Simulation of water use and herbage growth in arid regions

<p/>The and and semi-arid regions of the world, totalling about 30% of the land surface of the earth, are predominantly used for extensive grazing, as low and erratic rainfall presents too high a risk for arable farming. The population that can be sustained by the animal products -meat, milk or wool- is largely determined by the availability of primary production for fodder.<p/>An increase in primary production in these areas is possible on two levels:<p/>- without irrigation: either by an increase in the amount of available water or by a more efficient use of the moisture. The first may be achieved by controlling run-off or deep drainage. The latter through improvement of the botanical composition, mainly by legumes or by application of nitrogen fertilizer;<p/>- with irrigation: when water and nutrients are non-limiting, very high production levels may be attained because the environmental conditions, radiation intensity, temperature and length of the growing season are favourable in these regions.<p/>This study was done within the framework of the project 'Actual and potential herbage production under and conditions'. This project aimed at summarizing existing knowledge on the functioning of agro-ecosystems in dynamic simulation models, describing plant growth under semi-arid conditions. Such models were developed on three levels of interest:<br/>- optimum supply with water and nutrients;<br/>- optimum supply of nutrients, but limited moisture;<br/>- conditions where either water or nutrients were the limiting factor.<p/>Validation data for the models were collected in the field and relevant processes were studied under controlled conditions.<p/>This book deals mainly with the different aspects of the relation between water use and plant production under moisture limiting conditions.<p/>The validation experiments were carried out at the Tadmor Experimental Farm in the northern Negev desert of Israel (34°25' OL, 31°22' NB. The average rainfall in this area is 250 mm year ^-1, concentrated in winter (October-March); strong fluctuations between years occur with extremes of 42 mm in '61/'62 and 414 mm in '64/'65. Radiation intensity ranges from 1150 Jm ^-2day ^-1in December to 2750 Jm ^-2day ^-1in August, while average daytime temperatures range between 12.6°C in January and 26.8 °C in August. The soil consists of a sierozem, developed in a 10-20 m thick mantle of löss. Its physical properties are favourable for plant growth: high water-holding capacity and low resistance to root growth.<p/>The vegetation is an abandoned crop land vegetation and consists mainly of herbaceous annuals. The predominant species are the grasses <em>Phalaris minor, Hordeum murinum</em> and <em>Stipa capensis,</em> the crucifers <em>Erucaria boveana</em> and <em>Reboudia pinnata,</em> the compositae <em>Anthemis melaleuca</em> and <em>Centaurea iberica</em> and the legumes <em>Trigonella arabica</em> and <em>Medicago polymorpha.</em> The actual botanical composition shows a great variation both in time and place but this hardly influences the production potential.<p/>Experiments were carried out to determine the course of above ground dry matter production and of available soil moisture during the growing season. Dry matter production was determined via a doublesampling technique of visual estimates. In each treatment 200-400 estimations were carried out and each fifth sample was also harvested and weighed. The harvested samples provide a calibration curve, which is used to give an estimate of the average yield of the field.<p/>Soil moisture was measured with the neutron moderation technique. In each treatment 30-60 aluminium access tubes were installed and soil moisture was measured over two-weekly periods and after sufficiently heavy showers in 30 cm intervals. The upper 30 cm of the soil were at the same time sampled for gravimetric determination of the moisture content.<p/>Meteorological observations were obtained from a standard weather station of the Israeli Meteorological Service, located in Gilat, about 8 km from the experimental site. Rainfall however was recorded in rain-gauges at the experimental site.<p/>The results of the experiments, which were carried out in the fields fertilized with P and K as well as in fields that received NPK fertilization show that under the prevailing conditions actual production is more often limited by nitrogen shortage than by moisture availability. Nitrogen application increased the production up to 2-3 fold.<p/>Maximum growth rates of the natural vegetation reached values of 170 kg ha ^-1day ^-1under optimum conditions. A local wheat variety, used in another experiment attained the same value, which is also equal to the theoretical optimum calculated for the conditions.<p/>The efficiency of water use was also the same for the wheat and for the natural vegetation. However the nitrogen deficient fields showed a much lower efficiency of water use, due to either a lower photosynthetic capacity or greater losses through continuous break-down and rebuilding of nitrogenous compounds.<p/>Experiments carried out to determine the influence of grazing on primary production showed that under normal grazing pressures, the consumption by animals during the growing season is negligible compared with the growth rates. Only heavy overgrazing may affect the level of primary production.<p/>The simulation model 'ARID CROP' developed in the second part of this study, written in the simulation language CSMP/360, describes crop growth under water limiting conditions.<p/>A hierarchical approach is applied to incorporate the various processes when their time constant is too small. A detailed model of evaporation from a bare soil is described and the results are compared with both laboratory and field measurements.<p/>A simplified procedure to calculate the water use efficiency on a daily basis is worked out, based on the results of a detailed model, described elsewhere. The calculated values are compared with the results of pot trials carried out in the framework of the project. In practically all cases the agreement is within 20% which is considered reasonable in view of the assumptions.<p/>Validation of the model shows that reasonable agreement between measured and simulated values is obtained, both for above ground dry matter production and for soil moisture in spite of the limitations of the model. Special difficulties arise from the initialization due to the gross simplifications with regard to germination. Some examples for the application of the model have been described: potential production under moisture limiting conditions is calculated for a 15-year period at the experimental site, based on historical weather data. Comparison with the measured yields shows that mostly actual production was limited by nitrogen availability. Fertilization would not only increase the average yield, but also the fluctuations to about the same range as the rainfall. Management problems will then arise as short- term adaptation of herd size is not possible.<p/>These may either be overcome by adaptation of the grazing pressure to some average production level so that food reserves may be build up in good years or by adjustment of the size of the herd to a smaller grazing area. Part of the total available area may be sown with small grain, which will yield seed in favourable years, and may be used as additional grazing area in unfavourable years.<p/>Another conclusion derived from this example is, that the distribution of the precipitation over the growing season is of equal importance for the production as the total amount of rainfall. An unfavourable pattern-many small showers-increases losses through direct soil evaporation and limits the amount of moisture available for plant growth.<p/>Finally it is shown that spatial inhomogeneity of the soil, which is so troublesome during experimentation counteracts the fluctuations described earlier, mainly through increase of production in unfavourable years.<p/>In conclusion it may be stated that the use of dynamic simulation models to establish the production potential of new areas is preferable to the application of correlation models. Simulation models provide a better insight into the relevant processes and have a much greater potential for extrapolation. Moreover they supply a useful tool for the design of relevant experiments when new regions have to be explored