Using hyperspectral satellite imagery for regional inventories: a test with tropical emergent trees in the Amazon Basin M. Pape ¸ s, R. Tupayachi, P. Martı´nez, A.T. Peterson & G.V.N. Powell Abstract Questions: Understanding distributions of tree spe- cies at landscape scales in tropical forests is a difficult task that could benefit from the recent development of satellite imaging spectroscopy. We tested an application of the EO-1 Hyperion satellite sensor to spectrally detect the location of five im- portant tree taxa in the lowland humid tropical forests of southeastern Peru. Location: Peru, Departamento de Madre de Dı´os. Methods: We used linear discriminant analysis with a stepwise selection procedure to analyze two Hy- perion datasets (July and December 2006) to choose the most informative narrow bands for classifying trees. Results: Optimal channels selected were different between the two seasons. Classification was 100% successful for the five taxa when using 25 narrow bands and pixels that represented 440% of tree crowns. We applied the discriminant functions de- veloped separately for the two seasons to the entire study area, and found significantly nonrandom overlap in the anticipated distributions of the five taxa between seasons. Conclusions: Despite known issues, such as signal- to-noise ratio and spatial resolution, Hyperion ima- ging spectroscopy has potential for developing regional mapping of large-crowned tropical trees. Keywords: Hyperion; Hyperspectral sensor; Imaging spectroscopy; Remote sensing; Species identifica- tion; Spectra; Tropical rainforest. Introduction Remote sensing provides potentially powerful tools to address diverse questions in biology. The availability of reliable sources of information-rich satellite imagery like the Landsat platform has en- abled study of diverse topics, such as mapping habitat use by caribou (Bechtel et al. 2004), identi- fying migratory bird habitat (Sader et al. 1991), and evaluating effects of logging on tree diversity (Foo- dy & Cutler 2003). We are exploring ways in which these approaches can inform the study of tropical frugivorous animals via characterizing the avail- ability of resources in time and space (Fleming et al. 1987; Levey 1988; Loiselle & Blake 1991; Price 2004). To better understand the seasonal movements and habitat needs of these animals, much-improved knowledge of the distribution and phenology of tro- pical trees is necessary. The difficulty, however, lies in the broad, geographic scale of this challenge: detailed phenological studies across significant spatial extents are not feasible. A more desirable approach would integrate local field studies with remote sensing ap- proaches to scale local-scale results up to the needed landscape-scale perspective. Here, we attempt to apply remote sensing ap- proaches to understanding tree species distributions in a relatively little studied, but highly diverse, area of the Amazon Basin in southeastern Peru; the hy- perdiverse nature of these forests adds complexity to the task. Our goal is to identify canopy trees to spe- cies using satellite images collected by a hyperspectral instrument designed to acquire data in hundreds of narrow, contiguous spectral bands, termed imaging spectroscopy (Goetz et al. 1985). Hyperspectral data have been used to study differ- ent aspects of tropical forest ecosystems for over two decades. Treitz & Howarth (1999) reviewed ad- vances, mainly in the study of temperate forest ecosystems, after development of airborne sensors such as the Airborne Visible/Infrared Imaging Pape ¸ s, M. (corresponding author, monapapes@ gmail.com) & Peterson, A.T. (town@ku.edu): Natural History Museum and Biodiversity Research Center, The University of Kansas, Lawrence, KS 66045, USA. Tupayachi, R. (litotup@gmail.com) & Martı´nez, P. (paola.martinez.g@gmail.com): World Wildlife Fund Peru Program Office, Calle Trinidad Mora 853, Lima 14, Peru. Powell, G.V.N. (george.powell@wwfus.org): Conser- vation Science Program, World Wildlife Fund, 1250 24th Street NW, Washington, DC 20037, USA. Current address: Pape ¸ s, M., University of Wisconsin – Madison, Center for Limnology, 680 N Park St., Madison, WI 53706, USA. Journal of Vegetation Science 21: 342–354, 2010 DOI: 10.1111/j.1654-1103.2009.01147.x & 2009 International Association for Vegetation Science