Effects of tree species on soil nutrients
Long-term afforestation of different tree species significantly changed soil BD, pH, and SWC. The decrease of soil BD was caused by the active activities of more abundant soil animals and microorganisms after afforestation (Frouz et al. 2001). Except for P. hopeiensis, other four tree species plantations significantly decreased SWC (Table 2), which was consistent with the findings of Cao et al. (2007), who reported that large-scale afforestation resulted in soil drying and water depletion. Compared with grassland, woody plants consumed more soil water, and canopy and litter layers of plantations intercepted precipitation (Sato et al. 2004). And we also found the pH of alkaline soils decreased after long-term afforestation, because woody plants could produce more organic acids or anions to enter the soil (Wang et al. 2021b). A meta-analysis showed that afforestation promoted soil pH neutralization (Hong et al. 2018). In this study, SWC and pH of broadleaf tree species (C. korshinskii and P. hopeiensis) were higher than those of coniferous tree species (P. orientalis and P. tabuliformis) by 35.66%-151.18% and 0.50%-2.29%, suggesting that the impact of tree species on soil properties was species-specific (Hong et al. 2018), which may be due to differences in water requirements, plant traits, litter decomposition, root exudate, and soil ion exchange (Rötzer et al. 2017; Oleghe et al. 2019; Wang et al. 2021b). Thus, to a certain extent, the degree of changes in soil physicochemical properties after afforestation depended on different tree species.
Tree species significantly affected soil nutrients in dryland plantations, and decided the direction and degree of soil nutrient succession. Long-term afforestation increased soil TC and decreased soil TP and TK because the afforestation drove the accumulation of more tree litter and faster soil mineralization to meet tree growth (Zhao et al. 2007; Deng et al. 2017; Rai et al. 2016). The significantly different soil nutrients were found in the five plantations, and C. korshinskii improved soil nutrients, but P. orientalis and P. tabuliformis as the evergreen tree species reduced soil nutrients, which supported our hypothesis about the different effects of tree species on soil nutrients. The findings of Wang et al. (2017) were consistent with ours, who noted that the impacts of thirteen tree species on soil fertility existed interspecific difference, and P. tabuliformis had the lowest soil fertility. Because deciduous and evergreen tree species had significantly different litter (Sariyildiz et al. 2005; Liu et al. 2016). The litter mixing test of deciduous and evergreen trees revealed that the high litter carbon and lignin contents of evergreen trees led to low decomposition rate and slow nutrient release (Sariyildiz et al. 2005; Liu et al. 2016). Our results also showed that deciduous plantations (C. korshinskii and P. hopeiensis) had higher soil TC, OC, TN, AN, and AK contents than those in evergreen plantations (P. orientalis and P. tabuliformis) by 1.34%-166.87% (Figure 1). In addition, C. korshinskii was not only a deciduous tree species, but also a leguminous tree species. The leguminous tree species had higher nitrogen and carbon input from fixation (Wang et al. 2019). Gei and Powers (2013) compared the effects of legumes and non-legumes tree species on soil properties in Costa Rican dry plantations, who confirmed that legumes had higher soil TC, TN, and nitrate nitrogen contents. These findings indicated that deciduous tree species, especially legumes, played a more important role than evergreen tree species in dryland vegetation restoration (Wang et al. 2012; Zhao et al. 2018), which produced high-quality and easily decomposable litter, promoting nutrient cycling and accumulation (Bohara et al. 2020). Therefore, the differences in soil nutrients in different plantations may be caused by the litter chemical properties of different tree species.
Effects of tree species on litter chemical properties
The litter chemical properties were significantly affected by different tree species, due to the interspecific differences (Kooch et al. 2017). The significant difference in litter quality among different tree species has been reported in previous studies (Sariyildiz et al. 2005; Liu et al. 2016). Xu et al. (2020) illustrated that Robinia pseudoacacia, C. korshinskii, A. sibirica and P. tabuliformis had significant differences in nitrogen and phosphorus resorption efficiency, green leaf nutrients, and growth rates, which may lead to the production of different quality litters (Bai et al. 2019). In our study, high-quality litters (higher LN, LP, and LK with lower LC) were observed in C. korshinskii, A. sibirica and P. hopeiensis, while low-quality litters (higher LC with lower LN, LP, and LK) in P. orientalis and P. tabuliformis. We also found that the LC, LN, and LP under P. tabuliformis were significantly higher than those under P. orientalis by 3.32%, 34.20%, and 96.65%. However, Bai et al. (2019) noted that P. tabuliformis had lower LN and LP than P. orientalis, which may be caused by abiotic factors, for example, different temperature (10.4 ℃), precipitation (500-600 mm) and afforestation years (25 years), compared to this study. Broadleaf forests had lower carbon and higher nitrogen and phosphorus contents of litter (Zhang et al. 2017), whereas coniferous forests produced litter rich in carbon and poor in nitrogen, and had low litter decomposition rate (Xiao et al. 2019). Satti et al. (2003) pointed that the leaf litter nitrogen, nitrogen mineralization, and soil nitrogen in coniferous trees presented lower characteristic than those in broadleaf trees. Consequently, different plantations in our study were obvious variation in soil physicochemical properties and litter quality, illustrating that long-term afforestation of different tree species may establish diverse patterns of plant-soil feedback (Song et al. 2017; Zhang et al. 2018a).
Effects tree species on soil enzyme activities
Plantations could alter plant biomass, species diversity, litter properties, soil characteristics and microbial communities (Chen et al. 2016; Bai et al. 2016; Zhao et al. 2019), which played an important role in the changes of soil enzyme activities (Ushio et al. 2010; Li et al. 2020). Five plantations of this study had significantly different soil properties and litter quality. A study of soil microbial time dynamics confirmed that SWC, pH, soil phosphorus, leaf litter phosphorus, and the ratio of leaf litter carbon to phosphorus influenced soil enzyme activities by enhancing or inhibiting the growth of soil microbes (Bai et al. 2021). Our study showed that soil SC, UE, and ALP in C. korshinskii were higher than those in other four tree species by 3.97%-16.90%, 13.96%-36.99% and 42.56%-178.25%, indicating that tree species could directly or indirectly affected soil enzyme activities (Wang et al. 2012; Ren et al. 2016). In the C. korshinskii plantation, improved SWC, BD, pH created an environment conducive to the growth of microorganisms, and higher quality litter provided a good material basis for the microbial growth and enzyme synthesis. And high-quality litters and high soil nutrients drove soil animal and microbial abundance increasing (Kooch et al. 2017; Zhao et al. 2019), which enhanced soil enzyme activities (Singh et al. 2012; Noll et al. 2016). In addition, P. orientalis presented lowest soil SC, UE, and ALP than C. korshinskii, A. sibirica, P. hopeiensis, P. tabuliformis, and grassland. However, in semi-humid plantations, soil enzyme activities in P. orientalis were higher than that in Sophora davidii (a leguminous shrub) (Li et al. 2020), suggesting that P. orientalis may not be suitable for afforestation in arid areas (Zhang and Chen 2007), which was also supported by the lowest quality litter, SWC and soil nutrients of P. orientalis in this study.
Response of soil nutrients to litter chemical properties and soil enzyme activities
Vegetation played a major role in the improvement of soil nutrients, and the influence of different vegetation types on soil nutrients was controlled by litter (Sariyildiz et al. 2005; Wang et al. 2021a). Nutrients return to the soil through the litter decomposition was the main link in the material cycle of the ecosystem (Wang et al. 2008). The high-quality litter and high soil enzyme activity could accelerate the nutrient release of litter and improve the supply capacity of soil available nutrients for plant growth (Wang et al. 2008; Noll et al. 2016). Therefore, the litter properties and soil enzyme activity determined the level of soil fertility (Pan et al. 2013; Bohara et al. 2020). Our results showed that the variation in soil nutrients following long-term afforestation with different tree species were caused by soil TC, OC, TN, AN, AP, and AK, which were significantly positively affected by the LN, LP, UE, and ALP (table 3). Firstly, the LN and LP represented the litter quality, which decided the decomposition rate and microbial metabolism, thereby affecting the nutrient cycle (Prescott 1996; Suseela and Tharayil 2018). Laughlin et al. (2015) found that correlation coefficient between leaf litter nitrogen content and soil fertility reached a significantly positive level (0.80), because litter with high nitrogen content had the faster decomposition rate, nitrogen mineralization rate and nutrient release (Rai et al. 2016). Zhou et al. (2012) revealed that the loss and return of phosphorus increased significantly with the increase of total phosphorus concentration in litter. Thence, the LN and LP actively increased soil nutrients, and the synchronous changes of the litter chemical properties and soil nutrients in C. korshinskii and P. orientalis plantations proved that high-quality litter improved the soil nutrient succession in dryland plantations, and vice versa (Rai et al. 2016).
Secondly, the differences in soil carbon, nitrogen and phosphorus contents were similar to that in soil enzyme activities in different tree species plantations, especially C. korshinskii and P. orientalis, indicating that there was a strong interaction between soil nutrients and enzymes (Table 3). Some studies have reported that soil nutrients were closely related to soil enzyme activities (Zhang et al. 2018b; Zhao et al. 2018), and soil enzyme activities had the potential to estimate the decomposition rate and the availability of nitrogen and phosphorus (Sinsabaugh and Moorhead 1994). The leaf litter decay rate, soil dissolved organic nitrogen, and AP significantly increased with increased enzyme activities (Waring 2013; Ren et al. 2016). The improvement of soil physical and chemical properties was conducive to the synthesis of soil enzymes and the maintenance of soil enzyme activities (Ushio et al. 2010; Pan et al. 2013). Thus, soil enzyme activities had a positive interaction with soil nutrients (Table 3).
In addition, the significant positive correlation between LN, UE, TN, AN and LP, ALP, AP demonstrated that nitrogen and phosphorus existed an obvious coupling relationship, which was confirmed by Zhao and Zeng (2019), who noted that phosphorus addition strongly altered the impact of nitrogen addition on soil nitrogen and phosphorus transformations. Another nitrogen addition test revealed that soil specific acid phosphatase, N-acetyl glucosaminidase, and oxidative enzyme activities significantly increased with nitrogen levels increasing (Li et al. 2019), which suggested that improved soil nitrogen availability could promote the cycling of the phosphorus and other elements. Zhang et al. (2018b) and Zhang et al. (2021) pointed out that phosphorus limitation existed in the Loess Plateau and might become more severe in the future. Soil microbes promoted the extraction of restricted elements by regulating the production of soil enzymes and the efficiency of nutrient utilization (Bai et al. 2021). The significant positive correlation between TC, OC, TN, AN, LN and LP, OC, TN, AN, LN and ALP, and LN, SC, UE, and AP illustrated that soil microbes might tend to consume more carbon and nitrogen to increase the soil phosphorus availability, alleviate the limitation of phosphorus on the growth of plants and microorganisms, and meanwhile, improve other nutrients in the soil (Yan et al. 2020; Bai et al. 2021). It was also proved that in the C. korshinskii plantation with the highest LN, TC, OC, TN and AN, the LP, soil ALP and AP in whose was higher than that in other plantations and grasslands (Figure 1, 2&3). And the LP and ALP had a larger correlation coefficient with soil nutrients than LN and UE (Table 3). Therefore, these results revealed that the LP and ALP were the key factors driving soil nutrient changes in dryland plantations limited by phosphorus, which supported our prediction that the LP and ALP were important for soil nutrient dynamics.