Pollen limitation in the endangered Chinese endemic species Sinocalycanthus chinensis

Abstract Pollen limitation negatively impacts endangered and endemic plants with small fragmented populations, such as Sinocalycanthus chinensis, an endangered plant endemic to China. In this study, we analyzed the pollen limitation of the S. chinensis Damingshan (DMS) population in 2006, 2009, and 2010, and crossed plants with mates separated by different distances, both within and between populations. The DMS population exhibited strong pollen limitation in fruit set, seed set, and seeds per fruit in 2006, 2009, and 2010. The average accumulated pollen limitation (for fruit set times seeds per fruit) was 0.510 ± 0.180. Progeny crossed with pollen from intermediate neighboring plants within the same population (separated by 30–50 m from pollen recipients) had the lowest fitness. No optimal outcrossing distance was found within the DMS population. Progeny from crosses with the Shunxiwu (SXW) and Daleishan (DLS) populations performed relatively better, while those from crosses with Qingliangfeng (QLF) and Longxushan (LXS) populations performed worse. Compared with average reproductive success, outbreeding depression was found in progeny from crosses with the LXS and QLF populations. Reproductive success from pure self‐pollination indicated S. chinensis is self‐compatible. Geitonogamous selfing increased reproductive success. Based on geitonogamous selfing, the proportion of selfed offspring was relatively high. These results provide basic references for the conservation of this species.

Pollen limitation may be caused by a decrease in pollen quantity or quality (Aizen & Harder, 2007;Ashman et al., 2004). Pollen quantity limitation occurs when an insufficient quantity of pollen is deposited on the stigma by scarce or ineffective floral visitors (Aloso et al., 2013;Gómez, Abdelaziz, Lorite, Muñoz-Pajares, & Perfectti, 2010). Pollen quality limitation occurs when pollinators deposit incompatible pollen, self-pollen, or pollen from closely related individuals (Aloso et al., 2013;Fernández et al., 2012). The occurrence of pollen quantity limitation and quality limitation vary among species and populations and can be affected by the pollinator assemblage, plant population size and density, and other habitat variables (Fernández et al., 2012). The endangered plant Ottelia acuminate (Hydrocharitaceae) has been shown to suffer severe pollen limitation owing to a low pollinator visiting frequency (Xia et al., 2013). The endangered plant Disanthus cercidifolius Maxim.
var. longipes H.T. Chang (Hamamelidaceae) was verified to be prone to pollen limitation; however, the pollen source rather than the quantity of pollen had significant effects on the reproduction of this species (Xiao, Zeng, Li, Hu, & He, 2006). Understanding the pollen limitation situation and causes is particularly important for the management and conservation of endangered and endemic plants (Geerts & Pauw, 2012).
In plants, the spatial distance between mates may have negative consequences on progeny fitness in two ways. Crosses over short distances, such as occurs in self-pollination or crossing between close relatives, may result in inbreeding depression (Charlesworth & Charlesworth, 1987;Oostermeijer, Altenburg, & den Nijs, 1995). In contrast, crosses over long distances may result in outbreeding depression (Price & Waser, 1979;Waser, 1993;Waser & Price, 1994). For most plants, an optimal intermediate outcrossing distance between two mating plants is associated with an optimal degree of outbreeding (Billingham, Simões, Reusch, & Serrão, 2007;Grindeland, 2008;Price & Waser, 1979;Waser, 1993). However, no effect of pollination distance was found to be associated with the reproductive success of several species, including Hypochaeris radicata (Becker, Reinhold, & Matthies, 2006) and Lychnis flos-cuculi (Hauser & Loeschcke, 1994). However, few studies have investigated the effects of pollination distance on the reproductive success of endangered species.
Sinocalycanthus chinensis Cheng et S. Y. Chang (Calycanthaceae), a tertiary relict species, has been listed as the second most protected plant in China owing to its habitat deterioration and artificial overexploitation (Hu, 2002). Sinocalycanthus chinensis is entomophilous (Li & Jin, 2006), and its seeds are enwrapped by pericarp and dispersed only by gravity, which results in the limited dispersal of S. chinensis offspring (Li & Jin, 2006). However, there is some potential for self-compatibility at the late developmental stage, and controlled pollinations studies have shown S. chinensis to be self-compatible and have a mixed mating system (Li, Jin, & Gu, 2012;Zhao et al., 2011). Accordingly, we can hypothesize that pollen limitation, both pollen quality and quantity limitation, might have critical effects in S. chinensis populations. However, no empirical data support this hypothesis.
The size of the wild population of S. chinensis has contracted, and only a few limitedly distributed populations remain, which have been divided geographically into an eastern group and a western group (Zhang, Chen, Qiu, Li, & Jin, 2001;Zhou & Ye, 2002). Damingshan  Sinocalycanthus chinensis is a deciduous shrub that grows 1-3 m tall and is scattered under evergreen broad-leaved forests or mixed evergreen and deciduous broad-leaved forests around ravines. The distribution spans altitudes from 470 m to 1,200 m (Zhang & Jin, 2009).

| Pollen limitation
As a woody shrub, the age class of S. chinensis can be estimated from is the height of the tallest stem, b is the crown width along the longitudinal axis, and c is the width along the perpendicular axis (Yang, Zhang, Wu, Li, & Zhang, 2006). For S. chinensis in the DMS population, the age class ranged from 0 to 3 (Jin, Li, & Liu, 2012). In May 2006, 19 mature plants in age class III (2 < d ≤ 3) and with similar phenologies that were located in the center of the DMS population and separate from each other by distances of more than 10 m were chosen as pollen recipients. Pollen grains were also collected from 15 pollen donor plants in age class III and with similar phenologies that were located within the same population and separated from each other by a distance of more than 10 m by rubbing a toothpick against newly dehisced anthers. The pollen grains were then mixed in a small plastic vial and stored at 4°C on ice bags in a plastic box, and then the pollen grains were transferred to recipient stigmas within 2 hr (Irwin, 2001 The mean number of seeds per fruit was also measured.

| Optimal outcrossing distance
In  Table 1 by rubbing a toothpick against newly dehisced anthers. The pollen grains were then mixed in a small plastic vial and stored at 4°C on ice bags in a plastic box, and then the pollen grains were transferred immediately to recipient stigmas. According to a previous study by Zhang and Jin (2009), the pollen viability of S. chinensis lasts for 5 days after collection (no significant difference was detected among pollen of different days). The within-population pollen grains were used for hand-pollination within 2 hr, while the between-population pollen grains were used for hand-pollinations within 6 hr, without an observed decrease in the viability of the pollen grains. All hand-pollinated stigmas were saturated with pollen grains. After treatment, bagging was conducted on the flowers to avoid the contamination of other pollens.
To quantify levels of inbreeding depression and outbreeding depression, we calculated relative performance of crosstypes (R P ) for fruit set, seed set, and seeds per fruit (Weisenberger, 2012).

| Optimal outcrossing distance
Fruit set, seed set, seeds per fruit under the xenogamous pollen cross-treatment among populations showed that progeny in treatments crossed with the SXW and DLS populations performed relatively better. Based on these two xenogamous cross-results, outbreeding depression was not observed in treatments crossed with SXW (R Ph > 0, one-sample t test, t = 3.894, p = .030) and DLS (R Ph > 0, one-sample t test, t = 9.051, p = .012) populations, except for with respect to seeds per fruit (Table 3). Moreover, those comparisons also showed that progeny in treatments crossed with QLF and LXS populations performed relatively worse. Based on the results of these two xenogamous crosses, outbreeding depression was identified in the treatment crosses with QLF (R Ph < 0, one-sample t test, t = −4.436, p = .021) and LXS (R Ph < 0, one-sample t test, t = 9.051, p = .012) populations, except with respect to seeds per fruit (Table 3).

| Geitonogamous selfing
Experiments conducted to assess levels of self-compatibility showed that S. chinensis is self-compatible. Pure self-pollination crosses yielded 9.09% fruit set, 2.01% seed set, and 2.86 ± 0.86 seeds per fruit. In 2009, compared with naturally open-pollinated fruit set, seed set, and seeds per fruit, the geitonogamous pollination treatment yielded increases of 68.96%, 150.05%, and 47.91%, respectively; based on the geitonogamous pollination selfing treatment, the proportion of selfed offspring (s) according to fruit set, seed set, TA B L E 3 Fruit set, seed set, and seeds per fruit under the nine pollen supplementation treatments and the relative performance of crosstypes (R P ) for fruit set, seed set, and seeds per fruit, and fruit set times seeds per fruit in Sinocalycanthus chinensis and seeds per fruit were 0.4679, 0.6229, and 0.4557, respectively.
In 2010, supplementary treatment with autogamous pollen grains showed that the fruit set and seed set were 0.5690 and 0.7960, respectively.

| D ISCUSS I ON
In this study, we found the DMS population of the endangered S. chinensis exhibited strong pollen limitation, and the pollen supplementation, based on fruit set times seeds per fruit, increased reproductive output by 0.510 ± 0.180 seeds in 2006, 2009, and 2010.
Plant reproductive success often depends on pollination, mating systems, and population habitats, among other factors (Aizen & Harder, 2007;Rymer, Whelan, Ayre, Weston, & Russell, 2005). The reproductive success of S. chinensis might be affected by the floral structure of S. chinensis. Flowers of S. chinensis are protogynous with female organs maturing earlier than male organs within the same flower (Zhang & Jin, 2009). However, there is overlap between male and female reproductive stages (Zhang & Jin, 2009;Zhao et al., 2011). Thus, autogamous selfing is possible when pollen from matured stamens falls onto matured stigmas. In this study, we found the fruit set of pure selfing to be 9.09%, indicating the self-compatibility of S. chinensis. In general, self-compatible plants can be considered facultatively autogamous (Aguilar, Ashworth, Galetto, & Aizen, 2006). Although self-compatible species usually require animal pollinators to transport pollen from other conspecific individuals, either selfing (autogamous or geitonogamous crosses) or outcrossing (xenogamous crosses) can yield seeds (Aguilar et al., 2006). In addition, S. chinensis is entomophilous and pollinated by small insects with limited or no flight, which leads to flowers being visited mainly within the same individual plant by any one pollinator (Zhang & Jin, 2008). Protogynous flower traits can decrease the rate of selfing (Reusch, 2001), which might increase geitonogamous selfing versus autogamous selfing. The limited pollinators in the DMS populations (Zhang & Jin, 2008) and protogyny of flowers are the main factors underlying the strong pollen limitation in S. chinensis and why the reproductive success of geitonogamous selfing was high. Similar results have been observed in the endangered protandrous self-compatible species Dracocephalum austriacum (Castro et al., 2015).
In this study, we found that pollen limitation of the S. chinensis  (Table 1), which might have contributed to the natural high pollination success growing in years with lower mean precipitation totals and higher mean temperatures. Fernández et al. (2012) suggested that plants growing in areas with low rainfall are unable to respond to supplementary pollination by increasing their seed number. Haig and Westoby (1988) found that in populations with low rainfall, seed production is likely to be limited by water availability or by a combination of water availability and pollen. In addition, the high reproductive success in the control in 2006 could also be owing to more pollen naturally reaching S. chinensis stigmas in a year with low mean precipitation. Significant negative correlations between pollen influxes and relative humidity and vapor pressure in summer were found in majority arboreal pollen (Li, 2013). However, the reasons for the lowest reproductive success of the control and the highest pollen limitation occurring in 2009, given the intermediate annual mean precipitation and temperature, remain unknown. Forsyth (2003) found that the reproductive success (percent seed set) was significantly correlated with the number of plants flowering annually, which varied greatly among years. In addition, they found that plants flowering asynchronously were pollen-limited, whereas plants flowering synchronously were not (Forsyth, 2003). Further research should focus on the relationship between pollen limitation and the number of flowers and flowering synchrony in S. chinensis.
Determination of the optimal outcrossing distance provides support for the management of this endangered species. Waser and Price (1994) describe a demographic study of F1 progeny resulting from hand-pollinations of Delphinium nelsonii plants separated by a range of crossing distances and found that inbred and outbred progeny, resulting, respectively, from short-and long-distance crosses, performed more poorly than progeny from crosses over intermediate distances, both in terms of survival and eventual reproduction by flowering. The intermediate optimal outcrossing distances were also found in the monocarpic angiosperm Ipomopsis aggregata, with outcrossing distances of 10 m producing more offspring with higher fitness than those separated by outcrossing distances of 1 m or greater than 30 m (Waser, Price, & Shaw, 2000). However, in this study, we did not find such results at intermediate outcrossing distances in the three crossed treatment within the same populations. Similarly, several studies have found no evidence of an optimal outcrossing distance in Sabatia angularis (Dudash, 1990), Yucca whipplei subsp.
Third, inbreeding depression may be cryptic and difficult to assess in species with a long history of selfing (Weller, Sakai, Thai, Tom, & Rankin, 2005). Sinocalycanthus chinensis is self-compatible, with a mixed mating system. The long history of selfing of S. chinensis might weaken the effect of outcrossing distances on the reproductive success of this species.
Outbreeding depression is the population-level counterpart to mechanisms separating species or subspecies (Schierup & Christiansen, 1996). Progeny from crosses with the DLS population performed relatively better, and outbreeding depression was absent in treatments crossed with the DLS population. Zhao, Zhou, Liu, and Bao (2014) also found that outbreeding with the DLS population was apparently more dominant than that with the other treatments.
This ecological mechanism assumes that subpopulations are differentiated by adaptation to different environments and that crosses between sites are then expected to yield maladapted offspring, resulting in outbreeding depression (Schierup & Christiansen, 1996).
There is high genetic differentiation between the DLS and DMS populations . The heterosis produced by the increased interpopulational gene flow between the DMS and DLS populations might diminish any outcrossing depression. However, in this study, progeny in treatment crosses with QLF and LXS populations per- Nevertheless, there were many factors potentially affecting the results, such as the unknown level of pollen manipulation and the possible quality differences between supplementation and control treatments (Ashman et al., 2004), and thus, caution should be applied in interpreting these results. In addition, to more accurately assess the potential optimal outcrossing distance in S. chinensis, future research should increase the number of experimental populations separated by different distances, the number of populations in total, and the range of distances within individual populations.

ACK N OWLED G M ENT
This work was supported by the National Natural Science Foundation, China (No. 30870392).

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data were deposited in Dryad (https://doi.org/10.5061/dryad. j6q57 3n9t).