Forest crown density restoration and influencing factors in the burned area of northern Great Hing′an Mountains of China

Fire in the Great Hing′an Mountains in 1987 affected an area of more than 1.33×10⁶ hm², creating a mosaic of burn severities across the landscape, which strongly affected the postfire vegetation succession. In addition, undulate landform and anthropogenic disturbance inevitably influenced the postfire vegetation succession. In this paper, a typical area was selected for a case study, including two forest farms, covering more than 1.2×10⁵ hm². In order to reveal how the forest changed in 2000 (13 years after the fire) by comparing with 1987 (prefire) and to find out the relationship between the forest succession and the affecting factors, forest crown density was selected as the criterion, and forest type, fire severity, silviculture practice, elevation and topography gradients were designed as the affecting variables. With the support of GIS software, each variable was classified and entered into the multivariate regression model. The result showed that the forest crown density changed notably in 2000 compared with that of the prefire, and all the variables significantly affected the forest crown density. The most important affecting variable was elevation, which was positively correlated with the forest crown density. The next was fire severity, which was negatively related with the forest succession. The effects of topographic factors and silviculture practices on forest crown density were relatively small.     

Human intervened post-fire forest restoration in the Northern Great Hing’an Mountains: a review

The devastating fire in May 1987 in the northern Great Hing’an Mountains created a mosaic of burned severity. Subsequent log harvesting and tree planting complicated the restoration process. Based on intensive field work and GIS analysis for the burned area, we studied the landscape pattern change in relation with its influencing factors, the restoration of some ecosystem functions and the long-term effect of human planting on vegetation restoration. A post-fire vegetation restoration process was also established using spatial series instead of temporal series. The results indicated that coniferous forest, broad-leaved forest and mixed forest increased obviously in the burned area. Factors influencing the restoration process are ranked as the burned severity, way of restoration (planting, promoted restoration or natural restoration) and topographical factors. The latter was further ranked as the slope, elevation, slope position and aspect. Primary productivity, hydrological functions, habitats for wild animals and permanent frozen soil of the area have all largely been restored to the pre-fire level after more than 15 years. Parameters demonstrated a transitional character of the forest from the early succession stage to middle stage. LANDIS simulation for the long-term forest succession under pure natural restoration and human-intervened restoration indicated that post-fire tree planting largely influenced the age structure, spatial pattern and timber stock of dominant species such as Larix gmelini, Betula platyphylla and Pinus sylvestris var. mongolica. In general, the influence of post-fire human planting can remain for more than 200 years.     

Influencing factors on early vegetation restoration in burned area of Pinus pumila – Larch forest

Pinus pumila (Dall.) Regel, a rare conifer and key species in high latitude and high altitude mountains, has an important role in soil and water conservation. This evergreen shrub grows 3–6 m high in P. pumila – larch (Larix gmelini) open forest at altitudes of 800–1200 m in the Greater Kingan Mountain Range (Daxing’an Mountain). Forest fires are major natural hazards to P. pumila – larch forest. The unique ecological role of this community gives important theoretical and practical significance to research on P. pumila – larch forest restoration after fires. Literature concerning factors influencing early vegetation restoration in burned areas in this habitat is sparse. We studied these factors, especially those related to P. pumila seedling establishment. The results showed fires in P. pumila – larch forest usually resulted in severe burns. Typically almost no P. pumila survived after fires. Nearly all ground fuels were consumed. Second growth after fires exhibited low species richness. The dominant tree/shrub seedlings found after fires were birch (Betula platyphylla) and larch, with small number of P. pumila. Other shrub seedlings found were Ledum palustre, Vaccinium vitis-idaea, Betula fruticosa, and Rubus arcticus. The main herb species found were Deyeuxia langsdorffii, and Chamaenerion angustifolium. Important factors influencing early vegetation restoration after fires included seed dispersal, fire size, and site condition. Seed dispersal in birch and larch is higher than in P. pumila; more seedlings of birch and larch were found in burned areas than seedlings of P. pumila. Most seeds germinated in the first year following a fire. The extent of the burned area influences seedling distribution patterns, especially in species with limited seed dispersal ability. Birch and larch seedlings were evenly distributed in the entire burned area, while seedlings of P. pumila were found only at the fire edge. No P. pumila seedlings were found more than 50 m away from seed source trees. Site condition significantly influenced seed germination and growth in birch and larch; these seedlings only grew well in burned areas with good site conditions (shallow slopes, thick soils, etc.). They did not grow well in burned areas with poor site conditions (steep slopes, thin soils, etc.). However, P. pumila seedlings could grow well in burned areas with either good or poor site conditions. The strong vitality of P. pumila seedlings gives this species an enormous ecological advantage in soil conservation and environmental restoration and conservation. We conclude that the main factor influencing seedling establishment in P. pumila is its weak seed dispersal ability. Although the P. pumila seeds can germinate in all burned areas, natural regeneration rarely restores burned areas to the original P. pumila – larch forest. Planting seeds and/or seedlings may facilitate burned area restoration to P. pumila – larch forest. Because P. pumila seedlings grow very slowly, the restoration process may take decades.     

Quantitative classification and carbon density of the forest vegetation in Lüliang Mountains of China

Forests play a major role in global carbon (C) cycle, and the carbon density (CD) could reflect its ecological function of C sequestration. Study on the CD of different forest types on a community scale is crucial to characterize in depth the capacity of forest C sequestration. In this study, based on the forest inventory data of 168 field plots in the study area (E 111°30′–113°50′, N 37°30′–39°40′), the forest vegetation was classified by using quantitative method (TWINSPAN); the living biomass of trees was estimated using the volume-derived method; the CD of different forest types was estimated from the biomass of their tree species; and the effects of biotic and abiotic factors on CD were studied using a multiple linear regression analysis. The results show that the forest vegetation in this region could be classified into 9 forest formations. The average CD of the 9 forest formations was 32.09 Mg ha⁻¹ in 2000 and 33.86 Mg ha⁻¹ in 2005. Form. Picea meyeri had the highest CD (56.48 Mg ha⁻¹), and Form. Quercus liaotungensis + Acer mono had the lowest CD (16.14 Mg ha⁻¹). Pre-mature forests and mature forests were very important stages in C sequestration among four age classes in these formations. Forest densities, average age of forest stand, and elevation had positive relationships with forest CD, while slope location had negative correlation with forest CD.     

Predicting the wetland distributions under climate warming in the Great Xing’an Mountains,northeastern China

The wetland ecosystem is particularly vulnerable to hydrological and climate changes. The Great Xing’an Mountain is such a region in China that has a large area of wetlands with rare human disturbance. The predictions of the global circulation model CGCM3 (the third-generation coupled global climate model from the Canadian Centre for Climate Modeling and Analysis) indicated that the temperature in The Great Xing’an Mountain will rise by 2–4°C over the next 100 years. This paper predicts the potential distributions of wetlands in this area under the current and warming climate conditions. This predication was performed by the Random Forests model, with 18 environmental variables, which will reflect the climate and topography conditions. The model has been proven to have a great prediction ability. The wetland distributions are primarily topography-driven in the Great Xing’an Mountains. Mean annual temperature, warmness index, and potential evapotranspiration ratio are the most important climatic factors in wetland distributions. The model predictions for three future climate scenarios show that the wetland area tends to decrease, and higher emission will also cause more drastic shrinkage of wetland distributions. About 30% of the wetland area will disappear by 2050. The area will decrease 62.47, 76.90, and 85.83%, respectively, under CGCM3-B1, CGCM3-A1B, and CGCM3-A2 by 2100. As for spatial allocation, wetlands may begin to disappear from the sides to the center and south to north under a warming climate. Under CGCM3-B1, the loss of wetlands may mainly occur in the south hills with flatter terrain, and some may occur in the north hills and intermontane plains. Under CGCM3-A1B, severe vanish of wetlands is predicted. Under CGCM3-A2, only a small area of wetlands may remain in the north of the high mountains.     

The role of plant–soil feedbacks and land-use legacies in restoration of a temperate steppe in northern China

Plant–soil feedbacks affect plant performance and plant community dynamics; however, little is known about their role in ecological restoration. Here, we studied plant–soil feedbacks in restoration of steppe vegetation after agricultural disturbance in northern China. First, we analyzed abiotic and biotic soil properties under mono-dominant plant patches in an old-field restoration site and in a ‘target’ steppe site. Second, we tested plant–soil feedbacks by growing plant species from these two sites on soils from con- and heterospecific origin. Soil properties generally did not differ between the old-field site and steppe site, but there were significant differences among mono-dominant plant patches within the sites. While soil species origin (i.e., the plant species beneath which the soil was collected) affected biomass of individual plant species in the feedback experiment, species-level plant–soil feedbacks were ‘neutral’. Soil site origin (old-field, steppe) significantly affected biomass of old-field and steppe species. For example, old-field species had higher biomass in old-field soils than in steppe soils, indicating a positive land-use legacy. However, soil site origin effects depended on the plant species beneath which the soils were collected. The predictive value of abiotic and biotic soil properties in explaining plant biomass differed between and within groups of old-field and steppe species. We conclude that the occurrence of positive land-use legacies for old-field species may retard successional replacement of old-field species by steppe species. However, high levels of idiosyncrasy in responses of old-field and steppe plant species to con- and heterospecific soils indicate interspecific variation in the extent to which soil legacies and plant–soil feedbacks control successional species replacements in Chinese steppe ecosystems.     

Distribution,population density,and status of sympatric cercopithecids in the Campo-Ma’an area,southwestern cameroon

A study on species composition, distribution, and population density of cercopithecids in the Campo-Ma’an area, Southwestern Cameroon, was undertaken from December 1997 until August 2000. A total of 665.5 km of line transects was used for the census. Thirteen diurnal primate species including five endangered species (Gorilla g. gorilla, Pan troglodytes, Mandrillus sphinx, Colobus satanas, Cercocebus torquatus) were recorded in the Campo Forest, the greatest part of which is a logging concession.Cercopithecus nictitans (1.43 groups/km²),C. cephus (1.13 groups/km²),C. pogonias (0.81 groups/km²), andC. torquatus (0.51 groups/km²) occurred at medium frequencies compared to figures from other Central African study sites. Mandrill densities estimated (0.27 groups/km²) show that the area is very important for the conservation of this rare species. Guenon densities found inside areas with a high level of human activities did not differ significantly from densities estimated in areas with a lower level of human activities.C. torquatus densities were significantly higher in the areas with a low level of human disturbance and encounter rates withLophocebus albigena also indicate a preference of less disturbed areas. Mangabeys are thus likely to be adversely affected by human activities in the area. In the Ma’an Forest, which has not been logged yet, ten species were confirmed. The population densities of two guenons (C. nictitans andC. cephus) were estimated to be twice as high in the unlogged forest area as compared to the logged forest of Campo. Other species are rarer in the Ma’an Forest than in the Campo Forest. Although mangabeys are adversely affected by human activities, the results still indicate that selective logging may be compatible with the conservation of cercopithecids, if a reduced damage logging concept and antipoaching measures are implemented. Increased hunting activities following logging operations will definitely have a negative longterm impact on primate populations in the Campo-Ma’an area if no further, more effective conservation measures will regulate wildlife use in future.     

Spatial pattern of soil organic carbon of the main forest soils and its influencing factors in Guangxi,China北大核心CSCD

Aims Our objectives were to study the spatial distribution of soil organic carbon (SOC) density and its influencing factors in the main forest ecosystems in Guangxi. Methods A total of 345 sample plots were established in Guangxi, and the size of each plot was 50 m × 20 m. Based on the forest resource inventory data and field investigation, the SOC storage of the main forests in Guangxi was estimated. Geostatistics was applied to analyze the spatial pattern of SOC density and the main influencing factors on SOC density were also explored by principal component analysis and stepwise regression. Important findings The total SOC storage in the main forests in Guangxi was 1 686.88 Tg, and the mean SOC density was 124.70 Mg•hm2, which is lower than that of China. The best fitted semivariogram model of SOC density was exponential model, and the spatial autocorrelation was medium. The contour map based on Kriging indicated that northeastern Guangxi had high SOC density and northwestern Guangxi had low SOC density, which corresponded to high SOC density in non-karst region and low SOC density in karst region. The SOC density followed the sequence of bamboo forest > deciduous broadleaf forest > warm coniferous forest > mixed evergreen and deciduous broadleaf forest > evergreen broadleaf forest, and yellow soil > red soil >lateritic red soil > limestone soil. The dominant environment factors affecting SOC density included soil depth, longitude, latitude, and altitude. Soil depth was the most influential factor, which was mainly attributed to the karst landscape.     

Macrobenthos functional groups as indicators of ecological restoration in the northern part of China’s Yellow River Delta Wetlands

Freshwater releases to restore degraded wetlands are a globally recognized way to maintain the biodiversity and enhance the health of wetland ecosystems. To better understand the efficacy of freshwater releases in the northern part of China’s Yellow River Delta Wetlands, we used macrobenthos functional groups in spring (before freshwater releases), summer (during), and autumn (after) as indicators of the ecological responses. We also created abundance–biomass comparison curves and analyzed secondary production of each trophic level to evaluate the magnitude of the disturbance of the macrobenthos community. Abundance, biomass, and biodiversity of macrobenthos functional groups generally improved after the freshwater releases. In contrast with an intertidal (reference) area, the macrobenthos community in the ecological restoration area tended to be freshwater species. In the ecological restoration area, strong and moderate ecological disturbance of the macrobenthos community was evident during and after freshwater releases because the abundance curve remained above the biomass curve. Secondary production was in the order trophic level III > II ≅ IV in the summer, which indicates fragility of the macrobenthos community. The ecological restoration area had the highest sediment total organic carbon and moisture contents, but the lowest salinity and median particle size, and these differences were statistically significant. Our results suggest that adaptive freshwater releases, including a long-term freshwater release plan that more closely emulates natural flows and increasing the efficiency of freshwater utilization, will be necessary to achieve sustainable management of the wetland’s ecosystem and reduce the disturbance caused by the freshwater releases.     

Forest Landscape Restoration: increasing the positive impacts of forest restoration or simply the area under tree cover?

Restoring forest landscapes is critical in the face of continued global forest loss and degradation. In this article, we explore some challenges underlying the delivery of global commitments to restore forest landscapes. We propose that three fundamental questions need to be resolved upfront for the effective implementation of Forest Landscape Restoration and related commitments: (1) What social and ecological landscape objectives are being sought through Forest Landscape Restoration? (2) How are specific areas being selected for restoration? (3) How is success measured when restoring forest landscapes? We believe that there is an urgent need to adequately answer these questions to successfully implement political commitments for large‐scale forest restoration.     

Are biotic factors significant in influencing the distribution of halophytes in saline habitats?

The influence of biotic factors on the distribution and establishment of halophytes is being considered in this review. Physicochemical factors, such as salinity and flooding, often are considered to be the determining factors controlling the establishment and zonational patterns of species in salt marsh and salt desert environments. Sharp boundaries commonly are found between halophyte communities even though there is a gradual change in the physicochemical environment, which indicates that biotic interactions may play a significant role in deterining the distribution pattern of species and the composition of zonal communities. Competition is hypothesized to play a key role in determining both the upper and lower limits of species distribution along a salinity gradient. Field and laboratory experiments indicate that the upper limits of distribution of halophytes into less saline or nonsaline habitats is often determined by competition. There appears to be a reciprocal relationship between the level of salt tolerance of species and their ability to compete with glycophytes in less saline habitats. Halophytes are not competitive in nonsaline habitats, but their competitive ability increases sharply in saline habitats. Allelopathic effects have been reported in salt desert habitats, but have not been reported along salinity gradients in salt marshes. Some species of halophytes that are salt accumulators have the ability to change soil chemistry. Chemical inhibition of intolerant species occurs when high concentrations of sodium are concentrated in the surface soils of salt desert plant communities that are dominated by salt-accumulating species. Establishment of less salt-tolerant species is inhibited in the vicinity of these salt-accumulating species. Herbivory is reported to cause both an increase and a decrease in plant diversity in salt marsh habitats. Heavy grazing is reported to eliminate sensitive species and produce a dense cover of graminoids in high marsh coastal habitats. However, in other marshes, grazing produced bare patches that allowed annuals and other low marsh species to invade upper marsh zonal communities. A retrogression in plant succession may occur in salt marshes and salt deserts because of heavy grazing. Intermediate levels of grazing by sheep, cattle, and horses could produce communities with the highest species richness and heterogeneity. Grazing by geese produced bare areas that had soils with higher salinity and lower soil moisture than vegetated areas, allowing only the more salt-tolerant species to persist. Removal of geese from areas by use of inclosures caused an increase in species richness in subarctic salt marshes. Invertebrate herbivores could also inhibit the survival of seeds and the ability of plants to establish in marshes. Parasites could play a significant role in determining the species composition of zonal communities, because uninfected rarer species are able to establish in the gaps produced by the death of parasitized species.     

Are two strategies better than one? Manipulation of seed density and soil community in an experimental prairie restoration

Restoration practitioners have a variety of practices to choose from when designing a restoration, and different strategies may address different goals. Knowledge of how to best use multiple strategies could improve restoration outcomes. Here, we examine two commonly suggested strategies in a single tallgrass prairie restoration experiment: increased forb sowing density and prairie soil inoculation. We designed a study with two different forb seeding densities. Within these densities, we transplanted seedlings into 1‐m² plots that had been grown in either a whole prairie soil inoculum or sterilized prairie soil. After 4 years, we found positive effects of both high forb sowing density and inoculation treatments on the ratio of seeded to nonseeded plant cover in these plots, and negative effects of both treatments on nonseeded plant diversity. No effects of either treatment were seen on seeded plant diversity. Each strategy also affected the plant community in different ways: high forb sowing density increased seeded forb richness and decreased native nonseeded plant cover, while inoculation decreased non‐native cover, and tended to increase average successional stage of the community. These effects on restoration outcome were typically independent of each other, with the result that plots with both manipulations had the most positive effects on restoration outcomes. We thus advocate the combined use of these restoration strategies, and further studies which focus on both seeding and soil community manipulation in restoration.     

Limited effectiveness of artificial bird perches for the establishment of seedlings and the restoration of Brazil’s Atlantic Forest

Brazil’s Atlantic Forest biome is severely degraded and fragmented throughout its range. Developing effective techniques to restore pasture and agriculture back to native vegetation is therefore a priority for legal and conservation purposes. In this study, we evaluate the ability of artificial bird perches to enhance the arrival of new seeds and seedling establishment in a degraded, semi-deciduous seasonal portion of the Atlantic Forest in southern Brazil. Specifically, we assess the influence of previous land use and habitat types on the abundance, species richness and ecological traits of bird-dispersed seeds, as well as on seedling establishment. Eight sampling sites were established, each containing one unit with seed traps and restoration plots under artificial perches and one similar unit without the perches. These sites were located in pasture and agriculture, distributed between riparian and sub-montane areas. Monthly sampling was conducted over two years between December 2005 and November 2007, resulting in the evaluation of 25,755 seeds and 56 endozoochoric seed species. The most abundant species were the pioneers Cecropia pachystachya Trécul and Solanum americanum Mill. Experimental units with perches received significantly more seeds than control units. Moreover, seed arrival was higher in sub-montane areas and on former pasture sites. Species richness followed a similar pattern of higher seed arrival, but there was no effect of vegetation type. Ecological characteristics of seeds were associated with land use type: former pastures received more tree seeds and pioneer species than expected by chance. Seedling establishment was very low in all treatments, with only eight seedlings established in perch plots by the end of the experiment. We conclude that despite artificial perches significantly increasing the arrival of endozoochoric seeds onto degraded lands, seedling establishment is drastically limited in these areas, compromising the efficacy of this technique for restoration purposes.     

Approaching the serpentine factor at a local scale—a study in an ultramafic area in northern Greece

We explore factors responsible for vegetation differentiation in a small-scale serpentine area, and attempt to provide new insights in the complexity of the serpentine factor at community level. We sampled 49 quadrats. From each quadrat physical and chemical soil parameters were measured and species composition, altitude, inclination, aspect and coordinates were recorded. Quadrats were classified and ordination analyses were used to explore the environmental gradients and to estimate the explanatory power of the variables. Generalized linear models were used to investigate the response of species to environmental factors. Variance partitioning was applied to calculate the proportion of variance attributed to different groups of explanatory variables. The gradients revealed were related to soil texture, nutrient contents, calcium deficiency, chromium content, climatic parameters and grazing and disturbance intensity. Variance partitioning showed that the highest proportions of variance were attributed to the nutrients and physiographic (including soil texture) variables, while smaller but notable proportions of variance were attributed to geographical coordinates and to metal contents. Our study shows that vegetation differentiation at a local scale is determined by a complex factor of soil properties and climatic parameters, together with variation in disturbance and succession.     

Prevalence and genetic analysis of α-thalassemia and β-thalassemia in Chongqing area of China

Background and aims

Thalassemia is one of the most common hereditary disorders. This study aimed to investigate the prevalence of thalassemia and the mutation spectrum in Chongqing, the southern area of China.

Methods

A total of 1057 children were recruited from Chongqing. Hematological parameters were examined and globin genes were genetically analyzed.

Results

The total frequency of thalassemia carriers was 7.76% in this group of children. Among these, α-thalassemia was 5.20%, β-thalassemia was 1.99% and abnormal hemoglobin variant was 0.57%. Furthermore, 24 cases of α-triplication were detected, frequency of which was 2.55%. The true prevalence of silent α-thalassemia was first reported in this study. In addition, six novel mutations that give rise to α-thalassemia and two rare abnormal hemoglobin variants were first identified in Chinese population.

Conclusions

Our data suggested that the population in Chongqing are at high risk of α- and β-thalassemia. The findings will be useful for genetic counseling and the prevention of severe thalassemias in this area.     

Genome organization and variation in the 3′-partial sequence of garlic latent virus in China

Ten different isolates of a carlavirus were detected by degenerate PCR from 12 garlic samples collected from 6 provinces in China, and the complete genome sequence of the Zhejiang isolate ZJ1 and 3′-terminal sequences of 9 other isolates were determined. The RNA genome of isolate ZJ1 consisted of 8363nts excluding the 3′-poly (A) tail, and the genome organization was similar to other carlaviruses with 6 open reading frames encoding a replicase, TGB1, TGB2, TGB3, CP and NABP respectively. Sequence comparisons showed that all 10 isolates were Garlic latent virus (GarLV). The variations in the TGB2, TGB3 and NABP were more significant than those in the CP. High homology was also detected between those isolates and Shallot latent virus (ShLV). Phylogenetic analysis suggested that GarLV isolates from garlic can be divided into 4 main groups and Chinese isolates belonged to each group. This is the first reported molecular analysis of members of the genus Carlavirus in China.     

Reproduction,foraging and the negative density–area relationship of a generalist rodent

While many species show positive relationships between population density and habitat patch area, some species consistently show higher densities in smaller patches. Few studies have examined mechanisms that may cause species to have negative density–area relationships. We tested the hypothesis that greater reproduction in edge versus interior habitats and small versus large fragments contributes to higher densities of white-footed mice (Peromyscus leucopus) in small versus large forest fragments. We also examined vegetation structure and foraging tray utilization to evaluate if greater reproduction was a result of higher food availability. There were greater number of litters and proportion of females producing litters in the edge versus interior of forest fragments, which may have contributed to greater population growth rates and higher densities in edge versus interior and small versus large fragments. Data on vegetation structure and giving-up densities of seeds in artificial patches suggest that food availability may be higher in edge versus interior habitats and small versus large fragments. These results, in an area with few or no long-tailed weasels, provide a distinct contrast to the findings of Morris and Davidson (Ecology 81:2061, 2000) who observed lower reproduction in forest edge habitat as a result of high weasel predation, suggesting that specialist predators may be important in affecting the quality of edge habitat. While we cannot exclude the potential contributions of immigration, emigration, and mortality, our data suggest that greater reproduction in edge versus interior habitat is an important factor contributing to higher densities of P. leucopus in small fragments.     

Climate–growth analysis of Qilian juniper across an altitudinal gradient in the central Qilian Mountains, northwest China

In the context of global warming, it is of high importance to assess the influence of climatic change and geographic factors on the radial growth of high-elevation trees. Using tree-ring data collected from four stands of Qilian juniper (Juniperus przewalskii Kom.) across an altitudinal gradient in the central Qilian Mountains, northwest China, we compared the radial growth characteristics and climate–growth relationships at different elevations. Results indicated that there was little difference in the tree-ring parameters of the four chronologies. Correlation analyses both for unfiltered and 10-year high-passed data of monthly climatic variables and chronologies were presented to investigate the climatic forcing on tree growth, and results revealed that the correlation patterns were consistent among the four sites, especially for high-passed data. We employed the principal components analysis method to obtain the first principal component (PC1) of the four chronologies and computed the correlations between PC1 and climate factors. The PC1 correlated significantly with winter (November–January) temperature, prior August and current May temperature, and precipitation in the previous September and current January and April, indicating that tree growth in this region was mainly limited by cold winter temperature and drought in early growing season and prior growing season (prior August and September). However, the climate–growth relationships were unstable; with an increase in temperature, the sensitivity of tree growth to temperature had decreased over the past few decades. Considering the instability of the climate–growth relationships, climate reconstructions based on tree rings in the study area should be approached with more caution.