Changes in sea level and the formation of islands impact the distributions and abundances of local flora and fauna, with palaeo-environmental investigations providing a context for biological conservation. The palaeo-environmental knowledge of the north-west of Australia during the late Quaternary is sparse, particularly the impact of island formation on local faunas.In 1991 and 1993 Peter Veth and colleagues conducted archaeological surveys of the Montebello Islands, an archipelago situated 70 -90 km from the present-day coastline of north-west Australia. A group of three caves were found during this survey on the eastern side of Campbell Island. Two of the caves, Noala and Hayne's Caves, were analysed by Veth and colleagues in the early 1990s; the last cave, Morgan's Cave, remained unanalysed because it contained negligible archaeological material. It provides an opportunity to refine the interpretation of palaeo-environmental conditions, further information on the original pre-European fauna of the north-west shelf, the formation of the islands due to sea level rise, and the impact of sea level rise on local faunas.The fossil fauna assemblage of Morgan's Cave was sorted, identified to the lowest taxonomic level possible, and counted for analysis on relative abundance for paleoenvironmental interpretation. There are marked patterns of species loss and changing relative abundances in certain species, consistent with island formation due to sea level rise. This palaeoecological interpretation can be used in conservation efforts on nearby Barrow Island, a Class A ecological reserve, and management of faunas in conservation reserves and on islands.

Quaternary palaeontologists are often asked to identify and interpret faunal remains from archaeological excavations. It is therefore vital that both palaeontologists and archaeologists understand the limitations imposed by recovery methods, especially sieves. Fossil small mammals, particularly rodents, in Australian archaeological deposits are very significant sources of palaeoenvironmental information. Over the last half century, recovery techniques used by archaeologists in Australia have ranged from dry screening with 10 mm sieves to wet screening with nested sieves graded down to 1 mm. Nested 6 and 3 mm sieves have been a popular combination. Experimental investigations, using owl-accumulated mammal remains from two caves, show that sieves of 2 or even 3 mm (the metric equivalent of 1/8th inch) are fine enough to recover most of the dissociated complete jaw bones used to identify Australian native rodents (all Muridae). But they fail to retain first molar teeth that have become dissociated from the jaws by pre-depositional fragmentation, predator digestion or damage during excavation. A 1.63 mm sieve (the approximate metric equivalent of 1/16th inch) differentially recovers all isolated first molars of Australian ratsized rodents, but not small mice, or some first molars of large mouse species. Our results show that differential recovery statistically significantly biases the relative abundances of rodent species retained on a sieve, although there is considerable inter-species variability. The diagonal dimension of the mesh aperture is confirmed to be at least as important as the side dimension in determining what is lost through a sieve. The demographic structure of species can also be biased by differential loss of isolated molars from the youngest individuals, but only in species whose molar dimensions precisely span the threshold of retention on a particular sieve size. Some data suggest that a greater proportion of small objects will be lost through a sieve when immersed in water, than with sprayed water wet sieving or dry screening. Because highly fragmented faunal materials are characteristic of many Australian archaeological deposits, complete recovery of isolated first molars is essential for detection and identification of rodent species to produce the comprehensive assemblages needed for valid comparisons with accumulations by single predators such as owls, and for the multivariate analyses used in palaeoenvironmental interpretations. Complete recovery of first molars of Australian rodents requires a sieve mesh aperture no larger than 1 mm square. A recent archaeological collection made with such methodology provides an opportunity to demonstrate that a more meticulous standard is worth the time and effort to pursue.