What makes a grassland? The saga continues

This week’s issue of Science revisited the debate over the grassland-forest conservation tensions I examined earlier. If you recall, William Bond’s Perspective from January 8, “Ancient grasslands at risk,” argued that many grasslands that previous efforts have identified as the result of deforestation may actually be ancient ecosystems, preserved by fire and herbivory. Therefore, he claims, efforts to reforest grasslands may destroy these unique ecosystems.

The authors of one of the studies that Bond criticizes, the Atlas of Forest and Landscape Restoration Opportunities, responded with a letter in which they assert that the Forest and Landscape Restoration (FLR) project has no intention of converting ancient grasslands into forests in the name of forest conservation and that “A fundamental first step in the FLR process involves understanding the ecosystem at the landscape level, including its historical and cultural values, before making decisions on a restoration approach. FLR under this definition would promote exactly the same conservation of ancient grasslands that Bond advocates.”

This certainly sounds reasonable, but Bond replies in the same issue: “the western Serengeti and Kruger National Park are mapped as deforested instead of as ancient savannahs, and the natural species-rich montane grasslands of South Africa and southern Brazil are both mapped as deforested and/or degraded” in the Atlas of Forest and Landscape Restoration Opportunities (as, indeed, they are).

I am fascinated by this debate because it reveals a whole bunch of unresolved tensions in the fundamental principles of landscape science. Some I mentioned in my previous post: how should we value certain landscapes in order to apply conservation projects to them? Another I hinted at in my post about state-and-transition models: are these distinct landscape states such as “grassland” or “forest” meaningful? The Atlas authors mention that mosaics of forests and grasslands confound their attempts to classify, especially since the definition of forests that they use is not mutually exclusive with the definition of grassland that Bond uses.

Bond and the Atlas authors advocate their variants on the classification paradigm because it advances their interests: the preservation and continued study of grassland ecosystems on one hand and the reforestation of vast tracts of the globe on the other. My own interests in landscape dynamics push me towards an opinion that this whole project of classification is not incredibly useful. Rather than attempts to enumerate all the things you can see at a certain scale, leading to these ambiguous definitions of landscape states, I prefer approaches which build a representation of the environment at multiple scales, capturing not just the variability between different ecosystems but also the incredible amount of heterogeneity within a single ecosystem. I think it’s still an open question though how best to design environmental interventions which are aware of multiscale phenomena, and that’s certainly something I would like to know more about.