1. Introduction

Flowering phenology, which is critical for plant reproductive growth, has been shown to be influenced by climate warming (Parmesan, 2003; Boyle and Bronstein, 2012; Davies et al., 2013; Jiang et al., 2021). Accordingly, flowering phenology is an important biological indicator of current climate change (Fitter, 2002; Calinger et al., 2013) and has been widely used to evaluate the effects of global climate change on plants (Cleland et al., 2007; Ge et al., 2015). Changes to flowering phenology affect plant fitness and biodiversity by altering biological activities such as pollination, seed diffusion, seed germination, and seedling settlement (Davies et al., 2013; Jabis, 2020). Importantly, changes in flowering phenology increase the risk of plant-pollinator mismatches, threatening the stability of the community, and thus affecting ecosystem structure and functions (Franco-Cisterna et al., 2020). Because climate warming has the greatest impact on biodiversity hotspots in mountain regions (Dedieu et al., 2014), mountainous regions are an ideal "natural laboratory" for investigating how climate change affects flowering phenology.

The effects of climate change on plant flowering phenology across mountainous regions have attracted the widespread attention of ecologists. Several recent studies have been conducted to explore changes in plant flowering phenology in response to climate changes. For example, studies of flowering phenology in mountain plants have shown that mean flowering time is sensitive to temperature changes (Hart et al., 2014), elevation, temperature and precipitation changes along elevational gradients (Rafferty et al., 2020). Notably, studies on flowering phenology have primarily focused on temperate mountains such as the Alps, and found that spring phenology has become earlier over the last decades (Ovaskainen et al., 2013; CaraDonna et al., 2014; Vitasse et al., 2021). Furthermore, a recent study that explored the impact of climatic change on flowering phenology across the southern subtropical Nanling region in southern China found that spring flowering times became earlier in response to warming temperatures (Song et al., 2021). Still, evidence is lacking on the flowering phenology of different taxa in the mountains located in the southeastern Qinghai-Tibet Plateau (QTP) of China, the world largest and highest plateau. Furthermore, it remains unclear how flowering phenological changes in different seasons, and whether spring flowering is delayed or becomes earlier in the subtropical regions.

Previous studies have indicated that variation of flowering phenology is determined by changes in temperature and precipitation along with biotic factors (mainly indicated by plant functional traits) (Cortes-Flores et al., 2017). For example, the average flowering time of Rhododendron advances 2.27 days per degree (℃) increase in mean annual temperature (Hart et al., 2014). Flowering phenology has been shown to be influenced by plant growth forms (Fitter, 2002), flower color (Wang et al., 2020), and plant pollination types (e.g. wind-borne and animal pollination) (Willis, 2008; Rafferty et al., 2017). However, the mechanisms underlying flowering phenology changes in mountainous regions have only been reported sporadically, and no consistent conclusions have been reached for flowering phenology changes across different regions, different taxonomic groups, or different seasons.

Since the eighteenth century, herbaria have been recognized as repositories of biodiversity and centers of taxonomic expertise. Herbarium specimens provide accurate information on species distributions, morphological characteristics, and phenology (Espinosa and Pinedo, 2018). In recent studies, herbarium specimens have been used to quantify long-term changes in plant morphological traits (e.g. leaf size and plant size) (Law and Salick, 2005; Li et al., 2020), plant phenology (Miller-Rushing et al., 2006; Davis et al., 2015; Park et al., 2018; Everingham et al., 2022; Tadeo et al., 2022), and to detect the changing interactions between plants and their antagonists (Meineke et al., 2019). In general, herbarium specimens provide a wide range of biological information that can be used to study the changes of phenology in response to climate change in both temperate (Davis et al., 2015; Park et al., 2022) and tropical regions (Davis et al., 2022).

Mt. Gongga is a global biodiversity hotspot that has been recognized as the cradle and refuge of plant diversity in Southwest China (Wu et al., 2013). Mt. Gongga is located between the southeastern edge of the Qinghai-Tibet Plateau and the Sichuan Basin, and is one of the highest mountains in the Himalaya-Hengduan Mountains region and Southwest China (Zu et al., 2019). During the last few decades, Mt. Gongga has experienced severe climate change, resulting in a shift in seed plant distribution to suitable habitats along the elevational gradient (Zu et al., 2021). However, there is still uncertainty about whether and how past climate change has affected the flowering phenology of plants in this unique region. Understanding changes in flowering phenology in Mt. Gongga and the factors that drive these changes will enhance our understanding of the effect of climate change on floras, thereby aiding in conservation of plant diversity in this region. In this study, we used historical records and recent plot surveys of plants on Mt. Gongga to better understand how climate change affects flowering phenology. Specifically, this paper addresses two main questions. First, did the flowering phenology of angiosperms in the Gongga Mountains change over the past 70 years? If so, how? Second, what factors cause changes in flowering phenology of different species?

2. Materials and methods 2.1. Study region

Mt. Gongga is located in the subtropical range of China (29°21′N–29°54′N and 101°44′–102°10′E), connecting the Qinghai-Tibet Plateau and the Sichuan Basin, with elevations ranging from 1000 m to 7556 m (Fig. S1). It is the highest mountain in the Hengduan Mountains, which form the core of the Southwest China biodiversity hotspot (Zu et al., 2019). Due to the large elevation gradient, Mt. Gongga contains one of the most comprehensive elevation changes in vegetation belts, including evergreen broad-leaved forest (EBF), temperate coniferous broad-leaved mixed forest (CBM), sub-alpine coniferous forest (SC), alpine shrub and meadow (ASM), and alpine scree and nival zone (ASN) (Zu et al., 2019). The annual mean temperature in Mt. Gongga has increased 0.013 (±0.003) ℃/yr over the past 70 years (c. 1950–2018). In contrast, annual precipitation has decreased 1.368 (±0.98) mm/yr (Zu et al., 2021).

2.2. Species flowering phenology

Flowering phenology data of species consisted of two main components. First, we integrated all plant flowering phenology data in the four counties (i.e., Kangding, Luding, Shimian, and Jiulong) of the Mt. Gongga region from the National Specimen Information Infrastructure (NSII, http://www.nsii.org.cn/) and the Chinese Virtual Herbarium (CVH, http://www.cvh.org.cn). The specimen data were rigorously cleaned to remove specimens with incomplete information, including flowering date, location, and elevation. Because the collectors may reflect preferences for certain species, there are many repeated collections in some groups. To avoid the influence of collection preference, we eliminated all duplicate collection records of the same species collected on the same date and at the same location. Species names were standardized based on the taxonomic name resolution platform (TNRS, tnrs.iplantcollaborative.org/TNRSapp.html) and the Asian Plant Synonym Search (APSL, phylodiversity.net/fslik/synonym_lookup.htm). A total of 23, 044 plant specimens were collected between 1950 and 2017. Second, we conducted three comprehensive field surveys of vascular plants of Mt. Gongga from 2018 to 2019. These field surveys yielded 3000 flowering plant specimens from the entire elevational gradient (1000–5000 m). For each specimen, we recorded the species name, specific collection date, coordinates of collection site. We also estimated the timing of flowering for each species at each site. Specimens of flowering plants showing more than 50% of flowers in full bloom were considered to be in full bloom (Calinger et al., 2013). To assess changes in the flowering period of species, we divided the past ~70 years into two periods: 1950 to 1988 and 1989 to 2018. We chose 1988 as a dividing line because this is the year when the climate in the region showed clear signs of warming (Zu et al., 2021). The flowering phenology of the species occurrence records along the elevational gradients is shown in Fig. S2. To reduce the potential influences of small sample size on the estimation of flowering phenology changes of the species between the two periods, we selected species that occurred during both time periods, and had at least 10 occurrence records of flowering phenology within each time period. In total, 33 plant species with 1362 records in Mt. Gongga were obtained. The information on the taxonomic characteristics, the number of specimens obtained, and the elevation range of each species are shown in Table S1.

2.3. Environmental and species traits information

To study the influence of climate variation on the flowering phenology changes, we extracted climate data from the WorldClim database (http://www.worldclim.org) at a spatial resolution of 30 arc seconds (approximately 1 km²) (Hijmans et al., 2005). The extracted climate data include mean annual temperature (AMT) and annual precipitation (AP). To calculate the changes in temperature and precipitation along the elevational gradients, we extracted data on annual climatic variables at a spatial resolution of 1 × 1 km² from the Resources and Environmental Data Cloud Platform of the Chinese Academy of Sciences (http://www.resdc.cn/). This spatial interpolation is based on daily observations from more than 2400 meteorological stations across China (Hutchinson, 1998). We then computed the long-term average values of MAT and AP during 1980–1988 vs. 2007–2015. The specific method and results are described by Zu et al. (2021, 2022).

To determine whether the variation in flowering time of different species is related to its functional traits, we obtained data on the life forms, maximum plant height, sex system and seed dispersal types of each species (Table S2). These data were obtained from the Royal Botanical Seed Information Database (KEY, http://data.kew.org/sid/sidsearch.html), Flora of China (http://foc.eflora.cn/, and Flora Reipublicae Popularis Bulgaricae (http://frps.eflora.cn/). Seed dispersal methods were divided into two categories: those that depend on living organisms for dispersal and those that do not depend on living organisms. Flower color was mainly derived from the records of the comprehensive field surveys. In this study, we used the main color of the petal of a species if its flowers have multiple colors.

2.4. Estimation of flowering time

To compare changes in flowering phenology of different species at different times, the Julian days conversion method was used to calculate variation in phenological periods. We converted the date of flowering occurrence from year to year into the actual number of days, setting the first day of the first month of this year as day 1 (Büntgen et al., 2022). We created probability density plots of flowering time at different elevations and at different time periods for each plant species. We also compared the differences in flowering phenology in response to temperature among species with different elevational distributions, and then investigated the reasons for the differences from the perspective of species' individual traits.

We also assessed variation in flowering phenology along an elevation gradient in four vegetation types (EBF, CBM, SC and ASM). The study region was divided into 9 elevational bands with a vertical interval of 400 m: h1, 1000–1400 m; h2, 1400–1800 m; h3, 1800–2200 m; h4, 2200–2600 m; h5, 2600–3000 m; h6, 3000–3400 m; h7, 3400–3800 m; h8, 3800–4200 m and h9, 4200–4600 m (Table 1). The study region was restricted between 1000 m and 4600 m due to heavy disturbances caused by agriculture and deforestation in regions below 1000 m, and a lack of survey data for flowering species above 4600 m. We also divided the plants into three groups based on the mean flowering month of each species: spring-flowering (March–May), summer-flowering (June–August), and autumn-flowering (September–November). We calculated the flowering periods and their variation for species in each elevational belt.

2.5. Data analysis

To estimate the temperature sensitivity of phenology (days/℃) of each species, we calculated the mean temperature of each record along the elevational gradients across different years. The climate data was derived from the Resources and Environment Data Cloud Platform of the Chinese Academy of Sciences (http://www.resdc.cn/). We evaluated the relationship between flowering phenology and temperature of each species individually using linear regression models conducted with the base packages of the R software. We take the slope of the relationship between flowering phenology and temperature as the temperature sensitivity of flowering phenology of each species.

T tests were used to test the difference in flowering time changes between species groups (community, lifeform, elevational belts, and season). We extracted two climatic factors (AMT and AP) at different elevational belts to assess whether climate change influences flowering phenology at different elevations. Second-order polynomial ordinary least squares (OLS) regressions were used to test the relationships between the flowering phenology of species and different climate factors. In order to investigate the potential mechanism of the flowering phenology of different species, we used multivariate analysis to investigate the relationship between the flowering periods and the plant traits (i.e. maximum plant height, seed size, life forms, mode of dispersal, sex characteristics, flower colors).

To test the phylogenetic signal of flowering times and their changes, we used the published time-calibrated phylogenetic tree of Chinese vascular plants at genus level (Lu et al., 2018). This phylogenetic tree was reconstructed with four nuclear genes (atpB, matK, ndhF, rbcL) and one mitochondrial gene (matR), and contains 2665 genera of Chinese vascular plants. We first inserted all species of each genus into the tree as polytomies, and then used R package "stick-Tips" to randomize terminal branch lengths and generate 100 random trees (Peng et al., 2021). We then calculated the phylogenetic signals of species flowering times and their changes separately. We used Blomberg's K value to calculate lineage signals (R package 'phytools', Blomberg et al., 2003). Blomberg's K is the most commonly used index to calculate phylogenetic signals at present. Phylogenetic signal is stronger when K is closer to 1, with p < 0.05 indicating significant phylogenetic signal.

All analyses were conducted in R v.3.5.3 (R Core Team, 2019).

3. Results 3.1. Flowering time variation on Mt. Gongga

Most species on Mt. Gongga flowered from April to October (Fig. S3). In total, 35.6% of collected individuals flowered in May, 26.8% in June, and 20.1% in July. A small percentage of plants flowered in other mouths. Over the past 70 years, the mean flowering time of the 33 selected plant species was 171 (±1.41) days (Fig. 1a; Table 2). Flowering time on Mt. Gongga was later during the period from 1989 to 2018 than during the period from 1950 to 1988 (t = −8.55, p < 0.001) (Fig. 1b).

The mean flowering period varied for different vegetation types (Fig. 1c). For example, the mean flowering period was 181 (±11.95) days in evergreen broad-leaved forest (EBF), 166 (±11.31) days in temperate coniferous broadleaf mixed forests (CBM), 166 (±6.37) days in subalpine coniferous forests (SC), and 184 (±7.23) days in alpine shrubs and meadows (ASM). All four vegetation types showed delayed flowering times. In addition, the flowering times of species in ASM and CBM (t = 2.85, p < 0.005), ASM and SC (t = 5.34, p < 0.001) were significantly different. Specifically, EBF flowering was delayed for 46 days, followed by CBF at 26 days, SC at 14 days, and ASM at 20 days (Fig. 2, Table 2).

The average flowering time of species at different elevational belts was also different, with flowering time generally delayed in most species (Fig. 1d; Table 2). The flowering times of lowland species were more strongly delayed. For instance, the flowering time changes for species located above and below 3000 m were significantly different (t = −3.31, p < 0.001).

Flowering phenological varied between seasons, with spring phenology earlier by approximately 3 days, and summer and autumn phenology delayed by 6 days and 1 day, respectively. The flowering time changes were also significantly different among the three seasons (p < 0.001) (Table 2).

3.2. Climate factors affecting shifts in flowering times at different elevational belts

The mean flowering times at different elevational belts were strongly correlated with mean annual temperature (AMT) (R² = 0.58; Fig. 3a). Flowering time decreased as temperature increased, reaching the earliest time point at 7 ℃, before increasing again. However, there was no significant relationship between mean flowering times and annual precipitation (AP) at different elevational belts (Fig. 3b). Changes in flowering times at different elevational belts were strongly correlated with changes in temperature (R² = 0.61), and changes in precipitation (R² = 0.73), and showed a decreasing and then increasing trend with changes in temperature and precipitation (Fig. 3c, d).

The relationship between flowering time changes and temperature varied among species. Only seven species (21.2%) showed earlier flowering times with increasing temperature, with a mean value of 3.39 days/℃. In contrast, 26 species (78.8%) showed delayed flowering times with increasing temperature, with a mean value of −5.31 days/℃ (Fig. 4).

3.3. Effects of functional traits on flowering times and flowering shifts

Multiple regression analyses showed that the mean flowering times of species are mainly influenced by their life forms and flower color (Table 3), while flowering shifts among species are not significantly related to functional traits tested (Table S3). Of all functional traits tested, the highest contribution to the variance of the flowering times was life form (Table 3). Species with varying flower colors exhibited inconsistency with their flowering times. For instance, both herbaceous and woody plants had different flowering times. The flowering times of herbaceous plants were 190 days on average and those of woody plants were 162 days on average (Table 3; Fig. 5). We also found that there were no significant phylogenetic signals in species flowering times (k = 0.11, p = 0.78) or their changes (k = 0.17, p = 0.34).

4. Discussion

Changes in the flowering times of plants in response to climate change have attracted widespread attention (Lesica and Kittelson, 2010; Chen et al., 2017; Templ et al., 2017). However, studies on flowering phenology have rarely been conducted on plants in the mountains of Southwest China, even though this region has two biodiversity hotspots. In this study, we investigated changes in the flowering times of 33 species in response to climate change over the past 70 years in the Mt. Gongga region. Our results show that peak flowering time in Mt. Gongga is delayed for plants of several vegetation types and elevational zones. Flowering time has been delayed most in evergreen broad-leaved forest plants, while it was least delayed in subalpine coniferous forest plants. Flower phenology varied in plants at different elevations, with flowering at lower elevations more delayed, most likely due to temperature and precipitation. Flowering time also varied between seasons, with flowering time reduced in spring-flowering plants, while it was delayed in summer- and autumn-flowering plants. Because there were fewer winter flowering plant records, the phenological changes of winter flowering plants were not considered in this study.

We discovered a non-synchronous change in the flowering phenology of different species, as well as a nonlinear relationship between flowering times and climatic factors along an elevation gradient. The peak flowering time of different species was mainly influenced by climatic factors (e.g., temperature) and species traits (e.g., life type and flower color). The flowering time changes are mainly influenced by shifts in temperature and precipitation but not related to the tested traits. Asynchronous variation in flowering plant phenology may cause a mismatch in the phenology of co-occurring species, resulting in asynchronous ecological interactions and affecting the structure and function of montane forest ecosystems (Kharouba et al., 2018).

4.1. Factors affecting flowering phenology changes in Mt. Gongga

In general, flowering phenology on Mt. Gongga has been delayed. The variation in flowering time was significantly different across elevational belts, vegetation types, and seasons. Phenological differentiation among species revealed multiple strategies associated with growth form and pollination syndromes that could be crucial for understanding species coexistence in this highly diverse plant community (Cortes-Flores et al., 2017). We observed some important differences in flowering phenology between species. Firstly, the flowering phenological changes in species and their variations along elevational gradients did not show an increasing linear trend. In addition, the trends changed when temperature and precipitation reached their peak values. We hypothesize that this is largely a function of the degree of variation in species flowering phenology in relation to abiotic factors such as temperature and precipitation. Although higher temperatures promote earlier plant flowering (Büntgen et al., 2022; Ma et al., 2022), other abiotic factors (e.g., precipitation and elevation) may also affect plant physiology, delaying flowering phenology of species (Ahmad et al., 2021).

Secondly, flowering phenology varied by season. Specifically, we found that spring phenology occurs earlier, whereas summer and autumn phenology is delayed. These findings are consistent with those of previous studies, which found that leaf growth and flowering occur early in the spring, and are typically delayed in the autumn (Bertin, 2008). Previous phenological observations from 1960 to 2011 at various sites in China found that 90.8% of species flowered early in spring and summer, whereas 69.0% flowered late in autumn (Ge et al., 2015). Earlier spring blooming times increases the probability of frost damage in natural and agricultural systems (Menzel et al., 2020). Delayed flowering times in autumn influence the timing of seed ripening and dispersal (Büntgen et al., 2022).

Thirdly, in this study, we found that the flowering time of different species are also different, with flowering times mainly related to individual species traits. The flowering times of species with different life forms also vary. For example, flowering time was earlier in woody plants seem than in herbaceous plants. This could be because woody plants are more sensitive to temperature, and long-term adaptations lead trees to adopt an early flowering strategy (Wang et al., 2020). One explanation for differences in plant phenology between plants with different flower colors is that flower color is a key adaptive plant trait that determines reproductive success and is critical for attracting pollinators (Toleno et al., 2010). In this study, we found that species with purple and yellow flowers (e.g., Rhododendron concinnum and Meconopsis integrifolia) flowered later. Notably, these plants are mostly found at higher elevations, where temperatures are cooler.

4.2. Consequences of changes in flowering time

Changes in the flowering phenology of species have an important impact on the survival of species, plant function, and structure of ecosystem services (Piao et al., 2019). There are two major consequences of the change in flower phenology. First, this study found differences in the phenological response of different species to climate change in Mt. Gongga, which may lead to asynchronous ecological interactions and climate change, thus threatening the structure and function of the ecosystem (Thackeray et al., 2016; Pelayo et al., 2021). Second, early and late flowering of species affects the ability of plants to adapt to their environment, which in turn determines the survival and death of species (CaraDonna et al., 2018). If changes in flowering phenology of species can track the rate of climate change, then these species can adapt to the environment. However, if the change in flowering phenology of species cannot track the rate of climate change, these species may be eliminated.

Furthermore, the sustainable survival of species relies on the reproductive success of pollinators. Changes in flowering phenology may lead to a mismatch between pollinating insects and flowering plants. For example, a previous study showed that changes in the flowering time of Corydalis yanhusuo created a phenological mismatch with its pollinator, bumblebees, resulting in a lower seed setting rate (Kudo, 2013). An additional study has shown that plant phenology near the Rocky Mountains has advanced due to climate change over the past 30 years, while their pollinators have seen little change in bumblebee phenology, resulting in decreased synchrony between bees and plants (Pyke et al., 2016). The mismatch between flowering plants and their pollinators should be examined in future studies based on field observations and controlled laboratory experiments to explore and understand plant and pollinator phenological changes. The findings of this study contribute to a better understanding of how mountain plants cope with climate change, which is crucial for conserving biodiversity.

Although historical records have been used to quantify the impacts of past climate change on plant phenology, they can also provide important support for predicting the impacts of plants response to future climate change through the construction of plant-climate models (Hänninen et al., 2019). However, due to the complexity of species composition and the difficulty of field observations, constructing suitable phenological models for tropical and subtropical regions remains a major challenge (Fu et al., 2020). A comprehensive understanding of the effects of historical climate change on the flowering phenology of plants in Mt. Gongga may provide an important basis for the construction of plant-climate models and the prediction of future climate change in subtropical mountains. As the climate continue to change, it is important to pay attention to flowering phenology of plants that are sensitive to climate change. In the conservation planning of protected areas, the impact of climate change on species phenology should also be considered, so as to rationally plan the construction and management of protected areas and avoid the loss of biodiversity and ecosystem functions.

Acknowledgements

This work was supported by Jiangxi Provincial Department of Education Science and Technology Research Project (GJJ2200433), the Natural Science Foundation of Jiangxi, China (#20224BAB213033), the National Key Research and Development Program of China (#2018YFA0606104), National Natural Science Foundation of China (#32125026, #31988102), and the Strategic Priority Research Program of Chinese Academy of Sciences (#XDB31000000).

Author contributions

ZW and KZ conceived the idea; KZ and ZW collected the data and performed the analyses; KZ wrote the manuscript with inputs from ZW and FC. FC, YL, NW, XF, SA and GN edited, provided suggestions and polished the manuscript. All authors contributed to the writing and approved the final manuscript.

Data availability statement

The data that supports the findings of this study are available in this article and the supplementary information of this article.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2023.07.007.