1 Introduction

Dioecious plants are phylogenetically widespread among angiosperms (Renner 2014) and often play crucial roles in various ecosystems, especially forests (Chazdon et al. 2003, Lin et al. 2020, Wang et al. 2020). Dioecy prevents selfing and potentially improves reproductive efficiency, but also raises a greater risk of population decline driven by an imbalance in the sex ratio (Charlesworth and Charlesworth 1981, Henry et al. 2018), which may be exacerbated by climate change (Eckert et al. 2010, Petry et al. 2016, Hultine et al. 2016). Thus, it is imperative to undertake more thorough investigations into variation in the sex ratios of dioecious plants as well as the underlying mechanisms that drive variation in maturity and other life-history traits that differ between male and female individuals.
The majority of dioecious species are trees with a long life-span (Renner and Ricklefs 1995, Wang et al. 2021b), but studies of the sex ratios of these taxa suffer from several difficulties. One primary challenge is that the sex of immature or non-flowering individuals is difficult to identify through vegetative traits (Lloyd and Webb 1977, Barrett and Hough 2013). Most previous studies have therefore been able to use only flowering individuals (i.e., the operational sex ratio, OSR) in order to estimate the population sex ratio (Barrett and Hough 2013, Field et al. 2013). However, using the OSR could generate unreliable and biased results because the sex ratio of all non-flowering individuals is not necessarily the same as the sex ratio of flowering individuals. This problem is especially acute for long-lived tree species, since their populations often include a high proportion of these immature and non-flowering plants (Nanami et al. 1999, Ueno et al. 2007, Queenborough et al. 2007). To address this issue, sex-specific molecular markers can be used to provide more accurate and reliable information on population sex ratios, because they are not restricted to flowering individuals (Eppley 2001, Stehlik and Barrett 2005, Shelton 2010). These markers have mostly been developed for applied research in agriculture, forestry, horticulture, and medicine, and few studies have applied molecular approaches to provide accurate and comprehensive assessments of the sex ratios of dioecious trees in natural field environments (Ling et al. 2003, Zhang et al. 2016, Liao et al. 2017, Zhou et al. 2018).
Interestingly, sex ratios can change across plant life stages from juveniles to adults, the study of which is critical for addressing when and how sex ratio biases are established (Eppley 2001, Stehlik and Barrett 2005, Stehlik et al. 2007). Because trees are often most vulnerable to competition, herbivory, and pathogen attack at juvenile stages (Moles and Westoby 2004, Comita et al. 2014), the population sex ratio could change dramatically at these stages if one sex is more likely to survive than the other. However, the OSR cannot be used across a number of these significant stages, because the sex of juvenile trees (seeds, seedlings, and saplings) and other non-flowering individuals are not included in the OSR. Several studies have applied the molecular sexing approach to reveal shifts in the ontogenetic sex ratio (Taylor 1994, Eppley 2001, Stehlik et al. 2007, Shelton 2010). For example, Stehlik et.al (2007) found that the perennial herb Rumex nivalis had a slightly female-biased sex ratio using flow cytometry, which shifted to be even more female-biased in later life stages (Stehlik et al. 2007). However, these studies have been limited to herbaceous plants with a relatively short life-span, and variation in the sex ratios across the life stages of trees with a long life-span has not been investigated. Sex allocation theory predicts that the juvenile sex ratio is negatively correlated with the adult sex ratio in species with overlapping generations like trees (West 2009). In species with overlapping generations, individuals not only compete for mates within their own generation but also with individuals from the previous generation. For example, if females are less available among adult individuals, female offspring will face less competition for mates compared to males, resulting in an increased reproductive value for females. Consequently, when the sex ratio of adult individuals is biased towards males, a primary sex ratio biased towards females would be preferred (Werren and Charnov 1978, López and Domínguez 2003, Booksmythe et al. 2018). But, to our knowledge, this hypothesis has not obtained any empirical support from tree species.
Understanding the mechanisms that drive variation in sex ratio is critical, and can be broadly grouped into two categories: genetic and ecological (Stehlik and Barrett 2005, Che-Castaldo et al. 2015). A variety of genetic mechanisms could drive divergence from a 1:1 sex ratio at the seed stage, including segregation distortion, Y chromosome degeneration, and sex-selective abortion (Taylor 1994, Stehlik and Barrett 2005, Stehlik et al. 2007). The early life stages of dioecious trees, such as seedlings, can better reflect such gene-determined sex ratios. Variation in sex ratio at later life stages is mainly influenced by ecological mechanisms involved in sex-related differences in resource allocation trade-offs towards growth, reproduction, and defense (Queenborough et al. 2007, Timerman and Barrett 2019). For example, females generally allocate more energy to reproduction, resulting in later maturity, lower flowering frequency, and higher mortality of female individuals and thus a male-biased operational sex ratio (Lloyd and Webb 1977, Delph 1999, Barrett and Hough 2013, Field et al. 2013). Furthermore, if one sex has a different environmental preference compared to another due to different reproductive inputs, a biased sex ratio could occur in heterogeneous environments (Freeman et al. 1976), and may even lead to spatial segregation of the sexes (Nanami et al. 1999). It should be noted that spatial segregation between two sexes would also occur when inter-sexual competition is stronger than intra-sexual competition (Bierzychudek and Eckhart 1988). Multiple studies have compared the growth pattern and spatial distributions between sexes to explore mechanisms influencing the operational sex ratios in dioecious trees based on flowering individuals (Nanami et al. 1999, Queenborough et al. 2007, Jácome-Flores et al. 2016, Timerman and Barrett 2019). However, as stated above, these results might be biased because of the understandable exclusion of non-flowering and immature individuals.
To address this gap in current studies of the sex ratios of dioecious trees, we applied a molecular sexing approach to investigate variation in the sex ratio across life stages and to determine the underlying mechanisms in the dioecious tree Diospyros morrisiana (Ebenaceae) in the Heishiding 50-ha forest dynamics plots located in southern China. Specifically, we identified the sex of all 2,255 tagged individuals in the plot and 349 germinated seedlings through sex-specific molecular markers. Combined with analyses of growth dynamics and the spatial patterns of tagged individuals, we address the following questions:
(1) Does the field-observed OSR accurately reflect the overall population sex ratio of D. morrisiana , and how does the sex ratio change across life stages? We hypothesize that OSR differs significantly from the overall population sex ratio. Additionally, we expect considerable changes in the sex ratio of D. morrisiana across life stages. Specifically, the adult trees are more likely male-biased, but juvenile trees are more likely female-biased.
(2) What processes are responsible for the shift in the sex ratio ofD. morrisiana in later life stages? We hypothesize that the sex-specific difference in the growth–reproduction trade-off will lead to a slower growth rate and higher mortality in female trees, resulting in a sex ratio that is increasingly male-biased in later life stages; We also tested whether male and female trees are spatially separated and discussed the potential cause of spatial pattern.