Close agreement between pollen‐based and forest inventory‐based models of vegetation turnover

• Published in: Global ecology and biogeography 2015-08, Vol.24 (8), p.905-916
• Main Authors: Nieto‐Lugilde, Diego, Maguire, Kaitlin C, Blois, Jessica L, Williams, John W, Fitzpatrick, Matthew C
• Format: Article
• Language: English
• Subjects: Biodiversity; Climate change; climatic factors; data collection; edaphic factors; forest inventories; forest inventory; forests; fossil; fossils; generalized dissimilarity modelling; Inventory; palaeoecology; Pollen; Quaternary; soil properties; spatial variation; Taxonomy; temporal variation; turnover; uncertainty; vegetation
• ISSN: 1466-822X; 1466-8238
• DOI: 10.1111/geb.12300

AIM: Fossil records are being increasingly used to help understand the consequences of climate change for biodiversity. Pollen records from the late Quaternary are among the most commonly used fossil data, but pollen‐based inferences of biodiversity can potentially be confounded by spatial and taxonomic uncertainties and the influence of non‐climatic abiotic factors such as soils on vegetation–climate relationships. Using paired pollen and vegetation inventories, we assess the fidelity of pollen‐based estimates of compositional turnover of vegetation along environmental gradients given various sources of uncertainty. LOCATION: Eastern United States. METHODS: We used modern pollen records and forest composition data from Forest Inventory and Analysis (FIA) plots to fit generalized dissimilarity models. To address how uncertainties in pollen records affect estimates of turnover, we coarsened the vegetation data spatially from individual plots to 10‐ and 30‐arcmin resolution and taxonomically from species to genus. To determine whether soil properties influenced turnover, we used deviance partitioning between models including climate or soil variables versus models with a combination of both. RESULTS: Pollen‐based estimates of turnover were highly correlated with those based on FIA data, but tended to be lower, mainly due to differences in taxonomic resolution and secondarily to differences in spatial resolution. Neither spatial nor taxonomic uncertainty substantially reduced the correlation between pollen‐ and FIA‐based estimates of turnover. FIA data best matched pollen records when they were aggregated to genus and 30‐arcmin resolution. Vegetation–climate relationships were similar across datasets, although models sometimes differed. The influence of soil variables was negligible compared with climate variables and did not improve model fit. Pollen thresholds did not greatly affect the form and strength of pollen–vegetation relationships. MAIN CONCLUSIONS: Pollen can act as a robust proxy for vegetation turnover, thereby supporting the use of pollen‐based estimates of turnover to predict temporal changes in vegetation.