1. Introduction

The distribution of tree species in mountain regions is shifting in response to rapid climate warming (Lenoir et al., 2008; Chen et al., 2011; Lu et al., 2021). Disturbances shift forest composition and enhance forest dynamics by creating new habitats for regeneration or by establishing and altering connectivity for species movement (Seidl et al., 2017; Guo et al., 2018). Moreover, disturbances amplify forest responses to climate change, as has been observed in subalpine forests in the Spanish Pyrenees (Camarero et al., 2000) and in temperate–boreal forest ecotones in North America (Liang et al., 2018; Brice et al., 2020). Consequently, changes in forest composition may be better explained by disturbances than by climate change (Danneyrolles et al., 2019; Yu et al., 2021). In addition, disturbances (e.g., wildfire) have been shown to cause abrupt transitions in foundation tree species in some mountain forests (Calder and Shuman, 2017; Hansen and Turner, 2019; Falk et al., 2022). However, it is unclear how anthropogenic disturbances modify long-term changes of mountain forest composition under contemporary climate warming conditions.

The Qinghai-Tibet Plateau (QTP), a key area for global change and multi-sphere interaction studies, is best known for its high elevations, harsh climate and heterogeneous topography (Piao et al., 2019; Chen et al., 2021). The southeastern QTP, which is affected by summer monsoons, holds one of the largest forest areas in China (Li, 1985). However, responses of these subalpine forests to disturbances during the recent climate-warming epoch are still little understood (but see Wang et al., 2011; Lu et al., 2019). Before 1998, these forests experienced extensive logging at lower elevations (Zhao and Shao, 2002). In addition, wildfire frequency has increased in these forests over the past decades (Tian et al., 2007). These logging and fire disturbances provide an excellent natural experiment to test how disturbances modify forest composition and distribution on the southeastern QTP.

The objective of this study is to determine how the distribution and composition of subalpine forests respond to disturbances and climate warming on the southeastern QTP. Long-term forest inventory data and onsite field survey records were used to track the variations of area and biomass of different functional groups (spruce-fir, pine and broadleaved forests). Previous plot surveys (Lu et al., 2019) led us to hypothesize that increasing dominance of warm-adapted broadleaved tree species would thrive after disturbances and in response to contemporary climate warming.

2. Materials and methods 2.1. Study area and climate

The study area is situated in the southeastern part of the Xizang Autonomous Region of China, where forests account for more than 90% of the total forest area of the whole province (China forest resources reports 2019). Forests across the region mainly consist of spruce-fir (evergreen dark conifers: Abies, Picea, Tsuga), pine-larch mixed forests (Pinus, Larix) and broadleaved forests (Quercus, Betula, Populus, and Salix). Alpine treelines are located at 4250–4900 m a.s.l (Miehe et al., 2007; Liang et al., 2011; Lyu et al., 2019). The natural spruce-fir forests account for 33% of all forests in the region, often occupying the subalpine regions with a narrower niche breadth. Pine and broadleaved forests are dominant at lower elevations, where logging disturbances were common until 1998. In addition, fire frequency in this region has increased since 2001, especially in the dry spring and winter seasons (Tian et al., 2007). According to the China Forestry Statistical Yearbook, the burnt area in Xizang was about 1.7 × 10⁴ hm² during the past 40 years (1981–2010), accounting for approximately 0.1% of the total forest area. Moreover, the logging area before the appliance of the Natural Forest Protection Project accounted for 0.74% of the total forest area (about 1.3 × 10⁵ hm²; Zhang and Li, 2001).

The study area is characterized by a monsoon climate with a warm, humid summer and a cold, dry winter. Weather stations located in the study area (period of records, 1980–2018) indicate that January and July are the coldest and warmest months with mean temperatures of −0.99 ℃ and 15.97 ℃, respectively. The mean annual precipitation is about 600 mm, most of which falls during the monsoon season from June to September. Long-term (1961–2018) variations in the mean air temperature and precipitation during growing season (May to September) at Linzhi station (29.646°N, 94.363°E, 3004 m a.s.l.) were also considered and analyzed.

2.2. Forest inventory data analyses

The national forest inventory data were published in China Forest Resources Reports (e.g., China forest resources reports, 2019) which were updated each four or five years for each province or autonomous region. In this study, the data spanning nine 4- to 5-year long periods (1973–1976, 1977–1981, 1984–1988, 1989–1993, 1994–1998, 1999–2003, 2004–2008, 2009–2013, and 2014–2018) were collected and analyzed. Within the data, the species composition, and the corresponding stand age, distribution area and stem volume were all documented, which were systematically estimated based on 3145 forest gridded inventory plots. We divided the existing 19 species or groups (e.g., Abies, Picea, Tsuga, Larix, Pinus armandii, Pinus yunnanensis, Pinus densata, Pinus wallichiana, Quercus, Betula, Populus, Salix, and other soft broadleaved trees) into three aforementioned functional groups according to leaf habit: spruce-fir, pine-larch and broadleaved forests. Changes in distribution area and biomass for spruce-fir, pine-larch and broadleaved forests were then analyzed. The biomass expansion factor for different forest types and stand ages (young, middle-aged, near-mature, mature, and old forest) was used to convert timber volume to forest biomass (Fang et al., 2001).

2.3. Field investigations

In October 2018, we surveyed five plots (30 m × 30 m, 3711–3865 m a.s.l.) situated at different elevations in burned fir forests in the Sygera Mountains, southeastern QTP. Late-successional forest communities of subalpine forest are dominated by fir (Abies georgei var. smithii) in these mountains. The plots were placed along the maximum slope and the lower-left corner of each plot was set as an original point (coordinates x, y = 0, 0). Dendrochronological analyses indicated that this forest patch was burned by wildfire around 1942 (see Lu et al., 2019). At each plot, we measured the elevation and geographical coordinates of all trees by using GPS and recorded tree species. Cores, containing or close to the pith, were collected by using an increment borer at the base of adult trees. Cores were then air dried, progressively sanded in the lab and the ages of adult trees were obtained by counting the number of rings, which were visually cross-dated under the stereomicroscope. For young seedlings and saplings, we estimated their ages by counting the bud scars along the main stems. Post-fire recruitment series in 5-year resolution were developed for different tree species.

Soil temperature at 10 cm depth was recorded by data loggers equipped with a thermistor probe (Tidbit v.2, Onset Computer Corp., USA) at one of the burnt sites and a nearby undisturbed fir forest since October 1st, 2020.

2.4. Maximum entropy (MaxEnt) model simulation

We used the maximum entropy (MaxEnt) modeling, a robust species distribution modeling approach, to predict habitat suitability of forests (Phillips et al., 2006; Elith and Leathwick, 2009). This framework has multiple advantages, performing well for presence-only data, incomplete data, small sample sizes and gaps. The latitude and longitude of 175 forest sites (110 evergreen needle-leaved forests and 65 broadleaved forests; Fig. S1) across the southeastern QTP, as well as the corresponding 19 environmental variables downloaded from the WorldClim-Global Climate Data (http://www.worldclim.org/bioclim), were input in Maxent v.3.3.3 (Phillips et al., 2006; Philips and Dudík, 2008). Analysis of receiver operating characteristic (ROC) was applied for the accuracy test. The distribution of broadleaved and evergreen needle-leaved forests was predicted under current climate and future emission scenarios of four representative concentration pathways (RCP2.6, RCP4.5, RCP6.0, and RCP8.5) in the 2070s. More details about the application of the MaxEnt model in the study area can be found elsewhere (Ma and Sun 2018; Naudiyal et al., 2021).

2.5. Statistical analyses

The temporal variations of growing-season air temperature and precipitation at Linzhi were assessed using linear regressions. To detect shifts in forest types, differences between the area and biomass of spruce-fir, pine and broadleaved forests in 1973–1998 and 1999–2018 were assessed using Student's t test. These analyses were done using SPSS 15.0 (SPSS Inc., Chicago, USA).

3. Results 3.1. Climate variability

The southeastern QTP has been experiencing a warmer and drier climate over recent decades. The mean air temperature during the growing season (May to September) has significantly increased (Fig. S2; R² = 0.67, p < 0.05), while precipitation did not show a significant linear trend (Fig. S2; R² = 0.02, p > 0.05) between the years 1960–2018.

Soil temperature largely differs between the burnt fir site and the nearby unburnt forest. Soil temperature at the burnt site was about +2.7 ℃ higher during the growing season and +3.7 ℃ higher in winter than that of the natural forest (Fig. S3).

3.2. Shifts of forest structure on the southeastern Qinghai-Tibet Plateau

The area and biomass of forest on the southeastern QTP has increased over the past 50 years (Fig. 1). However, a rapid shift has occurred in forest type since 1999, characterized by increasing broadleaved and pine forests in association with a reduction of spruce-fir forests (Figs. 1 and 2). Spruce-fir subalpine forests accounted for the largest area and biomass before 1999. However, their area and biomass declined by 34.6% and 32.9% between 1999 and 2003. In contrast, the area and biomass of pine forests increased by 80.44% and 166.67%, while those of broadleaved forests increased by 85.37% and 77.5%, respectively. As a result, both total forest area and biomass increased in response to the shift from spruce-fir forests to pine-broadleaved forests after 1998 (Fig. 2).

3.3. Fire-induced shifts of forest structure in the Sygera Mountains as a case study

Severe fire events reshaped post-disturbance forest succession. As shown by the field survey, the post-fire recruitment of the burnt fir forest mainly included oak (Quercus aquifolides), spruce (Picea likiangensis var. linzhiensis) and juniper (Juniperus squamata) (Fig. 3). Moreover, oak recruitment rapidly increased, especially at low elevations.

3.4. Forecasting the distribution of broadleaved and needle-leaved forests under different climate warming scenarios

Future warming is expected to favor an expansion of broadleaved forests over needle-leaved forests. As shown by the model output, the habitat suitability for both broadleaved and evergreen needle-leaved forests will likely increase towards the 2070s under the four climate scenarios (Fig. 4). Forecasted changes in the suitable area (0.4 < habitat suitability < 1.0) are greater for broadleaved forests than for evergreen needle-leaved forests (RCP2.6, 164.89% vs. 24.98%; RCP4.5, 168.52% vs. 49.07%, RCP6.0, 173.87% vs. 45.03%; RCP8.5, 152.58% vs. 45.43%; Table S1).

4. Discussion

This study provides robust evidence in support of our hypothesis that anthropogenic disturbances modify the structure and composition of subalpine forests. Disturbances seem to amplify responses of subalpine forests to climate change, as shown by increasing dominance of broadleaved tree species in recent decades. Moreover, disturbances can reduce turnover and convergence time for such a transition (Brice et al., 2020). Nonetheless, we cannot completely rule out other drivers, such as climate warming. The influence of climate change was supported by our model projection as well as observed changes in subalpine forests on the QTP and over inner East Asia (Liang et al., 2016; Zhang et al., 2020; Liu et al., 2021). Temperate–boreal ecotones of eastern North America also provide evidence that transitions of forest type are mainly driven by disturbances and secondarily by climate (Brice et al., 2020).

Regional forest transitions increase forest carbon sequestration. Global total forest area has declined by 3% between 1990 and 2015 (Keenan et al., 2015). However, the southeastern QTP has undergone a net forest expansion since 1973 (Fig. 1). Despite intense logging practices and wildfires in natural spruce-fir forests in past decades (Zhao and Shao, 2002), natural tree regeneration and plantations quickly offset the loss of forest area and biomass. Consequently, there was a net forest gain after the period of 1994–1998. Such a forest transition coincided with the start of the Natural Forest Conservation Program of China, which indicates that nature-based solution frameworks are needed for forest ecosystem sustainability (Liu et al., 2022).

4.1. Disturbances mediated the shift of forest structure and composition

On the southeastern QTP, the area and biomass of broadleaved forests have surpassed those of spruce-fir forests since 1998. As inferred from the plot-based survey, such a transition was likely triggered by disturbances such as wildfire. Disturbance intensity may have profound impacts on post-disturbance successional trajectories. Low or moderate disturbances may not fundamentally change forest structure, allowing forests to fully recovery within decades to hundreds of years (Turner et al., 2019; Fu et al., 2020). In contrast, severe disturbances may trigger forest conversion by shifting species dominance (Calder and Shuman, 2017; Mekonnen et al., 2019; Xu et al., 2022). In particular, severe climate-mediated disturbances may even exceed the threshold or tipping point of forest resilience (Scheffer et al., 2012; Falk et al., 2022), hence leading to changes in forest types (such as from coniferous to broadleaved forests) and even from forest to shrubland in high elevation mountain ranges or very dry sites (Allen and Breshears, 1998; McIntyre et al., 2015; Millar and Stephenson, 2015; Xu et al., 2022).

After intensified disturbance such as fire, late-successional boreal conifer stands shift into broadleaved forests often dominated by pioneer species such as birch (Barrett et al., 2011; Hansen and Turner, 2019; Dulamsuren, 2021). Given the relatively slow process rate and long recovery time of spruce-fir forest (Fu et al., 2020; Maliniemi and Virtanen, 2021), disturbances can destabilize the system and shift their compositional trajectories. In addition, the clearing of the tree layer strongly modifies microclimate and soil conditions, which may alter the germination of seeds and the survival rates of tree seedlings (De Frenne et al., 2019). This type of habitat modification and consequent release in competition may facilitate the establishment of competitive species, such as shade-intolerant pine and oak species in our case. Our long-term forest inventory data also support that conclusion that land-use change interacts with climate to speed up elevational species redistribution (Guo et al., 2018). Yet, more research is required to investigate the mechanisms behind such distributional changes.

4.2. Consistent expansion of forest distribution under climate warming?

Climate warming appears to be reshuffling the structure and demography of subalpine forests on the southeastern QTP. Simulations based on RCP2.6 and RCP4.5 scenarios indicate that subalpine conifer forests on the eastern QTP will expand from 2016 to 2096; in contrast, simulations using the RCP8.5 scenario indicate that these forests will begin to decrease around the same time (Liu et al., 2021). In fact, climate warming is triggering an upward shift of alpine fir treelines on the southeastern QTP (Liang et al., 2016). This field result is consistent with our simulations that the suitable area tends to decline under more extreme warming scenarios (evergreen needle-leaved forests at RCP4.5, broadleaved forests at RCP6.0). However, broadleaved trees seem to be more favored by climate warming and succession than subalpine conifers, which are showing growth declines due to stand encroachment (Cao et al., 2021).

5. Conclusions

Our findings show that disturbances and climate warming have caused subalpine forests distribution on the southeastern QTP to expand and altered their composition. Disturbances can reduce competition and create habitats for species immigration, facilitating gains of warm-adapted broadleaved and pine species over spruce-fir forest. Post-disturbance decreases in the biomass of spruce-fir forests demonstrates the need to prioritize the conservation of large and intact natural spruce-fir forest areas, strongly supporting the timely efforts of Natural Forest Conservation Programs. However, the impacts of anthropogenic disturbances (e.g., logging and wildfire) on some forest types should be considered in developing sustainable management plans. To this end, this research shows the need for a new strategy of adaptive forest management and conservation, such as assisted post-disturbance successional trajectories.

Acknowledgments

This research work was supported by the National Natural Science Foundation of China (42030508), the Second Tibetan Plateau Scientific Expedition and Research (STEP) program (2019QZKK0301), and the Key technology research and development projects in Xizang Autonomous Regions (XZ202101ZY0005G). We are grateful to the staff in Southeast Tibet Observation and Research Station for the Alpine Environment of CAS and those in the Team of Ecosystem Patterns and Processes (EPP) for the field work and computer analysis.

Authorship contribution

LZ and EL designed the study, LZ, XL, HZ, SG and JS collected the data. All authors contributed to data analysis, interpretation, editing and writing of the manuscript.

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.2023.03.002.