Macquarie University

About: Macquarie University is a education organization based out in Sydney, New South Wales, Australia. It is known for research contribution in the topics: Population & Context (language use). The organization has 14075 authors who have published 47673 publications receiving 1416184 citations. The organization is also known as: Macquarie uni.

Do island populations have less genetic variation than mainland populations

Abstract: Island populations are much more prone to extinction than mainland populations. The reasons for this remain controversial. If inbreeding and loss of genetic variation are involved, then genetic variation must be lower on average in island than mainland populations. Published data on levels of genetic variation for allozymes, nuclear DNA markers, mitochondrial DNA, inversions and quantitative characters in island and mainland populations were analysed. A large and highly significant majority of island populations have less allozyme genetic variation than their mainland counterparts (165 of 202 comparisons), the average reduction being 29 per cent. The magnitude of differences was related to dispersal ability. There were related differences for all the other measures. Island endemic species showed lower genetic variation than related mainland species in 34 of 38 cases. The proportionate reduction in genetic variation was significantly greater in island endemic than in nonendemic island populations in mammals and birds, but not in insects. Genetic factors cannot be discounted as a cause of higher extinction rates of island than mainland populations.

SPICE: Semantic Propositional Image Caption Evaluation

Abstract: There is considerable interest in the task of automatically generating image captions. However, evaluation is challenging. Existing automatic evaluation metrics are primarily sensitive to n-gram overlap, which is neither necessary nor sufficient for the task of simulating human judgment. We hypothesize that semantic propositional content is an important component of human caption evaluation, and propose a new automated caption evaluation metric defined over scene graphs coined SPICE. Extensive evaluations across a range of models and datasets indicate that SPICE captures human judgments over model-generated captions better than other automatic metrics (e.g., system-level correlation of 0.88 with human judgments on the MS COCO dataset, versus 0.43 for CIDEr and 0.53 for METEOR). Furthermore, SPICE can answer questions such as `which caption-generator best understands colors?' and `can caption-generators count?'

Using the class 1 integron-integrase gene as a proxy for anthropogenic pollution

Abstract: Around all human activity, there are zones of pollution with pesticides, heavy metals, pharmaceuticals, personal care products and the microorganisms associated with human waste streams and agriculture. This diversity of pollutants, whose concentration varies spatially and temporally, is a major challenge for monitoring. Here, we suggest that the relative abundance of the clinical class 1 integron-integrase gene, intI1, is a good proxy for pollution because: (1) intI1 is linked to genes conferring resistance to antibiotics, disinfectants and heavy metals; (2) it is found in a wide variety of pathogenic and nonpathogenic bacteria; (3) its abundance can change rapidly because its host cells can have rapid generation times and it can move between bacteria by horizontal gene transfer; and (4) a single DNA sequence variant of intI1 is now found on a wide diversity of xenogenetic elements, these being complex mosaic DNA elements fixed through the agency of human selection. Here we review the literature examining the relationship between anthropogenic impacts and the abundance of intI1, and outline an approach by which intI1 could serve as a proxy for anthropogenic pollution.