The study area is located in the soil-water conservation science and technology demonstration zone in the feldspathic sandstone zone in Nuanshui Village Jungar Banner, Ordos, Inner Mongolia ( Figure 1 ). This region (39°42’N -39°50’ N, 110°25’E -110°48’ E, 96 km²) has complex terrains with extensive gullies and fluctuating girders, and suffers soil erosion and soil-water loss. It has an average altitude of 800 - 1590 m, a precipitation of 400 mm (concentrated in July and August), a duration of sunlight duration above 3000 d, an evaporation of 2093 mm, and a temperature of 6.2-8.7°C on the annual average. The vegetation of this region is dominated by artificial vegetation, including H. rhamnoides, Pinus tableulaeformis, Caragana korshinskii, Medicago sativa and Prunus sibirica.

In the demonstration zone, we chose H. rhamnoides plantation land that had basically consistent site conditions and forest compositions and were in the decaying stage as the experimental site, which were under the northwest slope face at slope of 4°. On the same slope face, H. rhamnoides were planted at the row space 0f 2m × 4m, and its tree age was 10 years. H. rhamnoides were stumped at early March 2020 in the experimental site. The stump heights (the stumping distance from the ground) were 0, 10, 15 and 20 cm, which corresponded to the treatments H1, H2, H3 and H4, respectively ( Figure 2 ). A stump-free plantation site was established as a control (CK). All treatment sample plots were 50m×50m in area. Each treatment was carried out in triplicate. Stumping was performed using electric saws and pruning shears, which ensured that the incisions were flat and smooth without burrs. The overall stumping mode was adopted. To decrease plant moisture dissipation, we painted stumped sites after stumping.

The leaf and fine root traits in five samples from the plots treated at different stump heights were measured in the middle growing season (middle June to lower August) of 2022 (The growth of H. rhamnoides at different stump heights was listed in Table 1 ). A sample involved at least five representative and healthy plants under basically consistent growth status. From each plant, 10 healthy leaves in medium size were collected and used to measure morphology, C and N concentrations. Fine roots were collected by the digging method. The weeds and litter around the basal of each chosen sample were cleaned away. After the surface soils were mostly cleaned, the peri-root soils were carefully cleared away to find the main roots. When root branches were encountered in the growing direction of the main roots, we further dug along the branch roots until we reached the end of the root system. During sampling, the loss of terminal low-level roots was avoided to ensure root completeness. Then fine roots in diameter smaller than 2 mm were chosen (Cornelissen et al., 2003; Mitchell et al., 2017).

The roots collected as they were put in labeled preservation bags, which were then placed in a refrigerator at 2-3° C on the same day and taken back to our laboratory. The leaves and fine roots were washed with deionized water and scanned using an 11000XL scanner (Epson, Tokyo, Japan). Together with WinRHIZO Pro 2012b (Regent Instruments Inc., Quebec City, Canada), leaf areas, leaf volumes, total fine root length, and fine root volumes were measured. Subsequently, the leaves and roots in each sample were placed in water and stored away from light at 4°C for 24 h (Ning et al., 2022). After the water was saturated, the surface water of the leaves or fine roots was sucked using absorbent paper, and then the saturated fresh weight of the leaves or fine roots was measured. Then the leaves and fine roots were dried at 60°C for 48 h, and the dry weight of the leaves and the dry weight of the fine roots were monitored at constant weight (Cui et al., 2019).

Then the specific leaf area (SLA), leaf dry matter content (LDMC), leaf tissue density (LTD), specific root length of fine roots (SRL), fine root dry matter content (FRDMC), and fine root tissue density (FRLTD) were computed. SLA and SRL are the ratio of area to dry weight of leaves and the ratio of area to dry weight of fine roots respectively. LDMC and FRDMC are the ratio of dry weight to saturated water weight of leaves and the ratio of dry weight to saturated water weight of fine roots, respectively. LTD and FRTD stand for the ratio of dry weight to volume of leaves, and the ratio of dry weight to volume of fine roots respectively (Jiang et al., 2021). The dried leaves and fine roots were ground into powder using a PULVERISETTE 5 high flow ball milling system (Fritsch, Munich, Germany), which was passed through a 0.149 mm sieve for chemical analysis. Carbon and nitrogen content in fine roots and leaves were measured using a Vario MACRO cube elemental analyzer (Elementar, Hanau, Germany), and the carbon and nitrogen ratios in roots and leaves were determined.

Data were analyzed in SPSS 26.0. The data of leaves and fine roots were sent to descriptive statistics and analysis of the variation coefficient (variation coefficient =standard deviation/mean value ×100%). Differences in the traits of leaves and fine roots of H. rhamnoides among different stump heights were statistically analyzed by one-way analysis of variance (ANOVA). Significance was tested using Fisher’s least significant difference method at the level p < 0.05. The leaf and fine root variation characteristics at different stump heights, the Pearson heatmaps of fine roots and leaves, and the PCA of root and leaf coordination were plotted in Origin 2021.

Neither leaf C content (LC) nor fine root carbon content (FRC) was significantly different among different stump height treatments (p > 0.05) ( Table 2 ; Figure 3 ). SLA, LTD, LDMC, LC, leaf nitrogen content (LN), leaf carbon to nitrogen ratio (LC : LN), SRL, FRID, FRMDC, FRC, fine root nitrogen content (FRN) and fine root carbon to nitrogen ratio (FRC : FRN) were very significantly different between treatments (p < 0.001).

The total coefficient of variation among different stump heights was 1.05%–19.26%. Total coefficients of variation in SLA, FRN, FRC : FRN, and SRL were greater than 10%, with a maximum in SLA (19.26%). The total coefficients of variation in LN, LC : LN, RDMC, RTD, LDMC, LTD, FRC, and LC were all larger than 10%, with the minimum values of 1.05% in LC and 3.22% in FRC. The total coefficient of variation among different treatments was classified as SLA > FRN > FRNC : FRN > SRL > LN > LC : LN > FRDMC > FRTD > LMDC > LTD > FRC > LC. The coefficients of variation in all indices of H1, H2, H3, H4 and CK were all lower than 10% ( Figure 3 ).

The SLA, LTD, LDMC, LC, LN, LC : LN, SRL, FRID, FRMDC, FRC, FRN, and FRC : FRN were all significantly different between H2 and H3 (p < 0.05). Compared to CK, SLA, LN, SRL and FRN improved significantly after stumping, but LTD, LDMC, LC : LN, FRTD, FRDMC and FRC : FRN significantly decreased. SLA, LN, SRL and LN ranked as H3 > H2 > H1 > H4 > CK. The changing rules of LTD, LDMC, LC : LN, FRTD, and FRDMC were the opposite and ranked as H3 < H2 < H1 < H4 < CK ( Figure 4 ).

The leaf traits and the fine root traits were both significantly correlated among different stump heights ( Figure 5 ). SLA in CK was negatively correlated with LTD, LDMC, and LC : LN, and positively correlated with LN. LDMC in CK was positively correlated with LTD and LC : LN (p < 0.05), and very significantly negatively correlated with LN (p < 0.01).The LN and LC : LN in CK were very significantly negatively correlated (p < 0.01, Figure 5E ). The correlations of SLA with LTD and LDMC were weakened after stumping, but the correlation between SLA and LN or LC : LN was enhanced. In particular, the SLA in H1, H3 or H4 was highly significantly correlated positively with LN (p < 0.01) and negatively with LC : LN (both p < 0.01). After stumping, LTD was positively correlated with LDMC and negatively with LN (p < 0.05), and LN was very significantly positively correlated with LC : LN (p < 0.01, Figures 5A–D ).

SRL in CK was negatively correlated with RTD and FRC : FRN (p < 0.05), and was very significantly positively correlated with FRN (p < 0.01). FRN and FRTD in CK were positively correlated with FRDMC (p < 0.05), and were very significantly negatively correlated with FRC : FRN (p < 0.01, Figure 5J ). SRL after stumping was negatively correlated with RTD and FRC : FRN (p < 0.05), but SRL in H1 or H3 was very significantly negatively correlated with FRC : FRN (p < 0.01). The FRN and FRTD after stumping were both positively correlated with FRDMC (p < 0.05), and the FRTD and FRDMC of H1, H2, H3 were positively correlated with FRC : FRN (p < 0.05).The FRN was still very significantly negatively correlated with FRC : FRN after stumping (p < 0.01, Figures 5F–I ), and the correlation was unchanged. Clearly, the correlations between leaf traits and between fine root traits differ at different stump heights.

There was a significant correlation between the leaf and fine root traits of H. rhamnoide ( Figure 6 , p < 0.05).SLA is negatively correlated with LTD, LDMC and LC : LN, and positively with LN. LTD is positively correlated with LDMC and LC : LN, and negatively correlated with LN. LN and LC : LN are very significantly positively correlated (p < 0.01). SLA and LN are both positively correlated with SRL and FRN and negatively correlated with FRTD and FRC : FRN. LDMC and LC : LN are positively correlated with FRTD and FRC : FRN respectively, and are both negatively correlated with SRL and RN. The correlations of the fine root traits are similar to those of the leaves. SRL is negatively correlated with FRTD and FRC : FRN, and very significantly positively with FRN (P < 0.01). Both FRTD and FRDMC are negatively correlated with FRN and positively with FRC : FRN. FRN and FRC : FRN are negatively correlated.

The accumulative contribution rates of variance on axis 1 and axis 2 in the PCA at different stump heights were both above 80%, which can well reflect the relationship between the functional traits of leaves and fine roots. CK PC1 is a structural axis decided mainly by SLA, LN, LDMC, FRC : FRN, and LTD, and its PC2 is defined by LC, FRC, and FRTD. The closest correlations in CK were found between LC : LN and LN, between FRN and SRL, and between FRC : FRN and SLA ( Figure 7E ). The indices in H1, H2, H3 and H4 after stumping did not change much on the PC1 axis, but LC and FRC axis 2 in all indices and the correlations between root and leaf traits were enhanced. In particular, the associations of SLA or SRL with LN, LC : LN, FRN, and FRC : FRN were closer ( Figures 7A–D ). Although the confidence groups overlapped slightly at different stump heights, the root and leaf traits after different treatments aggregated and were mutually separated. The CK root and leaf traits were mainly distributed in the right half of axis 2, and the root and leaf traits of H1, H2, H3 and H4 shifted along axis 2 from right to left, and H3 was mainly distributed in the right half of PC2 ( Figure 7F ).

The results showed that stumping significantly affected the morphological traits of both roots and leaves ( Table 2 ). The coefficients of variation in the functional morphological traits of the roots and leaves at different stump heights fell within 3.47%–19.25%, which were all smaller than 20%. The coefficients of variation in SLA and SRL were large and very significantly different among different stump heights. These results indicate that these functional traits are very sensitive to stumping. The tissue densities and dry matter content of both roots and leaves are relatively stable variables on the resource acquisition axis, and thus the coefficients of variation are small ( Figure 3 ).

According to leaf and root economics spectra (LES and RES respectively), SLA and SRL are two key traits reflecting the resource strategies of plants (Wright et al., 2004; Cheng et al., 2016). Specifically, the species with larger SLA had higher LN, leaf P content, faster photosynthetic rate, shorter leaf longevity, and lower LTD, which indicates the resource-acquisitive strategy. On the contrary, the species with lower SLA were conservative. The fine roots with large SRL, small diameter, low RTD and high N content are related to the low construction cost, fast respiration rate and high turnover rate. This pattern is similar to that of leaf trait correlations (Caplan et al., 2019; Reich, 2014).

The specific leaf area reflects the light receiving and capturing area per unit of leaf dry weight and indicates the abilities of plants to use environmental resources and to store the acquired resources. It is closely related to the rate of assimilation rate and the survival countermeasures of plants (Fajardo and Siefert, 2016). The specific fine root length is an important morphological structure that decides the water and nutrient absorption capacity of roots (Wang et al., 2019; Wang et al., 2020). Generally, the specific leaf area is larger in a resource-rich environment. Plants subjected to nutrient restriction or interspecific competition will increase the specific root length and root specific surface area to improve the acquisition capacity or competitiveness of nutrients (Kraft and Ackerly, 2010; Yang et al., 2014a). A decrease in tissue density or dry matter content in roots or leaves can rapidly accelerate plant growth turnover, so the loss of moisture and nutrients will decreased, thus improving the efficiency of use of moisture and nutrient and improving the defense force (Laughlin et al., 2021; Ning et al., 2022).

Our results showed that compared to CK, the SLA and SRL of H. rhamnoides were increased significantly after stumping, and LTD, LDMC, RTD, and RDMC were reduced ( Figure 4 ), which are basically consistent with previous research. The reasons are that during compensationary recovery and growth after stumping, the resource acquisition conditions of H. rhamnoides are altered, as the cost of leaf construction is lowered. Consequently, the evaporation of vegetation decreases and the root branching ability and the soil nutrient absorption ability are enhanced, which promotes H. rhamnoides to adjust the adapting strategy to enhance its viability and to provide moisture and nutrients for the growth and metabolism of plants (Yuan et al., 2020; Liu et al., 2022), forming a growing strategy of high SLA and SRL and low LTD, LDMC, RTD and RDMC. This validates our first hypothesis. SLA, LN, SRL and LN all rank as H3 > H2 > H1 > H4 > CK. The changing rules of LTD, LDMC, LC : LN, FRTD, and FRDMC are the opposite and rank as H3 < H2 < H1 < H4 < CK ( Figure 4 ).This compensation recovery and growth strategy was optimized at the stump height of 15 cm, above which the promoting effect was lowered.

The C element is the substrate and energy source of various physiological-biochemical processes of plants, and is the most important step to connect the external inorganic environment and organisms. C is also the basic framework of all organisms and is closely related to the photosynthesis and respiration of plants (Xu et al., 2018). N, a basic nutrient element of plants, and is an important compositional element and adjustment substance of diverse proteins and genetic materials (Cui et al., 2022). The C:N reflects the carbon assimilation ability of plants during the absorption of nutrient elements and indicates the use efficiency of nutrient elements. Generally, low C:N suggests that a plant grows fast (Hu et al., 2022).

We found that stumping significantly affected both N and C:N of leaves and roots, but leaf C contents were not significantly different among different stump heights ( Table 2 ), and the root and leaf C contents were highly stable ( Figure 3 ). This was because the organic carbon in the organs of plants is usually not directly involved in production activities, but acts as skeleton to provide plant activities with energy, and thus the organic carbon content in vivo is large and stable with low variation (Wang et al., 2018). Our results showed that the C : N was always larger in roots than in leaves at all stump heights, and the C : N in both leaves and roots was not significantly different among different stump heights ( Figure 4 ), indicating the rate of N utilization is larger in roots than in leaves. The C and N contents in leaves are always larger than in roots regardless of the stump height, which is because the efficiency of C and N use of leaves is lower than that of roots. Hence, when cell division occurs rapidly due to the rapid growth of the leaves, the leaves demand largely for nutrient. Consequently, the C produced from photosynthesis gradually accumulates, and the roots transport more N to leaves, which is used in the synthesis of proteins and nucleic acids (Craine, 2006; Morales et al., 2015), so the C and N contents in the leaves of H. rhamnoides are higher.

Leaf nitrogen is a key factor for photosynthetic material metabolism and plant growth, and is an important component for the synthesis of chlorophylls and relevant photosynthetic proteins. Leaves with high N contents usually have fast photosynthetic rate (Jaikumar et al., 2013). Our results showed the leaf and root N contents after treatments H1, H2, H3, and H4 were all significantly larger compared to the CK ( Figure 4 ). This was because the metabolism of roots was enhanced after the stumping, which promoted the nitrogen fixation ability of root nodules and improved the absorption and transportation of nitrogen nutrient elements by fine roots from the soils. As a result, the nitrogen contents in roots and leaves were significantly improved to maintain the rapid recovery and growth of plants (Yang et al., 2020; Liu et al., 2022; Liu et al., 2023). In addition, leaf area is an indicator of the photosynthesis ability of leaves. A larger leaf area is favorable for the interception of more solar light to produce organic matter (Osnas et al., 2013). Our calculations showed the leaf areas after stumping were significantly larger compared to CK. The increased leaf area and the increase nitrogen transport to leaves through roots after the stumping jointly promoted the efficiency of plant photosynthesis and accelerated the synthesis of abundant chlorophyll and photosynthetic proteins in leaves, which once again increased the leaf N content (Craine, 2006; Morales et al., 2015). This again validates our first hypothesis.

The C: N ratios in the roots or leaves after treatments H1, H2, H3, and H4 were all significantly lower than those without stumping ( Figure 4 ), suggesting H. rhamnoides at the stump height of 15 cm can grow faster. In all, stumping can not only control the external growing morphology of plants to some extent, but can also indirectly change the internal physiological processes by adjusting the needed resources and environment, which will considerably affect plant growth.

The functional traits of plants are not mutually independent, but are coordinated or traded off to promote plant growth. Then the correlations among the leaf functional traits were compared between CK and H1, H2, H3 or H4. Commonness was found, as SLA and LN were positively correlated, and LC : LN was negatively correlated with both SLA and LN. It is speculated the coupling between the 2 above traits is the most stable among the leaf functional traits of H. rhamnoides.

The correlations of SLA with LTD and LDMC were weaker after the stumping, and the correlations between SLA and LN, and between LC : LN and LN were strengthened ( Figures 5A–E ). These correlations between traits are similar to the findings on leaf functional traits in accordance with the leaf economics spectrum (Wright et al., 2004; Marechaux et al., 2020), which reflects that the associations between the leaf functional traits are universal. However, we found the correlations among leaf morphological traits were weakened after the stumping, and the correlations with chemical characteristics of the tissue were enhanced. Adjusting leaf morphological structure, growing pattern and nutrient element allocation strategy after stumping will alter the correlations between traits, so the stumped plants can rapidly recover the lost aground branches, which will improve the nutrient absorbing capacity of roots and the nutrient transport ability to leaves (Yang et al., 2020; Liu et al., 2022). The morphological traits of leaves and the absorption of nutrient resource by leaves are directly and closely associated (Ackerly et al., 2002; Wright et al., 2004). This is also the reason for the increased. correlations between the leaf morphological traits and chemical tissue traits after the stumping. The change between morphological traits may be indirect after the chemical tissue traits of leaves affect a certain morphological trait. Therefore, during the compensationary recovery and growth of leaves, the correlations between the morphological traits of the leaves are weakened.

We found the correlations between fine root traits of H. rhamnoides were similar to the correlations between leaf traits, especially SRL and SLA that reflect the resource acquisition abilities of fine root and leaves respectively (Reich, 2014; Laughlin et al., 2021). The results showed that SRL was negatively correlated with RTD, and FRC : FRN, and very significantly positively correlated with FRN. FRTD and FRDMC were both positively correlated with FRN. FRN and FRC : FRN were very significantly negatively correlated ( Figures 5F–J ). All of these results basically consistent with previous studies. The correlations of root morphological traits after the stumping with FRN, and FRC : FRN were enhanced, which is consistent with the correlations between leaf traits after the stumping.

In a dry environment, the functional traits related to nutrition contents and resource absorption are closely coordinated between leaves and fine roots. This is because the moisture and nutrient restrictions for plant growth require that the functions of fine roots (namely, moisture and nutrient absorption) must match the functions of leaves (namely photosynthesis and transpiration) (Carvajal et al., 2019). Our root and leaf correlation analyzes prove this view: SLA and LN are both positively correlated with SRL and FRN, and negatively correlated with FRTD and FRC : FRN. LDMC and LC : LN are positively correlated with FRTD and FRC : FRN respectively, and negatively correlated with SRL and RN ( Figure 6 ). Namely, nutrient absorption by the ground and underground organs is strongly associated. These results further prove that the aground traits and underground traits of H. rhamnoides are coordinated to some extent. Our results are consistent with previous studies on temperate grasslands (Craine et al., 2005), forests (Holdaway et al., 2011) and lawn (Liu et al., 2010; Zhao et al., 2016). This validates our second hypothesis.

According to the plant economics spectrum theory, functional traits of plants are important indices for measuring environmental resource trade-off strategies of plants (Reich et al., 2003; Freschet et al., 2010). When LTD, LDMC, RTD, RDMC, and C: N are low, SLA and SRL are large, and when the N contents in roots and leaves are large, the plants usually have a rapid photosynthetic rate and growth rate and become the ‘rapid investment–return’ type; otherwise, they approach the ‘slow investment –return’ type (Jiang et al., 2021; Pérez-Ramos et al., 2012). Compared with CK ( Figure 7E ), the LTD, LDMC, FRTD, FRDMC, LC: LN, and FRC : FRN significantly decreased, SLA, and SRL significantly rose, and LN and FRN were large after the stumping ( Figures 4 , 7A–D ). It is indicated H. rhamnoides in an environment with limited resources is indicated to optimally allocate resources between the functional traits of leaves and roots by using a trade-off strategy. Compared with CK, the root and leaf economics spectra of stumped H. rhamnoides are closer to one end of ‘rapid investment–return type’ species ( Figure 7F ), which indicates the shift to the resource trade-off strategy of ‘rapid investment–return type’. This validates our third hypothesis. Our findings are significant for the prevention and control of revegetation and soil erosion in feldspathic sandstone areas.