Factors promoting the natural regeneration of Larix principis-rupprechtii plantation in the Lvliang Mountains of central China

Study area

The study sites were located at high elevations in the forest zone of Guandi Mountain in the middle of the Lvliang Mountains, west Shanxi Province, China (37°45′−37°55′N, 111°22′−111°33′E) (Yang et al., 2017). The annual mean temperature is 4.3 °C, and the mean annual precipitation along this gradient ranges from 600 mm to 822 mm (Liu et al., 2007). The region has a temperate continental climate, with long, cold and dry winters and short, warm and rainy summers, the coldest month is January (−10.2 °C), and the warmest month is July (17.5 °C); the growing season generally runs from June to September. The soil is classified as brown soil (Chinese classification) with an average thickness of 70–80 cm, including a 10-cm humus layer (Yang et al., 2017). The forest communities are primarily composed of Picea, L. principis-rupprechtii, Betula, Populusdavidiana and Pinus tabulaeformis. Shrub communities primarily consist of Rosa xanthine, Hippophae rhamnoides, Vitex negundo and Heterophylla. Herb communities include Artemisia and Carex.

Community surveys

Community surveys were conducted in July 2019. A total of 18 plots (30 m ×30 m) were made, and each plot was divided into three forest types. Six plots had no regenerated larch seedlings, six plots had a few regenerated seedlings and six plots consisted of purely regenerated seedlings. Regeneration status was considered good if the number of seedlings > number saplings > number of trees and fair if the number of seedlings > number saplings. Regeneration was considered poor if the species was only present in the sapling stage and no seedlings were present or if there were no signs of regeneration despite the presence of adult trees (Ballabha, Tiwari & Tiwari, 2013; Ali et al., 2019).

The location of each plot (including elevation, longitude and latitude) was recorded using a navigation satellite time and ranging global position system receiver (TX35-S300, China) as well as the slope, slope position and slope aspect (Li et al., 2018). Variables measured in each plot were classified as stand structure data and included the following: (1) species, number, diameter at breast height (DBH), height and average crown; (2) identity and coverage of all shrub and herb species; (3) litter through the insertion of a metal ruler down to the soil surface at five points; and (4) tree stems, which were classified as either seedlings (height <50 cm), saplings (height > 50 cm and DBH(diameter at breast height) <5 cm) or adult trees (DBH > 5 cm) (Carrer, Soraruf & Lingua, 2013; Muhamed et al., 2018).

Soil environmental surveys

Five soil samples was obtained with a “X” shaped collection scheme at depths of 0∼60 cm in each plot, because the soil layer below our study area primarily consists of stones. To mitigate the influence of precipitation on soil moisture, the soil samples were not collected within 2–3 days after rainfall. Wet and dry weighing methods were used to determine soil water content. The air-dried soil samples were then milled and sieved with a 2-mm screen in order to remove the debris and roots from the soil. The chemical properties were measured in the laboratory, i.e., soil organic matter (SOM) was determined using the K₂Cr₂O₇ method; total nitrogen (TN) was measured using SKD-100 DigiPREP TKN Systems (Saacoo, Chongqing, China). Available nitrogen (AN) was extracted with KCl and quantified; on the other hand, total phosphorus (TP) was measured after extraction with HClO₄-H₂SO₄. In addition, NaHCO₃ solution was adopted to measure available phosphorus (AP) according to Olsen’s method and ammonium acetate solution was used to quantify available potassium (AK) (Li et al., 2018).

Statistical analysis

Environmental gradients were identified through sequencing analysis using CANOCO (version 5.0). A preliminary detrended correspondence analysis for the regeneration of all community plots indicated that the gradient lengths were less than 3.0, which demonstrated that most regenerated plots performed linear responses to potential environmental variation, justifying the use of linear multivariate methods (Li et al., 2018). Based on this theory, the redundancy analysis (RDA) method was selected to identify the dominant environmental factors impacting on the community, implemented in the CANOCO 5.0 software package. The ranking scale focused on interspecies correlations, and the significance of data was evaluated via Monte Carlo test (Chen, 2003). Then, regeneration data was transformed through logarithmic conversion so that the effects of extreme values were mitigated (Gazer, 2011).

Species were indicated by the number and height of regeneration seedlings. Environmental factors were divided into four categories: topographic factors (elevation and slope), stand structural factors (density, average breast diameter, tree height and crown width), soil properties (soil organic matter, total nitrogen, total phosphorus, available phosphorus, available potassium and ammonia nitrogen) and litter (thickness and accumulation). All data for these factors were square-root transformed to ensure that variances were homogeneous before statistical analyses.

Using ANOVA method, the logarithmic values after transformation were calculated. Duncan’s multiple range test (P < 0.01 or P < 0.05) was used to compare differences in topographic factors, stand structural factors, soil properties and litter. And SPSS 22.0 software was adopted to analyze descriptive statistical parameters and significance of all the indicators.

Effects of topography on regeneration

Micro-topographic factors, such as slope, greatly impact plant community structure and patterns of regeneration. The ordination results indicated that the slope was the main terrain factor affecting regeneration (Wang et al., 2016). The slope was positively related to aboveground biomass and herb coverage but negatively related to regeneration and stand density. However, the effects of topographic factors (−0.48) on regeneration were less significant compared with soil factors (−0.54). In our study area, the difference between the maximum and minimum elevations in the sampling plots was 124 m, which is not likely sufficient for causing significant variations in thermal conditions and thus affecting regeneration (Liu et al., 2012). Topographic factors affected both vegetation and soil nutrients. Topography can partially influence the accumulation and export of soil nutrients and thereby indirectly impact the distribution of plants. For example, previous studies have indicated that topographic changes can have strong effects on soil physical and chemical properties and soil moisture characteristics (Sariyildiz, Anderson & Kucuk, 2005; Dessalegn et al., 2014). Here, regeneration was negatively correlated with slope, suggesting that it is easier for seeds to accumulate on gentle slopes.

Effects of stand structure on regeneration

The relative contributions of different factors also varied depending on the life history stages of regenerated seedlings. The height structure of the regenerated seedlings is an important factor affecting the regeneration of trees, and reflects the status of stand regeneration (Kayes & Tinker, 2012). Stand structure factors also affect restoration and play important roles in regulating the natural regeneration of forest (Chen & Cao, 2014). The RDA analysis and SEM all showed that tree DBH was positively correlated with seedling regeneration. Parent trees with larger DBH generally obtain more resources to support their growth and occupation of a dominant position in the forest stand (Qian et al., 2014). The number of seeds produced by robust parent trees is higher in both quantity and quality and might explain the high number of regenerated seedlings. There was a positive correlation between stand crown and regeneration, but the strength of this correlation was relatively weak. Overall, the relationship between larch regeneration, DBH and stand density was highly consistent in our study site and is consistent with the results of other studies (Hébert, Krause & Plourde, 2016; Ali et al., 2019).

Effects of shrubs, herbs and litter on regeneration

Undergrowth vegetation is a critically important factor affecting forest ecosystems (Nilsson & Wardle, 2005), as it has an inhibitory effect on forest regeneration and, in turn, on forest dynamics (Montti & Campanello, 2011). However, shrubs were generally few and scattered in our plots, suggesting that they had little effect on stand regeneration. In regeneration plots, saplings grew well and were distributed in clusters, which inhibited the growth of herbs. Herbs did not affect the regeneration of seedlings, given that L. principis-rupprechtii is a shade-tolerant plant (Wagner et al., 2010; Lombaerde et al., 2019). Indeed, the shade provided by herbs protects the growth of the seedlings. Eventually, seedlings suppress the growth of herbs, which explains the decrease in herbs observed as a consequence of canopy closure (Soucy, Lussier & Lavoie, 2012).

L. principis-rupprechtii seeds were approximately 1–2 mm in length and were primarily concentrated in the litter layer. The heat preservation and water retention capacity of the litter layer provided prime conditions for seed germination. Many studies have shown that seed survival rate is significantly higher in litter than in bare habitats (Ibáñez & Schupp, 2002; Muhamed et al., 2018). This study also showed that litter thickness was closely related to the regeneration of larch, with an RDA contribution rate of 16.7% and an SEM path coefficient of 0.57. Other studies have shown that litter inhibits seed germination; for example, regeneration seedlings often failed to obtain sufficient nutrients given that the radicle could not reach the soil, resulting in increased mortality in regeneration seedlings (Nakagawa, Kurahashi & Hogetsu, 2003). The continuous accumulation would also impede the natural regeneration of vegetation, primarily via physical barriers, chemical effects, infestations of animals and microbial pathogenic effects. However, litter thickness was positively correlated with regeneration. This correlation might be explained by the moderate litter thickness in the study area, which did not accumulate and affect seed germination and seedling growth. However, this correlation also suggested that the litter is decomposed at an adequate rate and reflects an overall healthy ecological cycle in the study area. Litter layers of this thickness provided sufficient water for the seeds, facilitated the return of nutrients to the soil, promoted the growth of newer seedlings and also provided some protection for the seeds by reducing avian damage. Thus, this study suggests that the litter thickness in these plots represent thicknesses that would be ideal for stands of L. principis-rupprechtii in this region. In stands with thicker litter layers, litter could be managed to both facilitate the natural regeneration of the stand as well as nutrient cycling in the soil.

Relationship between soil nutrients and regeneration

Forest regeneration has a clear, predictive effect on soil nutrient contents (Fu, Qi & Chang, 2015); soil variables, such as AP, AK, ammonia nitrogen and soil organic matter; and the growth, reconstruction and development of vegetation (Qian et al., 2014). According to the SEM, ammonia nitrogen and soil organic matter made the largest contributions to explaining regeneration compared with other variables relating to soil properties (Wang et al., 2016). Because a large amount of litter is produced in the soil every year, the decomposition of litter increases soil organic matter. Consistent with this expectation, soil organic matter was positively correlated with average crown and negatively correlated with the height of regenerated seedlings; and high, but not significant, correlations were observed between average crown and seedling height. Thus, plants might increase soil organic matter content.

Similar to previous studies on forest regeneration (Chen, 2003; Bünemann et al., 2004), AP was one of the most important variables affecting variation in species richness, and its response curve was consistent with the response curve for regeneration. Ammonia nitrogen had the strongest effect on changes in species composition at our study sites, but the number of species varied little relative to variation in ammonia nitrogen. For instance, the number of species was at its maximum when ammonia nitrogen was relatively low (Liu et al., 2011). The richness and quantity of species varied in the RDA, demonstrating that both the diversity and numbers of species—specifically the ‘regeneration quantity’ and ‘valuable indigenous species’—were insufficient and that further restoration and cultivation are needed. High contents of ammonia nitrogen and soil organic matter inhibit regeneration; however, regenerated seedlings might have a higher consumption of ammonia nitrogen and soil organic matter (Xu et al., 2018).

Management implications

The most important factors affecting the regeneration of L. principis-rupprechtii plantations included ammonia nitrogen, soil organic matter, litter thickness, DBH and slope. These factors are also important in other systems (Puhlick, Laughlin & Moore, 2012). Thus, measurements of these variables could provide forest managers a means of assessing the state of regeneration in a given stand. The removal of litter to reduce its thickness could improve seed germination and promote regeneration. The removal of unhealthy trees in the forest could increase the space available to mother trees and promote their growth. However, increases in light and microbial activity accelerate the decomposition of litter, reducing its thickness and increasing soil organic matter. Overall, appropriate management actions can promote the robust regeneration of forest stands.