Changes in and evaluation of surface soil quality in Populus × xiaohei shelterbelts in midwestern Heilongjiang province, China

Jun Zhang · Yusen Zhao · Ying Xin

Abstract To examine changes in surface soil quality over time in Populus × xiaohei shelterbelts, we collected soil samples from f ive shelterbelts of diff erent ages and also from former cropland left fallow for 25 years. Twenty-one surface soil (0-20 cm) properties were measured, and variation in soil quality was assessed using one-way ANOVAs and multiple comparison tests. Based on this analysis, 16 soil indexes were used in a model evaluating soil quality,with each index given weight as determined by the correlation coeffi cient. Compared with the control, the postmature forest had greater soil moisture content but lower bulk density ( P < 0.05). The mature forest also had higher soil pH, total organic carbon, alkali-hydrolyzed nitrogen,available phosphorus, and biomass nitrogen content, but reduced nitrate-nitrogen and total phosphorus content than the control ( P < 0.05). Total porosity was highly positively correlated with aeration, nitrate-nitrogen, alkali-hydrolyzed nitrogen, available phosphorus, microbial biomass carbon and microbial biomass nitrogen. Soil total organic carbon,ammonium nitrogen, nitrate-nitrogen, alkali-hydrolyzed nitrogen, total nitrogen, available phosphorus, microbial biomass carbon and microbial biomass nitrogen were all strongly correlated. In the soil quality evaluation model,total organic carbon was assigned the highest weight and total potassium content the lowest. The soil quality index was lowest in the near-mature forest and greatest in the postmature forest. Generally, soil quality in Populus × xiaohei shelterbelts varied with age and was higher in the 10-20 cm versus 0-10 cm soil layer. After a single forest generation,surface soil quality was signif icantly improved.

Keywords Populus × xiaohei · Shelterbelts · Black soil area · Soil quality · Evaluation

Introduction

Black soils that contain plenty of organic matter are common in midwestern Heilongjiang Province, China, an important region for grain production. This region is also part of the“Three-North Shelterbelt Project” and also serves as a key area of ecological and environmental restoration. Here, a temperate continental monsoon climate dominates; wind erosion and its damage are prone to occur in both spring and autumn, harming agricultural production and livelihoods.The “Three-North Shelterbelt Project” has been ongoing for more than 40 years and is responsible for the creation of large-scale farmland shelterbelts. This project has created farmland-forest networks, forming a relatively stable, complex agroforestry system to reduce wind and sand damage.Shelterbelts have played an important role in improving the local microclimate, preventing wind damage and sand erosion, and increasing grain production (Zhang et al. 2019).By 2008, shelterbelts in northeast China covered an area of 1.16 × 10 5 km 2 , with the forest coverage rate reaching 19.9%(Zheng and Zhu 2013). In particular,Populusshelterbelts covered about 6.67 × 10 4 km 2 (Wu et al. 2018).

Previous studies of farmland shelterbelts have focused on potential agricultural benef its, such as their impact on crop yields (Zhang 2017), soil moisture distribution in the forest network and soil erosion (Zuo et al. 2018), as well as potential ecological benef it (Wang et al. 2014). However, few studies have evaluated the Effects of farmland shelterbelts on soil quality. Forest soils provide the necessary material basis for the survival and development of forest vegetation;conversely, the emergence and succession of forest vegetation aff ects the formation and development of soils (Song et al. 2005). Objective study of the potential soil changes and benef its following shelterbelt construction in the “Three-North” areas is therefore of great importance for sustainable management (Mao et al. 2009). In recent years, shelterbelts planted some 40 years ago have entered the regeneration stage, and stand degradation has occurred in many areas,resulting in decreased protection benef its. With shelterbelt degradation, the process of shelter forest growth has become a research focus, with studies examining, for example, precipitation and soil water thresholds associated with drought-induced mortality in farmland shelterbelts (Sun et al. 2019), the Effects of root excision on photosynthesis in poplar shelterbelts (Chen et al. 2017), optimal farmland shelterbelt spacing (Zhu et al. 2016), and diff erences in soil physical properties between shelterbelts and farmlands (Wu et al. 2016), Sun et al. ( 2018) examined soil nutrient content in poplar shelterbelts of diff erent ages in the black soil area of China, but only three forest ages were included and there was no evaluation of soil quality changes. Wu et al. ( 2018)studied soil changes over time in shelterbelts but quantif ied only a few soil indicators, such as soil organic matter, pH and bulk density, which would not fully ref lect any potential changes in woodland soils.

In general, shelterbelts created in the study region have been neglected after their establishment; no further forest management was typically undertaken after canopy closure. Additionally, due to problems with land contracting and management, and declining soil fertility over time inPopulusplantations, there have been challenges to proposals to expand forest areas or regenerate shelterbelts. The long-term Effects of shelterbelt creation on soil fertility are not clear. Nor are the mechanisms underlying changes in soil fertility. Therefore, understanding how soil characteristics change in shelterbelts over time may be critical to promoting shelterbelt growth and maximizing the benef its of forest regeneration (Lin and Shang 2009). Shelterbelt soil quality is a critical factor for forest health and may be ref lected in soil structure, soil fertility, soil microbial biomass and forest water conservation (Guo et al. 2012). Soil microbiological and biochemical properties can be used as indicators to ref lect changes in agroecosystem and soil productivity (Kennedy and Papendick 1995). Most researchers utilize diff erent metrics to evaluate soil quality. For example,Andrews et al. ( 2002) used multivariate statistical methods to select the components of MDS (minimum data set) and give each a weight, then used a nonlinear scoring function to convert index values into an evaluation index of soil quality,f inally evaluating diff erences in soil quality caused by different management methods. Moncada et al. ( 2015) treated soil moisture content (MC), bulk density (BD) and porosity as key indicators of soil physical quality. Costantini et al.( 2015) established soil evaluation indexes to quantify the Effects of ecosystem restoration in arid areas; in the shortterm, soil organic matter (SOM), pH, available phosphorus(AP), available nitrogen and porosity might be most important, while in the long-term (or over large geographical areas), slower-acting variables such as soil structure, porosity, layer thickness and water-holding capacity (WHC) merit the most attention. Although soil quality is much more than just nutrient content, to at least some extent, nutrient content does indicate soil fertility. As diff erent individual plants, and plant species have diff erent nutrient requirements, soil fertility can be relative. Certain nutrients may be most critical for some plants, but not necessarily for others. How to evaluate soil quality scientif ically is a topic of continuing debate.

At present, many evaluation methods of soil quality have been proposed, for example: MFI (multiple-factor index)(Wang et al. 2001), AHP (analytic hierarchy process) (Wang 2016), MVIK (multiple variable indicator Kriging) (Li et al.2011), LCA (life cycle assessment) (Mattsson et al. 2000),RSQI (relative soil quality index) (Karlen and Stott 1994),PCA (principal component analysis) (Wang et al. 2018), CC(correlation coeffi cient) (Li et al. 2007; Zheng et al. 2018),FMA (fuzzy mathematical analysis) (Wang 2016), CA (cluster analysis) (Jin et al. 2018), EW (entropy weighting), GRA(grey relation analysis) (Feng et al. 2018), BP-NN (BP-neural network algorithms) (Han et al. 2011), TOPSIS (technique for order preference by similarity to the ideal solution)(Zhang et al. 2016), GIS-SQA (soil quality assessment based on GIS) (Ma et al. 2004). Unfortunately, at present, no single method is consistently used to evaluate soil quality.

In this study, farmland shelterbelts forested withPopulus×xiaoheilocated in Baiquan County were used to examine how surface soil quality changes with forest age. Using a space-for-time substitution (Mao et al. 2009), surface soil (0-20 cm) physicochemical indicators were measured in shelterbelts of diff erent ages and these were compared to measurements taken from fallow cropland abandoned 25 years ago. Changes in soil properties were analyzed to elucidate the Effects of stand age inPopulus×xiaoheishelterbelts. The comprehensive soil quality evaluation model(Wang et al. 2001; Li et al. 2007; Zheng et al. 2018) was used to assess changes in soil quality. Our goal in this study was to contribute to a theoretical foundation for the rational and scientific management of surface soils in farmland shelterbelts.

Materials and methods

Study site

This study took place in Baiquan County of midwestern Heilongjiang Province, which is famous for its grain production (47°20′-47°55′N, 125°30′-126°31′E). In this region,the annual frost-free period lasts for 120-125 days, with a mean annual temperature of 1.3 °C and annual Effective accumulated temperature (≥ 10 °C) of 2454.5 °C; the mean number of hours of sunshine received annually is 2717.1 h.Average annual precipitation averages 496.7 mm and its seasonal distribution is uneven: precipitation taking place from April to May accounts for 49.9 mm, or only 10% of the annual average, while precipitation occurring from in June to August reaches 343.7 mm, or 69.2% of the total.Average annual evaporation is 1132 mm, making the region semi-humid. It is dry and windy in spring and hot and rainy in summer, with extremely high temperatures reaching up to 37.8 °C. Winters are cold, with temperatures dropping to a minimum of − 39.6 °C. In the growing season, the temperature averages 17 °C. Annual wind speed averages 3.0 m · s −1 and the wind direction is mostly from the southwest, averaging 202°30′ degrees azimuth. Winds above grade 5 occur 125 times per year on average, while winds above grade 8 occur 72 times; April and May are the windiest months.There are f ive soil types in the region: 67.9% of soils are black soils, 14.4% black calcium soils, 17.1% meadow soils,0.5% marsh soils and 0.2% salty soils, which are found at high latitudes, and in cold and dry agricultural areas (Sun et al. 2018). The study core area encompassed seven villages belonging to Chang’an Village, Fengchan Towns(47°37′18.1″-47°39′25.8″N, 125°43′29″-125°47′21.2″E).The main crops grown here are corn, soybean, sorghum and potato. The growing season is from April to September, lasting about f ive months. Beginning in the 1970s, local villagers began to build farmland shelterbelts. To date, 318 protective forest belts have been created, covering about 231.4 hm 2 ; the main tree species used in aff orestation werePopulus×xiaoheiandLarix gmelinii(Rupr.) Kuzen.

Sample plot selection

To implement the space-for-time substitution (Mao et al.2009), from May 7 to May 17, 2016, five shelterbelts belonging to diff erent age classes and a control plot were selected according to the Handbook of the Forest Resources Management (III) (Department of Forest Resource Management 2007) using data on aff orestation, f ield investigations and GPS coordinates. The shelterbelts were all well preserved, with good spatial homogeneity of land quality,far from man-made structures (thus subject to little human interference) and planted in the same direction. X09, X01,X96, X90, X84, and CK represented young forest (YF)(< 10 years), mid-mature forest (MMF) (11-15 years),near-mature forest (NMF) (16-20 years), mature forest (MF)(21-30 years), post-mature forest (PMF) (> 31 years), and unforested abandoned farmland for 25 years, respectively.According to the Chinese Soil Classif ication System, the selected study sites had typical Chernozem soils. For each shelterbelt, three repeated typical plots of 20 m × 8 m were built for each age-grade shelterbelt, the same was done for the unforested control plot, and a total of 18 typical plots were sampled. The boundaries of all sample plots were marked and recorded. In each of the typical plot, the height,diameter at breast height (DBH) and canopy density were measured for all trees. The basic features of all study plots are listed in Table 1.

Soil sampling in the f ield

In order to ref lect normal growth conditions, soil samples were taken early in the growing season, in July. In each typical plot, soil samples were collected from f ive individual sampling points (arranged in an S-shape) after removing surface litter and any weeds. Each soil prof ile was divided into two layers, an upper layer (0-10 cm) and a lower layer (10-20 cm); samples were taken in the middle of each layer. From each sample location, a 500 g soil sample was collected for the determination of soil pH,total organic carbon (TOC), total nitrogen (TN), alkalihydrolyzed nitrogen (AHN), total phosphorus (TP), AP,total potassium (TK) and available potassium (AK). A 100 g fresh soil sample was collected for the determination of indicators such as microbial biomass carbon (MBC) and microbial biomass nitrogen (MBN), ammonium nitrogen(AN) and nitrate-nitrogen (NN). Thus, a total of 360 bags of soil were collected (18 plots × 20 samples for each plot).Impurities such as roots and stones were removed before passing samples through 2 mm soil sieve. The soil samples from a given soil layer within a plot were combined(across sampling points) with all f ive samples represented evenly by weight. In total, 1 kg of mixed soil samples was collected, labeled and brought back to the laboratory for air drying and testing. After mixing, the fresh soil samples were stored separately at 4 °C in refrigerators and brought to the lab for further testing. Finally, undisturbed soil samples were collected from each layer using a ring knife and then weighed immediately for the measurement of soil bulk density, porosity and other indexes of physicalproperties. About 10 g of fresh soil was placed in an aluminum box and weighed immediately to calculate MC.

Table 1 Basic characteristics of the sample plots

Determination of soil physical and chemical properties

Soil moisture content was determined by drying, and the soil BD, saturation holding capacity (SHC), capillary holding capacity (CHC) and f ield holding capacity (FHC)were measured using the ring knife method. The soil noncapillary porosity (NCP), capillary porosity (CP), total porosity (T-P), aeration (Ae) were then calculated (Chen 2005). Soil pH was measured using a potentiometric pH meter (soil to water ratio was 1:2.5) (Chen 2005). The soil TOC was determined using an automatic CN-analyzer(Vario Max CN Macro Elemental Analyser, Elementar Analysen system GmbH, Hanau, Germany) with the dry combustion method (Gelaw et al. 2014; He et al. 2014).A 1 mol · L −1 KCl solution was used to determine AN and NN by leaching (AutoAnalyzer 3, SEAL Analytical GmbH, Norderstedt, Germany) (Schwarz et al. 2014),while AHN was measured using the NaOH-hydrolyzing,NH3-diff using, H3BO3-absorption method (Chen 2005).Soil TN was determined using the semi-micro-Kjeldahl method (Chen 2005). AP was measured by extracting a sample with a 0.5 mol · L −1 sodium bicarbonate solution and then analyzing the developed color at 700 nm with a spectrophotometer, while TP was determined using a sulfuric acid-perchloric acid solution and Mo-Sb colorimetry (Chen 2005). Following extraction with a 1 mol · L −1 NH4OAc solution, AK was measured using a f lame photometer (Chen 2005). TK was determined using hydrof luoric acid-perchloric acid solution f lame photometry (Chen 2005). Finally, chloroform fumigation was used to measure MBC and MBN (Wu et al. 1990).

Evaluation model of shelterbelt soil quality

SQI (soil quality index) method (Li et al. 2007) was used to evaluate the soil quality of the shelterbelts. The SQI of the evaluation model was calculated as follows:

where SQI is soil quality index,W iis the weight andf(xi) is the membership degree for each index.

Selection of evaluation indexes

As in previous studies (Wang et al. 2001; Li et al. 2007;Costantini et al. 2015, 2016; Moncada et al. 2015; Zheng et al. 2018), in this study, four factors were selected for inclusion in the model: water, air, fertilizer and microbes.These encompassed 16 evaluation indicators specif ically:BD, FHC, T-P, Ae, pH, TOC, AN, NN, AHN, TN, AP, TP,AK, TK, MBC and MBN.

To determine the weight of each evaluation index

There are many techniques used to determine the weight of evaluation factors, such as the analytic hierarchy process(AHP) and the Delphi process (Li et al. 2007), but these methods are highly sensitive to the subjective preference of scorers. To objectively determine the SQI, here, the correlation coeffi cient method was adopted to create the model. The specif ic steps taken were as follows.

Step 1 was to calculate the correlation coeffi cients (rij)among the index factors involved in the evaluation.

Step 2 was to take the absolute value of the correlation coeffi cients from the previous step, and then calculate the average of these absolute values for correlations between a given index and all other indexes

Step 3 was to calculate the index weights, a measure of an index’s contribution in the evaluation model of soil quality that is calculated as follows:

To determine the membership degree of each index

Since the units involved in the evaluation indicators were not uniform, they could not be directly compared without f irst being standardized. Next, as the changes in the soil evaluation indexes used here were continuous, membership functions were used in the evaluation model to adjust continuous variables. Considering both observed data on soil quality from this study and previous research results (Tang et al.2009; Sun et al. 2018; Zheng et al. 2018), the soil BD and pH were assigned to a descending membership function and the remaining indexes to an ascending membership function.

The functions were calculated as follows:

1. ascending type membership function

2. descending type membership function

wherex1andx2represent the minimum and maximum value of each indicator, respectively, andf(xi) is the membership degree of each indicator.

Statistical data analyses

The data were analyzed using Excel 2016 and SPSS 24.0.One-way ANOVAs and Duncan’s multiple comparison tests were used to examine the inf luence of shelterbelt age on soil properties and to test equality of means among groups at α = 0.05. Pearson bivariate correlation coeffi cients were used to analyze relationships among the soil quality indexes (α = 0.05 and 0.01). The Origin 2017 was used to draw graphics.

Results and analysis

Soil physical properties in shelterbelts of diff erent ages

Nine soil physical property indexes, ref lecting soil gases,liquids and solids, are illustrated in Fig. 1. The maximum values of MC, BD and Ae occurred in MF, YF and MMF,respectively, while all other physical properties were highest in NMF. Most indexes diff ered with shelterbelt age,except for NCP (both soil layers) and CP, T-P and Ae in the lower soil layer.

In the upper soil layer, a number of physical properties followed an inverse N-shape with increasing shelterbelt age (SHC, CHC, FHC, CP and T-P); soil BD f irst ascended, before continuously descending, while Ae continuously ascended before descending. Compared to the control, the soil MC was higher in the shelterbelts(P< 0.05, unless otherwise noted), with the exception of MF; meanwhile, soil BD was lower in the shelterbelts,with the exception of MMF. The SHC was higher in NMF and MF than in the control, while soil CHC and FHC were generally higher in the shelterbelts, except for MF. Soil CP and T-P did not diff er among sites. Ae was lower in YF and higher in MF versus the control but did not diff er between CK and MF.

In the lower soil layer, some physical properties followed a W-shape with stand age (MC, SHC, CHC, FHC and CP),while BD followed an M-shape. Ae ascended, descended and then ascended again. Compared to the control, the soil MC was lower in the shelterbelts, except for in YF and NMF.The soil BD was signif icantly higher in CK versus YF, NMF or PMF. The SHC, CHC and FHC varied with forest age and were highest in NMF. NCP, CP, T-P and Ae did not differ among sites, but the f irst three of these properties were highest in NMF.

Looking across both soil layers, soil properties that followed a W-shape with increasing shelterbelt age included MC, SHC, CHC and FHC. The soil BD followed an M-shape, while CP and T-P followed an inverse N-shape.The Ae f irst ascended before descending slightly. The soil MC was signif icantly higher in YF, NMF and PMF compared to the control. In contrast, the soil BD was greater in the control versus the shelterbelts, with the exception of in MMF. The SHC, CHC and FHC were lower in CK than in NMF and were higher in NMF versus MMF. The soil CP,T-P and Ae in all shelterbelts did not diff er from CK and reached a maximum in either NMF or MF.

Fig. 1 Soil physical properties in Populus × xiaohei shelterbelts of diff erent ages and in the control

Soil chemistry and microbial biomass properties in shelterbelts of diff erent ages

Ten soil chemical properties and two biological indicators are presented in Fig. 2. Eight out of twelve properties followed a Λ-shape as shelterbelt age increased, including TOC, AN, NN, AHN, TN, AP, MBC and MBN; for these properties, maximum values appeared in NMF, minimum values in YF, and values were higher in the upper versus lower soil layer. The soil pH was higher in the shelterbelts versus the control, but also tended to increase with forest age, though the rate of the change in pH slowed after NMF.Variation in the pH in the upper soil layer (from 6.60 to 7.37)was greater than in the lower layer (from 6.69 to 7.03). The soil AK and TK followed a V-shape with increasing shelterbelt age, and the inf lection point occurred in MMF for AK;the AK was higher in the upper versus lower soil layer. The soil TP continued to decline with forest age and was higher in the lower versus upper soil layer.

In the upper soil layer, soil pH was signif icantly higher in NMF than in YF or MMF, but was lower in NMF than in MF or PMF. Similarly, the soil TOC, AN, NN, AHN, AP,MBC and MBN were highest in NMF. Contrastingly, the TP was lower in NMF than in YF or MMF, but higher than in MF or PMF.

In the lower soil layer, soil pH was higher in NMF than in YF or MMF. Soil TOC, NN, AHN, TN, AP, MBC and MBN were highest in NMF. Meanwhile, the AP and the AK varied among shelterbelts, but the AN did not.

Averaging the chemical properties across both soil layers,the soil pH was higher in NMF versus YF or MMF, while soil TOC, NN, AHN, AP, MBC and MBN were all highest in NMF. The soil TP was higher in NMF than in MF or PMF,but was lower than in YF or MMF. The soil AK and TK were lower in NMF than in PMF. The soil TP did not diff er with shelterbelt age.

Correlations among main soil quality variables

Fig. 2 Soil chemical and biological properties in shelterbelts of diff erent ages and in the control

Bivariate Pearson correlation coeffi cients were calculated among 16 key soil variables; taken together, these variables ref lect the soil water content, aeration, nutrient levels, and microbial conditions. As shown in Table 2, soil BD was negatively correlated with all other soil variables except for TP (P< 0.05, Ae, AHN and T-P;P< 0.01, FHC, pH, TOC,AN, TN, AP, TP and AK). Soil FHC was correlated with all other variables (P< 0.01; except for NN,P< 0.05) except TK. Soil T-P was positively correlated with Ae, NN, AHN,AP, MBC and MBN (P< 0.01); similarly, the soil pH was correlated with AN, AHN, TN, AP, TP and AK. Soil TOC was correlated with all other variables (P< 0.01) except for TK. The following variables were all highly correlated:TOC, AN, NN, AHN, TN, AP, MBC and MBN. Finally, AK was correlated with TK (P< 0.05).

Index of soil quality

Figure 3 shows the distribution of the soil quality evaluation index weights for all 16 indicators. Weights for each indicator varied from 2.61 to 8.21%. The largest weight was assigned to TOC and the smallest to TK. Soil properties were divided into physical properties, chemical properties and biological properties: the average weight of the four physical property indicators was 5.70%, while the average weight of the ten chemical property indicators was 6.27% and the average weight of the two biological property indicators was 7.27%. The average weight of the biological indicators was signif icantly greater than that of the other two categories.

Soil quality index (SQI) ranged from 0.40 to 0.82 across sites, following a V-shaped pattern with increasing shelterbelt age (Fig. 4). The maximum SQI was calculated for PMF, while the minimum was for NMF. SQI was higher in the lower versus upper soil layer at all sites except MF.

In the upper soil layer, SQI ranged from 0.40 to 0.79 across sites. SQI was signif icantly higher in CK than in MMF or NMF, but lower than in MF or PMF; there was no diff erence between CK and YF. SQI was signif icantly lower in NMF than in MMF.

In the lower soil layer, SQI ranged from 0.55 to 0.82 across sites. Compared to CK, SQI was lower in NMF,higher in PMF, and showed no diff erence with YF, MMF or MF.

MBN MBC TK − 0.244 0.964** 1 AK 0.422* 1− 0.042 − 0.193 1 TP − 0.518** 1− 0.205− 0.286− 0.428** 0.028 AP 1 0.147− 0.156− 0.435** − 0.206 − 0.376* − 0.869** − 0.587** − 0.478** − 0.116 − 0.431** − 0.592** − 0.563** 1 0.786**0.887**TN 10.684**0.540**0.122 0.625**0.666**AHN NN 10.728** 1 0.531** 0.636**0.684** 0.926**− 0.090 0.085− 0.301 − 0.198 0.937** 0.825**0.908** 0.881**Table 2 Correlations among soil quality variables in shelterbelts and in the control AN 10.574**0.660**0.804**0.630**0.466**0.259 0.668**0.662**TOC 10.793**0.684**0.782**0.872**0.811**0.471**− 0.044 0.771**0.833**pH 1 0.514**0.132 0.532** 0.477** 0.632**0.503** 0.498** 0.435**0.585**0.645**0.304 0.279 0.414*Ae 0.346*0.421*0.247 0.079 T-P 10.743** 1 0.183 0.288 0.446** 0.281 0.423*0.463** 0.554** 0.607**0.199− 0.197 − 0.085 0.487** 0.342*0.502** 0.420*FHC 0.836**0.513**0.375*0.551**− 0.015 1− 0.741** 1 0.531**− 0.448** 0.642**Factor BD − 0.402*− 0.456** 0.451**− 0.622** 0.449**− 0.442** 0.564**− 0.430** 0.545**0.609**− 0.349*− 0.066− 0.549** 0.486**0.485**− 0.237 BD FHC T-P Ae pHTOC AN NN AHN − 0.392*TN AP TP AK TK MBC − 0.238 MBN − 0.308 BD, FHC, T-P, TOC, AN, NN, AHN, TN, AP, TP, AK, TK, MBC, MBN indicate bulk density, fi eld holding capacity, total porosity, total organic carbon, ammonium nitrogen, nitrate-nitrogen,alkali-hydrolyzed nitrogen, total nitrogen, available phosphorus, total phosphorus, available potassium, total potassium, microbial biomass carbon, microbial biomass nitrogen, respectively** P < 0.01, * P < 0.05

Fig. 3 Weights of soil property indexes

Fig. 4 SQI in both soil layers for forests of diff erent ages

Taking the average SQI values in the upper and lower soil layers, SQI was signif icantly higher in CK than in MMF or NMF, and signif icantly lower in CK than in PMF; there was no diff erence between CK and YF or MF. SQI was signif icantly lower in NMF than in MMF.

Comparison of shelterbelt growth in forests of diff erent ages

Tree growth rates (as measured by changes in height) inPopulus×xiaoheishelterbelts were most rapid from YF to MMF (Y-M) and slowest from MMF to NMF (M-N)(Table 3). With increasing forest age, i.e., from Y-M to MF to PMF (M-P), the growth rate in height f irst decreased andthen increased. Looking at changes in DBH, the most rapid growth also Y-M, with the slowest growth from NMF to MF stage (N-M). Thus, the growth rate for DBH f irst decreased and then increased. The changes of DBH lagged behind change in height. The changes in SQI decreased from YF to MMF and increased from NMF to PMF.

Table 3 Comparison of shelterbelts growth and SQI among forests of diff erent ages

Discussion

Changes in soil quality

The soil quality of woodlands is important for the growth of trees. Topsoil is generally considered to be the most active soil layer in terms of physical and chemical characteristics,while deeper soils are considered to be more inert (Rumpel and Kögel-Knabner 2011). Therefore, this study focused on the properties of surface soils in shelterbelts.

The physical properties of topsoils varied amongPopulus×xiaoheishelterbelts of diff erent ages. Soil MC, FHC,T-P and Ae can inf luence the formation of soil aggregates,soil nutrient cycling and microbial growth, which are indicators of soil quality. In this study, the YF shelterbelt had a closed canopy, reducing evaporation rates, as well as relatively weak root absorption; together, these factors may have resulted in the higher MC seen in YF versus the other shelterbelts. However, higher water consumption rates occurred due to continued forest growth in MMF and MF, resulting in low soil MC. These results are consistent with the f indings of Sun et al. ( 2018).

Changes in soil BD with forest age were likely due to the Effects of root activity, which loosens surface soils over time,increasing porosity and thus improving permeability. Root activity and growth were most vigorous in MMF; this rapid growth compacted the soil, leading a higher BD in MMF that was not signif icantly diff erent from that of CK. These results are consistent with the f indings of Wu et al. ( 2018).

Interestingly, soil SHC, CHC and FHC were basically the same across shelterbelts. Soil WHC was higher in NMF than in MF or PMF. One hypothesis for the decline in WHC in PMF may be related to tree physiological characteristics in the shelterbelts. In MF and PMF, trees were tall and had few branches on the lower trunk; this could produce well-ventilated forest belt structure, with rapid mineralization of dead leaves and humus, leading to a declined in SHC. In NMF,SHC was correlated with NCP, CP and T-P. In addition, tree growth was most vigorous in NMF; tree growth requires large amounts of water, which may explain improvements in WHC.

The chemical properties of topsoils varied amongPopulus×xiaoheishelterbelts of diff erent ages. The growth and development of forest plantations is an important driving force for the migration, enrichment, and redistribution of soil nutrients, which may, in turn, lead to changes in soil quality(Wang et al. 2007; Tang et al. 2009).

Soil pH may be aff ected by biological activity, soil parent material type, climatic conditions and human activity.As the study area is located in the transition zone between the Lesser Khingan Mountains and the Songnen Plain,groundwater levels are high and soils weakly alkaline. AsPopulus×xiaoheishelterbelts age, long-term transpiration increases, producing a continuous rise in soil pH; such further alkalinization of the soil is not conducive to the growth of either crops or trees (Tang et al. 2009). Here, the soil pH increased gradually with stand age and this trend was strongest in the upper soil layer. As the shelterbelts consume large amounts of water, salt from deeper soil layers may reach the surface with the f low of water. As the salt content of water transported aboveground by the roots is lower than that absorbed from deep soil layers, the soil pH increases with stand age. This phenomenon has been previously described by other authors (Tang et al. 2009; Ciadamidaro et al. 2014;Madejón et al. 2015).

Higher amounts of SOM can increase the ability of the soil to hold nutrients, increase nutrient availability, promote the formation of soil aggregate structure and raise permeability, an important index of soil fertility. Changes in SOM are usually ref lected by TOC, which is highly correlated with SOM (Chen 2005). In this study, TOC was correlated with several other soil properties and had the greatest weight assigned to it in the soil quality model (8.21%). In NMF, larger canopy size and greater abundance of primary branches might have Effectively reduced wind speeds in the forest, promoting the accumulation of SOM and topsoil nutrients. As forest age increased, eight soil indexes (TOC,AN, etc.) followed a Λ shape, which may be partly explained by the peak in TOC in NMF. The similarity of the pattern for the eight indexes also suggests that these properties are coupled in the shelterbelts, a suggestion supported by the strong correlations among the properties. Additionally, the values of all eight indexes decreased with soil depth, indicating that nutrients tended to aggregate at the surface, as reported by Sun et al. ( 2016).

Nitrogen is an essential element for plant growth. In this study, f ive measures of soil nitrogen increased with forest age in younger forests (i.e., from YF to MMF). This increase may be a result of interplanting nitrogen-f ixing leguminous crops (such as soybean or mung bean) in or near the forest shelterbelts, or perhapsPopulus×xiaoheiforests from associations with nitrogen-f ixing soil microorganisms. As forest age increased (MMF+), the nitrogen measures decreased gradually, likely due to large demands from tree growth(Sayyad et al. 2006). In PMF, growth began to decrease,and nitrogen content (except in the form of NN) returned to the control levels. In contrast, Wang et al. ( 2010) found that nitrogen levels decreased steadily as continuously planted poplar. This inconsistency may be a result of diff erences soil type, forest management methods and the planted poplar species. On the study area of Wang et al. ( 2010), soils were light yellow with poor fertility and forests were comprised ofPopulus deltoids I-69/55.

TP content was higher in the lower versus upper soil layers and decreased signif icantly with forest age; this decline was especially rapid after MMF. This revealed high phosphorus requirements and phosphorus leakage in woodland soils during the growth of thePopulus×xiaoheishelterbelts,as supported by a proposed theory of soil phosphorus leakage (Zhang et al. 2009).

In both MF and PMF, AK was higher than in the control,while TK did not show this pattern, a f inding inconsistent with previous research (Tang et al. 2009; Wang et al. 2010).This may be due to straw burning in the farmlands surrounding the shelterbelts: ash containing potassium might have blown into the shelterbelts.

Soil microorganisms play an important role in soil ecosystems, acting as a driving force for the transformation and circulation of organic matter and nutrients (Djukic et al. 2010; Li et al. 2013; Gryta et al. 2014) which makes them indispensable in terrestrial ecosystems. Soil microbial biomass directly ref lects the number of soil microorganisms, which inf luences soil fertility and the conversion and cycling of soil nutrients (Li et al. 2008; Rani et al. 2008;Zhang et al. 2013). In this study, changes in MBC or MBN were consistent with changes in forest age and soil nutrients, suggesting that microorganisms played a vital role in generating observed diff erences in soil quality across thePopulus×xiaoheishelterbelts. Microorganism abundance was lower in YF. As forest age increased and soils matured,soil microorganisms were most active in NMF. However, in PMF, microbial activity weakened, likely due to decreases in surface soil nutrients in surface soil, root aging, and increased pest and disease activity. This f inding was consistent with the results of Mao et al. ( 2009).

Evaluation of soil quality

In the above-mentioned evaluation method, there remain some disputes. For example, AHP is subjective in determining the weight of indicators. The evaluation process of MVIK is highly complicated and requires special software to run the algorithms. When there is a long growth cycle,LCA is not suitable for soil evaluation, while in RSQI, it is diffi cult to f ind the ideal soil for comparison. In evaluation,PCA has variable load loss, while CA ignores the synergy between evaluation indexes in their selection. Meanwhile,EW will assign a smaller value to an indicator with relatively concentrated data, and GRA ignores gradual changes in soil quality. The selection of learning samples in the BP-NN method remains challenging, and TOPSIS neglects inconsistencies in the direction of change among indicators.Lastly, GIS-SQE mainly ref lects only large-scale soil evolution and has a lag Effect when considering a short-term change. Therefore, in this study, a modif ied version of the correlation index synthesis method, which combines the advantages of fuzzy mathematical analysis and the index method, was used to objectively evaluate soil quality in the shelterbelts.

Although the correlations among soil quality indicators were statistically signif icant, this does not mean that they are necessarily ecologically signif icant. Nevertheless, a statistically signif icant correlation between indicators could highlight a future research direction to explore the synergy among indicators. In this study, there were some synergistic Effects among tested indicators: for example, the relationships between soil BD and TOC, between T-P and FHC,between Ae and MBC/MBN, between the various nitrogen forms and between MBC/MBN and the other indexes. These synergistic Effects indicate that the soil properties in the shelterbelt were dynamic, and played a role in the exchange and transfer of matter. Among the indicators, TOC had the greatest Effect on soil quality as indicated by its having the greatest weight assigned in the soil quality model, while TK was assigned the least weight. This was likely due to the high SOM content of shelterbelts in the black soil area. Soil TOC was a key factor in the soil quality evaluation model,consistent with the f indings of others (Sun et al. 2018; Wu et al. 2018).

From the SQI analysis, the planting ofPopulus×xiaoheishelterbelts in black soil areas did not cause any deterioration in surface soil quality, in contrast to the f indings of other studies of artif icialPopulusforests in southern China, which had relatively lower soil fertility but higher tree growth.Stark et al. ( 2015) suggested using fast-growingPopulustrees in heavily disturbed ecosystems to improve soil fertility and productivity, as also suggested by this study.

Surface soil quality changed as shelterbelt forests aged.SQI decreased from YF to NMF, where it reached its minimum, before increasing to reach its maximum in PMF.The decrease in SQI was likely due to the planting of grain crops in forests before canopy closure; grain plants may have absorbed nutrients from the surface soils, competing with demands from tree growth. In older forests, the increase in SQI could be due to decreases in root physiological functions, slower assimilation rates (leading to lower soil BD),higher T-P and FHC, improved Ae, the gradual mineralization of ground litter and the continuous accumulation of nutrient elements. Pan et al. ( 1997) proposed that changes in soil fertility could be measured by quantifying soil physical and chemical properties, soil nutrient element status,microbial populations and biochemical characteristics, but that ultimately changes in forest productivity were the best indicator of soil quality. In this study, maximum tree growth rates (as measured by changes in height and DBH) coincided with a rapid reduction in soil quality, suggesting that the growth ofPopulus×xiaoheicaused the decrease in soil quality. However, as forest age increased, this trend gradually decelerated, and soil quality had recovered in the PMF. Jiang et al. ( 1994) have suggested that the life cycle ofPopulusshelterbelts in northeast China generally lasts between 30 and 37.5 years, after which the protective benef its of the shelterbelts are diminished. Thus, from the perspective of maintaining soil quality and maximizing the comprehensive benef its of the shelterbelt, shelterbelt renewal should be carried out at the post-mature stage.

Suggestions for further research

Due to time limitations, the space-time substitution method was adopted in this study. Relatively fertile topsoils were collected from study shelterbelts, twenty-one soil quality variables measured, and SQI evaluated using 16 indicators.As a kind of living organism, soils have complex and diverse properties, among which soil microorganisms are the most active factor. However, in this study, only MBC and MBN were considered. The mechanisms underlying soil microbial Effects on forest soil quality remain unclear. In order to make a more accurate and objective evaluation of changes in soil quality over the time inPopulus×xiaoheishelterbelts,a long-term, f ixed-point experiment should be undertaken,with accompanying studies of soil depth and soil microorganismal gene function.

Conclusions

The following conclusions can be drawn from this study of the changes in surface soil quality inPopulus×xiaoheishelterbelts of diff erent ages.

1. Compared to control, there was higher MC and reduced BD in PMF, while NN and TP were lower; in MF, pH,TOC, AHN, AP, MBN were higher than in the control.

2. In the soil quality model, the greatest weight was assigned to TOC and the least to TK. SQI varied amongPopulus×xiaoheishelterbelts of diff erent ages, being lowest in NMF and highest in PMF; SQI was higher in the 10-20 cm soil layer than in the 0-10 cm layer. After a singlePopulus×xiaoheishelterbelt life cycle, the surface soil quality was signif icantly improved.

Acknowledgements We would like to thank the local government and organization for their help with f ieldwork, and our colleagues for their guidance in carrying out the lab analyses. Finally, we would like to give special thanks to the two anonymous reviewers and the editor for their helpful suggestions on the manuscript.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affi liations.