植物交配系统理论研究大都表明完全自交和异交才是进化稳定的，然而大部分植物在长期的进化过程中，形成了折中的混合交配系统。二型闭花受精植物，通过异花授粉的开放花和专性自花授粉的闭锁花这两种花型来完成有性繁殖，二型闭花受精植物的广泛存在表明混合交配系统很有可能也是进化稳定的。揭示二型闭花受精植物闭锁花和开放花的维持机制能帮助我们更好地理解二型闭花受精混合交配系统的维持以及植物交配系统的进化。目前对该系统的研究主要集中在开放花和闭锁花的相对频率等性状的调查，对闭锁花维持的传粉限制假说等缺乏直接的实验验证。本文以堇菜科堇菜属多年生二型闭花受精植物紫花地丁(Viola philippica Cav.)为研究对象，对多个地理种群的开放花的花部性状、传粉者种类组成及访花行为等传粉生物学特性进行观测，结合人工受控放蜂实验以及开放花人工异体和自体授粉实验，度量相应的雌性繁殖成功指标，衡量不同地理种群的开放花经历的波动传粉环境以及波动的传粉环境对雌性繁殖成功的影响，揭示开放花可能存在的传粉限制，从而探讨二型闭花受精植物闭锁花维持的波动传粉假说。

研究发现，紫花地丁不同地理种群间以及处于同一地理区域的不同局域种群间的开放花数量和花大小均存在显著差异；不同地理种群传粉者类群变化大，北京种群的传粉者种类包括紫壁蜂(Osmia jacoti)等6种，辽宁种群传粉者包括紫壁蜂、凹唇壁蜂(O. excavata)、黑纹食蚜蝇(Episyrphus balteatus)3种；而四川种群传粉者为2种蜂类和1种蝇类，安徽种群仅观察到1种传粉者；传粉者的访问频率也变化显著；不同种群的结实率和结籽率等雌性繁殖成功指标均差异显著。与此相对，闭锁花的结实率和结籽率均维持在较高的水平。这可能表明，紫花地丁开放花的花大小和花展示的变异，引起了传粉者类群和访问频率的波动，并进而导致开放花雌性繁殖成功的波动。

人工受控放蜂实验，比较了壁蜂、蜜蜂、熊蜂三种堇菜属常见传粉昆虫的传粉效率；发现，不同类型的传粉者有不同的访花行为；壁蜂的单花访问时间最长，蜜蜂次之，熊蜂最短；三种传粉者访花处理的结实率显著不同，壁蜂传粉效率最高，5只壁蜂访花60分钟结实率为0.162 ± 0.281，蜜蜂访问组结实率为0.094 ± 0.208，而熊蜂访问组结实率仅为0.033 ± 0.183。

人工授粉实验表明，在花粉数量和质量能够保证的情况下，开放花具有较高的结实率和结籽率。开放花异体授粉、开放花自体授粉、闭锁花闭花授粉，三种处理，闭锁花的受精率和结籽率均保持在较高水平，能提供较强的繁殖保障；开放花自交亲和，异体授粉和自体授粉在结实率和结籽率上差异不显著，但开放花异体授粉形成的成熟种子最大，闭花授粉形成的种子最小；这可能暗示着开放花异体授粉在子代萌发和早期生长等方面有适合度优势。

Most theoretical studies on mating system show that complete selfing and outcrossing are evolutionarily stable. However, most plants have exhibited a mixed mating system. Among them, the dimorphic cleistogamous plants can reproduce sexually by producing potential outcrossed chasmogamous flowers and obligately self-fertilized cleistogamous flowers. The widespread existence of this kind of plants proves that the mixed mating system is likely to be evolutionarily stable. Explaining why cleistogamous and chasmogamous flowers occurs simultaneously within a single plant in the dimorphic cleistogamous plants can help us understand the evolution of plant mixed mating system better. Previous studies mainly focused on the relative frequency of these two types of flowers, while the pollination limitation hypothesis of the maintenance of cleistogamous flowers has not been evidenced directly.

This study investigated the geographic variations in flower size and number, pollinators composition and behavior, and female reproductive success in 4 regions for the dimorphic cleistogamous plant Viola philippica Cav., the effects of pollinators on fruit set has been evaluated in a manipulated pollinator behavior experiment and hand pollination has been conducted to compare the reproductive success among cleistogamous and chasmogamous flowers when pollen quality and quantity were not limited. The aims were to demonstrate the fluctuated pollination environment of the dimorphic cleistogamous plants and their effects on female reproductive success to test the pollination limitation hypothesis for the cleistogamous flowers.

The results showed that there were significant differences in the number and size of chasmogamous flowers among geographical populations and local populations in the same geographical area. Different pollinators occurred in different geographical populations. The pollinators of Beijing population include 6 species such as Osmia jacoti, and the pollinators of Liaoning populations include 3 species such as O. jacoti, O. excavata and Episyrphus balteatus; the pollinators in Sichuan population include 2 bee species and 1 fly species, and only 1 species of pollinators was observed in Anhui population. The pollinators visiting frequency also varied significantly, consequently, the female reproductive success, as illustrated by fruit and seed set, varied significantly among populations. In contrast, the fruit and seed set of cleistogamous flowers were high across all the populations.

The pollination efficiency of three common pollinators of Viola was compared in the manipulated pollinators release experiment. It was found that pollinators of different types had different visiting behaviors. The visiting time on single flower of O. cornifrons was the longest, followed by Apis cerana and Bombus sp. The fruit set was 0.162 ± 0.281 for 5 mason bees visiting for 60 minutes, 0.094 ± 0.208 for honeybees, and 0.033 ± 0.183 for bumblebees.

Hand pollination showed that the chasmogamous flowers exhibited high fruit and seed set when pollen quantity and quality were guaranteed. The seed set of cleistogamous flowers were kept at a high level, which could provide strong reproductive assurance. The seeds were larger after outcross-pollination than those after self-pollination in chasmogamous flowers, and the seeds from the chasmogamous flowers were larger than those from cleistogamous flowers. The results may suggest that the offspring from chasmogamous flowers had the advantages in seed germination and seedling growth.

In conclusion, the results showed that there were significant variations in flower size and floral display, pollinators composition and visiting behavior among different geographical population. The manipulated pollinators release experiments showed that different pollinators had different pollination efficiency. Taken together, these results indicated that the V. philippica Cav. survived in fluctuated pollination environments. As a result, the female reproductive success of chasmogamous flowers varied dramatically across the geographic populations. When pollination was not limited, the chasmogamous flowers could set larger seeds, which suggested potential advantages of chasmogamous flowers in genetic recombination. So, this study provided evidence for the pollination limitation hypothesis for cleistogamous flowers.