Climate changes in China are basically consistent with global warming, the spatial distribution and temporal variation of precipitation variability are large , the relative humidity of the ground has decreased in most parts of China . Fog and haze weather, acid rain phenomenon, the shortage of water resources, land desertification and the reduction of forest biodiversity being caused by climate change have become an important topic in the world. The impact of rising temperatures on the environment and ecology is profound . Climate change has a significant impact on the distribution of Chinese plants . Vegetation distribution is determined by their parents distribution as well as by complex environmental factors [5,6]. Seedling-sapling may be the most important stage in the whole life history of plants . Compared with the adult trees, seedlings have stronger sensitivity to the environment . Environmental variables have spatial heterogeneity that can be expressed as horizontal and vertical zonality . In terms of forest vegetation, they mainly include temperature, light, soil moisture, soil properties, canopy , latitude, altitudinal gradient , longitude and slope habitat factors. While, the main cause of environmental and climate change is especially remarkable in expressing in temperature, precipitation and altitude. They have interrelationship with each other. Environmental factors significantly restrict seed germination of plant and availability of seedling growth opportunities . Therefore, a suitable growth environment is critical to species regeneration and succession, especially for seed and seedling.
However, finding out how the environmental variables affect the seedling-sapling distribution is important. Most vegetation models are state-of-the-art tools used in ecological and forest ecosystems. In the past decade, many different types of factors have been used to evaluate the response of vegetation to climate change . At the same time, the methods of developing equations to describe for environmental factors and their influence on plants distribution have evolved from simple least squares regression to seemingly unrelated regression (SUR) models [14,15]. The key to model fitting istoanalyze the main components of model. Principal component analysis (PCA) is a multivariate technique that analyzes a data table in which observations are described by several inter-correlated quantitative dependent variables . Principal components are often used as a dimension reduction device to obtain a smaller number of variables for input to another analysis . Another method is Gaussian mixture model (GMM), which is a good method for describing the situation of seedling-sapling distribution . The data can naturally be modeled as a low-rank plus a sparse contribution by important applications method. Robust principal components fitting our model are sought by all the statistical applications . However, few studies have applied PCA and GMM modeling to describe seedlingsapling distribution from our knowledge.
Spatial and temporal patterns of seedling survival suggest that exposure to high sunlight may exacerbate low-temperature and water stress in young conifer seedlings . Below ground invertebrates, soil nutrients and seedling distribution have mixed effects with them . Environmental conditions such as trees positions have multi effect on seedling distribution . Both topography and relative locations of habitat patches features affected bird behavior and seed dispersal and seedling distribution . A topographical moisture gradient has relationship with the distribution of seedling establishment shapes . Seeds dispersed in a forest floor mosaic of litter depths that form an obvious barrier to plant seedling recruitment . Therefore, we should give a good summary for the relationship between environmental variables and seedling-sapling distribution.
Dacrydium pierrei belongs to the family Podocarpaceae, and is distributed in Oceania, South America, Indo-China, and Southeast Asia. This species is the only representative of this genus in China. At present, it is distributed in the rainforests of Jianfengling, Bawanglingand Diaoluo Mountains, which are located in the southern part of Hainan Province [26,27]. This species is classified as a third-class rare and endangered plant in China . The wood of Dacrydum pierrei is an excellent material for high-grade architectural work and handcrafting . Since the 1960s, the abundant tropical forest resources of China have been severely damaged by excessive deforestation, typhoons, and other external forces, leading to a significant reduction in natural forest resources . As a result of deforestation, Dacrydium pierrei natural forests have been destroyed. The reduction in the size of forested areas threatens the natural evolution, survival and development of this species. Consequently, Dacrydium pierrei could become extinct in China.
Now, the research on Dacrydium pierrei is still in initial stage. Researchers have primarily focused on the forest community structure characteristics , population genetic diversity , palynological analysis and the origin of evolution . Several studies about environmental variables and seedling-sapling distribution have been conducted in Costa Rica, Canada and Finland [33,34,35]. A model was developed to predict tree response to a synthetic environmental variable generator . Providing quantitative estimates of changes in lake water chemistry, soil variables relating to past vegetational shifts [34,36]. From the perspective of landscape heterogeneity and habitat cover [37,38], there is little research like that in China. Thus, to provide relevant guidelines to improve the management of Dacrydium pierrei natural forests, the current study is aimed at finding how the environmental variables effect seedling-sapling distribution of Dacrydium pierrei. Besides, the usefulness of model depends strongly on their ability to predict the seedling-sapling distribution correctly under different environment scenarios. Thus, the major objectives of this study wasto: (1) develop a new model for modeling seedling-sapling distribution based on environmental factors and seedling trait; (2) simulate the distribution patterns of the seedling-sapling of Dacrydium pierrei in Hainan. (3) investigate the response of forest ecosystems to a changing environment through sensitivity analysis.
2 Materials and methods
2.1 Study area and sample selection
The research area is located in the National Nature Reserve of Bawangling, Hainan Island, China (109.189°–109.201° longitude and 19.089°–19.096° latitude) . Based on the meteorological observation data from 1970 to 2016 at the weather station of the Wulie forest farm. The area has a tropical monsoon climate with an annual mean temperature of 23.6 °C. The average temperature of the coldest month (January) is 18°C. The average temperature in the hottest month (July) is 28.6°C. The average annual rainfall is 1500–2000 mm. The precipitation is at most in July or August (626.2 mm), annual evaporation is 1596.8–1646.5 mm; average relative humidity is 76%–82%; average annual sunshine is 2000–2300 h (Figure 1). The forest area of Bawangling is large, land types are complex. There is a big difference between vertical and horizontal directions microclimate. You will be feeling the heat at the foot of the mountain, but it is cold on the top of the mountain. The forest will encounter typhoon, frost, coldair outbreak occasionally .
The research area has 2213 species of vascular plants, belonging to 967 genera and 220 families. There are 131 species of fern belonging to 73 genera and 36 families. There are 13 species of gymnosperms belonging to 5 genera and 8 families. There are 2069 species of angiosperms belonging to 886 genera and 179 families. Two of these species are first-level nationally protected plants, Cycas hainanensis and Hopea hainanensis. There are 18 second-level protected species within the following genera: Oncodostigma hainanense, Alseodaphne hainanensis, Horsfieldia glabra, Heritiera parvifolia, Cibotium barometz, Ceratopteris thalictroides, and B rainea insignis. Third-level nationally endangered and protected plants include Ixonanthes chinensis, Podocarpus imbricatus, and Dacrydium pierrei .
2.2 Field investigation
A comprehensive survey of Dacrydium pierrei individuals distributing in the two 2.5-km-long line on both sides of the ridge was acted in October 2013, Dacrydium pierrei was found only in the ridge and uphill, it was rarely found in the middle and lower slopes, few seedlings under the land, only one relatively concentrated area of Dacrydium pierrei seedlings was found, and then the area was investigated in March 2014. Taking the ridge line as the center line, extended 35m (horizontal distance) on both sides, setting up the sample plot with the horizontal area is 100m x 70m. The average elevation of plots is 1240 m, ridge direction is from northwest to southeast, and both sides of the ridge slope are about 40 degrees. The plots were divided into 70 continuous sampling with 10m x 10m, each samples were further divided into 4 small squares with 5m x 5m, numbering each sampling point, and recorded the spatial coordinates of the sample points (Figure 2).
We carry out wood seized feet for all Dacrydium pierrei individuals in the forest (diameter at breast height (DBH), tree height, crown width), recorded the coordinate data of each Dacrydium pierrei. According to DBH and tree height, we classified Dacrydium pierrei into (Height ≤ 50cm) seedlings and saplings (DBH ≤ 10cm, Height > 50cm) and mature tree (DBH > 10cm) 3 growth stages. Because of the temperature, light, humidity, moisture and rainfall anomalies or the barrier of forests or ground cover, the litter makes seeds difficult to take root and germinate, and then affect the distribution of seedlings and saplings. Five quadrant were randomly selected in each plot (10m x 10m), soil was sampled in 0–10 cm depth, the 5 duplicate samples are fully mixed and then loaded into a soil bag (about 200g) for laboratory analysis. Environmental factors (same as soil sampling scale factor) that thickness of litter, herbaceous layer coverage, canopy density, air temperatures, soil temperatures and soil moisture content and illuminance of four 5m x 5m plots were determined by hand held soil temperature and humidity measuring instrument (average value of four points selected randomly) etc. There were measured and calculated the average value by analysis process.
2.3 Laboratory experiment
We determinate soil chemical indicators in the laboratory, the available potassium acetic acid ammonia leaching absorption spectrophotometry with atomic, inorganic nitrogen, ammonium nitrogen, KCl extraction method was used to measure nitrate nitrogen, the extraction solution was determined by AA3 continuous flow analyzer. Inorganic nitrogen content was got by adding nitrate nitrogen to ammonium nitrogen. Organic matter was determined by potassium dichromate. Rapidly available phosphorus was measured by sodium bicarbonate leaching with molybdenum antimony colorimetric method. After that, eighteen environmental variables (litter thickness, herb coverage, canopy density, air temperature, illuminance, soil temperature, soil water content, inorganic nitrogen, ammonium nitrogen, nitrate nitrogen, available phosphorus, available potassium, organic matter, altitude, longitude, latitude, slope, aspect) were obtained. The 18 environmental variables are defined as X1-X18 respectively.
2.4 Factors measurement of Dacrydium pierrei
Dacrydium pierrei seeds mature in October. Mother trees were selected in the sample to collect healthy seedsin the October of 2014. We cleaned seeds manually and stored it in darkness with room temperature until sowing in a laboratory experiment conducted in 2015. A site was selected for collection of soil samples in October 2014. We sampled the soil using shovel. Samples were taken back to the laboratory. Soil samples were used in a control experiment for seedling emergence. Before the experiment, soil was sterilized in an oven for 72 h at 80°C to kill any plant seeds .
The experiment was conducted using a random sampling method to investigate the effects of moisture conditions on seedling emergence. A phytotron (E15 controlled environments. Ltd. Canada) was used to automatically control the relative humidity of the air. Moisture conditions consisted three treatments: moist (air relative humidity: 90%), dry (air relative humidity: 50%) and intermediate (air relative humidity: 70%). Two trays were used to contain bare soil (control) with ten replicates for each substrate treatment for each moisture treatment. There were 30 experimental units in three moisture treatments in total.
In November 2014, thirty plastic trays sized 19 x 13.5 x 7 cm3 were filled with thin sterilized soil (soil thickness: 2.01 ± 0.15 cm). The soil was saturated with distilled water. Thirty additional plastic trays were filled with sterilized soil (soil thickness: 5.03 ± 0.38 cm). Soil water content underthe various relative humidity conditions was measured using the conventional oven-dry method. Soil moisture states were divided into three classifications, moist, intermediate and dry, having soil water content of 58.4% ± 2.7%, 45.3% ±1.8% and 17.2% ± 1.4%, respectively.
Dacrydium pierrei seeds were disinfected using 0.5% concentration potassium permanganate for one hour washed with sterilized water repeatedly and dried, and then 100 seeds were randomly sown in each plastic tray by carefully placing them with forceps. The trays were kept without lids over the experimental duration to maintain adequate light and air circulation.
To maintain the three moisture levels, the plastic trays were sprayed with distilled water and a weak nutrient solution regularly. The trays were sprayed every second day for the moist treatment, and every four and six days for the intermediate and dry conditions respectively. Each time, a 150 ml volume of distilled water was used. The weak nutrient solution was sprayed once a half month. The experiment was terminated when no additional seedling emergence was observed, lasting for a total of two months. We counted all seedlings at the end of the experiment.
2.5 Statistical analysis
2.5.1 Principal component analysis (PCA)
The data were summarized using a variables matrix taken to analyze the habitat of Dacrydium pierrei seedling-sapling, variables included altitude, slope, aspect and canopy density. Principal component analysis (PCA) was used to examine the variation in habitat among plots (CANOCO 5.0; software for Multivariate Analysis of Ecological Data). Due to the great variation of the variables and different measurement units, the correlation coefficient matrix is chosen to be used in the PCA process .
Assuming the acquisition variables is (1)
The observation value of N group is χi=[χi1,χi2,…,χip]T, i=1,2,…,n.
The principal components are calculated as follows:
The sample covariance σij and sample correlation coefficient γij are calculated from the observed data that is (2) in the formula, Thereby the covariance matrix V and the correlation matrix R are obtained: (3)
For p eigenvalues of R, remember it as λ1≥λ2≥…≥λp≥0.
The number of principal components that need to been selected was solved by principal component contribution. The contribution rate of the i principal component remembered as (4) taking (5) as the cumulative contribution rate of the first m principal components.
Calculate unit characteristic vectors corresponding to the first M eigenvalues li, i=1,2,…,m.
The principal component of the first i sample of X is yi = X . Then, obtained principal component variables are (6)
2.5.2 Gaussian mixture model (GMM)
In order to clarify the influence degree of important variables on seedling-saplings distribution, we discussed seedling sensitivity by Gauss mixture model (GMM) , which can determine the significant response variables on the distribution of Dacrydium pierrei seedling-saplings.
A GMM is a combination of several individual Gaussian components: a 1-dimensional Gaussian mixture (Equation 1) can be represented in 2-dimensional space, and a 2-dimensional Gaussian mixture (Equation 2) can be represented in 3-dimensional Gaussian space .
In Eq. 7, f(x) is the occurrence probability function of a 1-dimensional trait belonging to a specific seedling number, which is also known as a 1-dimensional Gaussian function; x is the independent variable (i.e., trait); and μ is the mean value of the trait sample for a specific seedling number. σ represents the standard deviation of the sample.
In Eq. 8, f(x,y) is the occurrence probability function of 2-dimensional traits belonging to a specific seedling number, also known as a 2-dimensional Gaussian function. x and y are independent variables (i.e., traits). μ1 refers to the mean of the first trait dimension, and μ2 refers to the mean of the second dimension. σ1 and σ2 are the standard deviations of the sampled traits. r2 is the correlation coefficient between x and y.
A GMM can be expressed as in Eq. 9 (9)
Where p(Ck) is the Gaussian density of traits belonging to the Ck class; Jc is the number of components; and wj. represents the components’ weights, such that wj > 0, and Σwj = 1. f (θ, j) represents the jth Gaussian component.
The conducted research is not related to either human or animals use.
3.1 Environmental variables and distribution of seedling
In order to eliminate the influence of dimension before calculation, the original data was standardized. Formula is: ,
Si is sample mean square error.
To identify important variables affecting the distribution of Dacrydium pierrei, we extracted the principal components (PC) by PCA. The results of principal component analysis can be obtained from principal component analyzed data, the total variance analysis is shown in Table 1. The cumulative contribution rate of the first 6 principal components is 82.52%. We selected the first 6 principal components that reflect the required assessment information.
Get the eigenvector matrix of 6 principal components by squaring root of the eigenvalues of the 6 principal components, and then the eigenvector is multiplied with the normalized data to obtain the expression of 6 principal components:
F1=+0.227ZX1−0.134ZX2−0.446ZX3+0.182ZX4+0.205ZX5+0.021ZX6−0.393ZX7−0.197ZX8+0.209ZX9−0.163ZX10+0.174ZX11−0.136ZX12−0.357ZX13+0.397ZX14−0.179ZX15+0.206ZX16−0.588ZX17−0.216ZX18 F2=+0.059ZX1−0.344ZX2−0.007ZX3+0.439ZX4+0.225ZX5+0.321ZX6−0.269ZX7−0.306ZX8+0.173ZX9−0.103ZX10+0.023ZX11−0.136ZX12−0.013ZX13+0.011ZX14−0.418ZX15+0.387ZX16−0.135ZX17−0.149ZX18 F3=+0.312ZX1−0.474ZX2−0.205ZX3+0.187ZX4+0.016ZX5+0.214ZX6−0.487ZX7−0.541ZX8+0.027ZX9−0.266ZX10+0.574ZX11−0.002ZX12−0.206ZX13+0.009ZX14−0.148ZX15+0.061ZX16−0.212ZX17−0.126ZX18 F4=+0.647ZX1−0.129ZX2−0.015ZX3+0.344ZX4+0.061ZX5+0.155ZX6−0.174ZX7−0.226ZX8+0.108ZX9−0.204ZX10+0.611ZX11−0.017ZX12−0.021ZX13+0.108ZX14−0.204ZX15+0.033ZX16−0.112ZX17−0.129ZX18 F5=+0.065ZX1−0.027ZX2−0.411ZX3+0.553ZX4+0.112ZX5+0.208ZX6−0.315ZX7−0.152ZX8+0.016ZX9−0.167ZX10+0.305ZX11−0.125ZX12−0.135ZX13+0.271ZX14−0.166ZX15+0.073ZX16−0.106ZX17−0.184ZX18 F6=+0.095ZX1−0.287ZX2−0.065ZX3+0.116ZX4+0.158ZX5+0.006ZX6−0.557ZX7−0.623ZX8+0.592ZX9−0.311ZX10+0.137ZX11−0.365ZX12−0.069ZX13+0.017ZX14−0.248ZX15+0.013ZX16−0.185ZX17−0.387ZX18
And then, performing the stepwise regression analysis with the data while taking the number of seedling distribution (ZY) as the dependent variable (standardized data). Taking F1-F6 as an independent variable for multiple linear regression analysis (Table 2).
The regression model can be established by the stepwise regression analysis of seedlings distribution and the main components: ZY=2.76F1-4.246F3+1.882F5
3.2 Result of GMM methods
The intervals, mean etc. of the parameters estimated were listed in Table 3.
We tested all of the models listed in Table 4 and compared the results. The results demonstrated that: in all models, the GMM trained by the L3-H4-A5-I6, combination exhibited the highest accuracy (overall accuracy=74.74%; kappa coefficient =0.86); and the optimal number of trait was four, with this model showing higher accuracy than the 6-trait combinations.
The A1-C2-L3-H4-A5-I6 combination shows limited predictive ability when it is integrated into predict model. The C2-L3-H4-A5-I6 combination exhibited similar accuracy (overall accuracy = 72.23%; kappa coefficient = 0.85) as that of L3-H4-A5-I6 combination and could provide more parametric information about seedling distribution pattern. Therefore, the C2-L3-H4-A5-I6 combination was applied in training the GMM for the analysis of trait-environment relationships and the response of seedling distribution patterns to environment change, indicating that the trained GMM was sufficiently accurate for application in modeling seedling distributions of Dacrydium pierrei.
3.3 Sensitivity analysis
A changed air temperature shifts the seedling distribution boundaries. Seedling numbers exhibit an increasing trend with the increasing air temperature and illuminance, and then decreased as it increases. While, it was not significant in that of litter thickness, the canopy density have the opposite phenomenon in that changing, the seedling numbers are decreased with the canopy density increasing. There is a same regularity in herb coverage. With the increasing in altitude, there is not aorderliness in seedling number with increasing or decreasing.
3.4 Effect of soil moisture content on seedling establishment
Overall, accroding to laboratory experimen t(soil moisture was measured by gravimetric method) result, seeding emergence significantly differed (P < 0.05) among different moisture conditions of bare soil (Fig. 3). In the dry and intermediate moisture conditions, the emergence rate was significantly lower (P < 0.05) than that in moist condition (Fig. 3).
4.1 Environmental variables influencing seedling distribution
This study demonstrates that environmental variables have significant effects on the distribution of Dacrydium pierrei seedling. Similar studies have been conducted in other areas [44,45,46,47]. Based on the method of PCA, we found that among six factors, canopy density and soil moisture are associated with the PC1, having important factors that impact the abundance and distribution of Dacrydium pierrei seedling-saplings, and they explained the most variation (43%) in the variables. Previous studies have demonstrated that canopy density and soil moisture have a strong influence on seedling-sapling distribution [48,49]. The place with small canopy density has plenty of light, which is conducive to the improvement of soil temperature, while higher soil temperatures promote seeds germination and seedlings growth. Nitrate nitrogen and illuminance also have the same impaction. Claussen  researched on regeneration pattern of tropical rainforest species, and also confirmed that light and temperature had significant effects on seed germination and seedling growth. Ma et al.  found that in addition to salt, drought, high and low temperature, light intensity was too strong would affect the growth and development of plants, and ultimately threaten the growth and reproduction of plants. Gray et al.  suggested that gap size, within-gap position and canopy structure effects on conifer seedling establishment. Beckage et al.  also found that canopy gaps and shrub understories have a big impact on the tree seedling recruitment in southern Appalachian forests.
Available potassium and litter thickness are, correlated with the PC2, the critical biotic variables affecting the distribution and abundance of Dacrydium pierrei seedling-saplings. We also found that the input of litter significantly affected the growth of Dacrydium pierrei. Molofsky et al.  said the amount and distribution of leaf litter within a forest can influence patterns of plant establishment. Janišová et al.  found that tree-litter cover negatively affected in situ germination. Kisinyo  studied the effects of phosphate fertilizer on soil chemical properties (soil chemistry) and the growth of maize seedlings, the results indicated that available phosphorus promoted the growth of maize seedlings. In addition, previous studies have shown that microhabitat characteristics such as high temperature, high intensity light, low soil moisture and low soil nutrients could affected the seed germination, seedling growth and sapling establishment in the dry season [55,56].
In other words, the survival and growth of seedling-saplings are more sensitive to environmental change. Our results show that altitudes ranging from 1140 to 1300 m can provide a suitable habitat for seedling establishment of Dacrydium pierrei. So, different tree species have a specific elevation growth range with forests that growing in high altitudes having fundamental ecological roles. Consequently, the spatial and temporal dynamics of forest natural regeneration are correlated . For instance, altitude, age and competitive interactions modify the growth response to climate  affecting overall forest growth.
The seedling-sapling distribution of Dacrydium pierrei is topography-dependent. Slope and aspect significantly influenced the recruitment of Dacrydium pierrei seedlings-saplings. The slope was negatively correlated with Dacrydium pierrei seedling recruitment. Because the ability of slope to fix the seed weakens with slope increasing with a lower chance of rooting by seedlings on steeper slopes, the seedlings recruitment of Dacrydium pierrei decreased with slope increasing. The Dacrydium pierrei seedlings were mainly distributed on the ridge with almost no seedlings being distributed on other parts of the slope. Thus, the Dacrydium pierrei seedling-sapling recruitment has a specific slope position distribution.
Longitude and latitude significantly affect the survival and development of Dacrydium pierrei seedling-sapling in relation to changing solar radiation intensity from south to north and west to east. Longitude was positively correlated with Dacrydium pierrei seedling recruitment. The recruitment of Dacrydium pierrei seedlings increased with solar radiation intensity from east to west increasing. Latitude was negatively correlated with the recruitment of Dacrydium pierrei seedlings. The recruitment of Dacrydium pierrei seedlings decreased with solar radiation intensity from south to north increasing.
4.2 GMMs application and sensitivity analysis
GMMs have been successfully accepted and applied for the prediction of seedling distribution of plants. In this study, we improved the prediction accuracy to 74% based on the optimal trait-environment model for Dacrydium pierrei seedlings. Among that model with A1-H4-A5 has still low overall accuracy (53.62%) and kappa coefficient (81.16%) showing little improvement than before factors combination. The model with A1-C2-H4 has the highest accuracy (71.85%) coefficient of that is 82.95% of the three trait models. For the four trait models, the model with L3-H4-A5-I6 has the highest accuracy (74.74%) with the coefficient is 86.25%, which is also the most optimal trait-environment model of this study. While, the accuracy of A1-C2-L3-H4-A5-I6 model is not the highest (68.43%) of all. We suggest that the spatial distribution of seedlings-sapling was positively correlated with soil temperature, litter thickness and available phosphorus, having a negative correlation with canopy density, available potassium and nitrate nitrogen. In a word, more effective trait-climate relationships should be developed in the future.
4.3 Effect of temperature and precipitation on seedling-sapling distribution
Soil temperature is the representative of comprehensive effect of air temperature and rainfall intensity on seedling . Thus, soil moisture effects content on species distribution mirrors the influence of climate change on species distribution. Three different moisture conditions in the study area represent obvious climate change, which have a sharp impact on the distribution of species especially on seedling-sapling. The result shows that seedling emergence rate is low at dry moisture condition, and is becoming high with the increase of water in soil, and arrive at the highest in the most moist soil condition (Fig. 3). This is the same as a study by Gong et al. , who found that the higher the soil moisture, the higher the germination rate and germination energy of the seed in seed favorites humid conditions. We deduce that seedling emergence should having enough water sucked into the seeds and should have a suitable temperature environment. While the water content is controlled by precipitation. Previous studies have shown that site precipitation had an overall positive effect on seedling recruitment especially at intermediate precipitation levels, and seedling emergence and establishment have big difference with all temperature levels . Benigno et al.  found that increasing soil water retention with native-sourced mulch improves seedling establishment in post mine mediterranean sandy soils. Given the above, our results provide evidence for the research hypothesis that soil moisture content have strong impact on seedling distribution.
The adaptability of seed germination and seedling establishment to the heterogeneous environment is pivotal for population succession and species propagation. The seedlings distribution was affected by biotic and abiotic factors. We suggest that the spatial distribution of seedlings-sapling was mainly effected by soil temperature, litter thickness, available phosphorus, canopy density, available potassium and nitrate nitrogen. Besides, topographic characteristics have a relatively significant effect on the distribution of Dacrydium pierrei seedlings, such as longitude, latitude, slope and aspect. But these factors will not be the dominant factors of Dacrydium pierrei’s natural regeneration, because it has the specified survival area. The Dacrydium pierrei seedlings had a specific range of survival. What’s more, the C2-L3-H4-A5-I6 combination was applied in training the GMM is sufficiently accurate for application in modeling seedling distributions of Dacrydium pierrei. The soil moisture and precipitation play an important and crucial role in seed germination, seedling establishment and distribution of seedlings-saplings. All of these discoveries provide a scientific basis for the further understanding of Dacrydium pierrei population maintenance mechanism, the relationship between regeneration dynamics and habitat, natural regeneration barrier and ecosystem restoration.
This study was supported by the National Natural Science Foundation of China (31270678). The authors thank the Bawangling Natural Reserve of Hainan Island who provided the test site and experimental materials, and Rucai Li, Zhengchong Zou and Shaoming Han for them taking part in the field seed collection during laboratory experiments.
Luo C.D., Luo C., Wang F.J., A study of the climate change in China based on given climate data of 10 cities, Journal of Leshan Normal University, 2014, 29(12), 100-105 (in Chinese) Google Scholar
Ding Y.H., Wang H.J., Newly acquired knowledge on the scientific issues related to climate change over the recent 100 years in China, Chin. Sci. Bull., 2016, 61, 1029–1041 (in Chinese) Google Scholar
Lv J.J., The impacts of climate change on the distribution of rare or endangered species in China and adaptation strategies, Chinese Research Academy of Environmental Sciences, 2009 (in Chinese) Google Scholar
Li Z.C., Luo W., Chen Y.F., Hong X.J., Han W.T., Li X.C., The relationships between microhabitat heterogeneity and the spatial distribution of Dacrydium pectinatum in Bawangling, Hainan Island, Acta Ecologica Sinica, 2015, 35(8), 2545-2554 (in Chinese) Google Scholar
Barve N., Barve V., Jiménez-Valverde A., Lira-Noriega A., Maher S.P., Peterson A.T., et al., The crucial role of the accessible area in ecological niche modeling and species distribution modeling, Ecol. Model., 2011, 222, 1810–1819 CrossrefGoogle Scholar
Giménez-Benavides L., Escudero A., Iriondo J.M., Local adaptation enhances seedling recruitment along an altitudinal gradient in a high mountain Mediterranean plant, An. Bot.-Landon, 2007, 99, 723–734 Google Scholar
Wang Q.T., Zhao C.Y., Gao C.C., Xie H.H., Qiao Y., Gao Y.F., et al., Effects of environmental variables on seedling-sapling distribution of Qinghai spruce (Picea crassifolia) along altitudinal gradients, Forest Ecology and Management, 2017, 384, 54–64 CrossrefGoogle Scholar
Stuiver B.M., Wardle D.A., Gundale M.J., Nilsson M.-C., Seedling responses to changes in canopy and soil properties during stand development following clear-cutting, Forest Ecology and Management, 2016, 378(15), 31–43 CrossrefGoogle Scholar
Wei X.Z., Wu H., Meng H.J., Pang C.M., Jiang M.X., Regeneration dynamics of Euptelea pleiospermum along latitudinal and altitudinal gradients: Trade-offs between seedling and sprout, Forest Ecology and Management, 2015, 353(1), 232–239 CrossrefGoogle Scholar
Li N., Bai B., Lu C.H., Recruitment limitation of plant population: from seed production to sapling establishment, Acta Ecologica Sinica, 2011, 31(21), 6624–6632(in Chinese) Google Scholar
Wullschleger S.D., Epstein H.E., Box E.O., Euskirchen E.S., Goswami S., Iversen C.M., et al., Plant functional types in Earth system models: past experiences and future directions for application of dynamic vegetation models in high-latitude ecosystems, Ann. Bot., 2014, 114(1), 1–16 PubMedCrossrefGoogle Scholar
Bi H., Long Y., Turner J., Lei Y., Snowdon P., Li Y., et al., Additive prediction of aboveground biomass for Pinus radiata (D. Don) plantations, For. Ecol. Manag., 2010, 259, 2301–2314 CrossrefGoogle Scholar
Lei X., Zhang H., Bi H., Additive aboveground biomass equations for major species in over-logged forest region in northeast China. In: IEEE proceeding of IEEE 4th international symposium on plant growth modelling, simulation, visualization and applications, 2012 Google Scholar
Bhardwaj V., Singh D.S., Singh A.K., Water quality of the Chhoti Gandak River using principal component analysis, Ganga Plain, India, Journal of Earth System Sciences, 2010, 119(1), 117–127 CrossrefGoogle Scholar
Rencher A.C., Principal Component Analysis, Methods of multivariate analysis, Second Edition., 2003 Google Scholar
Ju J.C., Kolaczyk E.D., Gopal S., Gaussian mixture discriminant analysis and sub-pixel land cover characterization in remote sensing, Remote Sensing of Environment, 2003, 84(4), 550–560 CrossrefGoogle Scholar
Candes E.J., Li X., Ma Y., Wright J., Robust principal component analysis? J. ACM., 2011, 58, 11–37 Google Scholar
Zelikova T.J., Sanders N.J., Dunn R.R., The mixed effects of experimental ant removal on seedling distribution, belowground invertebrates, and soil nutrients, Ecosphere, 2011, 5 (2), 1–14 Google Scholar
Mamet S.D., Kershaw G.P., Multi-scale analysis of environmental conditions and conifer seedling distribution across the treeline ecotone of northern Manitoba, Canada, Ecosystems, 2013, 16(2), 295–309 CrossrefGoogle Scholar
Li N., Bai B., Wang Z., Luo F., Lu X.Z., Lu C.H., Avian seed dispersal and seedling distribution of the endangered tree species, Taxus chinensis, in patchy habitats, Plant Ecology & Diversity, 2015, 8(3), 407-414 CrossrefGoogle Scholar
Albrecht M.A., McCarthy B.C., Seedling establishment shapes the distribution of shade-adapted forest herbs across a topographical moisture gradient, Journal of Ecology, 2009, 97(5), 1037–1049 CrossrefGoogle Scholar
Zheng Z., Distribution of Dacrydium in south China during the late quaternary, Journal of Integrative Plant Biology, 1991, 33(2), 130–139 Google Scholar
Keppel G., Prentis P.J., Biffin E., Hodgskiss P., Tuisese S., Tuiwawa M.V., et al., Diversification history and hybridisation of Dacrydium (Podocarpaceae) in remote Oceania, Australian Journal of Botany, 2011, 59(3), 262–273 CrossrefGoogle Scholar
Su Y.J., Wang T., Chen G.P., Sun Y.F., Jiang Y., Deng F., et al., Postglacial population expansion of Dacrydium pierrei (Podocarpaceae) in Hainan, Southern China, Based on cpDNA trn L-F Nocoding Sequence Data, Acta Scientiarum Naturalium Universitatis Sunyatseni, 2005, 44(5), 70–74 (in Chinese) Google Scholar
Yang Y.C., Zhang W.Y., Lin R.C., Yang X.S., Study on structure and species diversity in post harvested tropical montane Rain forest Dominated by Dacridium pierrii in Bawangling, Hainan Island, Forest Research, 2008, 21(1), 37–43 (in Chinese) Google Scholar
Huang L., Deng Q., Li N., Su Y.G., Wang T., A set of microsatellite markers developed for Dacrydium pectinatum (Podocarpaceae), a vulnerable conifer in China, Conservation Genetics Resources, 2014, 6(1), 167–168 CrossrefGoogle Scholar
Lian J.Y., Yu S.X., Bawangling tropical mountain rainforest Dacrydium pierrei in Hainan Island—The floristic characteristics of branch line formation of Syzygium, Journal of tropical and subtropical plants, 2001, 9(2), 101–107 Google Scholar
O’brien J.J., Oberbauer S.F., Clark D.B., Whole tree xylem sap flow responses to multiple environmental variables in a wet tropical forest, Plant, Cell & Environment, 2004, 27(5), 551–567 CrossrefGoogle Scholar
Pienitz R., Smol J.P., Diatom assemblages and their relationship to environmental variables in lakes from the boreal forest-tundra ecotone near Yellowknife, Northwest Territories, Canada, In: Herman van Dam (Ed.), Twelfth International Diatom Symposium, 1993, 90, 391–404 Google Scholar
Cienciala E., Tumajer J., Zatloukal V., Beranová J., Holá Š., Hůnová I., et al., Recent spruce decline with biotic pathogen infestation as a result of interacting climate, deposition and soil variables, European Journal of Forest Research, 2017, 136(2), 307-317 CrossrefGoogle Scholar
Ansolin R.D., da Silva A.C., Higuchi P., Küster L.C., Ferreira T.S., Júnior F.B., et al., Environmental heterogeneity and floristic-structural variation in a araucaria forest fragment in Coxilha Rica, Santa Catarina State, Ciência Florestal, 2016, 26(4), 1201–1210 Google Scholar
Bakker K., Naugle D., Higgins K., Incorporating landscape attributes into models for migratory grassland bird conservation, Journal Article, 2002, 16(6), 1638–1646 Google Scholar
Qi X.M., Zeng X.Y., Ecological evaluation of the Hainan Bawangling National Nature Reserve, Central South Forest Inventory and Planning, 2011, 30(2), 61–64 (in Chinese) Google Scholar
Omacini M., Eggers T., Bonkowski M., Gange A.C., Jones T.H., Leaf endophytes affect mycorrhizal status and growth of coinfected and neighbouring plants, Functional Ecology, 2006, 20(2), 226–232 CrossrefGoogle Scholar
Pérez de Diego Y., Pellikaan H.C., Wubbolts F.E., Borchard G., Witkamp G.J., Jansens P.J., Opening new operating windows for polymer and protein micronisation using the PCA process, The Journal of Supercritical Fluids, 2006, 36(3), 216–224 CrossrefGoogle Scholar
Kerroum M.A., Hammoucha A., Aboutajdine D., Textural feature selection by joint mutual information based on Gaussian mixture model for multispectral image classification, Pattern Recognition Letters, 2010, 31(10), 1168–1174 CrossrefGoogle Scholar
Criminisi A., Shotton J., Bucciarelli S., Decision forests with long-range spatial context for organ localization in CT volumes, Miccai Workshop on Probabilistic Models for Medical Image Analysis, 2009, 9, 55–60 Google Scholar
Bače R., Svoboda M., Pouska V., Janda P., Červenka J., Natural regeneration in Central-European subalpine spruce forests: Which logs are suitable for seedling recruitment? Forest Ecology and Management, 2012, 266, 254–262 CrossrefGoogle Scholar
Chen D.M., Pan C.D., Liu C.L., Chang J., Yin K., Analysis on Micro-habitat Variables Affecting Natural Regeneration and Survival of Seedlings in Picea schrenkiana Stand, Journal of Xinjiang Agricultural University, 2005, 28(3), 35–39 (in Chinese) Google Scholar
Mitsopoulos I., Xanthopoulos G., Effect of stand, topographic, and climatic factors on the fuel complex characteristics of Aleppo (Pinus halepensis Mill.) and Calabrian (Pinus brutia Ten.) pine forests of Greece, Forest Ecology and Management, 2016, 360(15), 110–121 CrossrefGoogle Scholar
Kovač M., Kutnar L., Hladnik D., Assessing biodiversity and conservation status of the Natural 2000 forest habitat types: Tools for designated forestlands stewardship, Forest Ecology and Management, 2016, 359(1), 256–267 CrossrefGoogle Scholar
Ma Y., Zhang J.H., Li X.R., Zhang S.Y., Lan H.Y., Effects of environmental stress on seed germination and seedling growth of Salsola ferganica (Chenopodiaceae), Acta Ecologica Sinica, 2016, 36(6),456–463 CrossrefGoogle Scholar
Beckage B., Clark J.S., Clinton B.D., Haines B.L., A long-term study of tree seedling recruitment in southern Appalachian forests: the effects of canopy gaps and shrub understories, Canadian Journal of Forest Research, 2000, 30(10), 1617–1631 CrossrefGoogle Scholar
Janišová M., Škodová I., Hegedüšová K., Kochjarová J., Seed bank and seedling recruitment of endangered Tephroseris longifolia subsp. moravica (Asteraceae), Folia Geobotanica, 2017, 6, 1-18 Google Scholar
Kisinyo P.O., Effect of lime and phosphorus fertilizer on soil chemistry and maize seedlings performance on Kenyan acid soils. Sky Journal of Agricultural Research, 2016, 5(6), 97–104 Google Scholar
Zhang Y., Xue L.G., Gao T.P., Jin L., An L.Z., Research advance on seed germination of desert plants, J. Desert Res., 2005, 25(1), 106–112 ( in Chinese) Google Scholar
Qu X.X., Huang Z.Y., The adaptive strategies of halophyte seed germination, Acta Ecol. Sin., 2005, 25(9), 2389–2398 Google Scholar
Carrer M., Soraruf L., Lingua E., Convergent space–time tree regeneration patterns along an elevation gradient at high altitude in the Alps, Forest Ecology and Management, 2013, 15,1–9 Google Scholar
Primicia I., Camarero J.J., Janda P., Čada V., Morrissey R.C., Trotsiuk V., et al., Age, competition, disturbance and elevation effects on tree and stand growth response of primary Picea abies forest to climate, Forest Ecology and Management, 2015, 354(15), 77–86 CrossrefGoogle Scholar
Chen F.Q., Geng L.S., Zhao Y., Study on the effect of rainfall on soil moisture. Shandong Water Resources, 2015, 5, 14–17 (in chinese) Google Scholar
Gong, J.J., Yang X.F., Xing G.C., Lin H.S., Study on seed dormancy and germination characteristics of Portulacapilosa, Asian Agricultural Research, 2016, 8(11), 71-74 Google Scholar
Tingstad L., Olsen S.L., Klanderud K., VandvikV., Ohlson M., Temperature, precipitation and biotic interactions as determinants of tree seedling recruitment across the tree line ecotone, Oecologia, 2015, 179(2), 599–608 CrossrefPubMedGoogle Scholar
Benigno S.., Dixon K.W., Stevens J.C., Increasing soil water retention with native-sourced mulch improves seedling establishment in Postmine Mediterranean sandy soils, Restoration Ecology, 2013, 21(5), 617–626 CrossrefGoogle Scholar
About the article
Published Online: 2017-11-10
Conflict of interest: Authors state no conflict of interest.
Citation Information: Open Life Sciences, Volume 12, Issue 1, Pages 345–355, ISSN (Online) 2391-5412, DOI: https://doi.org/10.1515/biol-2017-0040.
© 2017 Chunyan Wu et al.. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. BY-NC-ND 4.0