Analysis of Changes in Forest Vegetation Peak Growth Metrics and Driving Factors in a Typical Climatic Transition Zone: A Case Study of the Funiu Mountain, China

Abstract

Phenology and photosynthetic capacity both regulate carbon uptake by vegetation. Previous research investigating the impact of phenology on vegetation productivity has focused predominantly on the start and end of growing seasons (SOS and EOS), leaving the influence of peak phenology metrics—particularly in typical climatic transition zones—relatively unexplored. Using a 24-year (2000–2023) enhanced vegetation index (EVI) dataset from the Moderate Resolution Imaging Spectroradiometer (MODIS), we extracted and examined the spatiotemporal variation for peak of season (POS) and peak growth (defined as EVI_max) of forest vegetation in the Funiu Mountain region, China. In addition to quantifying the factors influencing the peak phenology metrics, the relationship between vegetation productivity and peak phenological metrics (POS and EVI_max) was investigated. Our findings reveal that POS and EVI_max showed advancement and increase, respectively, negatively and positively correlated with vegetation productivity. This suggested that variations in EVI_max and peak phenology both increase vegetation productivity. Our analysis also showed that EVI_max was heavily impacted by precipitation, whereas SOS had the greatest effect on POS variation. Our findings highlighted the significance of considering climate variables as well as biological rhythms when examining the global carbon cycle and phenological shifts in response to climate change.

1. Introduction

As one of the most important components of terrestrial ecosystems, vegetation contributes significantly to their productivity and is essential for both regional and global ecological processes [1,2]. The term “vegetation phenology” refers to the periodic natural phenomena that plants display in response to environmental and other factors [3]. Dynamic changes in vegetation phenology could influence energy, water, and carbon exchange between the Earth’s surface and the atmosphere [4,5]. This interchange of ecological processes, in turn, exerts feedback effects on climate, thereby regulating climate change [6,7]. Investigating vegetation phenology therefore not only enhances our understanding of the response of vegetation to climate change, but also has significant implications for accurately assessing vegetation productivity and global carbon balance.

Currently, the rapid development of remote sensing technology provides substantial technological tools for the study of vegetation phenology on regional and larger scales. Vegetation indices are frequently employed as the main remote sensing data source for determining vegetation phenological parameters because of their great sensitivity to canopy and chlorophyll content [8,9]. Among studies of vegetation phenology, investigations into start of growing season (SOS), end of growing season (EOS) and length of growing season (LOS) have gained increased attention, and the significant effects of phenological shifts on vegetation productivity have been revealed. For instance, warming temperatures in the Northern Hemisphere have been linked to advanced SOS and delayed EOS, leading to enhanced productivity in both the early and late growing seasons, respectively [10,11,12]. Research conducted in the Northern Hemisphere indicates that the LOS has a greater impact on vegetation productivity [13] than either SOS or EOS do. Despite these findings, it is still apparent that the spatiotemporal relationship between productivity and phenology varies significantly across different climatic zones and types of ecosystem. For example, Xiao et al. [14] found changes in SOS were the dominant factor influencing variations in vegetation net primary productivity for both alpine meadows and steppes. A significant negative correlation was also observed between EOS and gross primary productivity (GPP) in the temperate grasslands of northern China [15]. Delaying the SOS would encourage the growth of plants in some areas with limited water resources, as evidenced by the positive correlation found between the two variables [16]. Though existing research has extensively explored the influence of spring and autumn phenology on productivity, a critical gap remains in our understanding of the mechanisms and spatiotemporal patterns in peak of season (POS) and vegetation growth amplitude, and their subsequent impacts on overall productivity. The POS indicates the time at which vegetation photosynthetic activity reaches its maximum value—usually in summer for most northern ecosystems [17,18]. Vegetation growth amplitude is another summer phenological metric that captures the peak of vegetation growth. This indicator represents the greatest potential for seasonal photosynthetic activity and is commonly quantified using the vegetation indices or GPP [19]. Since the vegetation POS and peak growth represent the date and extent of the year when vegetation has the best available growth resources, changes to these will further affect the carbon budget and seasonal dynamics of the carbon cycle throughout the year [20,21].

Understanding the vegetation POS and peak growth variations is crucial for deciphering vegetation dynamics and ecosystem responses to environmental changes. Generally, temperature and precipitation have a significant impact on the variability of phenological metrics [22,23]. For instance, Park et al. [24] discovered that there was a broad advance in POS in northern lands, and that this was linked to a rise in GPP as a result of improved carbon uptake early in the growing season. Liu et al. [25] reported that precipitation had the greatest impact on the maximum of vegetation index, while temperature had the largest effect on POS change in the continental United States. In addition, different vegetation types exhibit varying sensitivities to temperature and precipitation for POS. Hai and Bao [26] found that precipitation was the main factor influencing POS in desert and forest ecosystems, whereas temperature played a more dominant role in regulating the POS for grasslands. Studies have focused mainly on the effects of the climate, ignoring the innate rhythms within vegetation (i.e., phenological variables). These internal biological rhythms, which govern growth and development stages, are also important in determining the POS and peak growth [27,28]. To address this gap, a thorough knowledge of the dynamics of phenological metrics under changing environmental conditions will integrate both climatic parameters and vegetation phenological cycles.

The Funiu Mountain region, which is situated in the area where the northern subtropical zone and the southern mild temperate zone converge in the Chinese mainland, is particularly vulnerable to climate change. Compared to other areas, climate change manifests differently in this region, leading to substantial alterations in vegetation phenology and productivity [29,30]. This study examined the relationship between vegetation productivity and peak phenological metrics (POS and peak growth), and then investigated the response of peak phenological features to climate factors and vegetation phenology in the Funiu Mountain region from 2000 to 2023. The objectives of this study were (1) to quantify the temporal and spatial variations of vegetation POS and peak growth and their relationship with vegetation productivity in this region; and (2) to assess the relative effects of climate and phenological variables on vegetation POS and peak growth. We hope our findings will contribute to a better understanding of the carbon cycle and phenological changes in the vegetation of the Funiu Mountain region and provide some valuable information about phenology forecasting in the field of climate change for other typical climatic transition zones.

2. Materials and Methods

2.1. Study Area

The Funiu Mountains are located in the western part of Henan Province (110°34′~113°14′E and 32°48′~34°23′N) (Figure 1a), which belongs to the eastern extension of the Qinling Mountain, with elevations ranging from 90 to 2090 m. The average temperature ranges from 7.2 to 16.0 °C, and the annual precipitation varies between 646 and 960 mm, with the greatest precipitation concentrated between May and September. The Funiu Mountains represent a typical transitional zone between the second and third terraces of Chinese topography, characterized by intricate landforms and notable changes in the surrounding natural environment. Approximately 68.5% of the research area is covered by forest vegetation (Figure 1b). With clear vertical changes in flora and climate, the vegetation type changes from a northern subtropical evergreen and deciduous mixed forest to a southern mild temperate deciduous broad-leaved forest [31]. The dominant soil types include brown and yellow-brown soil, with relatively shallow soil layers and varied terrains, indicating relatively poor site conditions [29].

2.2. Datasets Acquisition

The Enhanced Vegetation Index (EVI) data used in this study are sourced from MOD13Q1, featuring a 250-m spatial resolution and a 16-day temporal resolution, and covering the period from 2000 to 2023 (https://ladsweb.modaps.eosdis.nasa.gov/search/, accessed on 3 April 2024). The Global 30 m Fine-grained Land Cover Dynamic Monitoring Product (https://data.casearth.cn/, accessed on 6 April 2024), which comprises 29 categories of land cover and is updated every 5 years, provided the land use data for this investigation [32]. Given the extensive time span of the study (2000–2023), data products from 2010 were selected to analyze spatiotemporal variations in forest areas, with the intention to mitigate the influence of land use changes on the results. Climate data, including temperature and precipitation information, were extracted from the National Tibetan Plateau Data Center (https://data.tpdc.ac.cn/, accessed on 22 May 2024). The data, which cover the years 2000 to 2022 only, have a monthly temporal resolution and a geographical resolution of 1 km. Based on high-resolution global climate data from WorldClim and meteorological data from CRU, this dataset was created using the Delta spatial downscaling method. Its reliability was confirmed by independent weather stations [33,34]. To align all data to the same resolution, both the land cover and climate data were resampled to match the resolution of the MODIS data.

2.3. Vegetation Phenology Extraction

The MODIS Reprojection Tools software (version 4.1) was employed to convert the format and transform the projection of the image data. The EVI time series was then smoothed using the Savitzky-Golay filter (SG) in R software (version 4.3.2). The SG filtering method, known for noise elimination while preserving the original signal’s shape and width, has been widely utilized for data smoothing and noise reduction [35]. Based on the reconstructed EVI time series, a logistic function with seven parameters was fitted (Equation (1)) [8,36],

$$y(t,m)=m_1+(m_2-m_{7}t)({\displaystyle \frac{1}{1+e^{m_3-m_{4}t}}}+{\displaystyle \frac{1}{1+e^{m_5-m_{6}t}}})$$

(1)

where y(t, m) is the modeled EVI at time t, and m (m₁, m₂, …, m₇) are the fitting parameters. Specifically, m₁ and m₂ are the minimum and maximum value of EVI; m₇ is the summer greendown parameter; m_3,4 and m_5,6 are parameters that adjust the shape of the sigmoid growth curve in spring and autumn, respectively. The maximum and minimum values of the first order derivative were respectively defined as the SOS and EOS (Figure 2). The time point when the fitted EVI curve reaches its maximum value was defined as the POS. We quantified the vegetative productivity using the cumulative EVI during the SOS and EOS periods.

2.4. Statistics and Analysis

Linear least-squares regression is widely employed for trend change analysis [23,27]. Using this method, we computed the vegetation phenological metrics change rates at each pixel for the period 2000–2023 in the Funiu Mountain region as follows (Equation (2)):

$${\theta }_{s\mathrm{lope}}={\displaystyle \frac{n\sum ^{\underset{n}{i=1}}i\times Ph{e}_i-\stackrel{\underset{n}{i=1}}{\sum i}\sum ^{\underset{n}{i=1}}Ph{e}_i}{n\sum ^{\underset{n}{i=1}}{i}^2-{\left(\sum ^{\underset{n}{i=1}}i\right)}^2}}$$

(2)

where n represents the total number of years in this study, Phe denotes the phenological metrics for the i-th year, and θ_slope indicates the inter-annual change rate of phenological metrics. The positive value and negative value represent delayed (increased) and advanced (decreased) trends respectively in vegetation phenology metrics, while a value of 0 indicates no significant change in vegetation phenology. All the statistical significance was determined at the 95% confidence level in this research.

Since a variable may correlate with multiple others, partial correlation analysis is necessary to investigate the relationship between two specific variables while controlling for the influence of additional quantitative variables. For example, given variables X, Y, and Z, to determine the partial correlation between X and Y, the influence of Z must be controlled [25,37]. The calculation method is shown in Equation (3):

$$r_{XY•Z}={\displaystyle \frac{r_{XY}-r_{XZ}{r}_{YZ}}{\sqrt{(1-r_{XZ}^2)(1-r_{YZ}^2)}}}$$

(3)

where $r_{XY•Z}$ denotes the partial correlation coefficient between X and Y after controlling for variable Z, and $r_{XY}$, $r_{YZ}$ and $r_{XZ}$ represents the correlation coefficient between X and Y, Y and Z, X and Z, respectively. This study employed partial correlation analysis to investigate the influence of the SOS and preseason climatic factors on POS and EVI_max. Generally, the month with average POS and the preceding month were used to establish preseason temperature and precipitation.

3. Results

Temporal and Spatial Changes of Forest Phenology in Funiu Mountain

The forest POS in the Funiu Mountain region showed a notable advancing tendency between 2000 and 2023, advancing at a pace of 5.5 days per decade. The latest POS occurred in 2001, approximately on day 192, while the earliest POS took place in 2022, roughly on day 164 (Figure 3a). Over the course of the period of this study, EVI_max displayed a substantial increase, growing at a rate of 0.064 per decade. The EVI_max value ranged from a minimum of roughly 0.52 in 2000 to a maximum of 0.72 in 2021 (Figure 3b). As shown in Figure 3c, there was a notable negative correlation between forest POS and EVI_max in the Funiu Mountain area, suggesting that as EVI_max increased, the POS tended to advance.

Spatially, the average POS in the research region was 177 days, showing a delay from the periphery to the center. The POS typically emerged before 177 days in the northern and southern regions, while in the middle and northwest regions, it appeared later, with the latest appearance at around 210 days (Figure 4a). The average EVI_max of forest vegetation in the study area was 0.64, with a spatial pattern showing a decrease from the center to the periphery (Figure 4b). The high EVI_max values (higher than 0.7) were concentrated mainly in the central and southwestern regions, whereas the low values (less than 0.6) were found mainly in the northern and northwest regions. In the Funiu Mountain area, the POS of forest vegetation showed a primarily advancing trend (88.38%, Figure 4c). About 36.73% of the study area’s pixels had a notable advancement, and these pixels were found mainly in the study area’s western, eastern, and northern parts. Areas with non-significant advancement were concentrated primarily in the central region of the study area. Pixels with delayed POS were found mainly in the central part, comprising only 11.62% of the study area. For EVI_max, almost the entire study area displayed a significant increase—the pixels accounting for 93.97%. Only 0.65% of the pixels depicted a decrease in EVI_max (Figure 4d).

Partial correlation analysis results between the POS and the productivity of the forest in the Funiu Mountain area indicated a dominant earlier peak–larger productivity pattern across 68.13% of the study area (Figure 5a). Approximately 8.71% of the total study area was made up of regions—found primarily on the periphery of the study area—with a substantial negative correlation. Pixels with a positive correlation (31.87%) were found primarily in the central part of the study area, with significantly positively correlated pixels making up only 1.02%. The partial correlation between vegetation EVI_max and productivity was mainly positive (96.28%) (Figure 5b), with regions of significant positive correlation covering about 54.57% of the study area, and located primarily in the northwestern and northeastern parts. Areas with a negative correlation were relatively small, accounting for only 3.72%, and were scattered across parts of the southern region of the study area.

The forest POS in the Funiu Mountain area exhibited predominantly positive partial correlations with SOS, temperature, and precipitation. For 98.05% of the research region, there was a substantial positive correlation between the SOS and forest POS. The western and central regions, which made up 70.88% of the overall area, were notably home to the most positive correlations (Figure 6a). About 56.85% of the research area, mostly in the northern, eastern, and southern regions, showed a positive correlation between forest POS and temperature (Figure 6b). Conversely, negative correlations were found mainly in the central and western regions, encompassing 43.15% of the total area. Specifically, only 6.52% of these pixels exhibited statistically significant negative correlations. Additionally, precipitation showed a primarily positive link with forest POS, covering 67.43% of the research region. The majority of these favorable relationships were found in the central, northern, and southern regions (Figure 6c). Negative correlations were found in 32.57% of the area, mostly in the northwest and southeast. Notably, both positive and negative correlation zones had a low percentage of statistically significant pixels.

Analysis of the spatial distribution of partial correlation coefficients between SOS and EVI_max revealed relatively equal quantities of positive and negative correlations, each covering about 50% of the research region (Figure 6d). SOS and EVI_max had negative correlations mostly in the northwest, northern, and eastern parts of the Funiu Mountain region, while only 4.75% of the research area had significantly positive associations. Positively correlated pixels were distributed mainly in the southern and northern parts of the study area, with significantly positively correlated areas accounting for 3.78% of the study area. For most of the study area (59.61%), EVI_max and temperature showed a negative correlation, especially in the northern and eastern parts (Figure 6e). By contrast, the remaining 40.39% of the region, located mainly in the southern, northwest, and northern regions, had positive correlations. Precipitation exhibited a negative correlation with EVI_max for approximately 90.21% of the pixels (Figure 6f), covering almost the entire study area. Only 9.79% of the research area’s pixels, mainly in the northern and northwest regions, were positively correlated.

Overall, changes in POS within the study area were influenced mainly by SOS, accounting for 87.4% of the total study area. These pixels were mostly evenly distributed throughout the entire region (Figure 7a). Pixels where POS changes were influenced by temperature and precipitation were mainly in the central, northern, and eastern parts of the Funiu Mountain. EVI_max, was controlled mainly by precipitation and SOS, with the areas of these pixels accounting for 47.7% and 38.0%, respectively. The region where precipitation predominantly controls EVI_max was mainly in the northern part of the study area, while pixels primarily influenced by SOS were mainly in the northwest and southern parts. Temperature had a limited influence on EVI_max changes, with only 14.3% of pixels in the northern and western regions exhibiting this effect.

4. Discussion

4.1. Phenological Metrics’ Trends and Their Impact on Vegetation Productivity

Based on the MODIS vegetation index, our results demonstrate that the EVI_max in the Funiu Mountain area exhibited substantial growth over the past two decades, with a noticeable improvement in vegetation growth during the summer. This result is consistent with prior EVI/NDVI-based research [30,37]. Additionally, we discovered that the advancement of POS coincided with the rise in EVI_max. The augmentation of vegetation production or sequestration of carbon was supported by the spatiotemporal variations of both components in this study. As shown in Figure 8, the shape differences of the mean EVI time series curves for different periods indicate that the increase in EVI_max and the advancement of POS have both contributed to an increase in the integral value of the EVI time series curve, which reflects enhanced productivity. Similar results have been reported in previous research. For example, studies at flux sites and regional scales have reported strong positive correlations between GPP_max and annual GPP [38,39], the main reason for this being that GPP_max greatly influences annual GPP [40,41], which means that raising GPP_max can raise total vegetation production. An earlier POS promotes vegetation carbon assimilation during the early growing season compared to a later POS, ultimately leading to increased productivity [42]. However, arid regions usually exhibit an opposite trend. Studies have showed that a delayed POS, coupled with a decrease GPP_max, leads to a reduction in vegetation productivity [21,43]. This phenomenon is probably caused by a decrease in GPP_max, which generates less vigorous growth later in the growing season and brings an early end to autumnal phenology [44,45].

4.2. Exploring Drivers of Regulating Observed Phonological Metrics

In this research, a partial correlation approach was used to evaluate the impact of climate variables, SOS, and their interactions on POS and EVI_max. Our findings reveal that SOS and POS positively correlated, which is consistent with observations for various vegetation types in the Northern Hemisphere [19]. Unlike in instances of preseason temperature and precipitation, SOS had a stronger correlation with POS. The changes in POS in the study area were influenced mainly by SOS, indicating that climatic factors may not be the decisive factor for changes in vegetation POS. Instead, the intrinsic genetic types of vegetation might be the true determinants of POS changes [20]. This finding further supports the notion that earlier events in the plant life cycle influence subsequent stages [28]. Recent research has corroborated this view, highlighting the stronger influence of spring vegetation growth (i.e., SOS) on summer and autumn vegetation dynamics than of climatic factors [27,46]. This dominant effect of preceding stages on later stages, coupled with the tight connection between growth and development, might explain the observed dominance of SOS in influencing POS trends [47]. A positive relationship was found between overall POS and both temperature and precipitation. This points to favorable preseason hydrothermal conditions possibly postponing POS, most likely because of the promotion of persistent plant growth. However, this discovery appears to contradict observations by Wang and Wu [42], who report an earlier POS related to rising temperatures in Northeast and South China. This discrepancy could be attributed to plant phenological strategies—an earlier POS could be a mechanism to escape the potential negative effects of summer heat stress [17].

An examination of regional correlations between SOS and EVI_max indicated somewhat similar proportions of positive and negative connections throughout the research area. In the northwestern region, an earlier SOS was associated with enhanced peak vegetation growth. This could be attributed to improved water and heat conditions during this period, which are more favorable for stimulating growth and maximizing productivity [48,49]. However, in the southern and northern regions, an earlier SOS did not result in higher EVI_max. This was most likely a purposeful adaptation to avoid early soil moisture depletion, ensure enough water availability for summer development, and reduce potential water stress [18]. Preseason temperature exhibited a generally negative correlation with EVI_max, suggesting that rising temperatures might not be conducive to achieving maximum canopy photosynthetic potential. While appropriate temperature and precipitation generally facilitated vegetation photosynthesis, excessively high preseason temperatures could cause moisture deficiencies, limiting vegetative activity and lowering productivity. The study also found that preseason precipitation had a negative link with EVI_max, indicating that an increased precipitation could lead to a decrease in EVI_max. This might be due to increased cloud cover associated with higher precipitation, which restricts solar radiation and consequently limits CO₂ absorption during photosynthesis, ultimately inhibiting vegetation growth [50,51,52].

4.3. Limitations and Future Work Perspective

Despite the promising results obtained in this study, it is vital to recognize several limitations. Firstly, the phenological metrics extracted–based on EVI and GPP–may differ. According to previous research, the seasonal variation patterns of the vegetation GPP curve match those of the vegetation index curve [53,54]. Some studies have utilized the integrated or cumulative values of vegetation indices during the growing season as proxies for vegetation productivity to investigate spatiotemporal changes and responses to driving factors [55,56]. However, there are still some differences in the peak phenological metrics retrieved from GPP and vegetation indices. For instance, studies have shown that both GPP_max and the maximum vegetation index value were strongly linked, regardless of vegetation type or inter-annual variations [17,57]. Nevertheless, the date of GPP_max occurrence differed from the peak date of the vegetation index, appearing either earlier (corresponding to NDVI_max) or later (corresponding to EVI_max) [58,59]. This discrepancy had a direct impact on the explanatory capacity of phenological parameters for vegetation GPP [60]. Secondly, when examining the factors influencing POS and EVI_max, we introduced the variable SOS. The results emphasized the crucial role of biological rhythms rather than climatic factors in regulating POS variations, perhaps improving our knowledge of carbon dynamics. However, variations in the phenological metrics can also be caused by solar radiation, nitrogen deposition, soil moisture, and the fertilization effect of CO₂, in addition to the SOS and meteorological elements (temperature and precipitation) taken into account in this work [4,57,61]. Future research could investigate the components that contribute to peak phenological features of vegetation in greater detail.

5. Conclusions

Based on MODIS EVI data from 2000 to 2023, this study extracted peak phenology metrics of forest vegetation in the Funiu Mountain region and examined their spatiotemporal fluctuations and links to vegetation productivity. Partial correlation analysis was then performed to investigate the links between SOS, preseason temperature and precipitation, and peak phenology traits. The results indicate that the advancement of POS was accompanied by an increase in EVI_max, and both variations could enhance vegetation productivity. Furthermore, the study showed that among SOS, preseason temperature, and precipitation, SOS had the greatest impact on POS variation. This was due mostly to the innate rhythm of the vegetation, which influences subsequent growth. The variation in EVI_max in the study area was negatively correlated with precipitation. This could be because severe precipitation caused an increase in cloud cover, which limited photosynthesis in the vegetation. Our results should be helpful in understanding the relationship between forest phenology and climate change in China’s climate transition zone.