A multi-scale assessment of hurricane impacts on agricultural landscapes based on land use and topographic features

Agricultural systems are increasingly vulnerable to the effects of extreme climate events. Yet strategies to reduce risk and vulnerability have not been greatly explored. Here, we examine the vulnerability of coffee agroforestry systems varying in management intensity (e.g. land use) and topographic features to disturbance related to Hurricane Stan in Chiapas, Mexico—a hurricane categorized by heavy rains andmild winds.Anapproximately50 kmareawaschosenwithinacoffee-growingregionwheredatawerecollected onavarietyof topographic and landscape features (aspect, slope, elevation,distance to river) andvegetation characteristics (canopy cover, vegetation structure, tree density) as predictive factors of vegetation, economic, and landslide damage at three distinct spatial scales. At the plot level, we collected vegetation data later compiled into a vegetation complexity index. At the farm level, we collected data to understand the effect of the hurricane on economic damage and farm area affected by landslides. We also recorded number and volume of roadside landslides as ameasure of post-hurricane disturbance.We then conducted a geo-spatial analysis to determine which factors contribute most to landslide occurrence at landscape scales.We found no effect of coffeemanagement on vegetation damage or on economic losses at the plot or farm scale. At the farm scale, increasing management intensity (i.e. reduction in vegetation complexity) correlatedwith increased proportion of farm area affected by landslides (P = 0.014). Additionally, reduction in vegetation complexity was correlated with increased number (P = 0.0224) and volume (P = 0.062) of roadside landslides at the landscape level. Topographic and landscape features, such as distance to river (P = 0.004) and wind exposure/aspect (P = 0.044) strongly influenced landslide frequency at the landscape scale. Forest proximity andproportion of forest cover didnot significantly influence the frequency or extent of landslide damage.We created hazardmaps using the vegetation complexity index, distance to river, and wind exposure as theheaviestweighted factors to assess areas of the terrainwith thegreatest vulnerability. These maps present a practical result of this study, and offer a template in which land management policy candevelop to lower regional vulnerability to landslide risk. These results show that farmersmaybe able to reduce vulnerability to extreme stormevents by carefullymanaging their farms. Although farmersmaynot be able to control negative topographic features of their farms, increasing vegetation complexity within farms may be an efficient strategy to reduce some susceptibility to hurricane disturbance. 2008 Elsevier B.V. All rights reserved. * Corresponding author. Tel.: +1 419 53