Effect Of Tree Crown And Leaf Litter Structure On Related Hydrological And Anti-erosion Functions Of Small Trees On A Process Basis In Beijing Mountainous Area, China

Previous researches have shown that vegetation cover had an impact on rainfall input, output and hillslope-soil erosion. However, most of those investigations concentrated on calculating or predicting the quantity of crown and litter interception, runoff and sediment yield, less attention has been paied on how to describe and modelling those phenomena on a process basis, specifically for the dynamic response between vertical structure of trees and the related hydrologic and erosion-control functions during rainfall. In order to address these issues, a series of process-based experiments were conducted using simulated rainfall, including rainfall interception by individual tree crown and leaf litter separately and jointly for for broadleaf (Platycladus orientalis, Pinus tabulaeformis) and needle tree species(Quercus variabilis, Acer truncatum) at five rainfall intensities (from 10 to 150 mm h-1), and runoff and erosion process in runoff plots (with an area of 4.5×2 m2å'Œ 4.5 ×1 m2) on hillslopes at five simulated rainfall intensities (from 5.7 to 75.6 mm h-1). Results indicated that:(1) Both crown and litter interception were dynamic processes which incorporated three phases:a repid-dampening phase, a stable-saturation phase, and a post-rainfall drainage phase. The average crown maximum interception storage (Cmax) and minimum (Cmm) interception storage were 0.66 and 0.40 mm, respectively; while the average litter Cmax and Cmm were 2.25 and 1.45 mm, respectively, revealing that almost 40% and 35% of intercepted water drained off from crown and litter after rainfall. Rainfall intensity only significantly affected litter Cmax (p< 0.05), such correlations were not observed in crown Cmax and Cmm, and neither litter Cmm. Crown structure such as leaf characteristics (i.e. leaf area, leaf area index, leaf biomass, minimum leaf density, distance between two neiboring leaves) and branch characteristics (branch area, branch biomass, branch count, minimum branch density, branch length) was significantly correlated with Cmax and Cmm (p< 0.05). Comparisons between leaf morphorology in Cmax and Cmm showed that needle crown Cmax were 1.9 times larger than the broadleaf crowns, while broadleaf litter Cmax and Cmm were 1.5 and 1.6 times larger than needle litter. Further, litter played a larger role compared with crown in rainfall interception, since litter Cmax and Cmm were 3.4 and 3.6 times larger than crown when the two processes were measured saperately, while the figure increased to 6.4 and 5.8 times in joint measurement. Moreover, the percentage of average joint Cmax and Cmm in gross precipitation were 18.6% and 9.5%, respectively.(2) Four models were proposed to depict crown and litter interception process, which included the cumulative crown interception during rainfall (CIDR) based on LAI and cumulative precipitation Pc: and the crown post-rainfall drainage (PRD) model: similarly, litter CIDR and PRD model parameters were litter mass M and Pc, litter CIDR: nd litter PRD: These models were able to depict and predict the entire interception processes well with MRE ranging from 8% to 27%.(3) Runoff can be reduced under crown and litter cover as a result of more infiltration. Compared with the bare soil plot (BSP), time to ponding and runoff were delayed in 5 to 10 minutes in crown and litter-covered plots (CLP), runoff rates were 73% and 85% of that in BSP, and infiltration rates were 1.5 and 2.3 times larger than in BSP, as a consequence, total runoff yield in CLP were 82% and 63% of BSP, and total infiltration yield were 1.5 and 1.6 times larger than BSP. Rainfal intensity (RI) significantly impacted runoff process in BSP and CLP, when RI increased from 5.7 to 75.6 mm h-1, total runoff yield increased 33 to 83 times. P. tabulaeformis with tree spacing 1.0 m× 1.0 m generated the least runoff yield due to its largest LAI and coverage, in contrast, litter runoff yield (Qg) may be controlled by RI and litter mass M:Qg=-7.24M+0.77RI (R= 0.97).(4) When the CIDR model was extrapolated to hillslope scale, results showed that the interception, runoff, and infiltration were all dynamic process with interactions during rainfall. Infiltration process was dominant in light rainfalls of 5.7 and 11.7 mm, while runoff process was dominant in heavy rainfalls of 49.8 and 75.6 mm, not surprisingly, a transition process from infiltration-dominant (30-45 minutes) to runoff-dominant (15-30 minutes) was observed in mediate rainfall of 25.2 mm. Gross water balance revealed that total infiltration was 47.0% of gross precipitation, and total runoff was 49.6%, total interception was only 3.4% in crown-covered plots regardless of rainfall intensities and tree species; In contrast, the total infiltration, runoff, and interception were 45.3%,44.1%, and 10.6% respectively in litter covered plots.(5) Infiltration and runoff process models were proposed and recalibrated. In crown covered plots, runoff dynaimic process (RDP) model was developed based on parameters such as LAI and cumulative precipitation Pc:Q1= (-0.09 LAI+0.55) Pc(0.04LAI+1.13), and light-rainfall infiltration process model (LIP):It= (-0.11 LAI+1.04) Pc(0.11LAI+0.74), also with heavy-rainfall infiltration process model (HIP): It= (0.44 LAI+1.25) PcLAI(0.17LAI-0.02). Similar models were developed in litter-covered plots based on litter mass M, litter species and Pc, which were RDP model:Qt= (-0.39 M+0.61) Pc(0.15M+1.08)(for Q.variabilis), Qt= (-0.43 M+0.64) Pc(0.17M+1.05)(for P. tabulaeformis), and infiltration dynamic process (IDP) model:= (1.44 M2-1.60 M+1.08) PcM(-1.05M+1.34)+0.40(for Q. variabilis), It= (-0.96 M2+1.94 M) PcM(-2.37M+2.72) (for P. tabulaeformis). These models can precisely simulate the the runoff and infiltration process with MRE raning from 13% to 27%.(6) Erosion process under crown and litter cover consisted of two phases:a rapid increase of sediment rate and concentration and a stable fluctuation afterwards. Outcomes implied that erosion can be significantly reduced in CLP, sediment rates in CLP were 63.3% and 16.4% of BSP, and total sediment yields were 58.8% and 16.8% of BSP. In addition, RI had an apparent impact on sediment, when RI increased from 5.7 to 75.6 mm h-1, total sediment yield incrased 30-77 times in all BSP and CLPs. Meanwhile, crown cover affected sediment loss rate (SLR) by the equation:SLR= e-0.02e. Further, total sediment was controlled by crown cover C and RI:S=-0.15 C+0.31 RI. Similarly, litter mass Mand RI showed combining effect on S:S=-0.46 M+0.02 RI. Comprehensively, P. tabulaeformis plot with tree spacing of and Q. variabilis litter-covered plot showed the greatest erosion-control function compared with other CLPs. Process-based sediment yield models were developed and calibrated in crown-covered plots:S,= (0.04 LAI2-0.21 LAI+0.60)Q1LAI(-0.16LAI+0.81), and litter-covered plots:S,= (-0.18M+0.18) Ql(-0.34M+0.90)for bare soil and Q. variabilis, and Sl= (-0.02 M+0.05) Ql(b=-0.70M+1.32)for P. tabulaeformis. These models well simulated the sediment process with MRE ranging from 28%-35%. Surface runoff and erosion processes on slope were simulated by the WEPP model (slope profile), on the basis of cumulative runoff volume simulation, the model showed a better prediction on total runoff with CE> 0.85 than on total sediment yield with CE over 0.50In summary, the conclusion above clarified the hydrologic and erosion-control functions of crown and litter on a process basis, the models based on crown and litter structure parameters reflected the ’structure-function’relationship, and provided a solid foundation not only on how to change the structure for specific funtions, but also on specis selection, spatial configuration, and structure optimization when implementing afforestation in Beijing mountainous area.