**4. Discussion**

### *4.1. Leaf and Soil Nutrient Concentration and Stoichiometry of Black Locust Forests of Different Ages*

The nutrient cycle and transformation process of C, N, and P between soil and plants during vegetation restoration is the basis of material circulation in terrestrial ecosystems [36]. Soil provides the necessary nutrients for plant growth, and plants impact the soil nutrients through litter restitution, root growth, and secretions. As a result of plant growth, this correlation would also change with stand ages [37]. Previous studies have shown that the artificial vegetation in the Loess Plateau may face a nutrient imbalance with the increase of stand age, mainly manifesting as P limitation in restored ecosystems [29]. In this study, the soil C and N concentrations increased by 209.9% and 139.2% (*p* < 0.05) from 10-year-old to 45-year-old black locust forests, respectively (Table 1), whereas the soil P concentrations showed no significant variation among the stand ages. The different variation of soil C, N, and P may have occurred because the accumulation of soil C and N are primarily driven by biological factors (e.g., decomposition of plant litter and dead roots), but P transformations in soil are mainly driven by biochemical mineralization (e.g., phosphate decomposition, which takes a long time [29]. Compared to P, the black locust trees can more effectively import N into the soil with symbiosis of N-fixing microbes [38]. Therefore, a significant increase in soil C:P and N:P ratios was found in the older black locust forests (*p* < 0.05) (Figure 1). Our previous study has found a significant decrease of soil-available P and increase of available N along with the increased stand age [39]; therefore, we can speculate that the soil may be relatively deficient in P in older black locust plantations on the Loess Plateau. This provides further support for conclusions from previous studies [29,31,40].

Nutrient concentrations and stoichiometry ratios in leaves change in different growth stages [40,41]. The results showed that the average concentrations of C and N in leaves from black locust forests of different ages were 436.98 g kg−<sup>1</sup> and 28.13 g kg−1, respectively (Table 2). The C concentration was similar to the average C content of leaves (438 g kg−1) in the Loess Plateau [42], whereas the N content was much higher than the average level in both the Loess Plateau (21.61 g kg−1) [43] and in China (18.60 g kg−1) [44]. The C and N concentration of green leaves increased and then decreased with the increase of stand ages. The concentrations peaked in the RP36 (Table 2). The overall change of C and N in black locust leaves showed the following trend: middle age forest > older forest > young forest, which was consistent with Ma et al.'s results [40]. This may be due to the accumulation of structural substances with more C in the middle- and older-aged plants compared with younger plants. The average P concentrations in green leaves from black locust forests of different ages was 2.01 g kg−<sup>1</sup> (Table 2), which was close to or slightly less than the global average P content (2.00 g kg−1) [45] and the average of leaves from the Loess Plateau (2.09 g kg−1) [43]. However, P concentration of the green leaves did not show significant differences with the change of stand ages. The N:P ratios of green leaves increased significantly from 10.83 to 16.72 along the chronosequence of black locust forests (*p* < 0.05) (Figure 1). Our results suggested that the black locust plantations would shift from relative N-limitation to relative P-limitation across the chronosequence based on the criteria proposed by Güsewell [22] (i.e., leaves N:P <14, 14–16, and >16 indicating N limitation, N and P co-limitation, and P limitation, respectively). In other words, the young black locust plantations were relatively more N-limited, and the older plantations were relatively more P-limited. Similar findings have also been reported in *Larix olgensis* A. Henry [9], *Larix kaempferi* (Lamb.) Carr. [10], and other plantations [46], which showed that long-term ecosystem development tended to cause a shift from N- to P-limitation. This may be a result of the relatively easy absorption of soil N (with N-fixing microbes in the rhizosphere), but relatively inadequate P absorption by the black locust [38]. However, our research only tested the soil properties in the top layer, thus further study on nutrient effectiveness and root absorption characteristics of deep soil should also be investigated.

Leaf litter is an important way for plants to return nutrients to the soil. Typically, at least some of the nutrients in senesced leaves will be transferred to the branches before falling, resulting in lower levels of nutrients in the senesced leaves compared to the green leaves [12,14,47]. The data obtained in our study also agreed with this rule. The average concentrations of C, N, and P of senesced leaves in stands of different ages were 398.07 g kg−1, 16.89 g kg−1, and 1.22 g kg−<sup>1</sup> (Table 2), respectively. Compared with the average level of nutrients of senesced leaves in broad-leaved forests in China (the C, N, and P concentrations was 479.9 g kg−1, 13.2 g kg−1, and 1.06 g kg−1, respectively) [48], our results showed relatively high concentrations of N and P, and low concentrations of C. Previous studies have concluded that the abundance of N and P in litter is conducive to microbial (especially bacterial) activity, thereby promoting rapid decomposition of litter, facilitating nutrient release [49,50]. In this study, the C:N and C:P of senesced leaves was 22.62–29.95 and 274.35–355.60 in black locust forests of different ages, respectively (Figure 1). These ratios are significantly lower than the average value of C:N and C:P in senesced leaves in the broad-leaved forests of China (36.36 and 452.73, respectively), demonstrating that the leaf litter in black locust forests are easy to decompose and are beneficial to nutrient release. In addition, our findings showed increasing N and P contents in senesced leaves in older stands, indicating that more N and P would be returned to the soil through litter decomposition.

### *4.2. N and P Resorption Correlated with Soil and Leaf Nutrients across the Plantation Chronosequence*

Plants absorb nutrients from the soil and allocate them to different organs to accumulate or participate in various life activities; the nutrients will be reabsorbed before the leaves fall, thus, prolonging the retention time of nutrients in plants [14]. In this study, the N and P resorption efficiency of black locust forests of different ages were 46.8%–57.4% and 37.4%–58.5%, respectively (Figure 2A); these efficiencies are significantly lower than that of global terrestrial forests (62.1% and 64.9%, respectively) [12], and this may be due to species-dependent differences. The NRE and

PRE were significant correlated with each other, and both of the NRE and PRE showed significantly increasing and then decreasing trends along the chronosequence of black locust forests (Figure 2), which was consistent with previous research in *Pinus massoniana* Lamb and *Metasequoia glyptostroboides* Hu et Cheng plantations of different ages [19,51]. This variation may be because young plants have rapid biomass production, thus need a higher N and P resorption efficiency to supply an appropriate amount of nutrients [19]. Black locust has greater N absorbing efficiency than P absorbing efficiency with symbiosis of N-fixing microbes around roots. Meanwhile, the rapid growth of young trees requires more P for production of genetic material. These reasons may likely result in a greater PRE than NRE in 10-year-old and 20-year-old forests [40]. A similar trend was also reported in *Eucalyptus urophylla* S.T. Blake × *E. grandis* W. Mill ex Maiden forests [20]. In the older growth stage of black locust forests, the PRE significantly decreased with the increasing stand ages, causing a deficit of P in the soil and trees. This disagrees with our first hypothesis, which may be because the wood biomass became an effective reservoir of P in the later stage of plant growth, thus reducing the P resorption from fallen leaves [52,53]. This indicated that nutrient resorption may not be the main strategy for the preservation of P elements in the older black locust forests. In addition, the NRE decreased with the age of the stand, which is contrary to our first hypothesis. Previous research does not support that tree growth depended more on N resorption to supply its N requirements with the development of the stand [18,37], but it is consistent with the view that decreased NRE was a result of sufficient soil N supply after decomposition [19]. The increased N concentration in soil and senesced leaves with the older stands and relatively low C:N of senesced leaves in our study also supported this view.

The nutrition resorption may be individually or collectively affected by soil nutrient conditions and nutrient contents of green leaves and senesced leaves [12,54]. There is still grea<sup>t</sup> uncertainty about the degree of influence and direction of these factors. For example, a study conducted by Aerts reported that there was no significant correlation between NuRE and nutrient concentrations in soil and green leaves, but a negative correlation was found with the nutrient content in senesced leaves [1,55]. However, a study conducted by Vergutz et al. based on data at a global scale concluded that the NuRE depends more on the nutrient state of green leaves than on the senesced leaves [12]. In addition, other studies have also shown that plants subject to N or P limitation should have higher N or P resorption efficiency [56,57]. Killingbeck proposed that nutrient transfer and resorption may be only an inherent feature of species [58], indicating that genetic differences in plants may be the main factor determining nutrient resorption [54,59]. Our results demonstrated that the NRE of black locust forests was significantly correlated with the C concentration in green leaves, P in senesced leaves, and C:N (*p* < 0.05). No significant correlations were found with soil nutrients (Table 3). This was consistent with a previous study that reported no impact of soil nutrients on NRE [11,55], whereas the PRE was significantly correlated with most of the nutrient content and C:N:P stoichiometry ratios of soil and leaves (*p* < 0.05). Intriguingly, the PRE was negatively correlated with P limitation indicators (i.e., N:P and C:P of soil and leaves) (Figure 3), suggesting that the PRE would be lower with greater P limitation. The decreasing PRE:NRE also indicated relatively lower rates of P resorption with increased plantation age. This was partly in accordance with our second hypothesis but in contrast with the "relative resorption hypothesis". The "relative resorption hypothesis" revealed that N and P would show an increase in resorption when plants are growing in N- or P-limited conditions [11,60], indicating that the leaves nutrients resorption may be not be the main strategy of black locust to cope the P-limited conditions. Similar results were also observed in *Metasequoia glyptostroboides* and *Larix gmelinii* Rupr. plantations [10,19]. These studies suggested that more efficient nutrient absorption by roots should also be considered.

**Figure 3.** Correlations between PRE and soil and the C:P (**A**) and N:P (**B**) ratios of green leaves in black locust plantations from stands of different ages.

Though the NRE and PRE respond differently to soil and plant nutrients, the results revealed that black locust plantations would alter the conservation and use strategy of nutrients in the ecosystem through a plant-mediated pathway in different stand ages. Based on the present study and previously published studies, we speculate that more nutrients stay in senesced leaves of trees in older forests, possibly prolonging the P cycle and increasing the risk of nutrition loss by leaching in the process of litter decomposition [17,61]. However, litter decomposition may induce increased incorporation of P into the soil microbial biomass, which can promote the recycling speed of nutrients and availability of P in ecosystems [62,63]. Recent studies in P-deficient deciduous forests in Europe have shown that the absorption and return of nutrients also depend on soil P availability [62,64,65]. Our study provided limited information about how the release of litter nutrients and root nutrient absorption affect and react to soil available nutrient change in black locust forests. Since plant nutrient absorption and utilization is a complex process [17,21], future experimental studies considering certain soil nutrient species (e.g., plant available/labile forms, organically-bound forms, etc.), root nutrients absorption, and soil microbes are needed to elucidate the central nutrient utilization strategy of black locust in response to a nutrient-poor environment and how these factors are involved in regulating NuRE in different growth stages.
