4 Discussion

To our knowledge, the present study represents a significant contribution to understanding variation in sex ratios across life stages and the underlying mechanisms in the dioecious trees. Sex-specific molecular markers were applied for the first time to the study of sex ratios in a fully mapped population of a dioecious tree species and revealed a female-biased sex ratio among all tagged and mapped individuals of Diospyros morrisiana in a 50-ha forest plot, which was significantly different from the male-biased flowering sex ratio (or OSR). More importantly, D. morrisiana was distinctly female-biased in early life stages, but gradually shifted to be male-biased in later life stages. We also tested several ecological mechanisms known to shape sex-specific differences in both life-history traits and spatial pattern. We found that the onset of reproduction had a larger impact on the growth of female trees than that of males, which may account for the ontogenetic shift in sex ratio. Moreover, there was no spatial segregation between male and female trees, except for the scale smaller than 2 m.
The OSR of flowering individuals of D. morrisiana in the HSD 50-ha plot was significantly male-biased (Figure 2a), consistent with previous findings in congeners (House 1992, Somanathan and Borges 2000, Venkatasamy et al. 2007). A systematic review also reported that the mean sex ratio based on flowering individuals of 88 dioecious tree species was significantly male-biased (Field et al. 2013). However, in our study we also identified the sex of all tagged individuals in the plot through sex-specific molecular markers and found that the population sex ratio (including all juvenile, immature, and non-flowering trees > 1 cm DBH) was actually female-biased (Figure 2a), confirming for the first time that the OSR does not represent the true overall population sex ratio in dioecious tree population. The disparity between the sex ratios of the OSR and the full population stemmed from sex-specific differences in flowering probability. We found that male trees were more likely to flower than females (Figure 3a, S1), which resulted in the D. morrisianapopulation with a female-biased population sex ratio displaying a male-biased OSR. In particular, saplings had a low flowering probability and were therefore extremely underrepresented in the OSR, but saplings were the most female-biased of all life stages. The gap between OSR and population sex ratio in D. morrisiana supports the results and inferences of previous studies on sex ratios in dioecious trees, showing male-biased flowering sex ratios and the effects of a greater reproductive investment by females in long-lived growth forms such as trees (Lloyd and Webb 1977, Field et al. 2013). Even so, it would be beneficial to apply sex-specific molecular markers more extensively when investigating sex ratios in dioecious plants, especially in trees.
Another important finding was that the sex ratio of D. morrisianashifted from female- to male-biased through later life stages (Figure 2b). The sex ratio in the seedling stage and for large trees was biased in opposite directions, which is consistent with predictions of sex allocation theory for species with overlapping generations (Werren and Charnov 1978, West 2009). Our study provides the first observational evidence supporting this prediction in dioecious trees. The proportion of female individuals gradually declined across life stages, likely resulting in more intense mate competition occurring among males than females. Thus, it could be suggested that the female-biased primary sex ratio might have been favored by selection. Through variation in the sex ratio across life stages, we could postulate how genetic and ecological mechanisms have jointly driven the ontogenetic shift of sex ratio. A distinctly female-biased sex ratio in seedlings may potentially suggest the involvement of genetic mechanisms at the early stage. As the genusDiospyros has an XY sex-determination system, restricted recombination between X and Y chromosome would lead to the accumulation of deleterious genes in Y chromosome, which could potentially affect the survival rate of malesn (Charlesworth and Charlesworth 1981, Ming et al. 2011, Akagi et al. 2014, Pilkington et al. 2019). Multiple studies have found that Y-bearing pollen would be less produced in meiosis or be inferior than X-bearing pollen in certation, while male ovules may also have a lower probability of development and germination (Correns 1922, Błocka-Wandas and Sliwinska 2007, Stehlik et al. 2007).
Different costs of reproduction between male and female trees likely explains much of why the population sex ratio was increasingly male-biased in later life stages (Figure 2). Female Diospyrosproduce large fruit, with a greater investment of carbon, nitrogen, and other nutrients than the male individuals invest in flowers and pollen. This differential cost prohibits many females from maturing at the same small size as males (e.g., Figure 3a, Figure S1), from flowering as frequently as males, and also leads to lower growth rates in larger mature females compared to similar-sized males (Figure S3). Specifically, in flowering individuals, male trees at the medium-tree stage that experienced more reproduction events had a significantly larger RGR than females at the same stage, while there was no difference in RGR between male and female saplings or small trees that seldom flowered (Figure S3). Our results are in line with previous studies of dioecious animal-dispersed trees in which females exhibit a greater reproductive investment (Cipollini and Whigham 1994, Obeso 2002, Queenborough et al. 2007, Barrett and Hough 2013, Field et al. 2013). Among flowering individuals, male trees had a higher probability of reflowering, so it was quite likely that males reproduced more times than females during the five-year census interval. However, the lower flowering frequency of female trees did not offset their higher reproduction costs, which suggested that the difference in the costs of one single reproduction event between males and females would be greater.
We may have underestimated the effects of this difference in the costs of reproduction on the population dynamics in this species. Mature females had lower growth than males, but a higher cost of reproduction in females could also lead to lower survival. Unfortunately, due to the timing of our sampling was too close to the second census, there were insufficient molecular-sexed individuals died during the intervening period. Therefore, we were unable to draw definitive conclusions about potential mortality differences between the sexes. Both lower growth and survival rate of females caused by higher reproductive costs would result in a decreasing proportion of females as size increases, and subsequent plot census will help us decouple the influence of sex-specific growth and survival on the shift in sex ratio. Besides, there are signs that sex-specific mortality unrelated to reproductive investment could be also responsible to the shift of sex ratio (Shelton 2010). We found that the sex ratio of individuals became less female-biased from saplings to small trees (Figure 2b), a period when flowering individuals was scarce and RGR between males and females were not significantly different. This shift suggests that female saplings (i.e., juveniles) may incur higher mortality than males, which was not driven by reproduction cost per se. Further monitoring of the saplings is needed to confirm a sex-specific difference in mortality at this stage.
Sex-based differences in life-history traits could lead to sex-specific habitat preferences, and may be manifested as sexual segregation in spatial pattern (Bierzychudek and Eckhart 1988). However, we did not detect any evidence of sexual spatial segregation among the tagged individuals, except for distances <2 m (Figure 4, Figure S1). Generally, because the scale at which environmental factors operate is always broader than the scale at which individuals interact, this small-scale spatial segregation is less likely to be attributed to sex-specific preferences for microhabitats, but was instead more likely caused by inter-sexual competition (Harms et al. 2001, Valencia et al. 2004, Wiegand and Moloney 2013, Timerman and Barrett 2019). Specifically, for D. morrisiana trees, the crown and underground root system spread far beyond 2 m, which indicates that environmental variation occurring only within 2 m would not lead to a difference in the response of male and female trees. In support of this inference, He et al. (2021) demonstrated that seedlings of D. morrisiana were more competitive when grown with opposite-sex neighbors than with same-sex ones in a greenhouse experiment, and suggested that differences in root exudates between the sexes may mediate the inter-sexual competition (He et al. 2021).