Sensitivity of First Leaf Date to Temperature Change for Typical Woody Plants in Guiyang, China

Abstract

The temperature sensitivity of plant phenology reflects how and to what extent plants respond to climate change and is significantly related to their ability to adapt to climate change. Previous studies on the temperature sensitivity of first leaf date (FLD) primarily focus on temperate regions, with relatively few studies conducted in subtropical areas. This study analyzed observational data on the FLD of 63 typical woody plant species from 1980 to 2019 in Guiyang, located in the subtropical zone of China. We quantified the trend of FLD and its sensitivity to temperature changes and then assessed the impact of sample size on the stability of sensitivity estimates. The results showed that (1) significant warming occurred in Guiyang during the study period, with the largest warming occurring in spring. (2) The FLD of the vast majority of plants (95.2%) showed an earlier trend during the study period (19.0% significantly at p < 0.05). The earlier trend of most species ranged from −3 to −1 days decades⁻¹. The median of trends for all 63 species investigated was 1.97 days decades⁻¹. (3) The interannual variation in FLD was significantly negatively correlated with the preseason average temperature (p < 0.05). Most of the temperature sensitivity of FLD was between −5 and −3 days °C⁻¹, with a mean of −4.53 days °C⁻¹. (4) The sample size significantly influenced the stability of the temperature sensitivity estimates. Using randomly selected 20-year data could limit the standard deviation of the sensitivity estimate to 0.3 days °C⁻¹. These results suggest that the leaf unfolding date of subtropical species could track climate warming closely like temperature species. The temperature sensitivity of FLD should be estimated based on long-term observation data.

1. Introduction

Global warming has become one of the most pressing environmental issues in the world today [1]. According to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC), the global mean surface temperature increased by approximately 1.09 °C between 2011 and 2020 compared to the period from 1850 to 1900 [2]. Changes in plant phenophases, as direct evidence of the impact of climate change on ecosystems, have received widespread attention [3,4]. Studies based on ground-based phenological observation data have shown that spring phenophases (such as first leaf date, first flowering date, etc.) of woody plants have predominantly exhibited an earlier trend in Europe [5,6], Asia [7,8,9], and most regions of North America [10,11]. This advancement has also been confirmed by changes in the start of the growing season based on remote sensing data [12,13]. These observed phenological changes are closely linked to climate change. The advancement of spring phenophases in temperate regions of the Northern Hemisphere could be mainly attributed to temperature increases [4,14]. Photoperiod and chilling also play a key role in regulating the spring phenophases of woody plants [15,16]. In arid ecosystems, water availability has also been shown to influence spring phenological changes significantly [17].

Although spring phenophases have generally advanced, there are significant differences in the degree of advancement among species. For example, the rate of advancement of spring phenophases in 232 plant species ranged from −13 to −2 days decades⁻¹ between 1985 and 2011 on the island of Guernsey [18]. In Canada, a study showed that the rate of change in first flowering date between 2001 and 2012 ranged from −20 to 2 days decades⁻¹ among 19 species [19]. Furthermore, the degree of advancement in plant spring phenophases also varies across different study periods. For instance, between 1971 and 2000, the average rate of advancement in first leaf date (FLD, defined as the date of the first appearance of fully expanded leaves on selected individuals) of European species was −3.94 ± 0.03 days decades⁻¹, whereas between 1951 and 2018, the advancement rate was −2.40 ± 0.02 days decades⁻¹, possibly due to the lack of active warming in 1951–1970 [3,20]. These interspecific differences were primarily attributed to varying sensitivities in response to temperature change. For example, in Xi’an, China, the temperature sensitivity of FLD for 42 woody species ranged from −5.48 to −1.47 days °C⁻¹ [21]. Such differences in temperature sensitivity have even altered the sequence of phenophases within the same year [22].

Temperature sensitivity of plant phenophases is closely related to the ability of plants to adapt to climate change [23]. Woody plants with higher temperature sensitivity are able to respond more quickly to warming by advancing flowering and leaf unfolding, thereby extending their growing season, which helps them better survive and reproduce under climate warming [24]. However, the advancement of spring phenophases may also make plants more vulnerable to late frosts, which can negatively impact their survival and reproductive success [25]. Plants with a lower temperature sensitivity of phenology may be unable to take advantage of the longer growing season for photosynthesis, which is detrimental to resource acquisition and reproduction [26]. Studying the temperature sensitivity of phenophases across different species helps identify plant species that are more resilient to climate change and provides scientific evidence for ecosystem management under future climate change scenarios [27,28].

To date, studies on the temperature sensitivity of FLD primarily focus on temperate regions, such as North China [27,28], Japan and Korea [29,30,31], Northwestern Europe [3,32], and the Northeastern United States [6,33]. However, research on the temperature sensitivity of FLD in subtropical humid climate zones is relatively limited. In light of this, we focused on the FLD of 63 woody plant species from 1980 to 2019 in a subtropical site (Guiyang, China). We applied a sliding correlation method to calculate the optimal period (OP) during which temperature influences FLD significantly for each species and used regression analysis to estimate the phenological trend and temperature sensitivity. The objectives of this study are (1) to reveal the temporal trend of FLD in Guiyang over the past 40 years; (2) to quantify the sensitivity of FLD in response to temperature; and (3) to assess the impact of sample size on the stability of sensitivity estimates.

2. Materials and Methods

2.1. Study Area

Guiyang is located in the central part of Guizhou Province, China, with an average elevation of around 1100 m. The phenological observation site is situated at Guizhou University (26°25′44″ N and 106°40′09″ E). This site has a subtropical humid and mild climate. During the phenological observation period (1980–2019), the annual average temperature was 14.2 °C, with January being the coldest month (average temperature of 3.7 °C) and July being the hottest month (average temperature of 22.6 °C) (Figure 1). The annual total precipitation was 1091.5 mm, with the majority of precipitation (63%) occurring from May to August (Figure 1).

2.2. Phenological and Temperature Data

The phenological data used in this study were obtained from the China Phenology Observation Network (CPON). We used FLD data for 63 woody plant species from 1980 to 2019 (

Table A1. Summary of plant species investigated in this study.

Number	Species	Family	Number of Observation Years	Mean First Leaf Date	Trends (Days Decades−1)	Sensitivity (Days °C−1)
1	Chaenomeles cathayensis	Rosaceae	26	11 February	−1.68	−5.91
2	Chaenomeles speciosa	Rosaceae	28	12 February	−1.15	−6.19
3	Chaenomeles sinensis	Rosaceae	32	12 February	−0.88	−5.48
4	Malus micromalus	Rosaceae	23	17 February	−0.88	−3.44
5	Liriodendron chinense	Magnoliaceae	23	17 February	2.03	−2.83
6	Salix babylonica	Salicaceae	30	19 February	−0.81	−4.75
7	Pyracantha fortuneana	Rosaceae	27	23 February	−2.60	−5.31
8	Forsythia viridissima	Oleaceae	28	25 February	−2.68	−4.11
9	Zanthoxylum simulans	Rutaceae	26	25 February	−0.14	−3.74
10	Eriobotrya japonica	Rosaceae	25	25 February	−3.49	−4.72
11	Swida paucinervis	Cornaceae	31	3 March	−1.86	−5.45
12	Armeniaca vulgaris	Rosaceae	30	5 March	−1.61	−5.01
13	Distylium dunnianum	Hamamelidaceae	26	5 March	−3.43	−3.78
14	Ilex fargesiivar.angustifolia	Aquifoliaceae	26	5 March	−2.86	−4.58
15	Cerasus subhirtella	Rosaceae	30	6 March	−2.28	−4.50
16	Acanthopanax trifoliatus	Araliaceae	29	6 March	−4.64	−4.78
17	Pterocarya stenoptera	Juglandaceae	32	8 March	−1.50	−2.83
18	Prunus salicina	Rosaceae	21	10 March	−2.18	−5.49
19	Caesalpinia decapetala	Leguminosae	29	11 March	−4.53	−5.02
20	Cercis chinensis	Leguminosae	31	12 March	−1.70	−3.86
21	Camellia oleifera	Theaceae	30	12 March	−3.32	−5.17
22	Pyrus pyrifolia	Rosaceae	21	13 March	−6.50	−4.37
23	Fraxinus chinensis	Oleaceae	30	13 March	−2.49	−3.92
24	Hibiscus syriacus	Malvaceae	27	13 March	−3.87	−4.95
25	Ulmus pumila	Ulmaceae	30	14 March	−0.77	−3.44
26	Populus yunnanensis	Salicaceae	29	15 March	−2.98	−3.99
27	Cedrus deodara	Pinaceae	31	15 March	−4.08	−4.62
28	Populus adenopoda	Salicaceae	25	16 March	−1.58	−3.55
29	Nerium indicum	Apocynaceae	26	16 March	−1.06	−5.01
30	Serissa japonica	Rubiaceae	27	16 March	−1.13	−4.69
31	Prunus persica	Rosaceae	24	16 March	−0.06	−2.18
32	Salix chaenomeloides	Salicaceae	30	17 March	−3.18	−4.40
33	Quercus acutissima	Fagaceae	30	17 March	−0.51	−3.83
34	Cinnamomum camphora	Lauraceae	29	17 March	−3.11	−5.03
35	Metasequoia glyptostroboides	Cupressaceae	32	18 March	−2.98	−4.04
36	Cinnamomum bodinieri	Lauraceae	32	19 March	−2.53	−3.85
37	Pistacia chinensis	Anacardiaceae	27	19 March	−0.05	−4.83
38	Camptotheca acuminata	Nyssaceae	32	20 March	−1.86	−5.14
39	Hibiscus mutabilis	Malvaceae	29	20 March	−5.74	−6.79
40	Celtis sinensis	Cannabaceae	31	20 March	−3.18	−5.26
41	Catalpa bungei	Bignoniaceae	31	21 March	−2.88	−6.29
42	Lindera communis	Lauraceae	31	22 March	−2.21	−3.87
43	Punica granatum	Punicaceae	32	22 March	−2.66	−5.83
44	Platanus acerifolia	Platanaceae	16	23 March	−2.85	−3.34
45	Robinia pseudoacacia	Leguminosae	32	23 March	−1.55	−4.40
46	Gleditsia sinensis	Leguminosae	31	23 March	−0.72	−3.65
47	Alangium chinense	Alangiaceae	27	23 March	−1.53	−3.82
48	Ginkgo biloba	Ginkgoaceae	30	23 March	−0.98	−2.95
49	Juniperus chinensis	Cupressaceae	26	24 March	1.55	−4.33
50	Cerasus yedoensis	Rosaceae	31	25 March	−1.01	−4.06
51	Broussonetia papyifera	Moraceae	31	25 March	−0.95	−4.99
52	Clerodendrum trichotomum	Verbenaceae	32	27 March	−1.89	−4.09
53	Wisteria sinensis	Leguminosae	26	28 March	−2.36	−4.55
54	Ligustrum lucidum	Oleaceae	31	28 March	−0.27	−4.11
55	Sophora japonica	Leguminosae	31	29 March	0.96	−4.05
56	Melia azedarach	Meliaceae	21	30 March	−5.11	−6.19
57	Hovenia acerba	Rhamnaceae	25	30 March	−4.09	−5.13
58	Lagerstroemia indica	Lythraceae	31	31 March	−2.02	−4.52
59	Sapium sebiferum	Euphorbiaceae	29	1 April	−2.29	−5.25
60	Catalpa ovata	Bignoniaceae	31	2 April	−1.33	−4.41
61	Firmiana simples	Sterculiaceae	32	4 April	−1.69	−4.19
62	Ziziphusjujuba	Rhamnaceae	31	6 April	−0.42	−4.38
63	Pinus massoniana	Pinaceae	30	12 April	−4.11	−5.70

). According to the observation standards of CPON, FLD is defined as the date of the first appearance of fully expanded leaves on selected individuals [34]. During the entire study period, phenological observations were not conducted in 1992 or from 1996 to 2002. Thus, a total of 32 years of data were analyzed (Figure A1).

To ensure sufficient sample size for accurate trend estimation, the selected plant species must meet the following criteria: (1) at least 15 years of observation records between 1980 and 2019; (2) at least 5 years of observation records in both the 1980–1995 and 2003–2019 periods. Based on these criteria, 63 woody plant species were selected (Table A1). The discontinuous nature of the phenological observations caused varying numbers of species observed in each year, ranging from 38 to 63 species (Figure A1).

The daily temperature data (1980 to 2019) at Guiyang Station were downloaded from the China Meteorological Data Service Center (http://data.cma.cn/).

2.3. Methods

To investigate the climatic and phenological changes in Guiyang, linear regression analysis was first conducted between the monthly average temperatures and years to calculate the temperature trends for each month. Next, linear regression analysis was applied to the FLD data of different plant species to estimate the phenophase trends for each species. Additionally, for each species, the FLD time series was transferred to an anomaly time series by subtracting the multi-year average. The median of the FLD anomalies for all species was calculated annually to obtain a time series of median FLD anomalies, which represents the overall trend of FLD for all plant species. Furthermore, to explore the influence of period selection on the estimate of phenological trends, we also calculated the trends in the median FLD anomaly for all 10-year and longer periods between 1980 and 2019.

To study the sensitivity of FLD to temperature response, we first determined the optimal period during which temperature influences FLD most significantly. Previous studies have shown that FLD was typically significantly correlated with temperature from a specific preceding period [21,35,36]. Using the multi-year average FLD as the endpoint (EP), Pearson’s correlation coefficient between temperature and FLD was calculated for each period [EP – 15 × i, EP] (i = 1, 2, …, 12), with a step of 15 days. The period with the highest absolute value of the correlation coefficient was selected as the optimal period (OP) influencing FLD. Subsequently, regression analysis was performed on the FLD of each plant species and the mean temperature of the optimal period:

$$\mathit{FLD}=b\times \mathit{TEM}+a+e$$

(1)

where FLD represents the leaf unfolding date, TEM represents the average temperature of the optimal period, a is the intercept, and e is the error term. b is the regression slope, which serves as the indicator of temperature sensitivity. For each species, the temperature sensitivity of FLD was estimated using all available data and Equation (1).

Furthermore, the average temperature time series for the optimal period of each species was converted to an anomaly time series by subtracting the multi-year average. The median of the OP temperature anomalies for all species was calculated, representing the overall interannual change of preseason temperature.

To investigate the effect of sample size on sensitivity estimation, we analyzed the 32-year time series of the median FLD anomaly and median OP temperature anomaly. First, the overall temperature sensitivity (b_all) was calculated using Equation (1). Next, a random number generation algorithm was employed to generate FLD and corresponding temperature series with different sample sizes. For example, for a 5-year sample size, 5 years were randomly selected from the available data and the corresponding temperature and FLD data were extracted to compute the temperature sensitivity of FLD. This process was repeated 1000 times to obtain a set of sensitivity estimates for a 5-year sample size. Similarly, we calculated sensitivity for different sample sizes (from 5 to 22 years) and compared the standard deviations of sensitivity estimates and the probabilities of results falling within the 95% confidence interval of b_all.

3. Results

3.1. Temperature Trend

From 1980 to 2019, Guiyang experienced significant climate change. The annual average temperature showed a significant increasing trend of 0.41 °C decades⁻¹ (p < 0.05, Figure 2b). Over the past 40 years, temperatures increased for all months (Figure 2a), with the temperature rise being statistically significant (p < 0.05) from February to October. Among these months, February exhibited the largest temperature increase of 0.85 °C decades⁻¹.

3.2. Leaf Unfolding Patterns and Changes Across Different Periods

The average FLD for all 63 species ranged from February 11 (Chaenomeles cathayensis) to April 12 (Pinus massoniana Lamb.). Nine species (14%) began leaf unfolding in February, while fifty species (79%) began in March and five species (8%) began in April (Table A1). FLD exhibited strong inter-annual variability in Guiyang, with an average standard deviation of 10.90 days (Figure 3a). The species with the smallest inter-annual variation (standard deviation of 7.05 days) was Liriodendron chinense (Hemsl.) Sarg., while the largest variation was observed in Hibiscus mutabilis L. (standard deviation of 16.38 days). Overall, the standard deviation of inter-annual variation in FLD was significantly negatively correlated with the average FLD (R = −0.36, p < 0.05), indicating that species with earlier FLD tend to show stronger inter-annual variation.

The FLD for various plant species in Guiyang revealed a predominant trend (Figure 3b). Among the 63 species, 60 species (95.2%) exhibited an earlier FLD between 1980 and 2019, with 12 species showing a significant advance (p < 0.05). Only three species showed a non-significant delayed trend in FLD. The species with the strongest advance in leaf unfolding was Pyrus calleryana Decne. (−6.50 days decades⁻¹), while the weakest trend was observed in Corylus avellana L. (−0.05 days decades⁻¹). The majority of species showed an earlier trend between −3 to −1 days decades⁻¹ (Figure 3b).

According to the time series of the median FLD anomaly for all the 63 species in Guiyang (Figure 3c), the FLD anomalies were predominantly late before 1995 (average anomaly of 3.73 days), with only one extreme early FLD in 1987 (anomaly of −18.45 days). From 2003 to 2019, the anomalies shifted towards earlier, with 9 out of 17 years showing negative anomalies and the average anomaly being −2.47 days (Figure 3c). Such change from late to early FLD resulted in an overall advance of 1.97 days decades⁻¹ over the 1980–2019 period, although this trend was not statistically significant. Figure 4 shows the changes in median FLD anomalies for different periods. The vast majority of periods longer than 10 years showed an earlier trend. Only in the periods starting between 1980–1983 and ending between 2009–2015, FLD showed a significant advance (p < 0.05). In the period from 1996–2010 to 2011–2019, the trend even showed a non-significant delaying trend. Considering the lack of observation data between 1996–2002, the slight delay in FLD mainly occurred after 2003.

3.3. Relationship Between FLD and Temperature

Figure 5a shows the length of the optimal period in which temperature influences the FLD most significantly for each species. The average length of the optimal period for all species was 52.8 days, with most species (71.4%) having an optimal period between 45 and 60 days. Figure 5b illustrates the temperature sensitivity of FLD for each plant species, based on the mean temperature during the optimal period. For all 63 species, the response of FLD to temperature during the optimal period was statistically significant (p < 0.05). The strongest response was observed in Hibiscus mutabilis L., with a sensitivity of −6.79 days °C⁻¹, while the weakest response was in Prunus persica (L.) Batsch, with a sensitivity of only −2.18 days °C⁻¹. In terms of the frequency distribution of temperature sensitivity, most species (62%) had temperature sensitivity between −5 and −3 days °C⁻¹. Regression analysis between the median FLD anomaly for all species and the median temperature anomaly for the optimal period revealed a significant correlation (R² = −0.80, p < 0.05). The overall temperature sensitivity (b_all) was −4.53 days °C⁻¹ (with a 95% confidence interval of −5.34 to −3.71), indicating that for every 1°C increase in temperature, the average FLD in Guiyang was advanced by 4.53 days (Figure 5c).

3.4. Impact of Sample Size on Sensitivity Estimation

The results of random sampling show that when the sample size was only 5 years, the standard deviation of temperature sensitivity from 1000 samplings was 1.24 days °C⁻¹, with a probability of 55.6% that the estimates fall within the 95% confidence interval of b_all (Figure 6). As the sample size increases, the standard deviation gradually decreases, and the probability of estimates falling within the 95% confidence interval of b_all increases. When the sample size reached 20 years, the standard deviation from 1000 samplings decreased to 0.32 days °C⁻¹, and 98.8% of the estimates fell within the 95% confidence interval of b_all. This result demonstrated that the sample size has a significant impact on the stability of sensitivity estimation.

4. Discussion

We found that the median FLD of all plant species in Guiyang showed a significant earlier trend between 1980 and 2010, but the trend became delayed between 2011 and 2019 (Figure 4). This result suggests that the trend of earlier leaf unfolding was weakening or even reversed in the recent decade. This result was consistent with the finding of a reduced response of spring phenology to temperature during warmer periods [37]. Moreover, the study period has a significant impact on the estimation of phenological trends. Only the periods starting from 1980–1986 and ending in 2009–2016 showed a significant earlier trend. If the time series began from 1997–2010 and ended in 2011–2019, the result may even indicate a delay in FLD (Figure 4). This is also in line with previous studies. For example, through a sliding trend analysis of long-term historical phenological data, it was found that phenological change trends varied among each consecutive 30 years [38,39]. Such a difference in phenological trends was due to the distinct climate fluctuations over various periods. When synthesizing phenological changes from multiple studies to derive national or global-scale phenological trends, it is crucial to consider the varying periods used in different studies, as the results depend on the time window chosen [36].

This study found that the FLD of all surveyed plants was significantly correlated with temperature, with an overall sensitivity of −4.53 days °C⁻¹. This value is higher than those reported for temperate regions. For instance, the average temperature sensitivity of FLD in Beijing, Xi’an, and Mudanjiang (all located in temperate regions) during 1963–2020 was −3.28 days °C⁻¹ [9], and for 21 countries across Europe, the average sensitivity for spring and summer phenology was −2.5 days °C⁻¹ [20]. This further confirms the fact that low-latitude regions exhibit higher temperature sensitivity compared to high-latitude regions [6]. Temperature sensitivity not only reflects how and to what extent phenological events respond to climate change, but also relates to the plants’ ability to adapt to climate change. Tree species with higher temperature sensitivity are able to respond more quickly to global warming, advancing leaf unfolding and flowering to lengthen the growing season. This adaptation helps these plants survive and reproduce in a warming climate [24]. In contrast, species that are less responsive to climate change may face growth limitations because they might miss out on crucial interactions with symbiotic organisms (e.g., plants that rely on insect pollination) or experience a shorter growing season [26,40]. Species whose phenology did not respond to climate change have significantly declined in their abundance within communities over the past 150 years [41]. Invasive species, which tend to have higher temperature sensitivity of leaf unfolding than native species, may contribute to species invasions at the community level [42]. Cleland et al. [40] synthesized data from 24 studies across 57 species and found that species more sensitive to climate warming showed increases in biomass, cover, and flower numbers. These studies highlighted the importance of the temperature sensitivity of leaf unfolding in determining how plant communities respond to global warming. Therefore, the results of sensitivity estimations can provide valuable insights for ecosystem management under future climate change scenarios. For example, in afforestation or landscaping projects, it might be beneficial to prioritize species with high phenological sensitivity, as these species would be better adapted to future climate changes.

Since temperature sensitivity was closely related to a species’ ability to adapt to future climate change, an accurate estimation of temperature sensitivity was crucial for interspecific comparison. There are usually two methods for estimating temperature sensitivity: long-term phenological observations and controlled experiments [43]. This study employed the first method to estimate the temperature sensitivity of FLD for 63 tree species in Guiyang. The results revealed that temperature sensitivity estimates were quite sensitive to the length of the time series. When the sample size reached 20 years, the standard deviation of the results from 1000 random samplings decreased to 0.32 days °C⁻¹, and 98.8% of the estimates fell within the 95% confidence interval of b_all. This suggests that when estimating temperature sensitivity in the future, longer time series should be used to achieve more stable and reliable results. Another method for estimating temperature sensitivity involves conducting controlled experiments in which plants are subjected to different temperatures. The temperature sensitivity was then determined as the ratio of phenological differences to temperature differences. However, significant differences in sensitivity estimates exist between the two methods for phenological events of the same species [6,14,43]. This discrepancy was likely due to the complex interactions of multiple environmental factors presented in long-term phenological observation data. Overall, long-term observational data can provide insights into how plants actually respond to climate variation over time, while controlled experiments offer a more direct measure of temperature sensitivity in a controlled environment. Ultimately, combining both methods could offer a more comprehensive understanding of plant responses to climate change.

5. Conclusions

This study investigated the changes in FLD and its temperature sensitivity for 63 plant species in Guiyang from 1980 to 2019. We found that the annual average temperature in Guiyang has significantly increased since 1980, with February temperature showing the greatest increase of 0.85 °C decades⁻¹. Of the 63 species, 60 species (95.2%) have shown an earlier onset of leaf unfolding, with 12 species showing a significant advance (p < 0.05). The overall trend for FLD from 1980 to 2019 was −1.97 days decades⁻¹, though a slight delay in leaf unfolding has occurred since 2002. During the study period, the inter-annual variation in FLD for all plants was significantly negatively correlated with the average temperature of the optimal period. The temperature sensitivity of FLD ranged from −5 to −3 days °C⁻¹ with an overall sensitivity of −4.53 days °C⁻¹. Sample size significantly affected the stability of temperature sensitivity estimates. To accurately compare the temperature sensitivity of FLD between species, sufficiently long time series (e.g., 20 years) are required to achieve stable estimates. These findings emphasize the importance of using long-term data for estimating temperature sensitivity and highlight how temperature increases are influencing the leaf unfolding date of species in subtropical regions.