Our study was a forest ecosystem in Jiufeng National Forest Park (116°28′E, 39°34′N) located in the northwest of Beijing, China (Fig. 1). Jiufeng National Forest Park is part of Beijing Forestry University and is used for teaching and scientific research. Elevation ranges from 60-1100 masl (above seal level). The climate is temperate continental with a mean annual precipitation of 630 mm, mean annual temperature 11.6 °C, and mean annual potential evapo-transpiration 19,000 mm. The dominant vegetation at elevations <800 masl was plantation of Platycladas orientalis, Pinus tabulaeformis, Quercus spp., Robinia pserdoacacia containing shrubs of Prunus armniaca and Vitex chinensis. Above 800 masl, Pinus tabulaeformis, Popular chinensis, Lespedeza bicolon, Spiraca trilobata, Caragana rosea dominated the sparse forest cover. The soil chemical and physical properties are described as sandy loam containing approximately 20% rock fragments and gravels. Its texture consists of sandy loam in forest soils to 0.60 m in depth. Its organic carbon (OC) content varies from 2.28 to 46.17 g kg^–1 in study site. The soil pH values range from pH 5.87 to 7.12 in the topsoil and subsoil (Table 1).

Six experimental plots were established within a 10 x 10 m quadrat situated in representative vegetation region at 260 masl. Experimental plots 1 and 2 were located in Sophora japonica Linn, experimental plots 3 and 4 in Platycladus orientalis Franco, and experimental plots 5 and 6 in Quercus dentata Thunb sections of the quadrat. Two replicates for each vegetation type were conducted in the quadrat. Fifty millimeters of enriched Brilliant Blue FCF dye solution (5g L^–1) was applied to the relevant experimental plots. The solution was uniformly applied to a 1.2 x 1.2 m area centered on the experimental trees to avoid border effects (Hagedorn & Bundt, 2002). Horizontal and vertical soil profiles were excavated when the solution had infiltrated the soil (Hu et al., 2013). Horizontal profiles were excavated from 0.5 x 0.5 m quadrats and vertical profiles with maximum dying depth were extracted respectively from points centered on the experimental trees one day after dye tracer application (Hagedorn & Bundt, 2002). For the horizontal and vertical sections, soil cores were replicated and extracted from preferential pathways and soil matrix to observe plant roots content at the soil profile scales. During field experiments, camera should be taken to record preferential pathways (stained areas) and soil matrix (unstained areas) (Hagedorn & Bundt, 2002) (Fig. 2).

Each experimental plot was excavated from horizontal sections in 10 cm depth increments 24 h after application of Brilliant Blue FCF dye solution. Undis­turbed soil samples were taken at each depth using soil corers (7 cm diameter, 5 cm height, 200 cm³ volume) with two field replications in preferential pathways and soil matrix. Samples were taken to a depth of 60 cm (0-5, 5-10, 10-15, 15-20, 20-25, 25-30, 30-35, 35-40, 40-45, 45-50, 50-55, 55-60 cm) in all experimental plots. Soil cores were stored at –2 °C (Castellanos et al., 2001) and soil was separated from plant roots using 5 mm sieves. When necessary, samples were placed in dishes with 4-5 mm deep water so that plant roots spread and soil particles could easily be removed. Soil-free roots were dried for 48 hours in an oven at 70 °C to constant weights (Helmisaari et al., 2007) and then weighed using an electronic balance (DV215CD). Fine root length density was measured using WinRHIZO (STD4800) (Yan et al., 2011). And fine root biomass was calculated by root dry weight on the basis of the cross-sectional area of soil cores.

Root content contribution to preferential flow was characterized by evaluating comparison of root content (root length density and root biomass) between soil preferential flow and soil matrix. Relevant equation was based on root content. Simplified equation is described by or , where η is root content contribution to preferential flow (%), α_PP is root length density (mm 100 cm^–3) or root biomass (g 100 cm^–3) in preferential pathways, α_SM is root length density (mm 100 cm^–3) or root biomass (g 100 cm^–3) in soil matrix. In general, will be appropriate when α_PP is higher than α_SM, while when α_PP is smaller than α_SM, will be appropriate.

Dye coverage: proportion of the dye-stained areas to sum of the dye-stained and non-stained areas (Kasteel et al., 2013) (equation (1)). In preferential pathways, water and solute flows through this way, bypassing the soil matrix.

One-way ANOVA was used to access differences in mean root length density and root biomass between preferential pathways and soil matrix and to characterize effects of root parameters on soil preferential flow. Comparison for the evaluation of plant roots content between the two regions was analyzed by SPSS software.

In this study, fine root length density ≤ 5 mm in diameter decreased with increasing distance from soil surface both in preferential pathways and soil matrix to some extent with respect to all experimental plots (Fig. 5). Fine root length density was greatest on the soil surface, with appropriately half root length density in the top 20 cm of the soil for experimental plot 1, and 15 cm for plot 2, plot 3, plot 4 and plot 6, and 10 cm for plot 5; 90% in the top 40 cm of the soil for experimental plot 1 and plot 3, and 25 cm for plot 5 and plot 6, and 30 cm for plot 2 and plot 4 whether in preferential pathways or soil matrix. Fine root length density (mm (100cm³)^–1) was also different from species (Table 2): Platycladus orientalis Franco > Quercus dentata Thunb > Sophora japonica Linn. The greatest plant roots concentration was found in the upper soil layer within a depth of 25-40 cm (topsoil). On the basis of all data from experimental plots, fine root length density ≤ 5 mm in diameter was 75.0-87.5% higher in preferential pathways than in soil matrix.

Meanwhile, we found that the contribution of fine root length density to soil preferential flow was significant also (Fig. 6). Among those data, reference value of root length density contribution to preferential flow (19.2%) was fully considered for all experimental plots. From those data which exceeded 19.2%, we supposed that root length density was higher in soil preferential pathways than in soil matrix and that soil preferential flow was positively correlated with root length density. Furthermore, such accounted for 63.3% of all experimental data. While with respect to those data which were below 19.2%, we stated that the impacts of root length density in preferential flow was ambivalent (positive/negative), because some experimental data illustrated that root length density was higher in preferential pathways than in soil matrix, while the others were on the contrary. Besides, the percentage ratio (root length density was higher in preferential pathways than in soil matrix or not) was 50% respectively among those experimental data which was below 19.2%.

In general, fine root biomass ≤ 5 mm in diameter also decreased with increase of soil depth in preferential pathways and soil matrix (Fig. 7). Fine root biomass was greatest on the soil surface, with appropriately 70-80% root biomass in the top 20 cm of the soil at the soil profile scale. Based on those data from experimental plots, fine root biomass ≤ 5 mm in diameter was, 66.7% for plot 1, 2, 4 and 5, 100% for plot 3, and 33.3% for plot 6, higher in preferential pathways than in soil matrix. Average fine root biomass was also different from species: Platycladus orientalis Franco (1.474 g (100cm³)^–1) > Quercus dentata Thunb (1.332 g (100cm³)^–1) > Sophora japonica Linn (0.837 g (100cm³)^–1). With respect to different soil depth, fine root biomass was 50% higher in preferential pathways than in soil matrix for soil depth 0-10 cm, and 83.3% for 10-20 cm and 66.7% for 20-40 cm respectively. Among all surveyed data, root biomass was 66.7% higher in preferential pathways than in soil matrix on the whole. Fine root biomass as a proportion of total tree biomass was 9% for plot 1, 14% for plot 2, 10% for plot 3, 30.5% for plot 4, 16.5% for plot 5 and 20.1% for plot 6 respectively.

In our study, we also realized that the contribution of fine root biomass to preferential flow was pivotal (Fig. 8). Like root length density, reference value of root biomass contribution to preferential flow (50%) was fully taken into account. There was no doubt that fine root biomass was higher in preferential pathways than in soil matrix as those data exceeded 50%. Therefore, we hypothesized that soil preferential flow was also positively correlated with fine root biomass as those data were considered. Meanwhile, such accounted for 55.6% of all surveyed data. As those data were below 50%, the influences of fine root biomass on preferential flow was ambivalent. And the percentage ratio that fine root biomass was higher in preferential pathways than in soil matrix was 20%, while the percentage ratio that fine root biomass was smaller in preferential pathways than in soil matrix was 80% among those data which was below 50%.

Our field dye tracing experiments conducted in forest ecosystems containing abundant rock fragments illustrated that soil preferential flow was more obvious and that finger flow was predominant in Jiufeng area. The results were in agreement with Noguchi et al. (1997) and Hagedorn & Bundt (2002). Dye coverage decreased with the increasing soil depth from soil surface which supported more studies (Öhrström et al., 2002; Kramers et al., 2009), while some studies also indicated that the amount of stained area per depth (dye coverage) decreased to a depth and afterwards increased (Bogner et al., 2013). Ecologists attribute the results to macropores density (e.g., root channels, rock soil interface and cracks). And sandy soil may also lead to the results. In the study site, forest soils contain abundant rock fragments and soil texture is not homogeneously distributed which makes higher soil spatial heterogeneity at the soil profile scale. Therefore, preferential pathways continuity is better because more plant fine roots on the soil surface. Particularly, the sandy loam will enhance plant roots into the macropores, meanwhile plant roots will also form corresponding channels to make soil water and solute transport through them. Lots of studies confirmed that degree of soil preferential flow in the upper soil layers was higher than that in the subsoil which led to smaller dye coverage, because preferential pathways were densely distributed in the topsoil. Maximum depth of dye infiltration was not similar greatly though experimental plots at the same site. At the soil profile scale, soil heterogeneity depended on preferential pathways continuity. In general, dye tracers could infiltrate into the deepest soil depth even groundwater levels which was not calculated by Richards equation, Mobile-immobile model and other mathematical models.

Our results indicate that fine root length density decrease with increasing distance from soil surface which supports early and recent studies (Mitchell et al., 1995; Tracy et al., 2013). Those results collected from fine root length density imply that plant roots content are higher in preferential pathways than in soil matrix, supporting studies like Bundt et al (2000, 2001), and the results are also in agreement with Bonger et al (2010) and Bengough (2012), because plant fine roots usually form root channels called preferential pathways (Li & Ghodrati, 1994; Jørgensen et al., 2002) to prompt water infiltration after their decay (Beven & Germann, 1982). Soil preferential pathways in forest soils include higher organic carbon content and microbial biomass compared with soil matrix (Backnäs et al., 2012). Plant fine roots themselves tend to release more organic matters to the soils as the rhizosphere decays. During plant roots decomposition, root channels are filled with organic matters derived from roots themselves (Ghestem et al., 2011). Plant fine roots constitute a dynamic component of forest ecosystems (Dastidar et al., 2012). By means of accumulating soil organic matters and redistributing nutrients at the soil profile scale, fine roots play significant role in their surrounding environment (Persson, 2000). Meanwhile, plant fine roots longevity is variable ranging from a few months to years because the process of plant roots turnover making new roots to replace decaying or dead roots is more related with soil nutrient availability (Yavitt et al., 2011). Changes of fine root length density is a complex network with millions of lateral branches associated with mychorrizal hyphae (Majdi et al., 2005).

Our field experiments were carried out during rainy season, particularly during heavy rain, soil water flow in preferential pathways and cracks increases, and fine roots may become asphyxiated even die. Clusters of fine roots are sometimes observed along or at the end of coarse roots and correspond to zones of major organic nutrients and water uptake. Fine roots have high decay and emission rates, and clusters may soak up water during rainy season and may contribute to decayed flow paths (Ghestem et al., 2011).

Furthermore, the highest fine root length density was found in the upper soil layer within a depth of 25-40 cm (topsoil). The result was in agreement with Lipiec et al. (2003), Bonger et al. (2008), Glab (2013). Fine roots on the soil surface are pivotal in forest nutrient cycling (Bengough, 2012). Despite forest nutrient cycling, fine roots may account for 50% of the Net Primary Production in forest ecosystems, and some studies proposed it up to 75% (Vogt et al., 1996) and 33% (Gill & Jackson, 2000), which was beneficial to examine forest dynamics. To better understand forest dynamics with climate changes, studies on fine roots dynamics tend to be necessary. However, characterizing fine root dynamics are tedious and time-consuming. Fine root length density as an index of root dynamics also changes with climate. Therefore, fine root length density dynamics could lead to carbon and nutrient fluxes during forest dynamics process although fine roots contribute little to total forests biomass (<5%).

And fine roots located in the soil surface layers could also improve soil physical and chemical properties (e.g., soil bulk density, porosity, organic carbon, citation exchange capacity and so on). For example, the topsoil tend to have higher recharge rate that brings the soil moisture close to saturation compared with the subsoil (Vogt et al., 1996; Schmid & Kazda, 2002). In the topsoil, fine roots which are short-lived and non-woody (Vogt et al., 1996) in forest ecosystems may have more preferential root channels. The conclusion is in agreement with Noguchi et al. (1997) who stated that decaying fine roots of alfalfa could create more stable preferential pathways than wheat. Meanwhile, the results indicate that 100% of all fine root length density are within a depth of 55 cm in study site which is agreement with Schenk & Jackson (2002).

In this paper, fine root biomass was densely concentrated in the upper soil layers. The result was in agreement with Ford & Deans (1977) and Raizada et al. (2013). Change of fine root biomass is more pronounced in the upper soil layers than in deeper soil layers because of the higher organic matters and nutrition in the topsoil. Some studies imply that fine root biomass increases in nutrient rich zones, while some species also have a high fine root biomass in nutrient poor zones. The average of fine root biomass of the forest ecosystems in this study (1.214 g (100cm³)^–1) was substantially lower than estimates from Safford (1974), but not all temperate forest ecosystems. It is likely that gravels constitute the large portion of soil profiles in Jiufeng area: 10% in the top soil of 15 cm, 15% to a soil depth of 30 cm and 20% at soil depths below 30 cm. However, gravels constituted only 9% of the soil profiles studied by McClaugherty et al. (1984) and Safford (1974) did not state gravels.

Our results indicated that fine root biomass accounting for the total tree biomass in forest ecosystems varied between 9% and 30.5%. Fine roots’ share of total biomass rarely represents more than 5% of total biomass of trees, while Santantonio et al. (1977) and Fogel (1983) reported that root biomass (coarse and fine roots) as a proportion of total tree biomass varied between 18 and 45%. Brassard et al. (2011) stated that coarse root biomass (diameter > 1 cm) could account for approximately 30% of total biomass in forest ecosystems. Maybe fine root biomass has been found to be variable in relation to forest stand characteristics (e.g., species, stand age, density, basal area and soil properties) or environmental factors (e.g., temperature, precipitation, geographical location and elevation) (Finer et al., 2007). Jackson et al. (1997) estimated fine root biomass and reported that live fine root biomass ranged from 130 g m^–2 in deserts to 950 g m^–2 in temperate grasslands.

Past results have shown that fine root biomass increases with stand age, for example, Grier et al. (1981) stated that there was an increase in fine root biomass with years, and Persson (1983) also. Those results are in agreement with us to some extent. With regard to Platycladus orientalis Franco located in plot 3 and 4 and Quercus dentata Thunb located in plot 5 and 6, we found that fine root biomass increased with years, while there was no significant differences in fine root biomass for Sophora japonica Linn, and such supported Vogt et al. (1985), Ruark & Bockheim (1987) and Finer et al. (1997).

Change in fine root biomass may alter organic nutrient availability in forest soils. In previous studies, changes in fine root biomass were found to be the results of internal factors (e.g., plant species) and external factors (e.g., soil properties, stand age and climate). As was mentioned, fine root biomass only accounted for little of the total forest biomass, but such may exerts significant influences on carbon and nitrogen cycling in forest dynamics. Meanwhile, forest dynamics may affects on root biomass dynamics correspondingly.

In summary, the highest plant root concentration was found in the upper soil layer (Bonger et al. 2008; Himmelbauer et al., 2010). More plant root channels would increase macropores network density and continuity (Shi et al., 2012). Not only plant roots grow into macropores but also create macropores (biopores, cracks and burrows) (Noguchi et al., 1997). More decaying or decayed plant roots as well as fine plant roots are distributed on the soil surface. Compared with living plant roots, decayed roots were more effective to create preferential pathways (Mitchell et al., 1995). Furthermore, plant roots could release complex organic compounds (e.g., amino acids and organic acids) into the soils to prompt plant growth (Bengough, 2012).