This thesis is about the dynamics of eastern grey kangaroo (Macropus giganteus) populations and their food supplies in temperate grasslands of south-eastern Australia. It is based on the study of three populations of eastern grey kangaroos inhabiting �warm dry�, �cold dry�, and �warm wet� sites within the Southern Tablelands climatic region. After a pilot survey and methods trial in early 2001, the main period of study was from August 2001 to July 2003. The study populations were found to have the highest densities of any kangaroo populations, 450 to 510 km-2. Their density was the same at the end of the two year study period as at the beginning, in spite of a strong decline in herbage availability due to drought. The eastern grey kangaroo populations were limited according to the predation-sensitive food hypothesis. Fecundity, as the observed proportion of females with late pouch young in spring, was high, in spite of the high kangaroo density and restricted food availability. Age-specific fecundity of a kangaroo sample shot on one of the sites in 1997 to avert starvation was the highest reported for kangaroos. Thus, limitation acted through mortality rather than fecundity. Population growth rate was most sensitive to adult survival but the demographic rate that had the greatest effect in practice was mortality of juveniles, most likely sub-adults. The combination of high fecundity with high mortality of immatures would provide resilience to low levels of imposed mortality and to fertility control. The normal pattern of spring pasture growth was not observed in the drought conditions and few of the recorded increments of growth were of the magnitude considered typical for sites on the southern and central tablelands. Temperature was necessary to predict pasture growth, as well as rainfall, over the previous two months. The best model of pasture growth (lowest AICc) included negative terms for herbage mass, rainfall over the previous two months, and temperature, and a positive term for the interaction between rainfall and temperature. It accounted for 13% more of the variation in the data than did the simpler model of the type used by Robertson (1987a), Caughley (1987) and Choquenot et al. (1998). However this was only 63% of total variation. Re-evaluation of the model based on measurements of pasture growth in more typical (non-drought) conditions is recommended. Grazing had a powerful influence on the biomass of pasture due to the high density of kangaroos. This is a marked difference to many other studies of the type which have been conducted in semi-arid environments where rainfall dominates. The offtake of pasture by kangaroos, as estimated on the research sites by the cage method, was linear on herbage mass. It was of greater magnitude than the more exact estimate of the (curved) functional response from grazedowns in high�quality and low�quality pastures. The widespread recognition of three forms of functional response is inadequate. Both the theoretical basis, and supporting data, have been published for domed, inaccessible residue, and power forms as well (Holling 1966; Noy-Meir 1975; Hassell et al. 1976, 1977; Short 1986; Sabelis 1992). Eastern grey kangaroos had approximately the same Type 2 functional response when consuming either a high quality artificial pasture (Phalaris aquatica), or dry native pasture (Themeda australis) in autumn. Their functional response rose more gradually than those published for red kangaroos and western grey kangaroos in the semi-arid rangelands, and did not satiate at the levels of pasture available. This gradual behaviour of the functional response contributes to continuous stability of the consumer-resource system, as opposed to discontinuous stability. The numerical response was estimated using the ratio equation, assuming an intrinsic rate of increase for eastern grey kangaroos in temperate grasslands of 0.55. There is indirect evidence of effects of predation in the dynamics of the kangaroo populations. This is demonstrated by the positive relationship between r and kangaroo density. Such a relationship can be generated by predation. A desirable future task is to compile estimates of population growth rate and simultaneous estimates of pasture, in the absence of predation, where kangaroo population density is changing, so that the numerical response can be estimated empirically. The management implications arising from this study are numerous and a full account would require a separate report. As one example, kangaroos in these temperate grasslands are on average smaller, eat less, are more numerous, and are more fecund, than would be predicted from other studies (e.g. Caughley et al. 1987). Thus the benefit of shooting each kangaroo, in terms of grass production, is less, or, in other words, more kangaroos have to be shot to achieve a certain level of impact reduction, and the population will recover more quickly, than would have been predicted prior to this study. Secondly, of much importance to managers, the interactive model which can readily be assembled from the products of Chapters 4, 5 and 8, can be used to test a range of management options, and the effect of variation in weather conditions, such as increased or decreased rainfall. For example, the model indicates that commercial harvesting (currently under trial in the region), at the maximum level allowed, results in a sustainable harvest of kangaroos, but does not increase the herbage mass, and only slightly reduces the frequency of crashes when herbage mass falls to low levels. (To demonstrate this with an ecological experiment would require an extremely large investment of research effort.) However, an alternative �national park damage mitigation� formula, which holds kangaroo density to about 1 ha-1, is predicted to increase herbage mass considerably and to reduce the frequency of crashes in herbage mass, but these effects would be achieved at the cost of having to shoot large numbers of kangaroos. Thus, aside from many specific details of kangaroo ecology, the knowledge gained in this study appears to have useful potential to illustrate to managers the dynamic properties of a resource-consumer system, the probabilistic nature of management outcomes, and the consequences of particular kangaroo management proposals.

In terms of acreage, forage is the number one crop in Virginia. The backbone of these forages has long been tall fescue (Lolium arundinaceum (Schreb.) S.J. Darbyshire). Knowledge of the plant species that make up a pasture and the relative amounts of each species present is important for interpreting potential animal performance. It is also important to know the relative amounts and types of weeds present and to monitor for the presence of poisonous plants or noxious weeds. An experiment was conducted in 2003 through 2005 to investigate botanical composition and yield of "Lakota" prairie grass (Bromus catharticus Vahl.), "Kentucky 31" endophyte-infected (KY31 E+), endophyte-free (KY31 E-), and "Quantum" tall fescue (non toxic endophyte infected) under grazing by stocker steers. Forage botanical composition and yield were determined by clipping three 0.25-m2 areas per treatment replicate. Prior to harvesting, the canopy height within each quadrate was measured with a disc meter. In 2005, productive ground cover was assessed using visual evaluation techniques, point quadrat method, and digital imagery quantified with terrestrial remote sensing. Forages were established September 2002 and grazing was initiated in July of 2003. Experimental design was a randomized complete block with four replications. Averaged over the three years the yield of KY31E+ was higher (p<0.05) than all other treatments. Lakota prairie grass had lower (p<0.05) yields than both KY31 E+ and Quantum tall fescue, however no yields did not differ between Lakota prairie grass and KY31 E-. Our results showed a typical forage distribution curve for all the treatments. Early spring, summer, and fall productivity of Lakota prairie grass was less than all the fescues, thus did not extend the grazing season. Forage persistence was greatest for KY31 E+ and Quantum and lowest for Lakota when averaged over all years. Among sampling methods for ground cover, terrestrial remote sensing was the most accurate, compared with visual evaluation and point quadrat methods. For estimates of all yield indirect methods of assessment had high errors; however the plate meter calibrated by sward density seemed the least variable of the methods tested. / Master of Science