Faculty of Mathematics and Natural Sciences - Biogeography

Faculty of Mathematics and Natural Sciences | Geography Department | Biogeography | Projects | Land-use & biodiversity Chaco | Land-use and biodiversity in the Chaco - Work package Description

Land-use and biodiversity in the Chaco - Work package Description

 

Work package 1: Image compositing

Work package 2: Reconstructing land system change

Work package 3: Bird community survey

Work package 4: Land use effects on bird communities

Work package 5: Trade-offs between agriculture and biodiversity

 

 

Work package 1: Image compositing

Rational – Satellite images have been used to map land use change in the Argentine Chaco in several studies (e.g., Steininger et al. 2001; Grau et al. 2005; Boletta et al. 2006; Zak et al. 2008; Gasparri et al. 2013, Hansen et al. 2013, Clark et al. 2010). The recent opening of the Landsat image archives, with images at relatively fine spatial resolution (30m), provides novel and exciting opportunities to reliably map many land use and land cover change (LUCC) processes, including logging or agricultural abandonment.

Analyses – As a basis for mapping land use and land cover change in the Argentine Chaco, we will generate Landsat image composites for the study region. Pixel-based image compositing (PCB) generates a single, homogeneous, and cloud-free image dataset by mosaicking image observations on a per-pixel basis from the entire pool of available images. Observations are selected based on a number of user-defined rules (e.g., acquisition year, acquisition day, distance to clouds) (Griffiths et al. 2012b; Griffiths et al. 2013a). We will derive early/mid-summer image composites (i.e., matched with the timing of the bird community surveys in WP3) for the focal years 1990, 2000, 2010, and 2015. The 2015 composite will use data from the Landsat 8 satellite. For a given target year, image acquisitions +/- 2 years from that year will be considered. Multiple images from within a vegetation period and from different years are beneficial to distinguish between different land use systems (Kuemmerle et al. 2008; Griffiths et al. 2012a; Prishchepov et al. 2012) and we will therefore also test the usefulness of wider acquisition windows (i.e., including autumn acquisitions). Moreover, we will derive three types of spectral variability metrics for each composite and spectral band: (1) variance, (2) coefficient of variation, and (3) range (10% and 90% percentiles to exclude outliers) for all available suitable observations in the Landsat archives. We have applied image compositing in a range of projects and a fully operational image processing chain for Landsat Pixel-Based Composites (PBC) is implemented at the Geography Department of HU Berlin.

Outcomes – This work package will make full use of the Landsat image archives to generate pixel-based image composites, which will form the basis for the LUCC mapping in WP2. The specific outcomes from this work package will be:

  • Landsat image composites for the years 1990, 2000, 2010, and 2015, and
  • Indices of spectral variability for each of these composites.

 

Work package 2: Reconstructing land system change

Rational - The extent and spatial patterns of LUCC processes in the Chaco like deforestation caused by agricultural expansion, the shift from grazing to agriculture, the intensification of grazing systems, and agricultural abandonment remain largely unquantified (Gasparri et al. 2008; Gavier-Pizarro et al. 2012a; Macchi and Grau 2012; Pelegrin and Bucher 2012).. Using the image composites from WP1, as well as and ancillary information, we propose to reconstruct fine-scale land use histories for the entire Argentine Chaco.

Analyses – We will conduct a wall-to-wall mapping of conversions between (1) three classes of natural vegetation, (2) low-intensity traditional grazing systems, (3) silvipastural grazing systems, (4) intensified ranching, and (5) agri-business cropping. We will first assess changes in land cover, and then combine these maps with ancillary land management data to map transitions between land use systems. To map changes in land cover, we will use the image composites from WP1 and employ automated change detection to separate the three natural vegetation classes, as well as the classes silvicultural grazing, intensified ranching and agri-business agriculture (mainly soybean farming). All LUCC maps will be rigorously validated using a stratified random sample of validation points. To translate the land use/cover maps from above into maps of land system change, we will take three steps. First, we will collate the classes forest, shrublands and grasslands into a natural vegetation class and distinguish fire disturbance from vegetation conversion to agriculture. For the latter, we will consider the shape of patches where forests are lost (agricultural fields are rectangular whereas fire scars are irregular) and use the MODIS burnt area and active fire products (http://modis-fire.umd.edu) to identify fire scars. Second, we will subdivide the natural vegetation class into a natural vegetation class without human land use and a traditional grazing system class. This will be done by assessing the spatial pattern of puestos (i.e., homesteads) for the entire Chaco in 1990, 2000, and 2010/15. Puestos are livestock concentration points and key indicators of grazing pressure (Macchi and Grau 2012). We will cross-check this layer against the image composites, and digitize emerging and abandoned puestos for each time period.  Third, in order to assess changes in stocking rates in silvicultural and ranching systems, we will obtain livestock statistics at the finest administrative level (i.e., departamentos, S~IGSA 2014) and use the areas of pasture and silvipastures from the land cover maps to calculate average stocking rates (using animal numbers from larger ,i.e., non-puestos farms only).

While the analyses outlined above do not allow for formal hypothesis testing, the following a-priori expectations about recent land use change in the study region can be assessed:

  1. Forest loss rates have accelerated substantially over the over the last 25 years.
  2. Intensive soybean farming and intensified cattle ranching have expanded drastically, mainly at the expanse of natural forests and low-intensity grazing systems.
  3. Much deforested land is initially used for intensive cattle ranching, but subsequently converted to intensive cropland (soybean farming).
  4. Where traditional gazing systems are abandoned, vegetation recovery takes place.

Outcomes – This work package will quantify recent land over change and transitions among land systems in the Argentine Chaco. The specific outcomes from this work package will be:

  • Maps of conversions in land cover based on Landsat image analyses for the years 1990, 2000, 2010, and 2015,
  • Maps of transitions between land systems, pertaining to natural ecosystems, traditional grazing, silvipastoral grazing, intensive ranching, and agri-business agriculture, and

Understanding of recent land system change in the Chaco.

 

Work package 3: Bird community survey

Rationale – The relationship of land use change and biodiversity in the Chaco remains poorly understood, mainly because primary biodiversity data collected for broad regions and across gradients of land use intensity is not available. In this work package, we will carry out a rapid bird community survey for a statistically representative sample of survey sites and we will derive bird community indicators pertaining to alpha and beta diversity, as well as focal species abundance.

Analyses – Birds are an excellent focal taxa for the purpose of this project, because they are commonly used surrogates to monitor biodiversity, particularly to study land use effects on biodiversity at landscape to regional scales (Donovan and Flather 2002; Meehan et al. 2010; Rittenhouse et al. 2010). This work package will carry out a rapid bird community survey to derive a range of community-level biodiversity indicators pertaining to alpha diversity, beta diversity and focal species population density. Specifically, this work package will deliver:

  • A database of bird community censuses for two subsequent years for the entire Chaco,
  • Indicators of alpha and beta diversity as well as focal species population density, and
  • Comparisons of these indicators across land use intensity and environmental gradients.
 
 

Rationale –  This work package will use the community-level indicators generated through the rapid bird community survey (WP3) to understand the relationship of avian biodiversity and land use change for effective conservation planning in the region.

Analyses – We will use regression models to assess the relative importance of a range of predictor variables and to test hypothesis explaining bird community variability across the Chaco. Further, we will use the land use and land cover maps from WP2 and derive forest fragmentation maps from them using morphological image segmentation (i.e., core, edge, islet, or matrix forest, Vogt et al. 2007). The effects of land use on wildlife populations can be substantially lagged, and contemporary biodiversity patterns may thus substantially depend on past land use/cover and changes therein (Foster et al. 2003; Wearn et al. 2012; Dullinger et al. 2013). To account for this, we will not only derive land use/cover variables for the contemporary situation, when bird surveys were implemented, but also for the last 20 years using the set of land change maps from WP2. In addition, we will generate variables capturing the timing and pace of habitat loss and habitat fragmentation. Including these variables in the regression models (see below) will thus allow to isolate the effect of land use legacies and the relative importance of contemporary vs. past land use patterns in determining bird communities in the Chaco (Gavier-Pizarro et al. 2012a). We have used all of the above approaches extensively in prior and ongoing work (e.g., Baumann et al. 2011; Sieber et al. 2013; Levers et al. 2014; St-Louis et al. 2014). Specifically, we will test five hypotheses:

  • H1: The species richness of taxa associated with natural vegetation declines with increasing land use intensity.
  • H2: Total species richness displays a hump-shaped relationship with land use intensity (i.e., as expected by the intermediate disturbance hypothesis).
  • H3: Focal species’ populations decline along a gradient of land use intensity.
  • H4: Land use and human disturbance variables are more important in determining the Chaco’s bird community patters than environmental variables.
  • H5: Past land use change have a significant effect on determining current bird communities.

This will yield predicted bird community maps for each time period covered by the land systems maps from WP1 and thus to allow assessing how habitat change affected biodiversity patterns in the Chaco. We will also compare the bird community maps to an expert-based assessment of high-conservation value regions by The Nature Conservancy (TNC 2005) to assess where recent land use changes (i.e., during the last 10 years) affected the high conservation value areas outlined there. Finally, we will compare the bird community maps to the recently drafted land use and conservation plan ‘Forest Law’ (Ley de Bosques), which will allow to identify high conservation value areas currently not considered for protection, and to understand how potential future land change (as planned in the Forest Law) will affect the region’s biodiversity.

Outcomes – This work package will carry out a rapid bird community survey to derive a range of community-level biodiversity indicators. These indicators will be used to understand the relationship of land use change and biodiversity patterns in the Chaco, and to identify past and potential future hotspots of threats to biodiversity. Specifically, this work package will deliver:

  • Models explaining patterns in bird diversity, evenness, and community structure, as a function of environmental, land use, and human disturbance variables,
  • Maps of bird diversity and community similarity for the region at a grain of 500m, and
  • Understanding of the past and potential future effects of land use change on bird communities in the Chaco.

 

 

Work package 5: Trade-offs between agriculture and biodiversity

 

Rationale – By understanding trade-offs between agricultural production and biodiversity, landscapes can be managed to promote both. Unfortunately, existing work has focused on small study regions, single species, or a limited number of land use types. This work package will build upon the results from WP1-4 to close this research gap by systematically exploring the trade-offs between agricultural intensity and biodiversity across the Argentine Chaco.

AnalysesThe assessment of trade-offs between agriculture and biodiversity will consist of two steps. First, we will link the bird diversity models (WP4) with a productivity model to identify trade-offs between biodiversity and agricultural production. Second, we will use simple optimization tools to identify the trade-offs between agricultural production and bird communities in the Chaco. Spatially explicit agricultural productivity metrics that are comparable across different land use systems can be joined with the bird diversity models from WP4 to map the trade-offs between bird communities and agricultural production. In a second step, we will use the formalized relationships of biodiversity indicators and agricultural productivity predictors to explore what characterizes landscapes that maximize biodiversity and agricultural production in the Chaco. To do so, we will use the newly developed software RobOff, which provides a rich toolkit to systematically analyze trade-offs between management alternatives (here: different land systems or conservation lands, each associated to different levels of agricultural production and biodiversity, Pouzols et al. 2012; Pouzols and Moilanen 2013).

The following a-priori expectations about trade-offs in the Chaco can be formulated:

  1. At the local level, trade-offs between agriculture and biodiversity are substantial and increase with increasing land use intensity
  2. At the level of the Chaco as a whole, existing trade-offs could be mitigated substantially via smart land use and conservation planning
  3. The current zoning law (Ley de Bosques) will only partly release synergies between agriculture and biodiversity conservation
  4. Different land use paradigms (e.g., sharing or sparing) can both lead to a better balance of agriculture and biodiversity

Outcomes – This work package will make use of the insights and datasets generated in WP1-4 to systematically assess the trade-offs between agriculture and biodiversity in the Chaco, providing the first, regional-scale assessment of its kind. Specifically, this work package will deliver:

  • Systematic assessment of the relationship between different agricultural production measures and the bird community indicators, and
  • Levels of different land uses that would minimize trade-offs between agriculture and biodiversity for a range of land use planning options.