*Article* **Temporal Changes in Litterfall and Nutrient Cycling from 2005–2015 in an Evergreen Broad-Leaved Forest in the Ailao Mountains, China**

**Shiyu Dai <sup>1</sup> , Ting Wei <sup>1</sup> , Juan Tang <sup>1</sup> , Zhixiong Xu <sup>2</sup> and Hede Gong 1,\***

	- Jingdong, Puer City 676209, China

**Abstract:** The study of litter can provide an important reference for understanding patterns of forest nutrient cycling and sustainable management. Here, we measured litterfall (leaves, branches, etc.) from a wet, evergreen, broad-leaved forest in Ailao Mountains of southwestern China on a monthly basis for 11 years (2005–2015). We measured the total biomass of litter fall as well as its components, and estimated the amount of C, N, P, K, S, Ca, and Mg in the amount of litterfall. We found that: The total litter of evergreen, broadleaved forest in Ailao Mountains from 2005 to 2015 was 7.70–9.46 t/ha, and the output of litterfall differed between years. This provides a safeguard for the soil fertility and biodiversity of the area. The total amount of litterfall and its components showed obvious seasonal variation, with most showing a bimodal pattern (peak from March to May and October to November). The majority of litterfall came from leaves, and the total amount as well as its components were correlated with meteorological factors (wind speed, temperate and precipitation) as well as extreme weather events. We found that among years, the nutrient concentration was sorted as C > Ca > N > K > Mg > S > P. The nutrient concentration in the fallen litter and the amount of nutrients returned showed a decreasing trend, but the decreasing rate was slowed through time. Nutrient cycling was influenced by meteorological factors, such as temperature, precipitation, and wind speed, but the nutrient utilization efficiency is high, the circulation capacity is strong, and the turnover time is short. Our results showed that although there was nutrient loss in this evergreen, broad-leaved forest, the presence of forest litterfall can effectively curb potential ecological problems in the area.

**Keywords:** litterfall production; elemental composition; nutrient cycle; subtropical forest; Ailao Mountain

### **1. Introduction**

In forests, organic matter from plants that is returned to the soil surface (e.g., fallen leaves, branches, floral and fruit parts) is generally referred to as litterfall [1,2]. The amount and quality of forest litterfall plays an important role in the development of soil and the cycling of nutrients. For example, growing plants absorb nutrients needed for their own growth from the soil, e.g., carbon, nitrogen, phosphorus, potassium, and other elements, and then return those elements back to the soil in the form of litterfall decomposition [3]. Among the different C fluxes of the forest ecosystem, canopy litterfall is the main aboveground organic C flux that reaches the soil, affecting C cycling as well as maintaining soil fertility globally [4]. Hence, litterfall acts as an important link between the aboveground production of trees and the soil organic C stock. At the same time, the litterfall production changes with climate, forest type, stand age, and season [5–7]. As a result, the quantity and quality of litterfall, as well as the environmental factors that influence them, regulate how these material cycles within ecosystems. However, there is variation in the turnover rates of different elements in a given ecosystem, as well as variation in turnover across regions in different climatic zones. Thus, it is useful to study the nature of forest

**Citation:** Dai, S.; Wei, T.; Tang, J.; Xu, Z.; Gong, H. Temporal Changes in Litterfall and Nutrient Cycling from 2005–2015 in an Evergreen Broad-Leaved Forest in the Ailao Mountains, China. *Plants* **2023**, *12*, 1277. https://doi.org/10.3390/ plants12061277

Academic Editor: Jianhui Huang

Received: 19 January 2023 Revised: 6 March 2023 Accepted: 8 March 2023 Published: 10 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

litterfall dynamics through time in order to gain a deeper understanding of patterns of variation in litterfall among seasons and across years, and how this contributes to variation in forest nutrient cycling.

Plants periodically shed parts of their biomass as litterfall, which transfers C and nutrients from plants back into the soils and is a key biogeochemical process within forests [4,8]. However, litterfall in forests is variable throughout the year, with less litterfall during the growing season than the non-growing season, The main seasonal pattern of presentation is that: unimodal, bimodal, or irregular pattern [9]. In deciduous forests, litterfall happens during the non-growing season when low temperatures stimulate plant leaf synthesis of abscisic acid, resulting in a high levels of leaf fall, this has been confirmed in many studies [10–12]. In addition to variation across the season and across different forest types, litterfall is also variable among years as a result of variation in climatic conditions (e.g., wind and snow) and forest age. As global climate change continues, studies on the within and among year variation in litterfall, and its role in nutrient cycling, will provide important baseline knowledge for understanding how these will change in the future [13–16].

Subtropical forests have high primary productivity and are also hotspots for biodiversity research., which play an important role in carbon storage in global terrestrial ecosystems [17]. The montane, moist, evergreen, broad-leaved forest in the Ailao Mountain Nature Reserve in Yunnan is currently the largest and most well-preserved subtropical evergreen broad-leaved forest in China. It is one of the valuable zonal vegetation. This is particularly urgent for understanding patterns of litterfall and nutrient cycling in primary forests which are rapidly disappearing. Hence, we can also better understand and utilize natural resources, thus improving the stability and sustainability of the ecosystem.

Here, we measured the total amount of litterfall, as well as its nutrient concentration, from monthly samples collected over an 11-year period (2005–2015) in the primeval forest of Ailao Mountain National Nature Reserve in Yunnan Province, southwestern China. We compared this variation at monthly, seasonal, and annual periods and examined how they were correlated with precipitation, temperature, wind speed, and extreme weather. Results from our study show considerable variation within and among years in the quantity and quality of litterfall in this forest, providing baseline data for studying forest nutrient cycles.

### **2. Results**

### *2.1. Dynamic Characteristics of Litterfall and Its Component Output*

### 2.1.1. Interannual Dynamics of Litterfall and Its Component Output

Between 2005 to 2015, we found that the amount of litterfall per year ranged from 7.70 to 9.46 t/ha a (Figure 1, Table 1), with an annual average of 8.11 ± 0.73 t/ha a. Across the observation period, we found that deciduous leaf litter was the greatest component of the overall leaf litter (representing 42–62% of the litterfall). Other components of litterfall, including fruit/flower drop, bark, moss/lichen, and other debris, were observed to an intermediate (21–26% of the litterfall) extent, and litterfall from branches the least (17–32% of the litterfall) (Table 1).

**Figure 1.** Interannual dynamics of litterfall from 2005 to 2015.

**Table 1.** Annual total litterfall productions and components**. Table 1.** Annual total litterfall productions and components.

**Figure 1.** Interannual dynamics of litterfall from 2005 to 2015.


<sup>2010</sup> 1.36 <sup>±</sup> 0.24 Aa 17.6% 4.80 <sup>±</sup> 0.84 Cb 61.5% 1.52 <sup>±</sup> 0.20 Aa 20.9% 7.76 <sup>±</sup> 1.20 <sup>2011</sup> 1.28 <sup>±</sup> 0.16 Note: Different uppercase letters indicate significant differences in litterfall output between different years of the same component (*p* < 0.05), and different lowercase letters indicate significant differences in litterfall output of different components in the same year (*p* < 0.05).

#### Aa 18.6% 4.24 <sup>±</sup> 0.56 Bb 59.5% 1.56 <sup>±</sup> 0.08 Aa 21.8% 7.12 <sup>±</sup> 0.52 2.1.2. Monthly Variation in the Output of Litterfall and Its Components within a Year

<sup>2012</sup> 1.60 <sup>±</sup> 0.20 Aa 18.9% 4.80 <sup>±</sup> 0.84 Cc 55.6% 2.20 <sup>±</sup> 0.28 Ca 25.5% 8.64 <sup>±</sup> 1.04 <sup>2013</sup> 1.32 <sup>±</sup> 0.20 Aa 17.3% 4.48 <sup>±</sup> 0.60 Bc 57.5% 1.96 <sup>±</sup> 0.28 Bb 25.2% 7.80 <sup>±</sup> 0.80 <sup>2014</sup> 2.12 <sup>±</sup> 0.52 Aa 22.5% 5.20 <sup>±</sup> 1.12 Cb 54.1% 2.24 <sup>±</sup> 0.28 Ca 23.3% 9.40 <sup>±</sup> 1.44 <sup>2015</sup> 2.52 <sup>±</sup> 1.40 Ba 32.0% 3.28 <sup>±</sup> 0.44 Ab 41.7% 2.04 <sup>±</sup> 0.32 Ca 26.3% 8.08 <sup>±</sup> 1.92 Total average 1.76 ± 0.2 21.7% 4.42 ± 0.18 52.0% 1.91 ± 0.06 23.6% 8.11 ± 0.27 Note: Different uppercase letters indicate significant differences in litterfall output between differ‐ ent years of the same component (*p* < 0.05), and different lowercase letters indicate significant dif‐ There was clear seasonal variation in the total litterfall in this forest (Figure 2a), but the variation was not always consistent among years. For example, most years had multiple peaks of litterfall, with one peak around April (ranging from March–May) and another later in the October–November. There was variation in these peaks, however. One year, 2015, was unique with the highest observed litterfall of the whole time series occurring in January. When we analyzed litterfall in its components, we found that leaf litter (Figure 2b), as the most abundant component, largely mirrored that of the total litterfall biomass. The other two components of litterfall, branches (Figure 2c), and 'other' (Figure 2d) were more variable throughout the year. The main source of litterfall production is leaves and branches, with the highest production of branches in January and February, and the highest production of leaves in other months (Table 2).

ferences in litterfall output of different components in the same year (*p* < 0.05).

2.1.2. Monthly Variation in the Output of Litterfall and Its Components within a Year

There was clear seasonal variation in the total litterfall in this forest (Figure 2a), but the variation was not always consistent among years. For example, most years had multi‐ ple peaks of litterfall, with one peak around April (ranging from March–May) and another later in the October–November. There was variation in these peaks, however. One year, 2015, was unique with the highest observed litterfall of the whole time series occurring in

est production of leaves in other months (Table 2).

January. When we analyzed litterfall in its components, we found that leaf litter (Figure 2b), as the most abundant component, largely mirrored that of the total litterfall biomass. The other two components of litterfall, branches (Figure 2c), and 'other' (Figure 2d) were more variable throughout the year. The main source of litterfall production is leaves and branches, with the highest production of branches in January and February, and the high‐

**Figure 2.** Monthly variation of litterfall and its component output in evergreen broadleaf forest in Ailao Mountain. (**a**) Total litterfall, (**b**) Leaf litter, (**c**) Branches litter, (**d**) Others. **Figure 2.** Monthly variation of litterfall and its component output in evergreen broadleaf forest in Ailao Mountain. (**a**) Total litterfall, (**b**) Leaf litter, (**c**) Branches litter, (**d**) Others.

**Table 2.** Intra‐year variation of litterfall productions and components. **Table 2.** Intra-year variation of litterfall productions and components.


Jul 0.16 <sup>±</sup> 0.32 Aa 26.58% 0.28 <sup>±</sup> 0.09 Aa 40.76% 0.22 <sup>±</sup> 0.34 Aa 32.66% 0.66 <sup>±</sup> 0.6 Note: Different uppercase letters indicate significant differences in litterfall output between different months of the same component (*p* < 0.05), and different lowercase letters indicate significant differences in litterfall output of different components in the same month (*p* < 0.05).

### 2.1.3. Correlation Analysis of Litterfall and Its Components with Climatic Factors

Overall, we found that the total litterfall was negatively correlated with the average wind speed, but positively correlated with temperature and precipitation (Table 3). These trends were mirrored by the 'other' category of litterfall, while only leaf litterfall was positively correlated with average monthly temperature (Table 3). For monthly data, we found that the total amount of litterfall and meteorological factors was positively

correlated with monthly precipitation in the first 1–2 months, but there were no other positive correlations between litterfall production and monthly climatic variation (Table 4).

**Table 3.** Correlation coefficients between litterfall productions of different components and various meteorological factors.


Note: \* Significantly correlated at level 0.05 (two-tailed). \*\* Significant correlation at level 0.01 (double-tailed).

**Table 4.** Correlation coefficients between meteorological factors from different months and monthly total litterfall productions.


Note: \* Significantly correlated at level 0.05 (two-tailed). \*\* Significant correlation at level 0.01 (double-tailed).

### *2.2. Dynamic Characteristics of Litterfall Nutrient Concentration*

### 2.2.1. Interannual Variation Characteristics of Litterfall Nutrient Concentration

We found high variation in the nutrient concentration of litterfall among years and in different components of the litterfall (Table 5). In general, the concentration of C, Ca, and Mg decreased from 2005 to 2010 and 2015, while the concentration of N, P, K, and S increased across this same period. While the nutrient concentration of litterfall was C > Ca > N > K > Mg > S > P in 2005 and 2010, the abundance of N increased relative to the other elements, such that C > N > Ca > K > Mg > S > P in 2015. Across years, we found higher nutrient concentration in leaves compared to branches.

**Table 5.** Characteristics of annual average nutrient concentration of litterfall in different years.



**Table 5.** *Cont.*

Note: Different uppercase letters in the same row indicate significant differences between different years (*p* < 0.05), and different lowercase letters in the same column indicate significant differences between different nutrient elements in the same year (*p* < 0.05).

### 2.2.2. Characteristics of Intra-Year Variation of Litterfall Nutrient Concentration

We illustrate the within year variation of each element in the litterfall in Figure 3, showing that each element has a signature variation. C concentration is higher in the first half of the year and declines to its lowest level in late autumn and early winter (Figure 3a). N, P, and S showed similar variation across the year, first declining through the first months of the year, peaking in the middle of the growing season, and declining again towards fall and winter (Figure 3b–d). The last three elements showed less distinct patterns and fluctuated around mean values throughout the year (Figure 3e–g). *Plants* **2023**, *12*, x FOR PEER REVIEW 7 of 18

**Figure 3.** Within‐year change of C, N, P, K, S, Ca and Mg concentrations in litterfall. (**a**) Within‐ year change of C element concentration. (**b**) Within‐year change of N element concentration. (**c**) Within‐year change of P element concentration. (**d**) Within‐year change of S element concentra‐ tion. (**e**) Within‐year change of K element concentration. (**f**) Within‐year change of Ca element concentration. (**g**) Within‐year change of Mg element concentration. **Figure 3.** Within-year change of C, N, P, K, S, Ca and Mg concentrations in litterfall. (**a**) Withinyear change of C element concentration. (**b**) Within-year change of N element concentration. (**c**) Within-year change of P element concentration. (**d**) Within-year change of S element concentration. (**e**) Within-year change of K element concentration. (**f**) Within-year change of Ca element concentration. (**g**) Within-year change of Mg element concentration.

2.3.1. Interannual Return Characteristics of Litterfall Nutrient Elements

We found that the return of leaf litter (except Ca) was higher than that of branches litter, and the return of C was much higher than that of other nutrient elements. In all, the returns were roughly sorted as C > N > Ca > K > Mg > S > P (Table 6). Overall, we found a

**2005 2010 2015**

**Leaf (g/kg)**

509.16 ± 16.44 Ad

17.60 ± 5.16 Dc

1.02 ± 0.50 Ca

4.95 ± 2.96 Cb

1.27 ± 0.17 Ca

**Branches (g/kg)**

481.84 ± 14.29 Ac

> 8.50 ± 2.13 Ab

> 0.73 ± 0.33 Ba

> 3.79 ± 2.49 Ba

> 0.74 ± 0.19 Aa

**Leaf (g/kg)**

500.20 ± 10.59 Ae

15.81 ± 2.30 Cd

0.98 ± 0.24 Ca

5.13 ± 2.23 Cb

1.25 ± 0.15 Ca

the return of C, N, and S decreased through time, while the return of P and K increased;

**Table 6.** Characteristics of annual average nutrient return of litterfall in different years.

**Branches (g/kg)**

491.33 ± 9.67 Ad

9.98 ± 2.42 Ac

0.66 ± 0.40 Ba

3.25 ± 2.90 Bb

0.76 ± 0.17 Aa

*2.3. Dynamic Characteristics of Litterfall Nutrient Element Return*

the return of Ca and Mg first decreased and then increased.

**Leaf (g/kg)**

553.96 ± 24.60 Bc

13.72 ± 1.66 Bb

0.66 ± 0.12 Ba

3.73 ± 0.86 Ba

1.18 ± 0.09 Ca

**Nutrient Element**

C

N

**Branches (g/kg)**

542.27 ± 25.34 Bc

10.30 ± 2.07 Ab

<sup>P</sup> 0.51 <sup>±</sup> 0.15 Aa

<sup>K</sup> 2.17 <sup>±</sup> 1.14 Aa

<sup>S</sup> 0.93 <sup>±</sup> 0.21 Ba

### *2.3. Dynamic Characteristics of Litterfall Nutrient Element Return*

### 2.3.1. Interannual Return Characteristics of Litterfall Nutrient Elements

We found that the return of leaf litter (except Ca) was higher than that of branches litter, and the return of C was much higher than that of other nutrient elements. In all, the returns were roughly sorted as C > N > Ca > K > Mg > S > P (Table 6). Overall, we found a downward trend of the annual mean return of nutrients from 2005 to 2015. In branches, the return of C, N, and S decreased through time, while the return of P and K increased; the return of Ca and Mg first decreased and then increased.


**Table 6.** Characteristics of annual average nutrient return of litterfall in different years.

Note: Different uppercase letters in the same row indicate significant differences between different years (*p* < 0.05), and different lowercase letters in the same column indicate significant differences between different nutrient elements in the same year (*p* < 0.05).

### 2.3.2. Characteristics of Intra-Year Return of Litterfall Nutrient Elements

In Figure 4, we show that the return of each element varies greatly throughout the year. The return of C is the largest, with a peak May. The return of N and S showed a multimodal distribution at the beginning of the year, mid-year, and at the end of the year, but there was a sudden decrease in July. K and P showed a peak in August. The return of Ca and Mg first decreased until around July and then increased.

In Table 7, we show that some, but not all, element returns are affected by meteorological factors. The returns of C, N, and S were not correlated with meteorological factors (temperature, precipitation, wind speed). The return of P and K was significantly negatively correlated with wind speed. The return of K was also positively correlated with precipitation, while the return of Ca was negatively correlated with precipitation. Finally, the return of Mg was negatively correlated with temperature and precipitation.

**Figure 4.** Within‐year nutrient return of each element. **Figure 4.** Within-year nutrient return of each element.


**Table 7.** Correlation coefficient between meteorological factors and nutrient return. **Table 7.** Correlation coefficient between meteorological factors and nutrient return.

Mg 0.001 \*\* 0.655 0.008 \*\* 0.527 0.247 0.131 Note: \* indicates significant correlation (*p* < 0.05), \*\* indicates very significant correlation (*p* < 0.01).

#### Note: \* indicates significant correlation (*p* < 0.05), \*\* indicates very significant correlation (*p* < 0.01). *2.4. Biological Cycle of Nutrient Elements in Evergreen Broad-Leaved Forests of Mount Ailao*

Nutrient cycling refers to the absorption of nutrients from the soil by plant, some of which is used for plant growth, while the rest is returned to the soil through litterfall, secretions and rainwater. This is given by absorption, retention, and return of the three links, where the cycle balance formula is: absorption = retention + return [18,19]. In our study, we were only able calculate the return of litterfall, thus underestimating the total cycle. Nevertheless, we can use a utilization coefficient, circulation coefficient,

and turnover time to estimate elements of the cycle [20]. The nutrient utilization coefficients were distributed from 0.23 to 0.29, with an average value of 0.25, which showed Ca > Mg > S, C > N, K > P. The circulation coefficients were distributed from 0.42 to 0.84, with an average value was 0.53, showing P > K > N > C > S > Mg > Ca. The turnover time was distributed from 8.40a to 14.14a, with an average value of 10.50a, which was manifest as Ca > Mg > S > C > N > K > P (Table 8).


**Table 8.** Biological cycle of different nutrients.

### **3. Materials and Methods**

### *3.1. Overview of the Study Area*

Our study area was located in the Xujiaba area (24◦320 N, 102◦010 E) within the Ailaoshan National Nature Reserve, Jingdong Yi Autonomous County, Puer City, Yunnan Province. The study area occurred at an altitude of 2400–2600 m, and the soil was fertile, acidic, yellow-brown soil [21]. The climate at the Ailao Mountain Forest Ecosystem Research Station is southwestern monsoon, with distinct dry and wet seasons (annual average temperature is 11.3 ◦C and the annual average precipitation is 1931 mm.), at the transition from central subtropical to south Asian tropics. Our study site is within the largest area of the primitive Zhongshan wet, evergreen, broad-leaved forest preserved in China, with a closed canopy and layered shrub layer with abundant epiphytes on the trees. The dominant tree species at the site include *Machilus bombycina*, *Populus rotundifolia*, *Schima noronhae*, *Castanopsis rufescens*, and *Castanopsis delavayi* [22].

### *3.2. Litterfall Sampling and Collection*

We established a litterfall collection grid within a 1-ha long term observation plot of the forest. Specifically, we divided the plot into 100 10 m × 10 m subplots. From those 100 subplots, we randomly selected 25 and placed a 1 m<sup>2</sup> litterfall collection basket in each. We constructed litterfall collection baskets out of steel frame boxes 1 m × 1 m × 0.25 m covered with 0.5 mm nylon mesh. We inserted the four corners of each basket into the soil such that the bottom of the basket was about 0.5 m from the ground.

We collected litterfall from each basket at the end of each month from 2005 to 2015. We sorted litterfall into categories, including branches, leaves, fallen flowers and fruits, bark, moss and lichen, and other debris, drying each in an oven at 65 ◦C to a constant weight, and then recorded the dry weight of each component.

### *3.3. Meteorological Data Observation*

Meteorological data, including precipitation, temperature, and wind speed, were collected from the Ailao Mountain Meteorological Station. Data were averaged monthly and according to season (Figure 5). The wet season is from June to October, while the dry season is from November to May.

*3.3. Meteorological Data Observation*

son is from November to May.

Meteorological data, including precipitation, temperature, and wind speed, were col‐ lected from the Ailao Mountain Meteorological Station. Data were averaged monthly and according to season (Figure 5). The wet season is from June to October, while the dry sea‐

**Figure 5.** (**a**) Ailao Mountain Meteorological Station has average monthly precipitation, average monthly temperature for many years, average monthly wind speed for many years, (**b**) tempera‐ **Figure 5.** (**a**) Ailao Mountain Meteorological Station has average monthly precipitation, average monthly temperature for many years, average monthly wind speed for many years, (**b**) temperature, precipitation and wind speed from 2005 to 2017.

### ture, precipitation and wind speed from 2005 to 2017. *3.4. Measuring Nutrients in the Litterfall*

*3.4. Measuring Nutrients in the Litterfall* We estimated important elements in the litterfall after drying by first grounding the We estimated important elements in the litterfall after drying by first grounding the littler into a fine powder, which was subsequently sieved through a 250-µm mesh. We

tion, and then explore the correlation between them.

We calculated the annual return of litterfall nutrients as follows:

littler into a fine powder, which was subsequently sieved through a 250‐μm mesh. We determined carbon (C) and nitrogen (N) using a carbon analyzer (EA3000 EuroVector, Milan, Italy) [23,24]. To prepare samples for phosphorus (P) and potassium (K), we first

lybdenum antimony colorimetry, potassium (K) was measured using plasma atomic emis‐ sion spectrophotometer [2], and the concentration of sulfur (S), calcium (Ca) and magne‐

We averaged the total litterfall output across the 25 collection baskets and took monthly and annual (sum of the 12 months) data for analyses. After using the Shapiro– Wilk test to test the normality of the data, one‐way ANOVA and LSD were used to com‐ pare the difference in the amount of litterfall in different parts of different years and its components, the concentration of nutrient elements, and the amount of return. The ANOVA is mainly used to test whether there is a significant difference in the mean of the components of litterfall between the interannual and intra‐annual periods. The use of mul‐ tiple comparisons can support a better understanding of the differences between them. In addition, we also use SPSS26.0 to linear fit the environmental variables (temperature, pre‐ cipitation, wind speed) and the output of components of litterfall and nutrient concentra‐

sium (Mg) was measured using a flame photometer and spectrophotometer [25].

*3.5. Statistical Analysis of Data*

determined carbon (C) and nitrogen (N) using a carbon analyzer (EA3000 EuroVector, Milan, Italy) [23,24]. To prepare samples for phosphorus (P) and potassium (K), we first digested samples in H2O2-H2SO4. We determined phosphorus (P) concentration using molybdenum antimony colorimetry, potassium (K) was measured using plasma atomic emission spectrophotometer [2], and the concentration of sulfur (S), calcium (Ca) and magnesium (Mg) was measured using a flame photometer and spectrophotometer [25].

### *3.5. Statistical Analysis of Data*

We averaged the total litterfall output across the 25 collection baskets and took monthly and annual (sum of the 12 months) data for analyses. After using the Shapiro–Wilk test to test the normality of the data, one-way ANOVA and LSD were used to compare the difference in the amount of litterfall in different parts of different years and its components, the concentration of nutrient elements, and the amount of return. The ANOVA is mainly used to test whether there is a significant difference in the mean of the components of litterfall between the interannual and intra-annual periods. The use of multiple comparisons can support a better understanding of the differences between them. In addition, we also use SPSS26.0 to linear fit the environmental variables (temperature, precipitation, wind speed) and the output of components of litterfall and nutrient concentration, and then explore the correlation between them.

We calculated the annual return of litterfall nutrients as follows:

$$L\mathbf{a} = \sum\_{i=1}^{12} \sum\_{j=1}^{25} Lij\mathbf{C}ij/100\tag{1}$$

where *L<sup>a</sup>* is the annual amount of nutrients returned by litterfall; *Lij* is the litterfall amount of the *j*th component in the *i*th month (kg/m<sup>2</sup> ); *Cij* is the nutrient concentration (g/kg) of the *j*th component of litterfall in the *i*th month [26].

The biocirculating coefficient mainly includes nutrient utilization coefficient, cycle coefficient, and turnover time. The nutrient utilization coefficient is the ratio of the elements absorbed by the plant per unit time and unit area to the existing elements of the plant, and the calculation method of nutrient utilization is mainly based on the Chapin index.

$$E = \mathbf{A}\_{\mathbf{P}} / \mathbf{M} \tag{2}$$

It can be seen from Equation (2) that *E* is Chapin index, M is plant biomass, and A<sup>p</sup> is nutrient storage (t/ha). In essence, Chapin index is the average content of plant body nutrients, which reflects the amount of nutrients consumed by the plant construction unit per unit biomass. However, Chapin index has a bias in overestimating the nutrient utilization efficiency of trees, and formula (2), revised from the perspective of nutrient cycle to obtain formula (3), can also better reflect the nutrient utilization status of trees.

$$R\_{\mathbf{e}} = \mathbf{F}\_{\mathbf{a}} / \mathbf{A}\_{\mathbf{p}} \tag{3}$$

where *R*<sup>e</sup> is the utilization coefficient, F<sup>a</sup> is the nutrient uptake (t/ha a), and A<sup>p</sup> is the nutrient storage (t/ha).

The nutrient cycle coefficient is a kind of index proposed based on the concept of the biological cycle, also known as the biological return coefficient. The method has certain limitations in reflecting the overall situation of forest nutrient cycling, since the calculation of the nutrient cycling coefficient does not involve the decomposition of forest litterfall, and the decomposition of litterfall is an important link in the nutrient cycle. The calculation is performed as follows:

$$R\_{\mathcal{S}} = \mathbf{F\_d} / \mathbf{F\_a} \tag{4}$$

where *R<sup>g</sup>* is the cycle coefficient, F<sup>a</sup> is the nutrient uptake (t/ha a), and F<sup>d</sup> is the nutrient return (t/ha a).

Turnover time is the time it takes for a nutrient element to go through one cycle.

$$T\_t = \mathbf{F\_d} / \mathbf{A\_p} \tag{5}$$

where *T<sup>t</sup>* is the turnover time, A<sup>p</sup> is the nutrient storage (t/ha), and F<sup>d</sup> is the nutrient return (t/ha a) [27].

### **4. Discussion**

Litterfall volume is a component of the forest ecosystem biomass, which reflects the primary productivity level as well as the functions of the forest ecosystem [25,28]. The research shows that average annual litterfall of evergreen broad-leaved forest is 6.96 t/ha [29], the average annual litterfall of Yuanjiang savanna ecosystem is 2.5–3 t/ha [30], and the average litterfall of Chinese grassland is 0.59 t/ha [31]. It can be seen that fallen forest materials play a very important role in the global ecosystem. Our results showed that the average annual litterfall from 2005 to 2015 was 8.11 t/ha. This is similar to that observed in evergreen broadleaved forests in other subtropical regions (e.g., Dinghushan South Asian tropical evergreen broadleaved forest (7–11 t/ha) [32], and Xiaokeng subtropical evergreen broadleaf forest (7.99–8.45 t/ha) [33]. Furthermore, Guan Xin summarized the research results of central, subtropical, evergreen broadleaved forests and found that the annual litterfall recovery ranged from 3.90 to 7.72 t/ha [34]. Thus, the amount of litterfall observed at our site was intermediate between the South Asian tropical monsoon, evergreen, broad-leaved forest and the central, subtropical, evergreen, broad-leaved forest; its litterfall production is much higher than other ecosystems.

When compared to evergreen, broadleaf forests in other subtropical regions, we found that the ratio of leaf litter to total litter was less than that of Wuyishan rice oak forest (77.03 ± 1.93%) [35], Guangxi Longgang National Nature Reserve (85%) [36], Baishan Zu evergreen broadleaf forest (51.34%) [37] and Zhejiang Ningbo Tiantong Mountain evergreen broadleaved forest (50.7%) [38]. The high variability we observed within and among years for total litter and its components was consistent with the results of Zou Bingzhang [39]. Likewise, we found that most litterfall came from leaves with less from other sources, which was consistent with the results of Wan Chunhong [40]. The size and dynamic changes of forest litterfall output are influenced by many factors and are the result of a combination of factors [41,42]. We found that the overall amount of litterfall on the ground is high, which represents positive feedback for the forest ecosystem. It provides abundant food sources for forest organisms, especially fruits and flowers, which play an important role in the survival and reproduction of rodents. This also indirectly ensures the survival and reproduction of birds that feed on rodents. On the other hand, the litterfall decomposes and releases nutrients, keeping the fertility of the research site at a high level, as well as changing the physical properties of the forest soil. Litterfall also has a strong water retention capacity, which can reduce water evaporation and maintain sufficient water storage on the forest surface, which is important for water conservation and maintaining soil environment stability [43,44]. The presence of litterfall plays a foundational role in the entire forest ecosystem. We also found that litterfall dynamics correlated with a number of features of the environment, including wind speed, precipitation, and temperature. In the one exceptionally unusual month year in our survey (January 2015), branch content was high as a result of an unusual amount of snow in this period (1129.2 mm), which created considerable tree and branch fall in this period. This indicates that, after the interference of extreme ice and snow weather, the leaves were violently shaken by strong external forces, resulting in non physiological shedding.

The nutrient concentration of litterfall is related to the characteristics of the plants and the soil nutrient content [45]. We found that the nutrient concentration of litterfall was roughly the same in different years and different litterfall components. Specifically, we found that C concentration was the highest, Ca and N concentration were second, and K, Mg, S, and P concentrations were relatively low, consistent with the results of Chen Jinlei and Xue Fei [46,47]. Our finding that different nutrients varied through time is similar

to the research results of Xue Fei and Zhao Chang [45,47]. Our results show that many elements are correlated with temperature and precipitation, which may be influenced by the relationship between these elements and plant growth [48]. For example, K is highly mobile, and P is easily affected by multiple factors, such as vegetative growth rhythms, rainfall leaching, microbial degradation, etc.

The return of litterfall nutrients to the soil is influenced by a combination of factors. We found that the return of nutrient elements in litterfall showed a decreasing trend through the year. Our annual nutrient return size of litterfall is roughly C > N > Ca > K > Mg > S > P, which differs from the results of Liu Yi, but is similar to those of Liu Lei and Gao Shilei [49–51]. Furthermore, the annual return amount of litterfall C in our study area (6.21 t/ha) was within the range of litterfall C return observed (0.05~7.50 t/ha) across the world's forested ecosystems [52]. Likewise, our results regarding the return of C, N, and P were higher than those of Mijiao natural forest in Sanming City, Fujian Province and Wuyi Mountain evergreen, broadleaved forest [53,54]. This suggests that evergreen, broadleaf forest literfall in at our study site in the Ailao Mountain plays a very important role in the carbon cycle of the soil. Additionally, the litterfall results in modifications in forest ecosystems, particularly in subtropical, evergreen, broad-leaved forests because the amount of litterfall can regulate the micro-climate in the soil, affecting the decomposition rate with the changes in microbial community, and soil microorganisms in the soil affect soil respiration [55].

The utilization coefficient, circulation coefficient, and turnover time are all important parameters in the nutrient cycling process, and the nutrient cycling parameters also vary due to the difference between nutrient uptake and return by different forest types [56]. The utilization coefficient is the ratio of absorption to storage, reflecting the storage rate of ecosystem elements; the larger the coefficient, the greater the storage capacity of plants and the lower the utilization efficiency. The nutrient utilization coefficient in our study area was 0.25, which was lower than that found in the karst peak cluster depression in Huanjiang, Guangxi (0.35) and the four-year-old Mazhan Acacia plantation (0.51) in the state-owned peak forest farm in Nanning City, Guangxi [20,57]. This indicates that the forest in our study system had high nutrient utilization efficiency and low storage capacity. The circulation coefficient is the ratio of plant return to absorption, and reflects the size of the residual amount of the element during the cycle; the larger the coefficient, the faster the rate of element circulation and the greater the fluidity. The cycling coefficient in our study (0.53) was higher than that of Gongga Mountain Natural Forest (0.474) [58], but lower than that of Dinghu Mountain Horsetail Pine Forest (0.68) and *Pinus tabulosis* forest (0.71–0.85) in the loess hilly area [19,59]. This indicates that the forest in the study area had high nutrient cycling capacity. Turnaround time is the ratio of a plant's total nutrient storage to return, indicating the time it takes for a nutrient element to go through the cycle. The longer the turnaround time, the longer the nutrients stay in plants. In this study, we found that the turnover time ranged from 8.40 a to 14.14 a (average value of 10.50 a), which was manifested as Ca > Mg > S > C > N > K > P, suggesting that Ca and Mg were inactive elements and P was the active element. Meantime, this also indicates that the plants in the research area absorb nutrients quickly, grow quickly, and have high yield and large biomass.

### **5. Conclusions**

In this paper, we present the results of an 11-year study investigating the dynamic changes of litterfall output, nutrient concentration, and return in a wet, evergreen, broadleaf forest in the Ailao Mountains of China to draw the following conclusions:

From 2005 to 2015, the total litter of evergreen, broadleaved forests in the Ailao Mountains was 7.70–9.46 t/ha a, and the interannual fluctuation of litterfall was large, with an average of 8.11 ± 0.73 t/ha, which was higher than that of Central Asian, thermal, evergreen, broadleaved forests. The presence of a large amount of litterfall provides nutrients to the study area, promoting the development and stability of the study area's ecosystem and ensuring the fertility of the soil and biodiversity. The output of litterfall was signif-

icantly different between different years. There are significant interannual and seasonal variations in the amount of litterfall, mostly bimodal (peak from March to May, October to November), with higher levels of leaf litter than other components in each year, some of which was correlated with meteorological factors (*p* < 0.05). The article only discussed the meteorological factor as the cause of the litterfall, indicating that the production of litterfall is the result of multiple factors. We generally found the nutrient concentration was sorted as C > Ca > N > K > Mg > S > P, except for the slight difference in Ca and N, this also conforms to the general pattern of changes in forest nutrient concentrations. The nutrient concentration and returned amount of litterfall are related to the growth of trees, as well as variation in meteorological factors, such as temperature, precipitation, and wind speed. Our results showed that although there was nutrient loss in the evergreen, broad-leaved forest area of the Ailao Mountains, forest litterfall could still maintain soil fertility in the area, maintaining the normal operation of the entire forest ecosystem.

**Author Contributions:** S.D. conceived and designed the study, oversaw data collection, Z.X. provide experimental materials and assist in the inspection and analysis of related chemical elements, and conducted all litterfall analyses; S.D., T.W., J.T. and H.G. analyzed the data and co-wrote the paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** The data are not publicly available due to the dataset is proprietary and the author currently does not have the rights to make the data public.

**Acknowledgments:** The data of this paper are collected and obtained by the staff of the Ailao Mountain Ecological Station of Xishuangbanna Botanical Garden of the Chinese Academy of Sciences for many years, especially the determination of nutrients and the acquisition of field materials. Here, I would like to especially thank Li Dawen for the sampling work.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

**Xingyu Zhou 1,†, Jiaxun Xin 2,†, Xiaofei Huang 1,3, Haowen Li <sup>2</sup> , Fei Li 1,3,\* and Wenchen Song 2,\***


**Abstract:** Plant leaf functional traits can reflect the adaptive strategies of plants to environmental changes. Exploring the patterns and causes of geographic variation in leaf functional traits is pivotal for improving ecological theory at the macroscopic scale. In order to explore the geographical variation and the dominant factors of leaf functional traits in the forest ecosystems of China, we measured 15 environmental factors on 16 leaf functional traits in 33 forest reserves in China. The results showed leaf area (LA), carbon-to-nitrogen ratio (C/N), carbon-to-phosphorus ratio (C/P), nitrogen-to-phosphorus ratio (N/P), phosphorus mass per area (Pa) and nitrogen isotope abundance (δ <sup>15</sup>N)) were correlated with latitude significantly. LA, Pa and δ <sup>15</sup>N were also correlated with longitude significantly. The leaf functional traits in southern China were predominantly affected by climatic factors, whereas those in northern China were mainly influenced by soil factors. Mean annual temperature (MAT), mean annual precipitation (MAP) and mean annual humidity (MAH) were shown to be the important climate factors, whereas available calcium (ACa), available potassium (AK), and available magnesium (AMg) were shown to be the important climate factors that affect the leaf functional traits of the forests in China. Our study fills the gap in the study of drivers and large-scale geographical variability of leaf functional traits, and our results elucidate the operational mechanisms of forest–soil–climate systems. We provide reliable support for modeling global forest dynamics.

**Keywords:** leaf functional traits; climate; soil; geographical variation; δ <sup>13</sup>C; δ <sup>15</sup>N; forest ecosystem

### **1. Introduction**

Leaf functional traits can not only reflect plant growth, metabolism and reproduction [1] but also represent plant adaptation strategies to different ecological environments [2,3]. In recent decades, patterns of geographic variation in leaf functional traits at large spatial scales have been paid much attention [4–6]. Understanding the geographical variation in leaf functional traits and its relationship with environmental factors can improve the predictions of vegetation changes [7–10], the large-scale mapping of plant function types [11], analysis of the community structure [12] and dynamic modeling of global vegetation [13,14]. At present, the mainstream view is that the variation in leaf functional traits is mainly affected by climate factors at large scale [15–21]. Temperature and precipitation were shown to be the most important climate factors, and they have been well known as the dominant factors that affect leaf functional traits, such as leaf area (LA), special leaf area (SLA), nitrogen mass (Nm), leaf phosphorus mass (Pm) and nitrogen-tophosphorus ratio (N/P) [16,22–24]. Humidity is also one of the climate factors affecting leaf functional traits (e.g., Nm, N/P) [25,26]. Furthermore, evapotranspiration has been

**Citation:** Zhou, X.; Xin, J.; Huang, X.; Li, H.; Li, F.; Song, W. Linking Leaf Functional Traits with Soil and Climate Factors in Forest Ecosystems in China. *Plants* **2022**, *11*, 3545. https://doi.org/10.3390/ plants11243545

Academic Editors: Jie Gao, Weiwei Huang, Johan Gielis and Peijian Shi

Received: 22 November 2022 Accepted: 12 December 2022 Published: 15 December 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

found to explain 30% of the variation in leaf area [27], and it is one of the limiting factors for some leaf functional traits (e.g., leaf dry matter content (LDMC), Pm, SLA, Nm) [28].

In recent years, soil factors have been shown to be one of the main factors affecting the functional traits at large scale [4,17–19,29,30], because the plant–soil interactions are typically the major determinants of changes in processes and functions in forest ecosystems [31]. Leaf functional traits, such as leaf mass per area (LMA), Nm and Pm, are strongly correlated with soil factors, such as available phosphorus (AP), available potassium (AK) available nitrogen (AN) and pH [16,32–34]. The carbon-to-nitrogen ratio of leaves (C/N) is associated with the soil salinity [35]. Besides, total nitrogen content (TN) is the main factor affecting LA and SLA, while the total organic carbon (TOC) in soil mainly affects the Pm and Nm of leaves [36]. Soil microorganisms play an important ecological role in soil formation and the protection and regulation of nutrient cycling [37], further changing the physiological adaptation of plants and making soil factors affect leaf functional traits in forest ecosystems [38].

The large latitudinal span of forest ecosystems in China covers all four major forest ecosystem types from north to south, namely, boreal coniferous forests, temperate deciduous broadleaf forests, subtropical evergreen broadleaf forests and tropical forests, making it a valuable sample place for studying geographic variation in leaf functional traits at large scale. Therefore, the variation patterns of leaf functional traits along the gradients of environmental factors (e.g., temperature, precipitation, soil elements and topography) have been widely studied in several climatic regions of China, such as the Tibetan Plateau region [39], the Sanjiangyuan region of northeast China [40], the Karst region of southwest China [35,41], the arid and semi-arid region of northwest China [15,23,42], the deciduous broadleaf forest region of southeastern China [15,43] and the Loess Plateau region of central China [36]. However, at the national scale, only a few reports have explored geographical variation, and most of these studies focused on several species [16,36,44]. Herein, an important question was asked: What are the main factors affecting leaf functional traits in China? The answer to this question will contribute to the global understanding of leaf functional trait variation.

To address this question, we explored the following two hypotheses:


In this study, we sampled soils and leaves from 33 forest reserves in China to test which of the two hypotheses are more realistic. The aim of this study is to link leaf functional traits with soil and climate factors in forest ecosystems in China.

### **2. Results**

### *2.1. Geographical Variation*

The leaf area (LA) and phosphorus mass per area (Pa) decreased with increasing latitude (Figure 1a,b); however, the respective curves did not show a good fit (R<sup>2</sup> = 0.37, *p* < 0.01; R<sup>2</sup> = 0.37, *p* < 0.01, respectively). The carbon-to-phosphorus ratio (C/P), nitrogen-tophosphorus ratio (N/P) and nitrogen isotope abundance (δ <sup>15</sup>N) increased with increasing latitude (Figure 1d–f). δ <sup>15</sup>N showed the strongest correlation with latitude (R<sup>2</sup> = 0.53, *p* < 0.01), whereas those of C/P and N/P were weak, albeit significant (R<sup>2</sup> = 0.24, *p* < 0.01; R <sup>2</sup> = 0.36, *p* < 0.01, respectively). In the range of 20–40◦ N, carbon-to-nitrogen ratio (C/N) decreased with increasing latitude, and in the range of 40–55◦ N, C/N increased slowly with increasing latitude (R<sup>2</sup> = 0.24, *p* < 0.05) (Figure 1c).

*Plants* **2022**, *11*, x FOR PEER REVIEW 3 of 14

**Figure 1.** Relationships between latitude and (**a**) LA, (**b**) Pa, (**c**) C/N, (**d**) C/P, (**e**) N/P and (**f**) δ15N, respectively. **Figure 1.** Relationships between latitude and (**a**) LA, (**b**) Pa, (**c**) C/N, (**d**) C/P, (**e**) N/P and (**f**) δ <sup>15</sup>N, respectively. **Figure 1.** Relationships between latitude and (**a**) LA, (**b**) Pa, (**c**) C/N, (**d**) C/P, (**e**) N/P and (**f**) δ15N, respectively.

The LA of leaves from the south was, on average, higher than that of leaves in the north (Figure 2a), which was consistent with the conclusions drawn in Figure 2a. The LA of leaves from the south had a significant but weak negative correlation with longitude (R2 = 0.33, *p* < 0.01). Leaves from the north showed a significant and stronger correlation with longitude (R2 = 0.62, *p* < 0.01). Between 105° E and 120° E, LA was negatively correlated with longitude, and between 120° E and 130° E, LA was positively correlated with longitude. The phosphorus mass per area (Pa) of southern leaves was generally higher than that of leaves in the north, which is consistent with the conclusion obtained from Figure 1b (Figure 2b). With increasing longitude, the Pa of the southern leaves first decreased and then increased, with the lowest value at 110° E (R2 = 0.46, *p* < 0.05). The Pa of the northern leaves increased with longitude and then decreased, with the highest value at 112 °E (R2 = 0.30, *p* < 0.05). The degree of variation of δ15N with longitude was not significant in the south and the north, but as a whole, δ15N showed a significant positive correlation with longitude (R2 = 0.32, *p* < 0.01) (Figure 2c). The LA of leaves from the south was, on average, higher than that of leaves in the north (Figure 2a), which was consistent with the conclusions drawn in Figure 2a. The LA of leaves from the south had a significant but weak negative correlation with longitude (R<sup>2</sup> = 0.33, *p* < 0.01). Leaves from the north showed a significant and stronger correlation with longitude (R<sup>2</sup> = 0.62, *p* < 0.01). Between 105◦ E and 120◦ E, LA was negatively correlated with longitude, and between 120◦ E and 130◦ E, LA was positively correlated with longitude. The phosphorus mass per area (Pa) of southern leaves was generally higher than that of leaves in the north, which is consistent with the conclusion obtained from Figure 1b (Figure 2b). With increasing longitude, the Pa of the southern leaves first decreased and then increased, with the lowest value at 110◦ E (R<sup>2</sup> = 0.46, *p* < 0.05). The Pa of the northern leaves increased with longitude and then decreased, with the highest value at 112 ◦E (R<sup>2</sup> = 0.30, *p* < 0.05). The degree of variation of δ <sup>15</sup>N with longitude was not significant in the south and the north, but as a whole, δ <sup>15</sup>N showed a significant positive correlation with longitude (R<sup>2</sup> = 0.32, *p* < 0.01) (Figure 2c). The LA of leaves from the south was, on average, higher than that of leaves in the north (Figure 2a), which was consistent with the conclusions drawn in Figure 2a. The LA of leaves from the south had a significant but weak negative correlation with longitude (R2 = 0.33, *p* < 0.01). Leaves from the north showed a significant and stronger correlation with longitude (R2 = 0.62, *p* < 0.01). Between 105° E and 120° E, LA was negatively correlated with longitude, and between 120° E and 130° E, LA was positively correlated with longitude. The phosphorus mass per area (Pa) of southern leaves was generally higher than that of leaves in the north, which is consistent with the conclusion obtained from Figure 1b (Figure 2b). With increasing longitude, the Pa of the southern leaves first decreased and then increased, with the lowest value at 110° E (R2 = 0.46, *p* < 0.05). The Pa of the northern leaves increased with longitude and then decreased, with the highest value at 112 °E (R2 = 0.30, *p* < 0.05). The degree of variation of δ15N with longitude was not significant in the south and the north, but as a whole, δ15N showed a significant positive correlation with longitude (R2 = 0.32, *p* < 0.01) (Figure 2c).

**Figure 2.** Relationships between longitude and (**a**) LA, (**b**) Pa and (**c**) δ15N, respectively. **Figure 2.** Relationships between longitude and (**a**) LA, (**b**) Pa and (**c**) δ15N, respectively. **Figure 2.** Relationships between longitude and (**a**) LA, (**b**) Pa and (**c**) δ <sup>15</sup>N, respectively.

#### *2.2. Environmental Factors Affecting Functional Traits of Leaves 2.2. Environmental Factors Affecting Functional Traits of Leaves 2.2. Environmental Factors Affecting Functional Traits of Leaves*

The RDA showed that the first axis explained 41.74% of the relationship of leaf functional traits and environmental factors, and the second axis explained 20.84%. The mean annual precipitation (MAP), mean annual temperature (MAT), mean annual humidity (MAH), available magnesium (AMg), available potassium (AK) and pH were The RDA showed that the first axis explained 41.74% of the relationship of leaf functional traits and environmental factors, and the second axis explained 20.84%. The mean annual precipitation (MAP), mean annual temperature (MAT), mean annual humidity (MAH), available magnesium (AMg), available potassium (AK) and pH were The RDA showed that the first axis explained 41.74% of the relationship of leaf functional traits and environmental factors, and the second axis explained 20.84%. The mean annual precipitation (MAP), mean annual temperature (MAT), mean annual humidity (MAH), available magnesium (AMg), available potassium (AK) and pH were more strongly correlated with leaf functional traits, followed by correlations with available aluminum (AAl), available phosphorus (AP) and electrical conductivity (EC). AMg, AK, pH, EC and

AP, all of which were positively correlated with the first axis, and AMg, AK and pH which were closely associated with the first axis. MAP, MAH, MAT and AAl were negatively correlated with the first axis, with MAT being the most closely related to the first axis. Total potassium (TK) was closely and positively correlated with the second axis, and total organic carbon (TOC) and AP were closely and negatively correlated with the second axis (Figure 3). AK and pH which were closely associated with the first axis. MAP, MAH, MAT and AAl were negatively correlated with the first axis, with MAT being the most closely related to the first axis. Total potassium (TK) was closely and positively correlated with the second axis, and total organic carbon (TOC) and AP were closely and negatively correlated with the second axis (Figure 3). We found that the factors influencing the functional traits of southern leaves dif-

more strongly correlated with leaf functional traits, followed by correlations with available aluminum (AAl), available phosphorus (AP) and electrical conductivity (EC). AMg, AK, pH, EC and AP, all of which were positively correlated with the first axis, and AMg,

We found that the factors influencing the functional traits of southern leaves differed significantly from those of northern leaves (Figure 3). Southern leaf functional traits were mainly influenced by climate factors (MAP, MAH, MAT) and to a lesser extent by soil factors (TK, AAl, TOC). Northern leaf functional traits were mainly influenced by soil factors (AP, TP [total phosphorus], TN [total nitrogen], EC, pH, AK, ACa, AMg) and to a lesser extent by climate factors (MAE). The area of the ellipse where the functional traits of southern leaves were located was significantly smaller than that of the ellipse where the functional traits of the northern leaves were located, thus it can be assumed that the intra-group variability among the functional traits of the southern leaves is smaller than that of the northern leaves. fered significantly from those of northern leaves (Figure 3). Southern leaf functional traits were mainly influenced by climate factors (MAP, MAH, MAT) and to a lesser extent by soil factors (TK, AAl, TOC). Northern leaf functional traits were mainly influenced by soil factors (AP, TP [total phosphorus], TN [total nitrogen], EC, pH, AK, ACa, AMg) and to a lesser extent by climate factors (MAE). The area of the ellipse where the functional traits of southern leaves were located was significantly smaller than that of the ellipse where the functional traits of the northern leaves were located, thus it can be assumed that the intra-group variability among the functional traits of the southern leaves is smaller than that of the northern leaves.

*Plants* **2022**, *11*, x FOR PEER REVIEW 4 of 14

**Figure 3.** RDA between leaf functional traits and environmental factors. **Figure 3.** RDA between leaf functional traits and environmental factors.

We performed pairwise Pearson's correlation analyses between 16 leaf functional trait variables and 14 environmental variables and produced heat maps (Figure 4). A total of 23 relationship pairs reached significance at the *p* < 0.01 level, and 25 relationship pairs reached significance at the *p* < 0.05 level. Horizontally, all 14 leaf functional traits, apart from potassium mass per area (Ka) and carbon mass (Cm), were significantly influenced by some of the many environmental variables, to varying degrees. Longitudinally, the remaining nine environmental variables, except for EC, TOC, TN, TP and AP, were significantly correlated with certain leaf trait indicators. Among all environmental variables, AMg, MAP, MAT and MAH had the most significant effects on leaf functional traits, and they were significantly correlated with 7, 9, 10 and 8 leaf functional trait variables, respectively. According to the division criteria of a previous study [47], among all significant correlations, only ACa showed strong positive correlations with leaf mass per area We performed pairwise Pearson's correlation analyses between 16 leaf functional traitvariables and 14 environmental variables and produced heat maps (Figure 4). A total of23 relationship pairs reached significance at the *<sup>p</sup>* < 0.01 level, and 25 relationship pairs reached significance at the *<sup>p</sup>* < 0.05 level. Horizontally, all 14 leaf functional traits, apart frompotassium mass per area (Ka) and carbon mass (Cm), were significantly influenced by some of the many environmental variables, to varying degrees. Longitudinally, the remaining nine environmental variables, except for EC, TOC, TN, TP and AP, were significantly correlated with certain leaf trait indicators. Among all environmental variables, AMg, MAP, MAT and MAH had the most significant effects on leaf functional traits, and they were significantly correlated with 7, 9, 10 and 8 leaf functional trait variables, respectively. According to the division criteria of a previous study [47], among all significant correlations, only ACa showed strong positive correlations with leaf mass per area (LMA) and nitrogen mass per area (Na), and most of the remaining correlations were moderately correlated in degree, with R<sup>2</sup> values generally ranging from 0.4 to 0.7, and a small number of correlations were weak.

(LMA) and nitrogen mass per area (Na), and most of the remaining correlations were moderately correlated in degree, with R2 values generally ranging from 0.4 to 0.7, and a

**Figure 4.** Heatmap of Pearson's correlations between environmental variables and leaf functional traits. Significant difference was indicated as follows: \* <0.05; \*\* <0.01. **Figure 4.** Heatmap of Pearson's correlations between environmental variables and leaf functional traits. Significant difference was indicated as follows: \* < 0.05; \*\* < 0.01.

#### **3. Discussion 3. Discussion**

### *3.1. Geographical Variation in Leaf Functional Traits*

small number of correlations were weak.

*3.1. Geographical Variation in Leaf Functional Traits*  The relationship between leaf physiological traits and latitude was not linear in any of our analyses (Figure 1), and this finding was consistent with those of a previous study [20]. The possible reason is that the presence of local habitat heterogeneity reduces the climate variation along the latitudinal gradient [4]. Leaf area (LA) decreased with latitude (Figure 1a), which was in line with the findings of a previous study [48]. Latitude did not have a significant relationship with specific leaf area (SLA) or leaf dry matter content (LDMC), in contrast to previous observations [20,48]. A significant positive correlation between the nitrogen-to-phosphorus ratio (N/P) and latitude was observed (Figure 2e), which was consistent with the observations of one previous study [38] but not with global patterns [45]. The law of N/P change with latitude suggests that, in China, nitrogen is restricted by low latitudes and mainly by phosphorus in high latitudes [49]. The reason for this pattern may be that the dependence of ectomycorrhizal fungi (EMF) on trees is higher at high latitudes than at low latitudes [38]. EMF can inhibit the mycorrhizal root colonization of neighboring arbuscular mycorrhizal herbs by promoting litter accumulation and limiting nutrient access [50,51], and they help trees absorb more N while enhancing their competitiveness, thus leading to higher N/P with increasing The relationship between leaf physiological traits and latitude was not linear in any of our analyses (Figure 1), and this finding was consistent with those of a previous study [20]. The possible reason is that the presence of local habitat heterogeneity reduces the climate variation along the latitudinal gradient [4]. Leaf area (LA) decreased with latitude (Figure 1a), which was in line with the findings of a previous study [48]. Latitude did not have a significant relationship with specific leaf area (SLA) or leaf dry matter content (LDMC), in contrast to previous observations [20,48]. A significant positive correlation between the nitrogen-to-phosphorus ratio (N/P) and latitude was observed (Figure 1e), which was consistent with the observations of one previous study [38] but not with global patterns [45]. The law of N/P change with latitude suggests that, in China, nitrogen is restricted by low latitudes and mainly by phosphorus in high latitudes [49]. The reason for this pattern may be that the dependence of ectomycorrhizal fungi (EMF) on trees is higher at high latitudes than at low latitudes [38]. EMF can inhibit the mycorrhizal root colonization of neighboring arbuscular mycorrhizal herbs by promoting litter accumulation and limiting nutrient access [50,51], and they help trees absorb more N while enhancing their competitiveness, thus leading to higher N/P with increasing latitude [52–55]. From a general perspective, δ <sup>13</sup>C showed no significant correlation with latitude, although fitting curves differed between north and south [56]. Leaf δ <sup>15</sup>N increased with increasing latitude (Figure 1f), probably because EMF colonization is also positively correlated with latitude [56]. EMF can supply relatively <sup>15</sup>N-enriched N to their hosts in the rhizosphere [57–59]; however, this was contrary to the global trend [46], and it also differed from the trend of increasing and then decreasing with latitude in the southern hemisphere [60].

Overall, most leaf functional traits showed no significant longitudinal trends and varied at random, as observed previously [61]. This may be due to the low number of our sampling sites and the lack of a significant trend in hydrothermal conditions between sample sites in the longitude direction [62]. The LA of leaves in southern China decreased significantly with increasing longitude (Figure 2a), which was also found in a study on southern *Taxus mairei* [61]. However, this trend cannot be explained exclusively by precipitation patterns [63]. In contrast, the LA of northern leaves first increased and then decreased with longitude ascending height (Figure 2a). A possible reason for this trend is that, over the longitude of northern sampling sites, the latitude first decreases and then increases, roughly corresponding to decreasing and then increasing warmth. From an overall perspective, the correlation between δ <sup>13</sup>C and longitude is not significant, which is consistent with the conclusions of previous studies [56,64]. δ <sup>15</sup>N showed a significant positive correlation with longitude (Figure 2c). A different study also found that δ <sup>15</sup>N was positively associated with longitude, but the correlation was not significant [56].

### *3.2. Factors Influencing Leaf Functional Traits*

### 3.2.1. Climate Factors

Mean annual temperature (MAT) showed a significant positive correlation with leaf area (LA) and phosphorus mass (Pm) (Figure 4), as observed previously [45,65]. Long-term monitoring results showed that leaf size increases with increasing temperature, which reflects plant adaptation [66]. The increase of MAT is conducive to the litter decomposition and nutrient circulation, increasing the mineralization rate of nitrogen in the soil, which favors the increase of available nitrogen (AN) in the soil and nitrogen mass (Nm) in the leaves [67]. However, a significant negative correlation between MAT and Nm was found in the present study (Figure 4), which probably occurred due to the presence of other soil elements limiting the growth of Nm [67]. Leaf carbon isotope abundance (δ <sup>13</sup>C) is markedly affected by MAT [38,64,68]. Temperature-related variables were found to exert stronger effects on δ <sup>13</sup>C than precipitation-related factors [12,69]. However, the direction and degree of influence of MAT differed between studies [64,70–72]. In the present study, MAT had a significant negative correlation with δ <sup>13</sup>C. MAT was previously shown to have a significant positive correlation with carbon mass (Cm) [73]; however, no such correlation was observed in the current study (Figure 5). No correlation between SLA and MAT was found, by contrast to other studies [20,74].

Mean annual precipitation (MAP) had a significant negative correlation with δ <sup>13</sup>C and LMA, as observed previously [64,75,76]. MAP is the strongest predictor of leaf δ <sup>13</sup>C among global climate variables, and it explains approximately half of the global variation in leaf δ <sup>13</sup>C [77]. When the soil water content and air humidity decrease due to insufficient precipitation, plants may reduce stomatal conductance or stomatal density, leading to improved water-use efficiency and positive leaf δ <sup>13</sup>C in plants [24,78–80]. Nm also had a significant negative correlation with MAP (Figure 4). This may be because, under water shortage, plants increase the allocation of N to the leaves, increase osmotic pressure in the cells, reduce the consumption of water by operating at lower stomatal conductance and improve water retention [81,82]. Leaf δ <sup>15</sup>N was negatively correlated with MAP (Figure 4), likely because high moisture levels reduce rhizomicrobial activity and the ability of mycorrhiza to obtain nutrition from decomposing soil organic matter [46,83,84]. However, no correlation of MAP and SLA (or LA) was observed, in contrast to previous studies [65,74,85,86]. This may be because MAP values at all sampling sites were high (>299 mm) and thus did not elicit plant stress.

Increasing water-table depth negatively and directly affects SLA [87], and mean annual evapotranspiration (MAE) is negatively correlated with water-table depth at a countrywide scale [88]. Thus, MAE should be positively correlated with SLA. However, in the current study, MAE and SLA were negatively correlated (Figure 4), which is contrary to

the prediction. This discrepancy occurs may be due to the different scales of regional studies [89]. MAH was significantly negatively correlated with δ <sup>13</sup>C (Figure 4), and Liu et al. obtained similar results using *Quercus variabilis* [75]. *Plants* **2022**, *11*, x FOR PEER REVIEW 7 of 14

**Figure 5.** Geographic distribution of the sampling sites. **Figure 5.** Geographic distribution of the sampling sites.

#### Mean annual precipitation (MAP) had a significant negative correlation with δ13C 3.2.2. Soil Mineral Elements

and LMA, as observed previously [64,75,76]. MAP is the strongest predictor of leaf δ13C among global climate variables, and it explains approximately half of the global variation in leaf δ13C [77]. When the soil water content and air humidity decrease due to insufficient precipitation, plants may reduce stomatal conductance or stomatal density, leading to improved water-use efficiency and positive leaf δ13C in plants [24,78–80]. Nm also had a significant negative correlation with MAP (Figure 4). This may be because, under water shortage, plants increase the allocation of N to the leaves, increase osmotic pressure in the cells, reduce the consumption of water by operating at lower stomatal conductance and AK had a significant positive effect on N/P (Figure 4), possibly because EMF colonization increases with increasing latitude [55,56], promoting tree roots to deposit their exudates into the soil to facilitate mineral transformation [90,91], thereby increasing AK content [92–94]. Higher AK concentrations are beneficial to soil EMF diversity [24,95,96], and EMF are considered to have a facilitating effect on N/P [55]. Nm was significantly positively influenced by ACa (Figure 4). Studies conducted in karst areas similarly found that alkaline soils with Ca2+ accumulation promote plant Nm [45]. AMg was positively correlated with δ <sup>15</sup>N and ACa with δ <sup>13</sup>C and LMA, which was observed previously.

#### improve water retention [81,82]. Leaf δ15N was negatively correlated with MAP (Figure 4), likely because high moisture levels reduce rhizomicrobial activity and the ability of **4. Conclusions**

mycorrhiza to obtain nutrition from decomposing soil organic matter [46,83,84]. However, no correlation of MAP and SLA (or LA) was observed, in contrast to previous studies [65,74,85,86]. This may be because MAP values at all sampling sites were high (>299 mm) and thus did not elicit plant stress. Increasing water-table depth negatively and directly affects SLA [87], and mean annual evapotranspiration (MAE) is negatively correlated with water-table depth at a country-wide scale [88]. Thus, MAE should be positively correlated with SLA. However, This study found that both climate factors and soil factors significantly affected the leaf functional traits of forests in China; these influences have obvious geographic variability. The leaf functional traits in southern China were predominantly affected by climate factors, whereas those in northern China were mainly influenced by soil factors. Mean annual precipitation (MAP), mean annual temperature (MAT) and mean annual humidity (MAH) were the major climate factors affected leaf functional traits, available magnesium (AMg), available potassium (AK) and available calcium (ACa) were the major soil factors. In this

in the current study, MAE and SLA were negatively correlated (Figure 4), which is contrary to the prediction. This discrepancy occurs may be due to the different scales of restudy, it is believed that climatic and geological variation processes dominate the geographical pattern of forest functional traits in China, and their impacts must be comprehensively considered in future studies.

### **5. Materials and Methods**

### *5.1. Soil and Leaf Sampling*

Leaf and soil samples were collected from 33 mountain forest reserves in China (Figure 1). These forest reserves are located in a latitude range of 21.40◦–53.56◦ N and a longitude range of 101.03◦–128.52◦ E. These areas were characterized by rich vegetation communities (tropical forest, subtropical forest, temperate deciduous broadleaf forest, temperate mixed coniferous–broadleaf forest and boreal forest), mean annual precipitation of 299–2210, mean annual temperature of −5.5–22.7 ◦C, mean annual humidity of 46.4%–80% and mean annual evapotranspiration of 604–1276 mm. In each forest reserve, 9–15 sampling plots (10 m × 10 m) were randomly selected along the same aspect of the mountain, and 5 topsoil samples (5 cm depth) were randomly collected in each plot and immediately stored in precooled polyethylene bags, at each plot (see Wang et al. [97] for details). Within each plots, 5–10 major species were selected and 20–50 leaf samples on individuals from 2–5 adult healthy trees of each species were collected (see Song and Zhou [56] for details). All meteorological data in the present study were downloaded from the National Meteorological Science Data Center of China (http://data.cma.cn; accessed on 14 April 2019).

### *5.2. Soil and Leaf Analysis*

Basic information on environmental variables and leaf functional traits is shown in Table 1. Soil pH and EC were measured using specific electrodes (Sartorius PB-10, Göttingen, Germany) and a soil suspension (soil and deionized water at 1:2.5). TOC was measured using a Shimadzu TOC Analyzer (TOC-Vcsh; Kyoto, Japan). The SmartChem Discrete Auto Analyzer (SmartChem 200; WESTCO Scientific Instruments Inc., Connecticut, USA) was used to measure TN and TP. Available Ca, Mg, Al, P and K were extracted using Mehlich-III solution and were measured through inductively coupled plasma optical emission spectrometry (ICP-OES, Optima 2100 DV; Perkin-Elmer, Waltham, MA, USA). TK was also measured by ICP-OES after sample digestion. The elemental content of C, N, P and K in the leaves was measured using TOC-Vcsh, SmartChem 200 and Optima 2100 DV instruments, respectively.


**Table 1.** Leaf functional traits and environmental variables.


**Table 1.** *Cont*.

After retrieving the leaf samples from the sample points, we performed the following experimental steps: (a) Samples were sorted and washed in distilled water. (b) Leaf thickness, root length and plant height were measured using a Vernier caliper. (c) The leaf area was calculated through Photoshop pixel analysis. (d) The samples were placed in a constant-temperature drying oven for 8 h at 75 ◦C. (f) An analytical balance was used to record plant mass.

To measure chemical properties, plants were ground with mortar and were sieved through 200 mesh. The values of C%, N%, δ <sup>13</sup>C and δ <sup>15</sup>N were measured using a Finnigan M.A.T 253 Isotope Ratio Mass Spectrometer and Flash 2000 EA-HT Elemental Analyzer (Thermo Fisher Scientific, Waltham, MA, USA). The measurement precisions for δ <sup>13</sup>C and δ <sup>15</sup>N were < <sup>±</sup>0.1‰ and < <sup>±</sup>0.2‰, respectively (see Song and Zhou [56] for further details).

The determination of δ <sup>13</sup>C and δ <sup>15</sup>N was based on the international standard Peedee Belemnite (PDB) formation and was calculated according to the following Equations (1) and (2):

$$\delta^{13}\text{C} = \left[\frac{\left(^{13}\text{C}/^{12}\text{C}\right)\_{\text{Sample}}}{\left(^{13}\text{C}/^{12}\text{C}\right)\_{\text{PDB}}} - 1\right] \cdot 1000\,\%\,\tag{1}$$

$$\delta^{15}\text{N} = \left[\frac{\left(^{15}\text{N}/^{15}\text{N}\right)\_{\text{Sample}}}{\left(^{15}\text{N}/^{15}\text{N}\right)\_{\text{PDB}}} - 1\right] \cdot 1000\% \tag{2}$$

where δ <sup>13</sup>C is the thousand percent deviation of sample <sup>13</sup>C/12C from the standard sample; ( <sup>13</sup>C/12C)Sample is the <sup>13</sup>C/12C of the leaf sample, and (13C/12C)PDB is <sup>13</sup>C/12C in Peedee Belemnite (South Carolina).

### *5.3. Statistical Analyses*

We first performed a DCA test on the leaf functional trait data, and the DCA1 value of axis lengths was 0.459, considerably smaller than 3.0, implying that the data distribution was consistent with the linear model. We therefore used an RDA to test correlations of leaf functional trait indices and environmental factors [98–100]. The 'vegan' package in RStudio (Integrated Development for R, RStudio Inc., Boston, MA, USA) was used to calculate RDA ordinations. The Pearson's correlation coefficient between leaf functional traits and environmental variables was calculated using IBM SPSS Statistics (IBM Inc., New York, NY, USA), and a respective heatmap was produced using Origin 2022 (OriginLab Inc., Northampton, MA, USA). The distribution map of forest reserves was produced using ArcGIS 10.4 (ESRI Inc., Redlands, CA, USA). Statistical analyses were performed using

Excel 2019 (Microsoft Inc., Redmond, WA, USA). Values are presented as means ± standard error of the mean.

**Author Contributions:** Conceptualization, W.S. and F.L.; methodology, X.Z.; software, J.X.; formal analysis, J.X., H.L. and X.H.; data curation, W.S.; writing—original draft preparation, J.X. and H.L.; writing—review and editing, W.S.; project administration, X.Z.; funding acquisition, F.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by National Natural Science Foundation of China, grant number 42127807.

**Data Availability Statement:** The data produced in this study are available in Refs. [38,56,97].

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

