Sustainability of reduced-impact logging in the Eastern Amazon Plinio Sist a, * , Fabricio Nascimento Ferreira b a Cirad (Forest Department, UPR37) Convenio Cirad-Embrapa, CENARGEN, Embrapa Recursos Gene ´ticos e Biotecnologia, Brazil b UFRA, Bele ´m, Para ´, Brazil Received 13 September 2006; received in revised form 10 February 2007; accepted 13 February 2007 Abstract Damage caused by reduced-impact logging (RIL) was assessed in 18 plots 1 ha each in a terra firme rain forest of Eastern Amazon (Brazil, Paragominas). Mean logging intensity was 6 trees ha 1 and the resulting commercial volume 21 m 3 ha 1 . On average, logging damage affected 16% of the original stand while skidtrails occupied 7% (661 m 2 ha 1 ) of forest soil area. Canopy openness doubled to a mean of 11%. Of the variables studied, ‘‘number of trees harvested or felled per plot’’ gave the best correlation to ‘‘proportion of damaged or destroyed trees’’. Damage to each diameter class was distributed in accordance with relative abundance of trees (dbh  20 cm) in the original population before logging, suggesting that all diameter classes were affected equally. The sustainability of timber management applying RIL was evaluated through the calculation of the recovery level of commercial trees in three different scenarios. In the most optimistic scenario (growth rate of 5 mm year 1 and 1% annual mortality), after 30 years, only 50% of the commercial stand would recover, provoking a drastic reduction of the harvesting intensity at the second felling cycle. Within a 30-year felling cycle (i.e. the legal felling cycle duration in the Brazilian Amazon) and even under RIL systems, the present logging intensity occurring in the study area (6 trees ha 1 ) is not compatible with sustainable yield production on a long-term basis. For the study area, only the implementation of silvicultural treatment ensuring that the remaining potential crop trees grow at 4–5 mm year 1 would guarantee a logging intensity of 3–4 trees ha 1 (10– 14 m 3 ha 1 ) 40 years after the first harvest. This study showed that in the Amazon, RIL alone is clearly not sufficient to achieve sustainable forest management. More sophisticated silvicultural systems must be urgently elaborated and implemented to ensure that the forest will still be sustainably managed on a long-term basis. This issue is particularly important in the case of Brazil as a new law allowing the creation of 500,000 km 2 of forest concessions by 2010 has been recently approved by the congress. # 2007 Elsevier B.V. All rights reserved. Keywords: Reduced-impact logging; Amazonian rainforest management; Tropical silviculture 1. Introduction The conservation potential of managed production tropical forests has promoted the implementation of timber harvesting practices generally referred to as ‘low-impact’ or ‘reduced- impact’ logging (RIL). These techniques act at the operational plan level by planning skidtrails, practicing carefully controlled felling and skidding and reducing damage to soils and residual trees (Dykstra and Heinrich, 1996; Sist, 2000). Numerous studies in the three major tropical rainforests (Latin America, Central Africa, South East Asia) have demonstrated that, under moderate logging intensity, these techniques can reduce the damage on the residual stand and soil by 50% (Dykstra and Heinrich, 1996; Sist, 2000). Unfortunately, in practice, most logging operations occurring in tropical rainforest still remain unplanned and very destructive to the stand. This unplanned logging is generally called predatory or conventional logging (CL). The hypothesis of RIL is that the degree to which biological diversity and ecosystem functions are retained is directly correlated to the degree to which the harvested forest retains its state immediately prior to harvesting (Fredericksen and Putz, 2003). For example, if gap-phase dynamics typify a forest’s disturbance regime prior to logging, then logging should mimic this process. However, it is unrealistic to pretend that RIL techniques will mimic natural disturbance as this operation, even when controlled, remains one of the most damaging interventions in the forests, if compared to natural disturbances (Sist and Brown, 2004). All experiments which compared RIL and CL clearly demonstrated that the www.elsevier.com/locate/foreco Forest Ecology and Management 243 (2007) 199–209 * Corresponding author. Tel.: +55 61 3448 4913; fax: +55 61 3448 4915. E-mail addresses: sist@cirad.fr (P. Sist), fabricio.nascimento@ufra.edu.br (F.N. Ferreira). 0378-1127/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2007.02.014