Report No: 808/1/01

August 2001



Several recent developments in South Africa have highlighted the need for better information on the water use of trees in riparian zones. The declaration of plantation forestry as a streamflow reducing activity (SFRA) has served to maintain interest in the issue of forest water use, and the extent to which their impact on catchment water yields may be minimized. One long-recognized option is to remove plantations a set distance away from streams, and promote indigenous, preferably herbaceous vegetation in its place. The trend towards forest certification has further encouraged growers to adopt this practice as a means of minimizing the hydrological impacts of their forests on catchment yields, and promoting the sustainability of their forestry enterprise.

It is, however, the problem of self-established alien invasive trees that has received the most publicity in recent years. Large scale funding by both the Government and the private sector continues in support of the Working For Water programme, which is active in many regions of the country. One justification for this programme of alien tree removal is the streamflow enhancement that is believed to follow the replacement of dense stands of invasive trees by indigenous, largely herbaceous or shrub-dominated plant communities. Often the densest stands of invader trees occur within riparian zones, where removal of trees in close proximity to stream channels is believed to strongly enhance streamflow.

Few data are available to quantify these assumptions. Portable weir experiments (Dye and Poulter, 1995; Prinsloo and Scott, 1999) document streamflow enhancement immediately following the clearance of trees, but do not take into account water use by the indigenous plant community that with time subsequently develops on the site. Scott and Lesch (1995) report changes in streamflow following clearance of riparian trees in four research catchments. Their paper illustrates how variable streamflow response may be to this treatment, and how important the characteristics of the post-felling vegetation are in determining this response.

Results from a number of research catchments have consistently shown that catchment-wide afforestation significantly decreases streamflow where the pre-afforestation vegetation was seasonally dormant mountain grassland or fynbos (Versfeld, 1994). The net difference in evapotranspiration (E7) between riparian thickets or plantations of alien trees and indigenous riparian vegetation may be quite different, however, due to the yearlong availability of soil water and enhanced plant growth in riparian zones, and the occurrence of many different types of indigenous riparian vegetation. The water use characteristics of thickets of alien self-established trees in South Africa remains largely unknown, adding further uncertainty to the effect of alien tree removal on streamflow.


The objectives of this study were:


A riparian fynbos site was chosen in the upper reaches of the Jonkershoek valley, close to the Eerste River that flows past Stellenbosch. A 12-month record of 20-minute evaporation rates from this site was recorded using the Bowen ratio energy balance (BREB) technique. A closed-canopy, mature stand of self established wattle in the Wellington area of the Western Cape was selected to provide comparative water use data. The heat pulse velocity technique was used to record hourly sap flow rates in six sample trees representing the range of tree sizes in the thicket. Total daily sap flow in all sample trees was closely correlated to a daily air humidity index defined as the product of mean daily vapour pressure deficit (M) of the air and the number of daylight hours. A wildfire destroyed the sample trees after seven months of data collection. However, the daily sap flow/VPD index relation was found to be constant over the entire data record, which spanned both wet conditions in late winter, and very hot and dry conditions in late summer, indicating an absence of stress due to soil water deficits. It was assumed therefore that the same relationship would hold throughout an entire year, and could be used to predict wattle transpiration at the Jonkershoek site using the VPD data recorded there by the BREB system.

A replacement A. mearnsii riparian site was chosen on the slopes of the Groot Drakenstein mountains close to Pniel in the Stellenbosch district. Hourly sap flow was monitored in six sample trees over a period of seven months, until the experiment was terminated by another wild fire. Daily sap flow over the entire period of measurement showed a distinct and progressive reduction over the second half of the dry summer, which is shown to be caused by stress due to soil water deficits. Sap flow rates recorded over the period prior to the commencement of transpiration reduction were found to be very similar to those recorded at the Wellington site. Both data sets were used to develop a simple model of daily sap flow (under conditions of no water stress) for this species on the basis of tree DBH and the product of mean daily VPD and the number of daylight hours.

This model of daily sap flow by A. mearnsii was used to predict the annual transpiration of a thicket of this species at the Jonkershoek site. Rainfall interception by the canopies was predicted from the number of rain days recorded at Jonkershoek, and daily rainfall interception estimates for this species provided by Schulze et al. (1995)

In the summer rainfall region, the annual ET of riparian grassland at Gilboa forest plantation in the Karkloof hills of the KwaZulu-Natal midlands was estimated from 20 minute Bowen ratio data over a full growing season. Six months of hourly sap flow data were recorded in five sample trees growing within a riparian zone situated in a mature A. mearnsii compartment on Mistley-Canema plantation in the Seven Oaks district. Instrument problems at the Gilboa site caused gaps in the data record during the mid-summer period. A model of daily ET based on daily solar radiation and an empirical leaf area function provided satisfactory patching. Problems were experienced with the heat pulse measurements at Seven Oaks. Sap flow measurements declined with time to unusually low levels. Examination of sectioned stems at the conclusion of the monitoring period revealed heavy resin impregnation of the region surrounding the probe positions. Comparison of sap flow rates recorded over the first four to eight weeks (presumed to be the least affected by the wound reaction), to those recorded in the Western Cape, showed the Seven Oaks trees to be transpiring at relatively low rates. Possible reasons for this are discussed, and the most likely is believed to be the result of heavy defoliation of the canopies by a high population of bagworm observed in the area.

In view of the difficulties associated with the Seven Oaks plantation data, the A. mearnsii model developed from the Western Cape sites was used, together with daily mean VPI) data from Gilboa, to predict annual transpiration of A. mearnsii at this site. Rainfall interception by the canopies was again estimated from the number of rain days recorded at the site, and daily rainfall interception estimates for this species provided by Schu1ze et al. (1995)

Table I summarises the annual ET recorded or modelled at each site.

TABLE 1 A summary of transpiration and rainfall interception estimates recorded or modelled at the study sites.
Locality Vegetation Annual water use (mm)
    Transpiration Rainfall Interception ET Difference
Jonkershoek A. mearnsii 1304 171 1475 143
  fynbos     1332
Gilboa A. mearnsii 1188 183 1371 535
  Grassland     836


The principle conclusions and recommendations arising from this project are as follows:


This project has been effective in building the technical capacity of previously disadvantaged members of the team. Mr Godfrey Moses gained much experience with meteorological and sap flow instrumentation, and attended a scientific symposium where a paper co-authored by him was presented. Mr Vilakazi and Mr Ndlela were trained in the installation and maintenance of heat pulse and Bowen ratio equipment, while Ms Royappen played a major role in analysing the heat pulse velocity data.