Phytoliths from modern and ancient habitats: toward a modern-based, quantitative approach to reconstruct vegetation change during the MMCO of Patagonia, Argentina

Abstract

Phytolith analysis has high potential for reconstructing past vegetation with higher spatial resolution compared other high-resolution proxies, such as pollen and spores. Phytolith assemblages are used in paleoecology to reconstruct changes in vegetation structure through time. In addition, spatial variability of the phytolith signal (across samples collected along a single stratigraphic level) is interpreted as indicative of habitat heterogeneity based on the notion that phytolith assemblages are derived from vegetation that died and decayed in place and therefore hold a local signal. However, this and other assumptions have not yet been tested directly in modern environments; current data are insufficient to establish modern calibrations for the deep time phytolith record, and thus understand the fossil phytolith records in different vegetation types. In Chapter 1 and 2 of this dissertation I aim at helping bridging this gap, by 1) defining an appropriate methodology to sample phytolith for modern analogue studies that is applicable to the deep-time phytolith record; 2) and by providing a modern reference study of soil phytolith along transects in two Neotropical vegetation types in Costa Rica: a rainforest and a dry forest. I investigate the following questions: 1) how many samples and from which part of the (phytolith-rich) soil A-horizon are needed to reflect accurately the standing vegetation? (Chapter 1); 2) are gradients in vegetation structure, composition, and diversity recorded in phytolith assemblages across transects in rainforest and dry forest soils? (Chapter 2); and 3) can we use one or more phytolith assemblages to characterize these two vegetation types, and distinguish them in the fossil record? (Chapter 2). In Chapter 3, I apply the lessons learned from Chapter 1 and 2 to the study of vegetation heterogeneity and vegetation change in Patagonia, at the onset of the Middle Miocene Climatic Optimum (MMCO) –the last global warming event taking place on Earth before the current one, between ~17 and 14.5 Ma. The MMCO is poorly documented in the Southern Hemisphere and at high latitudes. The Santa Cruz Formation (SCF), in southern Patagonia, is an exception, preserving one of the most diverse and well-preserved fauna assemblages on Earth. Fauna and stable isotope data from the SCF suggest that global warming associated with increased aridity favored heterogeneous habitats characterized by many ecological niches which were able to support abnormally high fauna diversity. The phytolith record of SCF has been so far poorly studied but constitute the best line of evidence for high resolution reconstruction of vegetation change through time as well as of spatial patterns of vegetation variability (heterogeneity). Using phytolith assemblages from the SCF I investigate the following questions: 1) How did vegetation structure change in response to the initial warming pulse of the MMCO? 2) How did grass community composition change in response to warmer and drier conditions at the onset of the MMCO? 3)Was the remarkably high diversity of the Santa Cruz fauna supported by habitats characterized by vegetation heterogeneity (i.e., a mix of forested and open vegetation areas) throughout the onset of the MMCO as would be predicted based on modern ecology and SCF faunal data? In Chapter 1 phytolith from modern soil assemblages from two Neotropical forests in Costa Rica (a dry forest and a rainforest) are studied to determine a sample strategy for future modern analogue studies that is applicable to the phytolith deep-time record. Results suggest that the typical approach in deep-time paleoecology of taking point samples from the lower A-horizon of paleosols is justifiable (at least for paleosols reflecting rainforest and dry forest soils), and should therefore be implemented in future phytolith modern analogues studies that aim at improving interpretations of the deep-time phytolith record. Thus, the results of Chapter 1 constitute the basis upon which the modern analogue study described in Chapter 2 was conducted In Chapter 2, additional soil phytolith assemblages collected along vegetation transects are used to investigate whether and how soil phytoliths reflect gradients in vegetation structure, composition and diversity across the two habitat types (dry forest and rainforest). In all, our results demonstrate that phytolith assemblages can definitely distinguish dry and wet forest habitats. In addition, our results also suggest that phytolith assemblage characteristics within vegetation types do not capture all aspect of environmental and plant community gradients. However, overall higher environmental heterogeneity of the dry forest results in higher heterogeneity of the phytolith assemblages. This result suggest that overall, spatial sampling (along a transect) and the analysis of phytolith assemblage composition allow to reconstruct some structural, and compositional aspects of habitat heterogeneity, and that that phytolith assemblage heterogeneity within a habitat might be indicative of habitat heterogeneity. In Chapter 3, phytolith assemblages from The Santa Cruz Formation (Patagonia) spanning the onset of the Middle Miocene Climatic Optimum (MMCO) are analyzed to reconstruct vegetation response to the climatic event as well as to reconstruct vegetation heterogeneity across two stratigraphic layers, representing two snapshots of the SCF vegetation at two different times. Results show that before the onset of the MMCO southeastern Patagonia was characterized by heterogeneous habitats with abundant pooid C3 grasses and a woody component represented by conifers, dicots, as well as palms in varying abundance. This habitat corresponded to woodland or open woodland/shrubland, including palm shrubland. In the upper SCF, at the onset of the MMCO (inferred from isotopic data to be drier), grass abundance decreased, and phytolith assemblages indicate that the landscape was dominated by a woody component of the vegetation. In addition, grass communities were dominated by C3 pooid grasses whereas grasses of the tropical PACMAD clade (which includes both C3 and C4 grasses) were only a minor component of grass communities. We interpret these trends as reflecting the expansion of dry-adapted woody vegetation in response to MMCO climate change, and to the detriment of a C3 grass community which was not adapted to dry conditions. Further, we suggest that PACAMD grasses at the SCF were likely primarily C3, and the expansion of dry-adapted C4 grasses and grass-dominated open habitats did not take place in Patagonia until after the early middle Miocene.