Forest Fragmentation in the Amazon Basin: Dendritic Road Networks

This article dealt with fragmentation in Amazon basin, studied through the lens of road-building by private sector loggers.  The simulations within are based on micro-economic theory and can be considered as an example of computerized “thinking,” a way of gaining theoretical insights into empirical phenomena in the face of complexity and large amounts of data.

Forest loss drivers as environmental processes are well-studied from a social science perspective.  Human drivers of how deforestation patterns emerge, causally and spatially, are less studied.  The social sciences, including geography, need to say more about the causal origins of Amazonian fragmentation.  Improved modeling from GIS/geostatistics and empirical knowledge from fieldwork (in combination with theory about agent behavior and social process) need to complement landscape ecology in this task.  The Amazon is the world’s largest contiguous rainforest, and it is disappearing quickly.   The authors of this article take road networks as representative of the spatial manifestation of fragmentation, which landscape ecologists have long implicated in deforestation.  Loggers, as the creators of these roads, represent an important agent of land-cover change because they possess the financial means to build roads in areas of primary forest where the government is usually absent.

There are 3 aspects to the importance of roads in this context:

1)   Where infrastructure is expanding rapidly in high biodiversity areas.

2)   Roads may be abandoned as loggers move to new areas but spontaneous colonists follow them to claim lands and farm.

3)   Private sector, local (loggers) individuals’ roads are much more of an issue than federal roads

a.     The current policy requires a management plan on private holdings or the opening of new lands through colonization where colonists legally sell rights to loggers

                                               i.     This imposes costs/uncertainty on the sector so it targets illegal areas – policy failure

                                              ii.     These are mostly small operations which free-ride on the state road system

b.     Logging numbers are expected to increase with the Brazilian government’s move to a system of concession logging in national/state forests in an effort to decrease illegal logging

                                               i.     This is a rationalized use of national forest lands

                                              ii.     It will bring regulation to policed areas

                                            iii.     Increasing capacity in the region will intensify incursions into lands outside these areas – policy failure

c.      This situation is worsened by the increasing size/profitability of the Amazonian logging sector

Prior authors identified 6 major pattern types of spatial deforestation, with 3 specific types within the Amazon basin.  The unconstrained, irregular road networks studied in this paper resulted in a new type of pattern labeled “dendritic” as a result of its spatial similarity to hydrologic flow/stream patterns. 

GIS was previously used in transportation/network systems but not in network construction.  The authors discussed previous models of network construction and their limitations, created various algorithms (volume of timber surface, logging sites, road building), compared the simulation against remotely-sensed logging roads input into a rasterized true network, and determined its statistical error.  The model matches empirical observations at 90% when 70% of the study area is covered, but at smaller areas has less accuracy.  The authors identified 5 potential sources of error – 2 sites being modeled into one, centroids (the “patio,” or holding area for timber before being transported to the sawmill) as being in the wrong location, river streams not being considered correctly due to improper map usage, incomplete logger information (“mateiros,” or local timber guides provide and loggers decide if it’s profitable, moving space to space in a piecemeal process), and the assumption of only one agent with one point of entry (the observed unconnected segments were modeled as due to one agent, whereas “territorialization” of space by loggers is common). 

The authors conclude 2 things:

1)   The premier future issue in the spatial modeling of Amazonian road networks is the incorporation of territorialization into modeling.

2)   The incorporation of soil type, vegetation, watercourses into road parametrization of build cost beyond simple height topography is a future issue.

Arima, Y. Eugenio, et al.  (2008).  The Fragmentation of Space in the Amazon Basin: Emergent Road Networks.  Photogramattic Engineering and Remote Sensing, 74, 699-709.