Representation of flora and vegetation in Quaternary fossil assemblages: Known and unknown knowns and unknowns (Review)

摘要

Paleoecological inference rests on a foundation of processes and relationships that lie between the target variables of interest and the proxy data extracted from the fossil record. Inattention to these processes can lead to overconfidence, reification of proxies, and " ignorance creep", a process by which assumptions become taken for granted and ultimately hidden or forgotten. Paleoecological inference can be strengthened, and its uncertainties identified and quantified, by developing forward conceptual models that incorporate the processes and relationships known to lie between target variables and proxy data. These may fall into four general categories: source, vector, diagenesis, and analysis. The uncertainties within each of these categories are somewhat unique, although interactions often arise among processes in different categories. I use inference of regional species occurrence from macrofossil assemblages in woodrat (Neotoma) middens to illustrate the effects of various processes and interactions on strength of inference and origin of uncertainties, and to identify how uncertainties can be characterized, reduced, or quantified. The kinds of issues discussed in the woodrat-midden example are pervasive in paleoenvironmental and paleoecological inference, and should be examined thoroughly for all proxies and archives on a regular basis.