Measuring Biodiversity Using Remote Sensing
The Regen Guest Author Series features technical experts in the various fields that compose Regen Network. In this piece, GIS Specialist Fauzi Ramadhoan offers his expertise in quantifying ecological variables to make our protocol more statistically accurate.
Introduction
Biodiversity is a crucial ecological indicator and needed to track the impact of various drivers such as land use change and climate change. Through developing biodiversity assessment protocols, the composition, function and structure, as well as management approaches of ecosystems can be analyzed. In biodiversity surveillance and monitoring, ecosystems are an essential link between species and populations on the one side, and land use and landscapes on the other. What can be measured in ecosystems potentially touches on all the major dimensions of biodiversity. Therefore strategic choices have to be made about what should be measured, and how and where to measure it. This blogpost will explain what the important ecosystem variables are, how remote sensing can be applied to measuring biodiversity and which biodiversity indices could be used in the development of a Regen Network biodiversity assessment protocol.
Ecosystem variables
There are generally three ways to measure ecosystem variables. First, most functional processes can be measured as fluxes, using IOT sensors. For precise monitoring of composition, abundance, extent and change is commonly done by in situ monitoring through habitat surveillance combined with vegetation plots. Last, structural change monitoring can be done using in situ combination with remote sensing from space, aircraft, or drone.
There are advantages and differences between the methodologies and one solution does not satisfy all data questions. Remote sensing technologies are increasingly becoming integrated with in situ measurements as new technologies become available for ecosystem research -or global applications this development is essential. Currently it is possible to employ an array of instruments to monitor ecosystem characteristics, from fixed sensors and in-situ measurement, to drones, planes and satellite sensors.
Traditionally, ecologists have been mapping biodiversity and ecosystems based on in-situ observation, perhaps generalized using aerial photography. Although currently remote sensing has improved quality of biodiversity assessments for large areas, the quality of an assessment is still heavily dependent on the availability and quality of field data. The data collected from a limited number of field plots is used for linking remote sensing measurements to the attributes of interests. For example, forest structure measured with laser scanning describes height and density of vegetation, but does not necessarily provide the same attributes important for biodiversity assessment, such as variation in tree size and species, or the amount of dead wood. Thus, prediction models are usually developed for biodiversity assessments.
However, some of the important biodiversity indicators that are required for building prediction models could be measured by using high resolution remote sensing data collected with UAVs. This information could supplement or replace part of the currently required field work to support biodiversity assessments.
How remote sensing works when measuring biodiversity
Remote sensing measures the energy that is reflected and emitted from the earth’s surface. Because the properties of materials commonly found on the surface (e.g., plants and soils) are known, remote sensing provides insight into the surface composition. There are also biodiversity-relevant situations which may not be directly observable with remote sensing but may be correlated through direct observation. This allows remotely sensed observation to act as a “proxy” for surface activities if sufficient surface measurements are available to establish the link. For example, sea surface height can be measured and is correlated with upwelling and therefore with higher nutrient concentrations that affect the ecosystem in a variety of important ways.
Remote Sensing and Biodiversity Indices
Assessment of biodiversity at local and regional scales has traditionally relied on the assessment of both local diversity (alpha-diversity) and species turnover (beta-diversity); the combination of these two measures leading to an estimate of the whole diversity of an area (Whittaker 1972; Lande 1996). A large number of indices have been used to estimate alpha-diversity (e.g. species richness, Simpson, berger-Parker, Shannon-Wiener, Brillouin, McIntosh, Pielou Indices, Table 1). Species turnover is generally assessed using information on species compositional distance measures among sampling units, and expressed using a measure such as the Sørensen index or the Jaccard index.
Species monitoring in relatively large area has always been a challenging task for ecologists, mainly because of the intrinsic difficulty in evaluating the completeness of the resulting species lists and in quantifying sampling effort. Additionally, ground surveys are time consuming and costly. Moreover, in many biodiversity-rich location, field surveys can be risky due to challenging environmental and socio-political condition.
Remote sensing technology covers large regions in a short period of time. It represents a powerful opportunity for ecologists to gain critical knowledge about the drivers of the spatial and temporal distribution of biodiversity (Rocchini et al. 2005). The relationship between spectral variability over space and species diversity might be of great importance for maximizing the inventory of species diversity giving priority to sites which are spectrally more different, hence more diverse in species composition.
Most of the research dealing with remote sensing-based estimates of species diversity has focused on mapping localized biodiversity hotspots (Tucker et al. 2004), based on the Spectral Variation Hypothesis (SVH, Gould 2000; Palmer et al. 2002; Rocchini 2007). Furthermore, the use of common spectral indices, such as the normalized difference vegetation index (NDVI) in some studies, have demonstrated an increase in the strength of relationship between species alpha-diversity and remotely sensed spectral heterogeneity when using additional spectral information (e.g. Landsat 7 shortwave IR-band 5, from 155 to 175 nm and band 7, from 209 to 235 nm (Rocchini, 2007; Nagendra et al, 2010).
In addition to the importance of having the correct measure or spectral band/index for relating spectral and species diversity at local scale (alpha-diversity), different species diversity measures can lead to differences in the type and strength of the relationship between spectral and species diversity. For example, Oldeland et al. (2010); dealing with plan species diversity in African savannas, relied on relative abundances of species, as measured by the Shannon index to quantify the difference in the relative proportion of each species. They demonstrated that accounting for species relative abundances improves the capability of local species diversity estimation with hyperspectral remotely sensed data, with R2 values obtained up to five times higher than those achieved by only considering species richness (R2 values of 0.62 and 0.12, respectively). This is mainly because the Shannon index is less affected than species richness by the presence of rare species, which represents a relatively incidental set of species of more ‘disperse’ origin. (Riccota et al. 2008)
While alpha-diversity is related to local variability, species turnover (beta-diversity) is a crucial parameter when trying to identify high biodiversity areas (Baselga 2013). In fact, for a given level of local species richness, high beta-diversity indeed leads to high global diversity of the area. This is one of the basic rules underpinning the concept of irreplaceability of protected areas (e.g. Wegmann et al. 2014). In general, beta-diversity is assessed by plotting the compositional similarity among sites measured in the field versus their spatial distance. The higher the slope of the resulting curve, the higher the beta-diversity of the area. In others terms, the higher the decay in the similarity of species composition. Therefore, it is expected that species turnover should increase with increasing spatial extent. The curvilinear nature of this relationship, however, means in practice that the validity of extrapolation will depend on the sampling effort, that is, the extent of field knowledge. (Ferreier et al. 2007)
Recent advances in biodiversity indices mapping are based on the processing of high spatial resolution imaging spectroscopy and use an original approach to test the validity of SVH for the estimation of alpha-diversity in tropical rainforests (Féret and Asner, 2014a). One original aspect of this method is that it takes advantage of both high spatial and spectral resolution to arbitrarily assign a ‘spectral species’ identity to each individual pixel of the image, using unsupervised clustering. It consecutively performs pixel inventories over all individual surface units of a given size across the image. The pixel size can be adapted depending on the spatial resolution of the image and expected patterns of biodiversity. This method is based on the hypothesis that species or groups of species can be identified across the landscape based on their spectral signature (Clark et al. 2005).
Féret and Asner developed a fully unsupervised method to process hyperspectral images acquired over various sites in Peruvian Amazonian rainforest. They successfully mapped spatial variations in species composition and Shannon diversity index for various sites in Peruvian Amazonian rainforest using a preliminary spectral species mapping derived from repetitive K-means clustering.
This method was compared to various other methods which were relying on SVH and proposed in the literature, and dramatically outperformed indicators such as variations in NDVI and mean distance from center. In the second study, Féret and Anser (2014b) analyzed variations in both alpha- and beta-diversity related to changes in microtopography derived from digital elevation models obtained with airborne LiDAR (Light Detection and Ranging) acquisitions. Furthermore, they proposed a way to take advantage of combination of imaging spectroscopy and LiDAR acquisitions in order to map biodiversity and relate the spatial variations in species composition to environmental and physical factors.
Novel approaches integrating multi sensor acquisitions can help to improve understanding of the various environmental, physical, climatic, and human factors influencing biodiversity, by monitoring spatial and temporal variations in species composition.
Adding a further confounding factor, the relationship of beta-diversity with environmental heterogeneity is also scale dependent, perhaps even more than alpha-diversity. Areas of ecological transition, where the factors influencing patterns of biodiversity distribution change at different spatial scales, represent therefore a particular challenge for field monitoring. Yet, it is in precisely such areas where remote sensing may be especially helpful, enabling swift and easy computation of proxies of vegetation heterogeneity at different spatial scales, to generate hypotheses about scales at which such ecological may be taking place. For instance, Mairota et al. (2015) found differences in models of the association between remotely sensed values and biodiversity across scales, with plant diversity being most appropriately measured at the patch scale, while bird and insect diversity showed stronger associations with remotely sensed variables at the landscape and plot level respectively.
Conclusion
Regen Network can provide a useful framework for improving potential of remote sensing for predicting and monitoring species diversity. Remote sensing-based analysis needs to be conducted at multiple spatial scales using approaches such as texture analysis at different window sizes, moving windows, and/or pixel aggregation, to assess the scale most suitable for biodiversity monitoring of specific taxa, in specific contexts (Mairota et al. 2015). Further sensitivity studies on environmental parameters derived from remote sensing for biodiversity mapping need to be undertaken to understand the pitfalls and impacts of different data collection processes and models. Such information is crucial for a continuous global biodiversity analysis and an improved understanding of the impact of an ecological change of state.
References
Baselga, A. (2013, 01). Multiple site dissimilarity quantifies compositional heterogeneity among several sites, while average pairwise dissimilarity may be misleading. Ecography, 36(2), 124–128. doi:10.1111/j.1600–0587.2012.00124.x
Berger, W. H., and F. L. Parker. 1970. Diversity of planktonic foraminifera in deep-sea sediments. Science 168, 1345–1347.
Clark, M., Roberts, D., & Clark, D. (2005, 06). Hyperspectral discrimination of tropical rain forest tree species at leaf to crown scales. Remote Sensing of Environment, 96(3–4), 375–398. doi:10.1016/j.rse.2005.03.009
Féret, J., & Asner, G. P. (2014a, 09). Mapping tropical forest canopy diversity using high-fidelity imaging spectroscopy. Ecological Applications, 24(6), 1289–1296. doi:10.1890/13–1824.1
Féret, J., & Asner, G. P. (2014b, 09). Microtopographic controls on lowland Amazonian canopy diversity from imaging spectroscopy. Ecological Applications, 24(6), 1297–1310. doi:10.1890/13–1896.1
Gould, W. (2000, 12). Remote Sensing Of Vegetation, Plant Species Richness, And Regional Biodiversity Hotspots. Ecological Applications, 10(6), 1861–1870. doi:10.1890/1051–0761(2000)010[1861:rsovps]2.0.co;2
Jaccard, P. (1912, 02). The Distribution Of The Flora In The Alpine Zone.1. New Phytologist, 11(2), 37–50. doi:10.1111/j.1469–8137.1912.tb05611.x
Jordan, C. F. (1969, 07). Derivation of Leaf-Area Index from Quality of Light on the Forest Floor. Ecology, 50(4), 663–666. doi:10.2307/1936256
Lande, R. (1996, 05). Statistics and Partitioning of Species Diversity, and Similarity among Multiple Communities. Oikos, 76(1), 5. doi:10.2307/3545743
Lennon, J. J., Koleff, P., Greenwood, J. J., & Gaston, K. J. (2001, 11). The geographical structure of British bird distributions: Diversity, spatial turnover and scale. Journal of Animal Ecology, 70(6), 966–979. doi:10.1046/j.0021–8790.2001.00563.x
Mairota, P., Cafarelli, B., Labadessa, R., Lovergine, F., Tarantino, C., Lucas, R. M., . . . Didham, R. K. (2015, 05). Very high resolution Earth observation features for monitoring plant and animal community structure across multiple spatial scales in protected areas. International Journal of Applied Earth Observation and Geoinformation, 37, 100–105. doi:10.1016/j.jag.2014.09.015
Maurer, B. A., and B. J. McGill. 2011. Measurement of species diversity in A. E. Magurran and B. J. McGill, eds. Biological diversity: frontiers in measurement and assessment. Oxford University Press.
Mcintosh, R. P. (1967, 05). An Index of Diversity and the Relation of Certain Concepts to Diversity. Ecology, 48(3), 392–404. doi:10.2307/1932674
Nagendra, H., Rocchini, D., Ghate, R., Sharma, B., & Pareeth, S. (2010, February 02). Assessing Plant Diversity in a Dry Tropical Forest: Comparing the Utility of Landsat and Ikonos Satellite Images. Retrieved from https://doi.org/10.3390/rs2020478
Oldeland, J., W. Dorigo, L. Lieckfeld, A. Lucieer, and N. Ju¨rgens. 2010. Combining vegetation indices, constrained ordination and fuzzy classification for mapping semi-natural vegetation units from hyperspectral imagery. Remote Sens. Environ. 114, 1155–1166.
Palmer, M. W., Earls, P. G., Hoagland, B. W., White, P. S., & Wohlgemuth, T. (2002). Quantitative tools for perfecting species lists. Environmetrics, 13(2), 121–137. doi:10.1002/env.516
Pielou, E. (1966, 12). The measurement of diversity in different types of biological collections. Journal of Theoretical Biology, 13, 131–144. doi:10.1016/0022–5193(66)90013–0
Rocchini, D., Butini, S. A., & Chiarucci, A. (2005, 08). Maximizing plant species inventory efficiency by means of remotely sensed spectral distances. Global Ecology and Biogeography, 14(5), 431–437. doi:10.1111/j.1466–822x.2005.00169.x
Rocchini, D., C. Ricotta, and A. Chiarucci. 2007. Using satellite imagery to assess plant species richness: the role of multispectral systems. Appl. Veg. Sci. 10, 325–331.
Shannon, C. E., and W. Weaver. 1948. A mathematical theory of communication. Bell Syst. Tech. J. 27, 623–656.
Simpson, E. H. 1949. Measurement of diversity. Nature 163,688.
Tucker, C. J., Grant, D. M., & Dykstra, J. D. (2004, 03). NASA’s Global Orthorectified Landsat Data Set. Photogrammetric Engineering & Remote Sensing, 70(3), 313–322. doi:10.14358/pers.70.3.313
Wegmann, M., Santini, L., Leutner, B., Safi, K., Rocchini, D., Bevanda, M., . . . Rondinini, C. (2014, 04). Role of African protected areas in maintaining connectivity for large mammals. Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1643), 20130193–20130193. doi:10.1098/rstb.2013.0193
Whittaker, R. H. (1972, 05). Evolution and Measurement of Species Diversity. Taxon, 21(2/3), 213. doi:10.2307/1218190
Wilson, M. V., & Shmida, A. (1984, 11). Measuring Beta Diversity with Presence-Absence Data. The Journal of Ecology, 72(3), 1055. doi:10.2307/2259551
