time of CH and CL flowers and the CH/CL ratio in V. caspia, but also to a gradual switch from seasonal cleistogamy (occurring in nature) to simultaneous CH and CL flower production. There were conspicuous differences in temperature between the Live Collection of Iranian Violets (LCIV) and greenhouse conditions. In LCIV, the daily temperature varied throughout the year (0.42 • C in March, 25.83 • C in September); in the greenhouse, it was more stable and did not drop below 20 • C. The daily temperature influenced the variable CH/CL ratio of V. caspia in LCIV and greenhouse conditions. The photoperiod, one of the main factors inducing CH/CL flowering in nature, was the same in the LCIV and greenhouse conditions (artificial light was not supplied in the greenhouse), thus not influencing the differences in the CH/CL ratio in V. caspia.
Our results add to the research on the impact of environmental variables on the development of the two flower morphs in cleistogamous species. Following the studies of Viola pubescens which clearly indicated that, in nature, seasonally variable environmental factors can drastically modify the mixed breeding system of this species [11], we showed that the experimental manipulation of environmental variables could preferentially induce each bud type. These results open the possibility of experimental modification of the environmental factors influencing the temporal 'gap' between CH and CL flowers in species with seasonal cleistogamy, such as V. caspia, and indicate that the developmental program of cleistogamous species in nature might not be absolute/directional. The results from the greenhouse conditions clearly show that the daily temperature significantly affected the CH/CL flower ratio, with a clear predominance of CL flowers. This allows for long-term predictions of how global climate change (warming) will affect the CH/CL ratio of species with this reproductive system, which would have an impact on the reduction in population genetic differentiation.
Our results documented constant differences between CH and CL flowers at different stages of their development and at fruiting, allowing for the easy recognition of fruits derived from CH or CL flowers. CH flowers attract insect pollinators and promote crosspollination and potential genetic diversity within populations [34][35][36][37][38][39][40]. In contrast, obligate self-pollinated CL flowers produce pure lines. For breeders, selecting lines with important characteristics, especially in crops and medicinal and ornamental plants, is crucial. Viola caspia is a very variable species in terms of both extreme phenotypic plasticity and habitat ecology (steppe to creek sides to deep beech forests) [41,42]. Its oil of high medicinal value, practically derived from soaking and preserving violet petals in olive or sesame oil, but is not an original seed oil [43]. The most important is pure oil extracted from seeds. For this purpose, the genetically stable seeds of CL flowers are more suitable than CH seeds. The production of high amounts of CL seeds in a growing season allows for the extraction of pure oil from the seeds.
Distinguishing CH and CL fruits of violets with seasonal cleistogamy in the field is not difficult because both flower types and fruits are temporally separated in the flowering season as in V. reichenbachiana and V. caspia in nature. In simultaneous cleistogamy, induced in V. caspia under greenhouse-controlled conditions, the flowering and fruiting of CH and CL occurred at the same time (on the same plant, on one branch), and the distinction between CH and CL fruits was based on a constant pistil trait (curved style of CL pistil vs. straight in CH) which persists until fruiting. The traits of the CH and CL generative organs of V. reichenbachiana and V. caspia are not variable and not influenced by environmental conditions (temperate climatic zone vs. subtropical), similar to the stem and peduncle anatomy.
Modification of CH and CL bud fate by manipulation of environmental variables in experimental conditions opens a new area for further deep insight into the genetic regulation of chasmogamous/cleistogamous systems and could enable the control of outcrossing and self-fertilization in cleistogamous species.

Study Species
Specimens of V. reichenbachiana, early dog violet, were grown in an artificial forest in Cracow (Poland; 50 • 04 42.8 N 19 • 59 08.9 E). This species with seasonal cleistogamy served as a reference against V. caspia with the labile seasonal/simultaneous cleistogamy system.
Plants were observed over several seasons (2014)(2015)(2016)(2017), and plant material was collected for determination of the location of floral meristems and flower structure.
The number of V. reichenbachiana specimens used in this study depended on the analysis: for CH flower meristem/buds location,~10 plants were dug up, and the same plant number was used for CL meristem/buds location. For CH and CL flower morphology and microstructure, 4-5 flowers were collected from different plants, and 3-4 petioles or stem fragments were analyzed for organ anatomy.
Forty specimens of V. caspia were randomly collected in their natural habitat in the Gilan-Saravan forest, located 13  The localization of CH and CL floral meristems/buds in V. caspia has not been studied or documented because it is the same as that for V. reichenbachiana. Both species belong to the subsect. Rostratae of Viola section, in which floral features are conservative.

Morphology of CH and CL Flowers
All stages of CH and CL flowers of both species were analyzed during the season, and the differences in pistil shape, style and stamens between CH and CL flowers were documented with a stereoscopic microscope (Nikon SMZ1, Tokyo, Japan). Photographs were taken with an Olympus microscope (Olympus CX31, Tokyo, Japan) under 40× and 400× magnifications.

Anatomy of Petiole, Stem, Peduncle of CH and CL Flower Elements
The anatomy of V. reichenbachiana stem, peduncle, petiole and CL flowers was analyzed on acetocarmine-stained handmade cross-sections as well as on microtome-made paraffin cross-sections stained with Heidenhain's hematoxylin combined with Alcian blue according to the procedure described in [39].
Viola caspia stems and peduncles were fixed in FAA (10 mL  Twenty plants were cultivated in the LCIV and twenty plants were grown in pots (30 × 30 cm) containing garden soil-a mixture of sand, soil and rotten manure in equal proportions-in controlled greenhouse conditions. The photoperiod was supplied naturally according to the conditions of the season. A gradual increase in temperature and in the photoperiod from late winter to early spring as well as gradual cooling in autumn were more evident in nature than in the greenhouse, especially the gradual increase in temperature (Table S3). The data (mean parameters of daily temperature and humidity in the period of 2015-2020 in nature) were obtained in August 2021 from the Meteorological Department of Zanjan Province (http://www.zanjanmet.ir/, accessed on 27 August 2021).
Observation of CH and CL flowering and fruiting was carried out on plants growing in the greenhouse and in the LCIV throughout the season for five years (2015-2020). The number of CH and CL flowers per plant was counted in each month of the season. Every day, the open CH flowers and CL flowers were marked with a red label to avoid counting the same flowers in the following days.
The correlations (r, Pearson) between four parameters (temperature, humidity, number of CH flowers and number of CL flowers) in both conditions (greenhouse and LCIV) were counted in the stats package in R v. 3.6.3 (R Core Team, 2020).

Conclusions
(1) Environmental variables influence the timing of CH and CL flowering and the CH/CL ratio. The controlled greenhouse conditions may allow manipulation of this mixed mating system and even induce simultaneous flowering of CH and CL flowers in species with temporal separation of these two flower morphs in nature.
(2) Determination of the location of floral meristems is important for further molecular research on the differential expression of genes involved in the development of CL and CH flowers.
(3) The differences in fruit morphology between cross-pollinated CH and obligate self-pollinated CL flowers are important in population genetic diversity and in breeding programs of Viola.