Trade-off strategies between drought resistance and growth rate of dominant tree species in karst forests within heterogeneous habitats

Study site

The Guangxi Nonggang National Nature Reserve (The Nonggang Reserve), situated in the eastern Longzhou County and northern Ningming County, Guangxi Zhuang Autonomous Region, China (106° 42′ 28″−107° 04′ 54″ E, 22° 13′ 56″− 22° 33′ 09″ N), encompasses a total area of 10,080 ha and is distributed in a northwest-southeast elongated strip. Located at the northern edge of the tropics, the Nonggang Reserve experiences an annual average temperature of 22 °C and is characterized by a tropical monsoon climate with distinct wet and dry seasons. Annual rainfall ranges between 1150 and 1550 mm, with 76% of precipitation occurs from May to September, leading to seasonal water shortages and vegetation stress due to drought. The Nonggang Reserve features exposed karst landforms, including multiple peak-cluster depressions and peak-forest valleys. Over a short elevation change of 250–300 m from the depression bottoms to the hilltops, significant variations in soil moisture and fertility, light intensity, rock exposure, and vegetation community types are observed. Notably, the ecological distribution patterns in peak-cluster depressions are similar across different locations^50,51. Based on environmental factors and vegetation characteristics, the Nonggang Reserve can be classified into three habitats. (1) Depression (elevation < 230 m): These areas are shaded by surrounding mountains, receiving weak sunlight, and maintaining a consistently humid environment with potential flooding during the rainy season. Rock exposure is approximately 10%, and the soil layer is relatively thick, with a depth of up to 60 cm. The dominant plant species include moisture-loving and shade-tolerant species like Parashorea chinensis, Saraca dives, and Ficus hispida, with a canopy layer height of approximately 35–45 m. (2) Middle slope (230 m < elevation < 310 m): These areas receives ample sunlight and has moderate soil moisture; rock exposure is about 80%, with a thin and discontinuous soil layer ranging from 30 to 60 cm in thickness. Dominant plants include Excentrodendron tonkinense, Garcinia paucinervis, and Diplodiscus trichospermus, with a canopy layer height of 20–30 m. (3) Hilltop (elevation > 310 m): These areas experience direct sunlight throughout the day and are relatively arid, with summer surface temperatures exceeding 60 °C. Rock exposure is as high as 95%, and soil is scarce, with a thickness less than 30 cm. Dominant species include drought-resistant and light-loving plants such as Sinosideroxylon pedunculatum, Boniodendron minus, and Pistacia weinmanniifolia. Due to environmental constraints, the forest height is only 5–6 m, featuring crooked tree trunks, extremely small leaves, and hard textures⁵⁰.

This study was conducted within a 15-hectare monitoring plot in the Nonggang Karst Seasonal Rainforest (hereinafter referred to as the “Nonggang Plot”). The Nonggang Plot spans 500 m east to west and 300 m north to south, featuring a central depression surrounded by steep slopes, thereby constituting a holistic karst ecosystem with heterogeneous habitats encompassing “depression-middle slope-hilltop” terrain (Fig. 6). Following the global forest biodiversity monitoring protocol established by the Center for Tropical Forest Science (CTFS), the Nonggang Plot has been thoroughly divided and marked⁵¹. The plot was divided into 1,500 quadrats of 10 m×10 m, with each corner of each quadrat permanently marked with aluminum tubes. For more detailed research, each 10 m×10 m quadrat was further subdivided into four 5 m×5 m sub-quadrats. Field surveys were conducted using the 10 m×10 m quadrat as the basic unit. During these surveys, each tree individual (Diameter at breast height (DBH) ≥ 1 cm) was tagged, surveyed, and located. The survey content included tree species number, species name, DBH, tree height, crown width, spatial coordinates, and growth dynamics. Additionally, various environmental factors surrounding each tree individual were measured and calculated, including elevation (ELE), slope (SLO), transformed aspect (ASP), convexity (CON), rock exposure rate (RER), soil water content (SWC) and altitude above channel (AAC) etc. The initial survey of the Nonggang Plot was completed in 2011, followed by two additional surveys in 2016 and 2021, respectively.

15-hectare monitoring plot in the Nonggang Karst seasonal rainforest (the Nonggang Plot). (a) The landscape of karst peak-cluster depressions in the Nonggang Plot; (b) the topographic map of the Nonggang Plot; (c) soil water content of three different habitats in the Nonggang Plot.

Selection of dominant tree species

Within the Nonggang Plot, a continuous slope encompassing three habitats of “depression-middle slope-hilltop” was selected for the study. The top 10 dominant tree species in terms of importance value from each habitat were chosen as study subjects. Specifically, there were 12 species in the depression habitat, and 13 species in both the middle slope and hilltop habitats (Table 5). For each species, 10 healthy mature sample trees with DBH ≥ 10 cm were selected for leaf collection.

Leaf collection and anatomical trait measurement

Leaf sampling was conducted during the growing season (September to October) on sunny days. Branches were harvested from five different orientations (east, south, west, north, and center) of each sample tree using high branch cutters, and 5 mature functional leaves were collected from each branch. To avoid the influence from the main vein, small pieces of 0.5 cm × 0.5 cm were cut from the middle of each leaf to the edge at the 1/2 position and immediately placed in test tubes containing glutaraldehyde fixative for fixation.

Subsequently, the samples were transported to the laboratory and processed through a series of steps, including gradual dehydration with ethanol, critical point drying, and gold plating. They were then observed and photographed under a vacuum electron scanning microscope (ZEISS EVO18). The AxioVision SE64 Rel. 4.9.1 scanning microscope software was utilized to measure multiple anatomical traits of the leaves, encompassing leaf thickness (LT), cuticle thickness (CT), upper epidermis thickness (UET), lower epidermis thickness (LET), palisade mesophyll thickness (PT), spongy mesophyll thickness (ST), stomatal length (SL), stomatal width (SW), and stomatal count (SC). Additionally, several indices were calculated, including leaf palisade to spongy mesophyll thickness ratio (P/S = PT/ST), cell tense ratio ( CTR = PT/LT ), spongy ratio (SR = ST/LT), stomatal aperture (SA = SW/SL), stomatal density (SD = number of stomata per unit area in the field of view), and the stomstal area (SAR=(Π × SL×SW)/4). Each anatomical trait was measured more than 5 times to ensure accuracy throughout the measurement process. Ultimately, a large dataset comprising leaf anatomical trait data was obtained. It included 38 tree species, with 10 sample trees per species, 5 leaves per sample tree, and 5 repetitions per leaf, totaling 10,250 data points.

Concurrently, the serial numbers of each sample tree within the Nonggang Plot were recorded, and the environmental factor data surrounding the sample trees were retrieved (Table 5). Within the Nonggang Plot, the changes in elevation gradient around individual trees reflect the comprehensive variations in water and heat conditions of their respective habitats, and there is a negative correlation between elevation and soil moisture content^50,51.

Drought resistance evaluation

The drought resistance, which encompassing multiple traits, was evaluated using the membership function value (MFV) derived from fuzzy mathematics theory⁵². The evaluation steps are as follows: Firstly, a membership function is applied to assign a corresponding value within the closed interval [0,1] to each trait, termed as the “single-factor membership function value,” which serves to assess each trait individually. Secondly, a weighted arithmetic mean of all single-factor membership function values (MFVs) is computed, i.e. the “comprehensive membership function value (CMFV)”. The weights are determined by the ratio of each principal component’s eigenvalue to the sum of all eigenvalues obtained from the principal component analysis (PCA). Finally, the resulting “comprehensive membership function value (CMFV)” is denoted as the membership function value for drought resistance (MFVD) of the test material. A higher MFVD indicates greater overall drought resistance of the test material^53,54.

The formula for calculating the MFVD of tree species is presented as follows:

In the formula, U_ij represents the drought resistance membership function value of a certain leaf trait (j) of a tree species (i); DC_ij represents the drought resistance coefficient of a certain leaf trait (j) of a tree species (i); DC_jmax represents the maximum drought resistance coefficient among all tree species for a certain leaf trait (j); DC_jmin represents the minimum drought resistance coefficient among all tree species for a certain leaf trait (j); U_i represents the average drought resistance membership function value of all traits of a tree species (i); p_m represents the weight coefficient of the m-th principal component; n represents the number of measured traits. A higher U_i indicates stronger drought resistance of a tree species (i).

Growth rate quantification

To accurately quantify the growth of individual tree species, the growth rate (GR) of each species is defined as the average annual increase in diameter at breast height (DBH) between two consecutive surveys, specifically between 2011 and 2016 and 2016–2021^7,55. The specific calculation formula is: GR = (DBH_final–DBH_initial) / (t_final–t_initial), where t represents the survey time. During the data processing stage, abnormal data points resulting from human or other external factors (where DBH_initial≥DBH_final) were excluded to ensure the accuracy of the results. Ultimately, 310 individuals were retained for the subsequent statistical analysis, out of the original data comprised 38 species×10 individuals = 380 individuals.

Measurement of light response curves

Given the intricate terrain and sampling challenges of the Nonggang Plot, this study selected representative tree species for the measurement of photosynthetic physiological indicators. These measurements were conducted on sunny days between September and October, which corresponds to the growing season of the karst seasonal rainforest. Specifically, measurement were conducted from 9:00 AM to 12:00 PM, as this time period correspond to the most active photosynthesis. The LI-6400XT portable photosynthesis system (manufactured by LI-COR, Inc., USA) was utilized, which was equipped with a red-blue light source (Li-6400-02B). Fourteen decreasing light intensities, ranging from 2000 to 0 µmol·m⁻²·s⁻¹ (specifically, 2000, 1800, 1500, 1200, 1000, 800, 500, 300, 200, 150, 100, 50, 20, and 0µmol·m⁻²·s⁻¹) were set to measure the net photosynthetic rate of tree species leaves under different light intensities. During the measurement, the temperature was maintained at (25 ± 1) °C, relative humidity at (70 ± 5)%, and CO₂ concentration at (400 ± 2) µmol·mol⁻¹. For each tree species, measurements were taken from three sample trees. Subsequently, the obtained light response curves were fitted using the rectangular hyperbola modified model, enabling the derivation of key indicators such as apparent quantum efficiency (AQE), light saturation point (LSP), maximum net photosynthetic rate (P_nmax), light compensation point (LCP), and dark respiration rate (R_d).The modified rectangular hyperbola model is expressed as follows⁵⁶:

In the formula: P_n denotes the net photosynthetic rate; I represents the light intensity; α represents the slope of the plant photosynthesis light response curve at I = 0 (dimensionless), representing the initial slope of the light response curve; P_nmax denotes the maximum net photosynthetic rate; β is termed the light inhibition term; γ which has a value equivalent to α/P_nmax, is referred to as the light saturation term ; R_d represents the dark respiration rate.

Data analysis

This study employed R statistical software to investigate elevational adaptation mechanisms in dominant tree species through integrated analytical workflows. Data collection involved photosynthetic light response curves fitted using Photosyn Assistant 4.1.1⁵⁷. Linear regression analysis revealed elevation-driven patterns across leaf anatomical traits, drought resistance indices, and growth parameters. We fitted the model using the GLS function in R’s nlme package⁵⁸, incorporating the corAR1 function to account for temporal autocorrelation in trees’ repeated-measures data. Model parameters were used to analyze the magnitude of variation in leaf traits across elevational gradients quantitatively.

One-way ANOVA detected interspecific differences in light response parameters across habitats, while Redundancy Analysis (RDA) implemented through the R vegan package quantified environmental drivers influencing functional traits. Structural Equation Modeling (SEM) comprised three sequential phases: (1) collinearity diagnosis using the R collinear package (VIF < 5 threshold)⁵⁹, (2) variable selection retaining key environmental (Elv: elevation, BRR: bedrock ratio, SWC: soil water content) and drought-resistance indicators (CT: cuticle thickness, SA: stomatal area, ST: stomatal density), and (3) model construction via the R lavaan package using maximum likelihood estimation⁶⁰. The SEM demonstrated excellent fit indices (χ²(11) = 10.408, p = 0.494; RMSEA = 0.032)⁶¹, confirming elevation-mediated drought adaptation trade-offs with growth constraints.