Contrasting Performance of Lidar and Optical Texture Models in Predicting Avian Diversity

Contrasting performance of Lidar and optical texture models in predicting avian diversity in a tropical mountain forest

Christine I. B. Wallis, Detlev Paulsch, Jörg Zeilinger, Brenner Silva, Giulia F. Curatola Fernández, Roland Brandl, Nina Farwig, Jörg Bendix

1.  Forest habitat change

Figure S1: Forest loss between 2001-2014 for the study area derived from Landsat forest classifications (Hansen et al., 2013); classification data has been downloaded as “year of gross forest cover loss event” for the granule with top-left corner at 0N, 80W (Source: Hansen/UMD/Google/USGS/NASA). Rectangular buffers relate to the extent of the textural approach of optical and Lidar metrics. Forest loss within the Reserva Biológica San Francisco and the Podocarpus National Park was probably caused by landslide events which are very common in this region. Forest loss next to non-forest areas was probably caused by slash-and-burn events. Only one bird point count was affected by forest loss in 2011. Since Lidar data was acquired in 2012 and this point was not an outlier within the Lidar models, we conclude that habitat change had only a minor impact on our Lidar models.

2.  Bird diversity proxies

Table S1. Explained variance of the first four principal components of the PCA of bird abundance matrix. Each principal component represents the distribution of bird species along its ordination axis. We chose the first principal component as a proxy for a typical bird assemblage in our study area because it explains most variance in community composition.

Table S2. Loadings of first principal component based on the covariance matrix.

1

1

Figure S2: Scatterplot matrix of diversity proxies. Correlations are based on the Pearson correlation moment.

3.  Bird diversity models

Figure S3: Response plots of all six models fitted by either Lidar or optical texture metrics. A leave-one-out validation was performed.

1

Table S4: Predictors for each diversity measure selected owing to filtering in backward selection. For each predictor the base layer of the texture approach, the texture algorithm, and the window size applied for calculating the image texture is given. Window sizes define the surrounding of each pixel in which image textures were calculated. Given window sizes refer to the pixel size of the sensor. ** Both window sizes were selected

4.  Benefit of multi-sensor models

Other research areas, such as carbon stock estimations and forest stand inventories, also benefit from the combined use of different sensor data (Baccini et al., 2012; Kellndorfer et al., 2010; Saatchi et al., 2008; Walker et al., 2007). Some studies have explored the beneficial use of coupling image textures with elevation or habitat type (St-Louis et al., 2009, 2006), but to our knowledge, no study to date has explored the benefit of multi-sensor models fitted by combined Lidar and texture metrics to address different aspects of species diversity. Elevation metrics in addition to optical-based metrics might be beneficial for modeling bird species richness (Sheeren et al. 2014).

Consequently, we tested for all bird diversity proxies (a) the benefit of models fitted by textural image metrics and complex Lidar metrics and (b) the benefit of models fitted with elevation and slope in addition to texture metrics of optical images. We only found a benefit among bird community models, where Lidar and texture metrics together yielded a LOO R² of 0.85 and texture models with elevation and slope in addition yielded a LOO R² of 0.81. All other models did not perform better than the best single-sensor models based on a comparison of LOO R² and RMSE values.

References

Baccini, A., Goetz, S. J., Walker, W. S., Laporte, N. T., Sun, M., Sulla-Menashe, D., … Houghton, R. A. (2012). Estimated carbon dioxide emissions from tropical deforestation improved by carbon-density maps. Nature Climate Change, 2(3), 182–185.

Hansen, M. C., Potapov, P. V., Moore, R., Hancher, M., Turubanova, S. A., Tyukavina, A., … Townshend, J. R. G. (2013). High-resolution global maps of 21st-century forest cover change. Science, 342(6160), 850–853.

Kellndorfer, J. M., Walker, W. S., LaPoint, E., Kirsch, K., Bishop, J., & Fiske, G. (2010). Statistical fusion of lidar, InSAR, and optical remote sensing data for forest stand height characterization: A regional-scale method based on LVIS, SRTM, Landsat ETM+, and ancillary data sets. Journal of Geophysical Research: Biogeosciences, 115(G2), G00E08.

Saatchi, S., Buermann, W., ter Steege, H., Mori, S., & Smith, T. B. (2008). Modeling distribution of Amazonian tree species and diversity using remote sensing measurements. Remote Sensing of Environment, 112(5), 2000–2017.

Sheeren, D., Bonthoux, S., & Balent, G. (2014). Modeling bird communities using unclassified remote sensing imagery: effects of the spatial resolution and data period. Ecological Indicators, 43, 69–82.

St-Louis, V., Pidgeon, A. M., Clayton, M. K., Locke, B. A., Bash, D., & Radeloff, V. C. (2009). Satellite image texture and a vegetation index predict avian biodiversity in the Chihuahuan Desert of New Mexico. Ecography, 32(3), 468–480.

St-Louis, V., Pidgeon, A. M., Radeloff, V. C., Hawbaker, T. J., & Clayton, M. K. (2006). High-resolution image texture as a predictor of bird species richness. Remote Sensing of Environment, 105(4), 299–312.

Stotz, D. F., Conservation International, & Field Museum of Natural History (Eds.). (1996). Neotropical birds: ecology and conservation. Chicago: University of Chicago Press.

Walker, W. S., Kellndorfer, J. M., LaPoint, E., Hoppus, M., & Westfall, J. (2007). An empirical InSAR-optical fusion approach to mapping vegetation canopy height. Remote Sensing of Environment, 109(4), 482–499.

1