A multiple proxy approach to reconstruct landscape-scale vegetation in steppe-like environments : a case study from southern Kazakhstan
Résumé
Assessing landscape-scale vegetation mosaics based on pollen data is a challenge. In temperate environments, vegetation mosaics primarily correspond to the relative proportions of grasslands, crops and different types of forest cover. In steppe-like environments such as those of Central Asia, these mosaics consider the relative proportions of grasslands, Artemisia steppes, Chenopodiaceae steppes and semi-deserts, riverbanks and land-use categories. Most of these land cover types reflect open and semi-open landscapes, which increases the difficulties to identify some types of mosaic patterns. This leads us to the questions:
Do pollen assemblages provide reliable information about landscape mosaics in steppe-like environments? and
Can a multiproxy approach help us to disentangle such information?
To answer these questions, we conducted a vegetation study in the diverse semi-arid Ili River basin in southeast Kazakhstan. This area includes two national parks (Altyn Emel and Charyn Canyon) well known for their desert and canyon landscapes, biodiversity and red-listed species. It is characterized by a variety of desert, semi-desert and steppe landscapes. We sampled surface modern soils at 37 randomly selected locations within a radius of 55 km. At each site, we undertook botanical surveys within a 100 m radius from the soil samples following the protocol of the Crackles Bequest Project, as well as within a radius of 5 km from samples. A study of “relative pollen productivity estimates” is in progress. Maps of the major land cover types for the 37 sites have been created by using Sentinel-2 optical satellite images at a spatial resolution of 10 m, obtaining an iterative supervised classification approach within the remote sensing software SNAP. We extracted and analysed pollen, phytolith and biomarker assemblages from each soil sample. We compared the observed and proxy-based information using modelling and statistical methods. Indices of plant, pollen and phytolith composition and diversity, and biomarkers (distribution of GDGTs and n-alkanes in sediments), are used for comparison.