Scientific research now shows that humans are pervasive in earth ecosystems, and have been so for many thousands of years. So it may seem strange to argue against a concept that has been so hard won and is now empirically incontrovertible. What is starting to seem stranger is that, to paraphrase anthropologist Tim Ingold, the best way we have to describe our role in the world is to take ourselves out of it. As human influence on the Earth and its processes increases, we face the profound paradox that most of our intellectual weapons in the environmental area - from prehistoric fire debates to projections of climate change - have maintained a separation of humans and nature. This is an argument based not on semantics but on false separations that are adversely affecting the quality of our research and practice, including the ways we attempt to integrate archaeology and palaeoecology. While 'human impacts' may be applicable to a subset of human activities, it is neither conceptually nor empirically strong enough for the detailed networks of human and non-human others now evident. We need to articulate a broader concept of agency, both human and otherwise, and to develop explanations that focus on associations and relationships rather than separations and essentialisms.