Pollen serves as a robust indicator for discerning spatial and temporal variations in species assemblages and interspecific interactions, rendering it well-suited for high-throughput evaluation of global ecological shifts (Hornick et al., 2021). Pollen analysis remains the predominant method for recovering historical data regarding vegetation, climate, and the environment (Prentice et al., 1998; Xu et al., 2007; Zhao et al., 2015; Birks, 2019; Chevalier et al., 2020; Zheng et al., 2021). The majority of investigations in this domain hinge upon the ecological significance of pollen, specifically its representation of vegetation (Davis, 1963; Parsons and Prentice, 1981; Sugita, 1994).

Understanding the complex relationship between pollen and vegetation is a fundamental aspect of pollen analysis. Influenced by various processes, including pollen production, transport, and deposition, this relationship is complex (Xu et al., 2007; Xu et al., 2016). To elucidate the pollen-plant connection and enhance the pollen-vegetation calibration, several approaches have been introduced and applied, including pollen representation value (R value), extended pollen representation value (ERV), relative pollen productivity (RPP), and relevant source area of pollen (RSAP) estimation (Davis, 1963; Andersen, 1970; Prentice and Parsons, 1983; Prentice, 1985; Prentice and Webb, 1986; Jackson and Kearsley, 1998; Sugita, 2007a, b). With the increasing accessibility of data related to pollen R values and RPP for pollen-vegetation calibration (Li et al., 2018; Wieczorek and Herzschuh, 2020; Xia and Ni, 2024), the pollen-vegetation relationship has gained substantial recognition. Chinese palynologists have also been dedicated to investigating this relationship, including R value, RPP estimation, and quantitative aspects of the pollen-plant connection (Yu et al., 2001; Xiao et al., 2011; Zheng et al., 2015; Xu et al., 2016; Li et al., 2017; Cui et al., 2020). Although uncertainties persist, regression correlations tend toward a linear curve in specific regions following calibration with R values and RPP (Xu et al., 2016). However, limited attention has been directed toward whether pollen can adequately reflect the role of a plant within a community. This represents a minor barrier between community ecology rooted in investigative approaches and pollen-based paleoecology (Seddon et al., 2014).

Pollen percentage constitutes the most commonly utilized data type in calculating pollen representation, RPP, and pollen source areas, whereas the type of plant data employed varies across different studies. For example, Andersen (1970) considered plant frequency, tree crown area, or tree basal area within a given plot, whereas subsequent forest-related studies frequently incorporated tree percentage (abundance) (Prentice and Webb, 1986; Jackson and Kearsley, 1998). In the context of steppe vegetation, coverage has been the prevailing metric for studying pollen-vegetation relationships (Xu et al., 2007). This choice is justified because counting individual plants is more challenging than estimating coverage. From a community ecology perspective, both abundance and coverage are crucial, as they provide quantitative insights into individual plants and the spatial extent they occupy (Song, 2016). However, abundance and/or coverage sometimes fall short of depicting a taxon’s functional role within the community. Instead, dominants or edificators play a pivotal role in shaping the community structure and maintaining its stability and ecosystem services. In vegetation ecology, dominants or edificators typically exhibit high importance values, amalgamating several measurements such as abundance, frequency, dominance, and height (Curtis and Mclntosh, 1951). Whether the pollen percentage aligns with numeric measures of taxon importance remains an open question.

To address these issues, we conducted extensive fieldwork on Jinhua Mountain, located in eastern subtropical China. Furthermore, pollen analysis was performed on the soil samples obtained from the quadrats. A vegetation map of Jinhua Mountain was compiled in 2017. Over the course of 2017-2018, plant quadrats from different vegetation types and plant communities were investigated to study the quantitively relationship between soil pollen percentage and plant community. Soil samples collected before the investigation were subjected to pollen identification. Our primary objectives were to compare discrepancies between pollen signals and the contemporary vegetation community in Jinhua Mountain and to determine whether pollen signals could effectively reflect significant ecological components within the subtropical Chinese plant community.

Pollen analysis is a significant botanical and ecological tool for studying the dynamics of biota and biotic systems (Prentice et al., 1998; Seddon et al., 2014; Kong et al., 2018; Birks, 2019). By comparing pollen analysis with quadrat investigations, our results unequivocally demonstrated that pollen signals can accurately reflect the major plant community composition and emphasize the significance of specific plant taxa. Consequently, pollen analysis in subtropical forests holds ecological significance as it bridges the gap between current and paleoecology to some extent, bolstering our confidence in studying long-term plant community and vegetation composition changes using fossil pollen in subtropical forests.

In agreement with previous pollen analyses in subtropical China (Liu et al., 2001; Yu et al., 2002; Zheng et al., 2002; Zheng et al., 2007; Zheng et al., 2008; Ding et al., 2011; Fang et al., 2015; Wan et al., 2020), pollen composition in surface soil from Jinhua Mountain closely aligns with modern vegetation, indicating that pollen signals vary according to different plant communities. The landscape of subtropical China is characterized by moderate mountainous and hilly terrain, where mountain vegetation exhibits distinct variations along altitude gradients. On Jinhua Mountain, both vegetation investigations and pollen analyses reveal vertical variations in plant composition signals. In summary, a higher prevalence of evergreen broad-leaved taxa has been observed in the lower regions, whereas cold-tolerant coniferous taxa tend to be dominated at higher elevations ( Figure 2 ). This phenomenon corresponds to pollen analyses conducted on Tianmu Mountain in Zhejiang Province (Guo, 2021), Yushan Mountain in Taiwan (Yu et al., 2002), Xishan Mountain in Jiangxi Province (Li et al., 2016), and Shennongjia Mountain in Hubei Province (Liu et al., 2001). Pollen composition in surface soil effectively reflects the different vegetation communities, such as pine forests and Schima forests, and their geographic distribution closely mirrors that of actual vegetation. The prevalence of Pinus pollen, particularly in pine forests, is evident in nearly all the samples. Considering that Pin. taiwanness forest is situated at higher altitudes and that Pin. massoniana can also be found sporadically in other plant communities on Jinhua Mountain (Fan et al., 2019), the high presence of Pinus pollen in our study region is justified. This finding is consistent with previous researches indicating that a high proportion of Pinus pollen is indicative of pine forests or isolated occurrence within the study region (Li and Yao, 1990; Xu et al., 2007; Xiao et al., 2011). In contrast to Pinus pollen, a high content of Taxodiaceae pollen is primarily associated with the cedar forests ( Figure 2 ). This may be attributed to both low pollen productivity and the limited dispersal ability of Cu. lanceolata within modern vegetation (Xiao et al., 2011; Li et al., 2018).

Furthermore, our results illustrate that pollen percentage data aligns with the IVs derived from quadrat-based surveys of different plant communities. Pollen percentages provide quantitative information about pollen composition, enabling complex numerical analyses to reveal potential dynamics and driving forces. Therefore, this method is the most convenient and widely applied approach for pollen analysis. In synecology, IV serves as a comprehensive quantitative indicator of a given taxon’s status and role in the community (Song, 2016), synthesizing relative density, dominance, and frequency (Curtis and Mclntosh, 1951). Owing to its simplicity and clarity, IV has been a common tool in community ecology for many years (Song, 2016). According to the protest results, there is a significant correlation between the pollen percentages and IVs ( Figure 5 ). This correlation is reasonable because dominant and abundant taxa are more likely to produce a greater number of pollen grains than rare taxa within the community (Andersen, 1970). For example, Cr. japonica, a dominant and edificator species in JH4-JH5, corresponds well with the high Cryptomeria pollen percentage in the soil samples. Similarly, high IVs for Cu. lanceolata, Castanea, and Liquidambar are consistent with the elevated pollen percentages of Taxodiaceae, Castanea, and Liquidambar in the same quadrats. This suggests that the commonly used pollen percentage inherently preserves the quantitative characteristics of the source plants and can effectively reflect the importance of dominant taxa in the plant community.

It is essential to note the discrepancies between pollen composition and plant signals in our study, including overall compositional levels and individual numeric features ( Figures 2 , 5 , 6 ). Our results revealed that plant taxa do not precisely align with pollen composition. This discrepancy is particularly evident for certain rare taxa such as Alangium Lam., Rubiaceae Juss., and Verbenaceae J. St.-Hil. Conversely, some pollen taxa, including Alnus, Salix, and Pterocarya, were absent in the plant investigation. We propose that these mismatches stem from the limitations of the quadrat area and dispersal ability of pollen (Andersen, 1970; Jackson and Kearsley, 1998). In subtropical China, the RSAP is estimated to range from approximately 290-500 m (Chen, 2019; Jiang et al., 2020; Guo, 2021). Consequently, a forest quadrat area of 30 × 30 m may not be sufficient to capture all existing taxa (Fan et al., 2019). Concerning individual numeric features, pollen percentages do not always align with the IVs for specific taxa. The taxa absent in the pollen data exhibited an average IV of 8, contrasting with an average percentage of 1.8% for unique pollen taxa not recorded in the plant investigation ( Figure 6 ). Strikingly, Lauraceae Juss. possesses an average IV of 5.8 in 24 quadrats, yet its pollen is absent in the quadrat soils ( Figure 6 ). Lauraceae is common in tropical and subtropical forests (Tan et al., 2023), and many plants in this family are insect-pollinated. In our dataset, the Lauraceae plants mainly comprise Lin. glauca, Lin. erythrocarpa, and Lin. Reflexa, all of which are insect-pollinated characterized by small, fragrant flowers with floral glands. Similarly, Sch. superba is a dominant species and edificator in evergreen broad-leaved forests on Jinhua Mountain, but its pollen percentage does not align with IVs ( Figure 5 ). Like Lauraceae, Sch. superba is a xenogamous plant that is highly depended on insect pollinators for seed production (Bai et al., 2022). In contrast to Lauraceae and Schima pollen, Pinus is the dominant species in pine forests, yet its pollen can be found high percentages in nearly all samples. These observations underscore another critical issue: the different pollen dispersal strategies of plants might affect the relationship between pollen percentages and plant IVs (Ackerman, 2000). In subtropical and tropical forests, the high frequency of insect-pollinated taxa (Wang et al., 2023) can introduce substantial uncertainties in the relationship between pollen and vegetation, leading to the underestimation of insect-pollinated taxa, such as Lauraceae, in pollen analysis.

The application of recently published RPP and R values to pollen percentage data yielded limited improvements in our results. Fossil pollen grains preserved in soil and other archives are subject to varying pollen productivity, transportation processes, and preservation conditions, resulting in nonlinear relationships between pollen and vegetation, and uncertainties in pollen percentages when representing the importance of selected taxa (Sugita, 1994; Bunting et al., 2004). RPP estimations for selected taxa have been reported (Li et al., 2018; Wieczorek and Herzschuh, 2020), as have regional variations in pollen R values in China (Xia and Ni, 2024). We checked the RPP results from tropical and subtropical forests (Chen, 2019; Jiang et al., 2020; Guo, 2021). We found that applying these RPPs to Pinus, Castanea, Castanopsis/Lithocarpus, and Moraceae pollen percentage data led to changes and introduced additional biases. Given that pollen-plant abundance data were collected from regions with different forest compositions, this calibration should be avoided (Sugita, 1994). A similar situation arises with R values in China, which exhibit significant regional variability (Xia and Ni, 2024). As a result, we refrained from this calibration and focused on the raw pollen percentages. Nevertheless, our results indicated a significant consistency between pollen percentages and plant IVs ( Figure 5 ), which may hold significance for pollen analysis across different vegetation types within a small region.

Despite the inherent uncertainties, the consistency between pollen percentages and plant IVs in subtropical forests from Jinhua Mountain is superior to that observed in open landscapes from the Northeast China Transect studied by Tan (2020). We propose that the main reason for this disparity lies in the different RPP and RSAP values of the two regions. For instance, Artemisia and Amaranthaceae Juss. (Chenopodiaceae) are indicative pollen taxa for typical grasslands in northern China, characterized by high RPPs and RSAPs that can extend over 1000 km (Xu et al., 2016; Cao et al., 2017). The influx of long-distance pollen into local pollen rain undoubtedly introduces bias into the relationship between pollen percentages and plant IVs Tan (2020). In the current study, the plant investigation data did not include herbaceous and fern species because of their limited coverage and biomass. The average percentages of herbaceous pollen and ferns in the pollen data are 0.1% and 1.6%, respectively. Therefore, we propose that our data exhibits regional representativeness for subtropical forests. Consequently, pollen analysis based on lake or mountain peat in subtropical China, which often maintains a high percentage of herbaceous pollen, may have underscored the development of local grasslands/wetlands while underestimating forest coverage.

In summary, our results demonstrated a significant correlation between pollen percentages and plant IVs, reinforcing the utility of pollen analysis in reflecting the ecological importance of specific taxa within a plant community. Discrepancies between pollen composition and plant signals exist, which probably stem from the limitations of the quadrat area and pollen dispersal ability. Despite inherent uncertainties, the consistency between pollen percentages and plant IVs in subtropical forests from Jinhua Mountain is superior to that observed in open landscapes. These relationships may also hold promise in paleoecology, where pollen analysis can be combined with geochemical and molecular information from natural archives to reconstruct ecological and evolutionary systems deep into the past.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

KL: Formal analysis, Funding acquisition, Methodology, Writing – original draft, Writing – review & editing. BT: Investigation, Methodology, Writing – review & editing. ML: Writing – review & editing, Methodology. JN: Conceptualization, Funding acquisition, Supervision, Writing – review & editing.