Use of remote sensing in native grass biomass modelling to estimate range productivity and animal performance in a tree-shrub savanna in southern Zimbabwe

Abstract

Herbage and cattle production in semi-arid regions are primarily controlled by climate variation particularly rainfall variability and secondarily by disturbances such as drought, grazing and fire. These factors interact at different spatial and temporal scales in a complex manner difficult to observe or comprehend and, reduce availability and quality of herbage and cattle productivity. Variables for quantifying rangeland productivity are thus rarely available and unreliable yet options for sustainable management are limited. Grazing experiments have provided useful insight about ecological and management factors involved in rangeland functioning, but they have limited scope to deal with high environmental variation. This highlights the need for a systems approach for monitoring rangeland and cattle productivity at the appropriate spatial and temporal scales to enable productivity to be maximised whilst risk to climate variation is minimised. This study explored two broad objectives: to determine the ranch-scale impacts of rainfall variability and drought on herbaceous aboveground biomass (AGB) using optical remote sensing; and to parameterise, evaluate and apply a systems model, the Sustainable Grazing Systems (SGS) whole farm model to complement grazing experiments in assessing the effects of grazing strategies on beef cattle production. To determine rainfall variability impacts, twenty regression models were firstly developed between measured herbaceous AGB and, classical and extended multispectral vegetation indices (MVIs) derived from a Landsat 8 image. End-of-season herbaceous AGB was predicted with high accuracy (r2 range = 0.55 to 0.71; RMSE range = 840 to 1480 kgha-1). The most accurate model was used to construct a regression between rainfall and AGB derived from peak-season Landsat images available between 1992 and 2017. Standardised precipitation index and standardised anomalies of herbaceous AGB production were then used in a convergence of evidence approach to determine the response of AGB to rainfall variability and drought intensity. Total wet season rainfall revealed high variability (33 to 41 % CV) and subsequent herbaceous AGB production were 18 to 35 % more variable. Spatial heterogeneity of AGB production across herbaceous communities were high and deviated from mean AGB by 51 to 69 %. Landscape-level temporal variation of AGB production remained stable despite the increase of climate variability experienced in the region in the past 50 years. Climate inputs and parameter sets for upper-, mid- and foot- slope land types and key grass species, Urochloa mosambicensis and Eragrostis curvula were developed by integrating spatial data with previous soil surveys and extensive reviews of published experiments. A simulation experiment was conducted between 1992 and 2017 for all combinations of land types and grass species to analyse the extent of improvement resulting from parameter adjustments. The SGS model predicted the growth pattern known for grasses native to dry regions of southern Africa. The model represented measured herbaceous biomass moderately well (r2 = 0.57), at low average error (RMSE, 820 kg DM ha-1) despite huge discrepancies in summary statistics for measured (mean, 3877 kg DM ha-1) and simulated (mean, 3071 kg DM ha-1) biomass and residuals. Model predictions were also significantly correlated with remotely sensed AGB (r2 = 0.46) at reasonable overall performance error (RMSE, 981 kg DM ha-1). The integrated workflow developed for parameterising and calibrating the SGS pasture-simulation model can benefit model users in data-constrained environments. Animal growth parameters specific to Brahman weaner steers were defined in the SGS model to enable evaluation of impacts of recommended (10 haLU-1) and other three stocking rates (7, 15 and 20 haLU-1) and multi-paddock grazing systems (2-, 3- and 4- paddocks per herd) on rangeland productivity. Overall, there were no observable differences in herbage production and dry matter intake irrespective of stocking rate and multi-paddock grazing system. But stocking rate effects on animal production were more pronounced compared to multi-paddock grazing systems. To maximise cattle productivity in semi-arid rangelands, management should be emphasised on manipulation of stocking rates over multi-paddock grazing systems.

Keywords Rangeland monitoring, climate risk, sustainability, animal productivity, grazing strategies