Australian mammals have exhibited exceptionally high rates of decline since European settlement 230 years ago, especially in arid and semi-arid regions. In recent decades there has been increasing concern over the current declines of small mammals in northern tropical savannas. In these systems, little scientific attention has been given to the suite of large macropods, family Macropodidae, (common wallaroo [Osphranter robustus], antilopine wallaroo [O. antilopinus] and agile wallaby [Notamacropus agilis]), some thought to be declining by Aboriginal Traditional Owners. These species may be impacted by feral herbivores and contemporary fire regimes, threats that are both linked to small mammal decline and impact forage quantity and quality. A multi-scale approach – from landscape- to local-scales – was utilized with the overall objectives of exploring the nexus between fire, forage and herbivory in two regions of northern Australia and examining the applicability of the concept of pyricherbivory, the spatial and temporal interactions of fire and grazing, to Australian savannas. The North Kimberley bioregion in Western Australia and Arnhem Land in the Northern Territory are regions of largely intact tropical savanna, each with well-established feral herbivores and active Indigenous fire management. Arnhem Land, where water buffalo (Bubalus bubalis) is the dominant feral herbivore, has a continual history of Aboriginal fire management, in contrast to the North Kimberley, where fire management was disrupted by European colonial activity and with a more recent establishment of the dominant feral herbivore, cattle (Bos spp.). Arnhem Land has also been the focus of several feral herbivore surveys in the recent past, unlike the North Kimberley. My thesis begins by using aerial surveys, road transects and remote camera trapping to examine the effects of feral cattle and fire on the distribution and abundance of large macropods in the North Kimberley. Density and biomass of feral cattle exceeded that of macropods regardless of survey technique utilised. Density estimates for cattle were up to 125 times higher (0.3-10.0 km\(^{-2}\)) than estimates for macropods (0.08-0.49 km\(^{-2}\)). Cattle biomass, based on the aerial surveys (corrected for perception bias), were 15 and 95 times higher than macropods for infertile (279 versus 19 kg km\(^{-2}\)) and fertile savannas (518 versus 5 kg km\(^{-2}\)), respectively. Proximity to the nearest pastoral property was a significant predictor of the aerial sightings of feral cattle, suggesting cattle are continuing to expand throughout the area from adjacent pastoral lands. Abundance and foraging activity of cattle were positively associated with recently burnt areas. In contrast, camera trapping showed agile wallaby and wallaroo occurrence and foraging behaviour were associated with unburnt areas. Agile wallabies and wallaroos were negatively associated with cattle and showed substantial diurnal and seasonal separation consistent with an antagonistic interspecific interaction. These multi-scale surveys suggest that recent landscape changes such as altered fire regimes and introduced herbivores have negatively impacted large grazing macropod species. Isotopic analysis of faecal samples was conducted in the North Kimberley to evaluate how macropods and feral cattle utilise such variable forage resources throughout the dry season. Feral cattle, wallaroos and agile wallaby utilised forage resources differently, consistent with previous related studies comparing buffalo and macropods in Arnhem Land. The contribution of grass to the diet of the agile wallaby was unrelated to substrate fertility, fire or dry season period (early, mid-, late dry season) which aligns with their known status as a mixed feeder. Wallaroos had the highest contribution of grass to their diet, compared to cattle and agile wallaby, with higher proportions of grass consumed on fertile substrates, regardless of fire or dry season period. Cattle diets incorporated more grass in burnt, fertile areas and in the early and late dry season, with a decrease in grass utilisation during mid-dry season, most likely related to the availability of alternative food resources. Cattle diets were highly variable in response to forage quality, ranging from 0-91% grass compared to wallaroos, ranging from 36-100% grass. Diets of wallaroos and cattle were correlated with live forage fibre content (R\(^2\) = 0.41 and 0.56, respectively). As fibre content increased with herbaceous biomass curing, grass intake decreased, precipitously so for cattle, demonstrating their diet flexibility. To understand the underlying mechanisms of local-scale competition in both the North Kimberley and Arnhem Land, we used remote camera trapping and grazing exclosures to examine how herbivory, fire, seasonality, and soil fertility affect forage quantity and quality, in turn affecting herbivore distribution. We found that, even at low herbivore densities, grazing reduced forage quantity (live, dead and total standing herbaceous biomass) and increased forage quality (crude protein content of live herbaceous biomass). Fibre content of live herbaceous biomass was significantly lower in burnt vs. unburnt sites and fertile vs. infertile sites. There was no significant distinction between early and late dry season fires. However, sites burnt in the late dry season had the highest crude protein content of live biomass, highlighting the importance of heterogeneous fire regimes in providing high-quality forage throughout the dry season. Field sites in Arnhem Land had lower measures of overall forage quality (lower crude protein and higher fibre contents of live biomass) but supported more feral herbivores and wallaroos than sites in the North Kimberley. One possible explanation is that water buffalo and feral cattle may exert different levels of competitive pressure on native herbivores. Results suggest that pyricherbivory strongly influences the feeding behaviour of feral herbivores and some native herbivores. In summary, I found that the biomass of introduced feral herbivores far outweighs that of native herbivores at the two study locations in Australia’s northern savannas. Macropods are not behaving as predicted by the theory of pyricherbivory, which contends that herbivores are drawn to, and utilise, recently burnt, highly nutritious forage. Macropod foraging was associated with unburnt areas of low forage quality, in contrast to both historical and contemporary records of Traditional Ecological Knowledge regarding burning for macropod management. This discrepancy is possibly due to competition with feral herbivores for limited high-quality forage. Feral cattle and water buffalo have a wider dietary breadth than large macropods in the savannas and are thus better able to cope with forage of lower quality by varying their relative intake of grass and browse. The largest native grazing herbivores, wallaroos, are less capable of such dietary plasticity. Collectively, my results point to the importance of ongoing control of large feral herbivores in conservation areas and the incorporation of more late dry season fires in addition to early dry season burning to maintain forage quality for macropods throughout the dry season, more closely replicating the historical Indigenous fire regime. Further surveys of grazing macropods elsewhere in the tropical savannas of Australia are necessary to confirm regional population declines and advance our understanding of the complex relationships between native and feral herbivores, forage, fire and Indigenous wildlife management.