Land-cover changes during the past 50 years in the semi-arid tropical forest region of northern Peru

Abstract

Dry forests are under much higher pressure from deforestation than humid forests, yet less attention has been paid to these forests. To develop better management strategies, it is necessary to analyse the dynamics of these forests at different scales and to assess their potential to recover from natural or human disturbances. This study combines broad and fine scales to investigate temporal changes: at broad scale we use aerial photos (from 1949, 1962, and 1999) and GIS to document land-cover change during a long-term period of 50 years; at fine scale we use spatial gradient analyses to document changes in species composition during a short-term period of four years. Re-sampling of the same elevational gradient (100–460 m a.s.l.) provides both a spatial and temporal gradient. Plausible causal links between human impact and documented changes are evaluated by means of ordination and Monte Carlo permutation tests. Density and diversity of seedlings, juveniles, shrubs, and trees were investigated in permanent plots from 2003 to 2007 along an elevational gradient. Approximately 17% of the forest has been lost during the last 50 years; giving an annual forest decline of 0.33%. The forest degradation during the 50 years in northern Peru was mainly due to agricultural expansion. During 1949–1962 forest declined by only 1.7%, while a greater forest decline took place during 1962–1999 (15% decline). Reasons for the greater forest loss during 1962-1999 are discussed. A new agrarian land reform from 1969 and population increases are likely reasons. The gradient analyses show a clear increase in species richness and spatial turnover in species composition along the elevational gradient. The ordination analysis reveals a significant change in species composition from 2001 to 2005, and we hypothesise that this may relate to browsing and logging activities. Although the number of faeces and signs of logging did not have a significant correlation with the ordination axes, the Monte Carlo permutation test in redundancy analysis (RDA) reveals a significant effect of browsing on species composition. The changes in species richness and species composition with elevation may be due to climatic differences between the lowland and upland mountain study area, rather than different intensity of human land use. Browsing and logging activities take place quite evenly over the entire study area, rather than being restricted to a certain area. Re-sampling of the permanent plots reveals that the lowland species Prosopis pallida was the only species able to survive from seedling stage to juvenile stage during the four-year investigation period. About 30% of the Prosopis pallida seedlings developed into juveniles, which indicates that this species may be better adapted to drought than other species in this environment.