CORDIS - EU research results

Ecosystem function and Diversity in Amazonia

Periodic Reporting for period 2 - E FUNDIA (Ecosystem function and Diversity in Amazonia)

Reporting period: 2020-08-01 to 2021-07-31

The forests of Amazonia are celebrated for both their exceptional biodiversity and their vital role in the global carbon and water cycles, which contribute towards climate change mitigation. However, our understanding of the spatial organisation of biodiversity and its interaction with ecosystem function at large scales is limited, due primarily to the laborious nature of collecting biodiversity data in Earth’s largest tropical forest. The Carnegie Airborne Observatory (CAO) is a unique research platform, consisting of a custom-built aircraft providing airborne laser-guided spectral profiles of forest canopies at an unprecedented level of spatial and spectral resolution at large spatial scales (e.g. >100 km2). This ‘spectranomic’ approach presents a new paradigm in tropical forest ecology, as spectra reflect the chemical composition of leaves. When combined with careful field calibration, spectranomic techniques can be used to measure specific leaf traits (e.g. leaf N concentration), and potentially map the distribution of individual species at large spatial scales. The goal of E FUNDIA is to integrate state of the art, high fidelity imaging spectroscopy and extensive field plot networks in order to quantify the relationship between biodiversity and ecosystem function in lowland Amazonian tree communities at large spatial scales

The specific objectives of the project are:

1: Map the spatial distribution of >50 common tree species at landscape and regional scales in lowland Amazonia.
2: Determine the extent to which taxonomic diversity drives functional diversity across environmental gradients in lowland Amazonia.
3: Quantify the role of hyperdominant taxa in driving patterns of functional diversity across environmental gradients in lowland Amazonia.

Following the completion of the project the following conclusions have been made:

1. Most hyperdominant tree species in West Amazonian forests cannot be identified from airborne remotely sensed data with a 2m pixel size, this is because their crowns either cannot be seen at all from above, or cannot be isolated from the surrounding forest canopy.
2. Those tree species that dominate the uppermost canopy and emergent strata (i.e. those that are potentially visible in remotely sensed data) are different from those species that dominate lower strata. Moreover these canopy and emergent hyperdominant species are phylogenetically clustered in just a few key plant families.
3. While airborne spectranomic approaches have been used to successfully map variation in community level functional traits, using this data to observe trait data associated with specific tree species presents a major challenge and 2m ground resolution is likely insufficient in most instances.
4. The higher spatial resolution (<10 cm) of UAV data offers greatly enhanced potential for mapping hyperdominant tree species (particularly those with distinctive crown shapes) and potentially for measuring functional trait profiles if UAV-based hyperspectral sensors become more available.
5. More species rich forests do not appear to be necessarily more functionally diverse than more species poor, due to high functional redundancy in diverse Amazonian forests. Functional diversity is increased by including a range of different forest types, regardless of their species richness.
Objective 1

Extensive fieldwork was undertaken to geolocate 50 individuals for 20 common tree species at two locations in Amazonian Peru. Initial analysis of these data showed that 2 meter resolution is likely insufficient for isolating most hyperdominant tree crowns for species mapping proposes in Amazonian forests.

UAV imagery with 2 cm spatial resolution was also collected for 100 ha of forests at each of the two field sites in lowland Amazonia. Initial analysis show that this data can be used to develop landscape scale species maps of some hyperdominant tree species, including valuable fruit producing palm species. These maps are currently being prepared for publication.

An alternative approach for examining the distribution of hyperdominant tree species was developed by establishing a new network of forest inventory plots across Amazonia that uniquely examined patterns of hyperdominance across Amazonian forests strata. Results from these analyses demonstrate that different tree species are dominant in different forest strata, and these dominance patterns vary among Amazonian regions. Furthermore, hyperdominant species in the understory of Amazonian forests consist of a range of species that belong to different Amazonian lineages, while those species that dominate the forest canopy are concentrated in a few key lineages. This is important because it is only these species that are hyperdominat in the canopy and emergent strata that are visible to remote sensing approaches.

Objective 2

Tree functional trait data has been assembled to characterise the key functional distributions across environmental gradients in lowland Amazonia. Additional floristic data has been assembled to quantify taxonomic diversity across the same environmental gradients. These data are currently being examined to test the extent to which taxonomic diversity drives functional diversity in lowland Amazonia. Results suggest that more species rich forests do not appear to be necessarily more functionally diverse than more species poor, due to high functional redundancy in diverse Amazonian forests. Functional diversity is increased by including a range of different forest types, regardless of their species richness.

Airborne imaging spectroscopy data has been assembled across sites in lowland Amazonia and across the global tropics more broadly. An unsupervised method for estimating taxonomic diversity from this imaging spectrscopy data has been developed and is being deployed across sites.

Objective 3

Functional trait data for hyperdominant tree species have been assembled. The functional trait space occupied by these hyperdominant tree species is being quantified and compared to that of entire tree communities in order to assess the role of hyperdominant tree species in driving patterns of functional tree diversity.
The proposal is expanding the state of the art on several fronts: First, the new network of forest plots developed in work package one is providing entirely new insights into the compositional structure of Amazonian forests and the organization of hyperdominant species across forest strata . Second, the UAV approaches to map hyperdominant tree species at the 100 ha scale is providing a novel methodological approaches (deep learning networks) for quantifying the distribution of palm species at this scale. Third, The unsupervised approach for estimating tree alpha diversity using remote imaging spectroscopy data in work package two will provide a new method for measuring the most fundamental aspect of biodiversity at potentially global scales.

The project has generated a new understanding of the processes that govern the spatial organization of dominant tree species. As these species together account for half the trees in Amazonian forests, these species level understanding for monitoring and eventually predicting how these forests are responding to global change.
Example of hyperdominant species mapping with UAV imagary