Using tree species inventories to map biomes and assess their climatic overlaps in lowland tropical South America

• Published in: Global ecology and biogeography Vol. 27; no. 7/8; pp. 899 - 912
• Main Authors: de Miranda, Pedro Luiz Silva, Oliveira-Filho, Ary T., Pennington, R. Toby, Neves, Danilo M., Baker, Timothy R., Dexter, Kyle G.
• Format: Journal Article
• Language: English
• Published: Oxford John Wiley & Sons Ltd 01.08.2018; Wiley Subscription Services, Inc
• Subjects: Amazon Forest; Atlantic Forest; Biodiversity; Cerrado; Chaco; Climatic data; cluster analysis; Composition; Dry Tropical Forest; Ecosystem management; Ecosystems; Environmental assessment; Forests; Heterogeneity; Imagery; Interpolation; Inventories; Inventory management; NeoTropTree; Remote sensing; Savanna; Spatial distribution; Species composition; Species diversity; Tropical forests; Wildlife conservation; Woodlands; South America
• ISSN: 1466-822X; 1466-8238
• DOI: 10.1111/geb.12749

Aim: To define and map the main biomes of lowland tropical South America (LTSA) using data from tree species inventories and to test the ability of climatic and edaphic variables to distinguish amongst them. Location: Lowland Tropical South America (LTSA), including Argentina, Bolivia, Brazil, Ecuador, Paraguay, Peru and Uruguay. Time period: Present. Major taxa studied: Trees. Methods: We compiled a database of 4,103 geo-referenced tree species inventories distributed across LTSA. We used a priori vegetation classifications and cluster analyses of floristic composition to assign sites to biomes. We mapped these biomes geographically and assessed climatic overlaps amongst them. We implemented classification tree approaches to quantify how well climatic and edaphic data can assign inventories to biomes. Results: Our analyses distinguish savanna and seasonally dry tropical forest (SDTF) as distinct biomes, with the Chaco woodlands potentially representing a third dry biome in LTSA. Amongst the wet forests, we find that the Amazon and Atlantic Forests might represent different biomes, because they are distinct in both climate and species composition. Our results show substantial environmental overlap amongst biomes, with error rates for classifying sites into biomes of 19–21 and 16–18% using only climatic data and with the inclusion of edaphic data, respectively. Main conclusions: Tree species composition can be used to determine biome identity at continental scales. We find high biome heterogeneity at small spatial scales, probably attributable to variation in edaphic conditions and disturbance history. This points to the challenges of using climatic and/or interpolation-based edaphic data or coarse-resolution, remotely sensed imagery to map tropical biomes. From this perspective, we suggest that using floristic information in biome delimitation will allow for greater synergy between conservation efforts centred on species diversity and management efforts centred on ecosystem function.