2 Materials and Methods

2.1 Study site and study species

The study was conducted in a 50-ha (1000 × 500 m) forest dynamics plot within the Heishiding (HSD) Nature Reserve, Guangdong Province, southern China. The region has a subtropical moist monsoon climate with an average annual temperature of 19.6°C. The mean annual precipitation is 1743.8 mm, the majority of which falls from April to September. The HSD 50-ha plot was initiated in 2011 and all woody stems with diameter at breast height (DBH, 1.3m) ≥ 1 cm were tagged, identified, measured, and georeferenced following the field protocols of the ForestGEO network (Condit 1998). A second census was conducted from January 2016 to March 2017, and all living stems and new recruits were surveyed as above.
Diospyros morrisiana Hance ex Walp. (Ebenaceae) is a dioecious deciduous shrub or small tree inhabiting slopes, ravines, or streamsides in East Asian monsoon subtropical forests between 100 and 1000 m a.s.l (Wu and Raven 1997). It is a dominant species in early and mid-successional subtropical forests in southern China and can grow up to 20 m in height (Abbas et al. 2021). The flowers are small (about 7mm long), white, and urn-shaped. Male plants have axillary cymose inflorescence with 2–3 flowers, whereas female flowers are solitary. The fruits are fleshy berries, yellow when ripe, with a sweet flavor. The flowering season is from May to June, and fruits are ripe in November. The flowers are bee-pollinated, whereas seeds are dispersed by civets (Paguma larvata and Viverricula indica , Viverridae) (Corlett 1996, 2001).

2.2 Phenotypic sexing and molecular sexing

In May–June and September 2017, all tagged individuals of D. morrisiana in the plot were censused. Flowering individuals were recorded and leaves of all individuals were simultaneously collected for molecular sexing. Phenotypic sexing process in the field was undertaken by visual identification, based on flower structure and pedicels or fruit stalks remaining on the branches. Flowering individuals were revisited in May–June 2018 and their reproductive condition was recorded.
To obtain the sex ratio at the seedling stage, we collected seeds from the canopies of 16 female D. morrisiana in the HSD 50-ha plot in December 2017. After soaking in distilled water for 1 h and then sterilizing with 25% perhydrol for 1 h, a total of 2,202 seeds from 16 female trees were planted in plastic seedling trays (28 cells, 7 × 7 × 14 cm each) filled with peat soil in March 2018. The plants were routinely irrigated with Hoagland solution. After 3 months, we recorded the number of germinated seedlings for each parent tree, collected the leaves of all seedlings for molecular sexing.
We used a molecular marker to determine the sex of each individual tree and seedling as follows. Total genomic DNA of D .morrisiana was extracted from the silica-gel dried leaves using DNA Plantzol (Lifefeng, Hangzhou, China) according to the manufacturer’s protocol with slight modifications. Silica-dried leaves (0.3 g) were ground by a high-through tissue grinder (Scientz-48, Ningbo, China) with magnetic beads and 1% PVP for 400 s, and mixed with 850 μL Plantzol. After incubating at 65°C for 30 min at 600 rpm in MTH-100 Shaker Incubator (Thermo Scientific, Waltham, USA), an equal volume of phenol:chloroform (24:1) was added, and the tubes were turned upside down for 80 times and centrifuged (12,000 rpm, 20 min, 4°C). DNA was precipitated by mixing the aqueous phase with 520 μL isopropanol, turned upside down gently for 100 times and centrifuged (10,000 rpm, 20 min, 4°C). The DNA pellet was washed with 800 μL 75% ethanol, dried and dissolved in an 80 μL TE buffer [10 mM Tris-HCl (pH 8.0), 1mM EDTA]. After isolation, DNA quality was checked by gel electrophoresis and qualified samples were stored in a -20°C refrigerator.
For molecular sexing, we used the primers OGI-candF1(5’-CACAGTAGTCATATATTTTTAGC-3’)/ OGI-spR(5’-CTGGCACACAAAATATTTTCAACCCT-3’) developed by Akagi (Akagi et al. 2014). The PCR reaction mixture contained a total volume of 20 μL including 1 μL template DNA, 1 μL forward and reverse primers (TsingK, Beijing, China), and 10 μL 2 × EasyTaq PCR SuperMix (+dye) (Transgene, Guangzhou, China). Amplifications were performed using the following program: initial denaturation step at 94°C for 3 min; 30 cycles consisting of denaturation step at 94°C for 30 s, primer annealing step at 58°C for 30 s and primer extending step at 72°C for 90 s; and final extending step at 72°C for 7 min. Additionally, we designed and used the primers OGI_500-F (5’-ACATACAACCAAGCGGAACTG–3’) andOGI_500_R (5’-AACAGTGCCACCTTCCTTGA -3’), specific to the homologous regions of the OGI and MeGI genes. Notably, MeGI is an autosomal homolog of OGI and present in both male and female individuals. The reaction mixture and program were the same as described above except for the temperature of the primer annealing step (52°C) and extension time (60 s). The PCR amplification products were examined on a 1.0% agarose gel contaning ethidium bromide. The gels were imaged using a Tangunon 4100 Gel Image Analysis System (Tanon Science49 & Technology Co., Ltd.). The PCR products of OGI-candF1/OGI-spR showed an approximately 1000bp band only for males, while the PCR products ofOGI_500_F/OGI_500_R showed bands for both sexes. Using two primer sets simultaneously limits false negative results and thus improves the credibility of the sex identification. We examined the results of molecular sexing for reproductive individuals, and found it was in full agreement with field observation, which proved the feasibility of our molecular approach.

2.3 Sex ratios and sexual dimorphism in life-history traits

To track the ontogenetic shift in the sex ratio of this population ofD. morrisiana , we chose 5 and 10 cm as cut-off points to divide the tagged individuals of D. morrisiana alive in the second census into three DBH classes, and trees ≤ 2 cm DBH were further divided out because most flowering individuals are > 2 cm DBH (98.7%), resulting in the following five stages: seedlings, saplings (1 < DBH ≤ 2 cm), small trees (2 < DBH ≤ 5 cm), medium trees (5 < DBH ≤ 10 cm), and large trees (DBH > 10 cm).
Following the results of determining the sex of individuals using phenotypic and molecular methods, we calculated the OSR and population sex ratio for the whole population and for each size class. Sex ratios were expressed as the proportion of male plants to all individuals. To infer the sex ratios in population of D. morrisiana , we used Bayes’ theorem to estimate the posterior distribution of sex ratios based on binomial likelihood and a uniform prior. We calculated the mean value, standard deviation, and 95% credible interval to summarize the posterior distribution and determine whether each sex ratio is biased. Calculations were performed with the statistical software R (R Core Team 2020) in the package rethinking (McElreath 2020).
To assess for sex-related differences in flowering, growth, and survival, we compared the flowering proportion, DBH, relative growth rate (RGR, log difference of DBH in two censuses divided by 5 years), and mortality between the two sexes. The flowering proportion was expressed as the number of flowering individuals as a proportion of all individuals of a given sex. For flowering individuals, we compared flowering proportion in 2017 between the sexes. In addition, we used a generalized linear model to evaluate the effects of sex and DBH on whether individuals flowered or not. Further, we compared DBH and RGR for males and females among flowering individuals and molecular-sexed individuals, as well as RGR between flowering and non-flowering individuals using a Kolmogorov–Smirnov test. We also compared RGR of males and females in each size classes and used a generalized linear model to evaluate the effects of sex and DBH on RGR.

2.4 Spatial association between the sexes

To infer processes responsible for the sex ratio pattern of D. morrisiana population from its spatial distribution, we performed spatial point pattern analyses for all tagged individuals. We first tested if there was spatial segregation between male and female trees through a mark connection function. We calculated the mark connection function p(r) , which gives the probability that two trees separated by a distance r are of the same sex or opposite (Illian et al. 2008). It follows that the sum of all mark connection functions equals one. If the sexes are randomly distributed over the trees, the mark connection functions are constants derived from the proportions of male, pM , and female, pF , and pMM(r) = pM ,pFF(r) = pF ,pMF(r) = pFM(r) =pM · pF . To test if the observed pattern deviated from the expected value, we compared the mark connection functions for the observed pattern to a random labeling null model which fixed the locations of all trees and shuffled the sexes. It indicates the existence of spatial segregation of the sexes ifpMF(r) is smaller than the expected value of null model. To detect small-scale spatial structure more precisely, we additionally used a local random labeling null model that switched only sexes of trees located within 30 m, a scale somewhat larger than the typical range of plant–plant interactions that is widely affirmed in tropical forest and our study sites (Harms et al. 2001, Uriarte et al. 2004, Wiegand et al. 2007, Wang et al. 2021a).
To determine whether the observed spatial patterns deviated from the null models, we calculated point-wise simulation envelopes with a significance level α = 0.05 from the 25th highest and 25th lowest values of 999 Monte Carlo simulations of each null model. To avoid the Type I error inflation caused by simultaneous inference, we also constructed a global envelope to jointly evaluate whether observed pattern deviated from the null models significantly (Myllymäki et al. 2017). All spatial point pattern analyses were performed in the softwareProgamita (Wiegand and Moloney 2013).