The rain-runoff response of tropical humid forest ecosystems to use and reforestation in the Western Ghats of India

The effects of forest degradation and use and establishment of tree-plantations on degraded or modified forest ecosystems at multi-decadal time-scales using tree-plantations on the streamflow response are less studied in the humid tropics when compared to deforestation and forest conversion to agriculture. In the Western Ghats of India (Uttar Kannada, Karnataka State), a previous soil hydraulic conductivity survey linked with rain IDF (intensity-duration-frequency) had suggested a greater occurrence of infiltration-excess overland within the degraded forest and reforested areas and thus potentially higher streamflow (Bonell et al., 2010). We further tested these predictions in Uttar Kannada by establishing experimental basins ranging from 7 to 23 ha across three ecosystems, (1) remnant tropical evergreen Forest (NF), (2) heavily-used former evergreen forest which now has been converted to tree savanna, known as degraded forest (DF) and (3) exotic Acacia plantations (AC, Acacia auriculiformis) on degraded former forest land. In total, 11 basins were instrumented (3 NF, 4 AC and 4 DF) in two geomorphological zones, i.e., Coastal and Up-Ghat (Malnaad) and at three sites (one Coastal, two Up-Ghat). The rainfall-streamflow observations collected (at daily and also at a 36 min time resolutions in the Coastal basins) over a 2-3 year period (2003-2005) were analysed. In both the Coastal and Up-Ghat basins, the double mass curves showed during the rainy season a consistent trend in favour of more proportion of streamflow in the rank order DF > AC > NF. These double mass curves provide strong evidence that overland flow is progressively becomes a more dominant stormflow pathway. Across all sites. NF converted 28.4 +/- 6.41(stdev%) a of rainfall into total streamflow in comparison to 32.7 +/- 6.97(stdev%) in AC and 45.3 +/- 9.61(stdev%) in DF. Further support for the above trends emerges from the quickflow ratio Q(F)/Q for the Coastal basins. There are much higher values for both the DF and AC land covers, and their rank order DF > AC > NF. The quickflow response ratio Q(F)/P is also the highest for the DF basin, and along with the Q(F)/Q ratio, can exceed 90%. The corresponding delayed flow response ratios. Q(D)/P clearly show the largest Q(D) yields as a proportion of event precipitation from the Forest (NF1). The application of linear model supported these differences (e.g. 10-36% difference between NF and DF, p < 0.001) in the storm hydrologic response of the Coastal basins. The exception was Q(F)/P where there was a higher uncertainty connected with inter-basin mean differences. Cross-correlation plots for rain-streamflow and corresponding lag regression models for three storm events in the Coastal basins suggested the existence of alternative stormflow pathways with multiple lags with peaks between similar to 12 and 24 h in NF, compared to respective bimodal peaks at similar to 1 and 16 h in AC and similar to 1 and 12 h in DF. The long time lags for NF are suggestive of deep subsurface stormflow and groundwater as the contributing sources to the storm hydrograph. The short time lags in DF and AC are indicative of overland flow and so 'memory' of the previous degraded land cover is retained in AC as supported by previous hydraulic conductivity data. As potential and actual evapotranspiration is likely to be depressed during the monsoon, differences in streamflow and run-off responses between land-cover types is largely attributed to differences in soil infiltration and hydrologic pathways. Enhancing infiltration and reducing run-off in managed ecosystems should be explored in the terms of the context of other ecosystem services and biodiversity. (C) 2012 Elsevier B.V. All rights reserved.