Isabel M. D. Rosa

Predictive Modelling of Contagious Deforestation in the Brazilian Amazon

Tropical forests are diminishing in extent due primarily to the rapid expansion of agriculture, but the future magnitude and geographical distribution of future tropical deforestation is uncertain. Here, we introduce a dynamic and spatially-explicit model of deforestation that predicts the potential magnitude and spatial pattern of Amazon deforestation. Our model differs from previous models in three ways: (1) it is probabilistic and quantifies uncertainty around predictions and parameters; (2) the overall deforestation rate emerges “bottom up”, as the sum of local-scale deforestation driven by local processes; and (3) deforestation is contagious, such that local deforestation rate increases through time if adjacent locations are deforested. For the scenarios evaluated–pre- and post-PPCDAM (“Plano de Ação para Proteção e Controle do Desmatamento na Amazônia”)–the parameter estimates confirmed that forests near roads and already deforested areas are significantly more likely to be deforested in the near future and less likely in protected areas. Validation tests showed that our model correctly predicted the magnitude and spatial pattern of deforestation that accumulates over time, but that there is very high uncertainty surrounding the exact sequence in which pixels are deforested. The model predicts that under pre-PPCDAM (assuming no change in parameter values due to, for example, changes in government policy), annual deforestation rates would halve between 2050 compared to 2002, although this partly reflects reliance on a static map of the road network. Consistent with other models, under the pre-PPCDAM scenario, states in the south and east of the Brazilian Amazon have a high predicted probability of losing nearly all forest outside of protected areas by 2050. This pattern is less strong in the post-PPCDAM scenario. Contagious spread along roads and through areas lacking formal protection could allow deforestation to reach the core, which is currently experiencing low deforestation rates due to its isolation.

Changes in size of deforested patches in the Brazilian Amazon.

Different deforestation agents, such as small farmers and large agricultural businesses, create different spatial patterns of deforestation. We analyzed the proportion of deforestation associated with different-sized clearings in the Brazilian Amazon from 2002 through 2009. We used annual deforestation maps to determine total area deforested and the size distribution of deforested patches per year. The size distribution of deforested areas changed over time in a consistent, directional manner. Large clearings (>1000 ha) comprised progressively smaller amounts of total annual deforestation. The number of smaller clearings (6.25-50.00 ha) remained unchanged over time. Small clearings accounted for 73% of all deforestation in 2009, up from 30% in 2002, whereas the proportion of deforestation attributable to large clearings decreased from 13% to 3% between 2002 and 2009. Large clearings were concentrated in Mato Grosso, but also occurred in eastern Pará and in Rondônia. In 2002 large clearings accounted for 17%, 15%, and 10% of all deforestation in Mato Grosso, Pará, and Rondônia, respectively. Even in these states, where there is a highly developed agricultural business dominated by soybean production and cattle ranching, the proportional contribution of large clearings to total deforestation declined. By 2009 large clearings accounted for 2.5%, 3.5%, and 1% of all deforestation in Mato Grosso, Pará, and Rondônia, respectively. These changes in deforestation patch size are coincident with the implementation of new conservation policies by the Brazilian government, which suggests that these policies are not effectively reducing the number of small clearings in primary forest, whether these are caused by large landholders or smallholders, but have been more effective at reducing the frequency of larger clearings.

Thresholds of species loss in Amazonian deforestation frontier landscapes.

In the Brazilian Amazon, private land accounts for the majority of remaining native vegetation. Understanding how land-use change affects the composition and distribution of biodiversity in farmlands is critical for improving conservation strategies in the face of rapid agricultural expansion. Working across an area exceeding 3 million ha in the southwestern state of Rondônia, we assessed how the extent and configuration of remnant forest in replicate 10,000-ha landscapes has affected the occurrence of a suite of Amazonian mammals and birds. In each of 31 landscapes, we used field sampling and semistructured interviews with landowners to determine the presence of 28 large and medium sized mammals and birds, as well as a further 7 understory birds. We then combined results of field surveys and interviews with a probabilistic model of deforestation. We found strong evidence for a threshold response of sampled biodiversity to landscape level forest cover; landscapes with <30-40% forest cover hosted markedly fewer species. Results from field surveys and interviews yielded similar thresholds. These results imply that in partially deforested landscapes many species are susceptible to extirpation following relatively small additional reductions in forest area. In the model of deforestation by 2030 the number of 10,000-ha landscapes under a conservative threshold of 43% forest cover almost doubled, such that only 22% of landscapes would likely to be able to sustain at least 75% of the 35 focal species we sampled. Brazilian law requires rural property owners in the Amazon to retain 80% forest cover, although this is rarely achieved. Prioritizing efforts to ensure that entire landscapes, rather than individual farms, retain at least 50% forest cover may help safeguard native biodiversity in private forest reserves in the Amazon.

Using landscape history to predict biodiversity patterns in fragmented landscapes

Landscape ecology plays a vital role in understanding the impacts of land-use change on biodiversity, but it is not a predictive discipline, lacking theoretical models that quantitatively predict biodiversity patterns from first principles. Here, we draw heavily on ideas from phylogenetics to fill this gap, basing our approach on the insight that habitat fragments have a shared history. We develop a landscape ‘terrageny’, which represents the historical spatial separation of habitat fragments in the same way that a phylogeny represents evolutionary divergence among species. Combining a random sampling model with a terrageny generates numerical predictions about the expected proportion of species shared between any two fragments, the locations of locally endemic species, and the number of species that have been driven locally extinct. The model predicts that community similarity declines with terragenetic distance, and that local endemics are more likely to be found in terragenetically distinctive fragments than in large fragments. We derive equations to quantify the variance around predictions, and show that ignoring the spatial structure of fragmented landscapes leads to over-estimates of local extinction rates at the landscape scale. We argue that ignoring the shared history of habitat fragments limits our ability to understand biodiversity changes in human-modified landscapes.

The transparency, reliability and utility of tropical rainforest land‐use and land‐cover change models

Land-use and land-cover (LULC) change is one of the largest drivers of biodiversity loss and carbon emissions globally. We use the tropical rainforests of the Amazon, the Congo basin and South-East Asia as a case study to investigate spatial predictive models of LULC change. Current predictions differ in their modelling approaches, are highly variable and often poorly validated. We carried out a quantitative review of 48 modelling methodologies, considering model spatio-temporal scales, inputs, calibration and validation methods. In addition, we requested model outputs from each of the models reviewed and carried out a quantitative assessment of model performance for tropical LULC predictions in the Brazilian Amazon. We highlight existing shortfalls in the discipline and uncover three key points that need addressing to improve the transparency, reliability and utility of tropical LULC change models: (1) a lack of openness with regard to describing and making available the model inputs and model code; (2) the difficulties of conducting appropriate model validations; and (3) the difficulty that users of tropical LULC models face in obtaining the model predictions to help inform their own analyses and policy decisions. We further draw comparisons between tropical LULC change models in the tropics and the modelling approaches and paradigms in other disciplines, and suggest that recent changes in the climate change and species distribution modelling communities may provide a pathway that tropical LULC change modellers may emulate to further improve the discipline. Climate change models have exerted considerable influence over public perceptions of climate change and now impact policy decisions at all political levels. We suggest that tropical LULC change models have an equally high potential to influence public opinion and impact the development of land-use policies based on plausible future scenarios, but, to do that reliably may require further improvements in the discipline.

The Environmental Legacy of Modern Tropical Deforestation

Summary Tropical deforestation has caused a significant share of carbon emissions and species losses, but historical patterns have rarely been explicitly considered when estimating these impacts [1]. A deforestation event today leads to a time-delayed future release of carbon, from the eventual decay either of forest products or of slash left at the site [2]. Similarly, deforestation often does not result in the immediate loss of species, and communities may exhibit a process of “relaxation” to their new equilibrium over time [3]. We used a spatially explicit land cover change model [4] to reconstruct the annual rates and spatial patterns of tropical deforestation that occurred between 1950 and 2009 in the Amazon, in the Congo Basin, and across Southeast Asia. Using these patterns, we estimated the resulting gross vegetation carbon emissions [2, 5] and species losses over time [6]. Importantly, we accounted for the time lags inherent in both the release of carbon and the extinction of species. We show that even if deforestation had completely halted in 2010, time lags ensured there would still be a carbon emissions debt of at least 8.6 petagrams, equivalent to 5–10 years of global deforestation, and an extinction debt of more than 140 bird, mammal, and amphibian forest-specific species, which if paid, would increase the number of 20th-century extinctions in these groups by 120%. Given the magnitude of these debts, commitments to reduce emissions and biodiversity loss are unlikely to be realized without specific actions that directly address this damaging environmental legacy.

Spatial and temporal dimensions of landscape fragmentation across the Brazilian Amazon

The Brazilian Amazon in the past decades has been suffering severe landscape alteration, mainly due to anthropogenic activities, such as road building and land clearing for agriculture. Using a high-resolution time series of land cover maps (classified as mature forest, non-forest, secondary forest) spanning from 1984 through 2011, and four uncorrelated fragmentation metrics (edge density, clumpiness index, area-weighted mean patch size and shape index), we examined the temporal and spatial dynamics of forest fragmentation in three study areas across the Brazilian Amazon (Manaus, Santarém and Machadinho d’Oeste), inside and outside conservation units. Moreover, we compared the impacts on the landscape of: (1) different land uses (e.g. cattle ranching, crop production), (2) occupation processes (spontaneous vs. planned settlements) and (3) implementation of conservation units. By 2010/2011, municipalities located along the Arc of Deforestation had more than 55% of the remaining mature forest strictly confined to conservation units. Further, the planned settlement showed a higher rate of forest loss, a more persistent increase in deforested areas and a higher relative incidence of deforestation inside conservation units. Distinct agricultural activities did not lead to significantly different landscape structures; the accessibility of the municipality showed greater influence in the degree of degradation of the landscapes. Even with a high proportion of the landscapes covered by conservation units, which showed a strong inhibitory effect on forest fragmentation, we show that dynamic agriculturally driven economic activities, in municipalities with extensive road development, led to more regularly shaped, heavily fragmented landscapes, with higher densities of forest edge.