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

https://doi.org/10.1016/j.agee.2008.04.016Get rights and content

Abstract

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 and mild winds. An approximately 50 km2 area was chosen within a coffee-growing region where data were collected on a variety of topographic and landscape features (aspect, slope, elevation, distance to river) and vegetation 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 a measure 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 coffee management 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) correlated with 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 and proportion of forest cover did not significantly influence the frequency or extent of landslide damage. We created hazard maps using the vegetation complexity index, distance to river, and wind exposure as the heaviest weighted factors to assess areas of the terrain with the greatest vulnerability. These maps present a practical result of this study, and offer a template in which land management policy can develop to lower regional vulnerability to landslide risk. These results show that farmers may be able to reduce vulnerability to extreme storm events by carefully managing their farms. Although farmers may not 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.

Introduction

There is great urgency to understand the impact of global climate change on agriculture. These top issues may be directly linked by hurricanes. In recent years hurricane frequency and intensity has increased, and although there is considerable debate about the causes, some attribute these patterns to global climate change (Emanuel, 2005, Webster et al., 2005, Shepherd and Knutson, 2007). Agriculture is a main driver of habitat destruction, covers half of arable lands (McNeely and Scherr, 2003), and may increase susceptibility to hurricane damage (e.g. Perotto-Baldiviezo et al., 2004). Yet, tropical agroecosystems vary widely in management practices. For example, coffee agroecosystems fall along a management intensification gradient, from farms cultivated under a native forest shade canopy (most vegetatively complex) towards farms characterized by lower density, diversity, and height of shade trees, lower canopy cover, and higher coffee density (least vegetatively complex) (Moguel and Toledo, 1999). Vegetatively complex coffee farms are prized for contributions to ecosystem services such as biodiversity preservation, pest control, pollination, and erosion control (Beer et al., 1998, Perfecto et al., 2007). Because coffee often grows on steep, largely deforested, mid-elevation mountain ranges, vegetation complexity may be especially important in buffering against the devastating effects of hurricanes. Here, we investigate environmental and economic impacts of hurricanes on coffee agroecosystems varying in vegetation complexity at three spatial scales.

Hurricanes dramatically affect forests, but very few studies examine the role of agricultural landscape management on vulnerability to hurricane-related environmental and economic damage. To our knowledge, only three studies have examined whether differences in agricultural type alter hurricane impacts. Uriarte et al. (2004) examined hurricane impacts on plant communities in abandoned pastures, mixed plantations, abandoned cacao, and forests. Areas with more complex vegetation were least resistant to hurricane damage losing larger and more trees. Yet, land uses with low human impact (cacao and forest), experienced faster increases in species richness and tree density following the hurricane. A second study demonstrated that hurricane impacts (erosion, economic loses, and vegetation damage) were less severe in farms using trees, live fences, terraces, and contour planting in agricultural fields (Holt-Gimenez, 2002). Perotto-Baldiviezo et al. (2004) examined landscape scale relationships between landslides, topographic features, and land use. Landslide probability was significantly higher in steeper areas, especially those in active or recently harvested swidden agricultural fields. Thus, what little evidence exists indicates that farms managed with increased tree cover, and with agroecological practices better resist and recover from the effects of hurricanes.

Research on landslide frequency has shown that both storm events and anthropogenic disturbance can increase risk. Landslide disturbance severely affects economic stability of agricultural regions due to damage to crop plants, arable land, and on-farm infrastructure, as well as disturbance of public services (Laing, 2003). In Puerto Rico, the leading cause of landslides has been attributed to intense and prolonged rainfall (Larsen and Simon, 1993). In the Luquillo Experimental Forest in Puerto Rico, landslides and uprooted trees triggered by Hurricane Hugo covered roughly 1.25% of the study area (Scatena and Larsen, 1991). Torrential rains accompanying Hurricane Mitch in 1998 triggered thousands of landslides throughout Central America (Crone et al., 2001), with as many as 11,500 landslides in east-central Guatemala (Bucknam et al., 2001).

Landslides attributed to anthropogenic disturbance are generally associated with changes in land use and vegetation structure. Landslide frequency in tropical mountains in Puerto Rico is five times greater within 85 m of roadsides than outside road zones (Larsen and Parks, 1997). Soil mass movement associated with roads can be 300 times greater than rates in undisturbed forest (Sidle et al., 1985), leading to greater rates of landslides. Other disturbances, like logging, are also associated with increased frequency of landslides (Guthrie, 2002).

Hurricane Stan hit the coffee-growing Soconusco region of Chiapas, Mexico in October 2005, just as the coffee harvest was beginning. The hurricane timing particularly affected the agricultural economy disrupting all aspects of the harvest, from preventing the arrival of migrant laborers, and slowing the harvest, to severely affecting transportation. The extensive rainfall (>500 mm) and wind (130 km/h) (Pasch and Roberts, 2006) led to widespread fruit drop before workers could reach the fields for harvest. Additionally, the onslaught of precipitation (>10% of the annual average in a few days) created massive landslides and floods, heavily damaging roads and bridges and submerging entire villages. Up to 50% of the Soconusco coffee harvest was lost, an estimated 170,000–280,000 ha of land were damaged, and 50–90% of farm infrastructure was affected (Perez, 2005). The extent of damage caused by this and other recent storms raises questions concerning risk and vulnerability for those dependent on agricultural livelihoods. Specifically, there is increased interest in understanding those factors that may reduce hazards and increase resilience of agricultural landscapes to the vagaries of extreme climate (Turner et al., 2003, Dale, 1998).

Despite the importance of hurricanes and landslides for farms and for farmer livelihoods, we lack a systematic evaluation of hurricane impacts in a range of agricultural management types at both farm and landscape scales. At the farm level, the potential for vegetation damage, soil erosion, infrastructural, and economic losses are evident, whereas a landscape perspective may be necessary for assessing the influences of topography and land use on landslide probability. In this study, we examine several key factors that may influence hurricane impacts in coffee-growing regions. We worked at three distinct scales to examine the importance of land use and several topographic factors including elevation, slope, aspect, and distance to rivers, in determining environmental and economic impacts of hurricanes. First, using plot level data from coffee farms and forests, we examined how vegetation characteristics of the coffee farms before the hurricane influenced vegetation damage and landslide occurrence as a result of the storm. Second, using data from government surveys and local field surveys, we examined how overall farm-level vegetation management relates to economic losses and total farm area affected by landslides. Finally, using topographic map data, satellite imagery and roadside landslide data, we examined which topographic and land use factors are most important in predicting the occurrence and size of landslides.

Section snippets

Study region

We worked in the Soconusco region of Chiapas, Mexico in a mosaic of small forest fragments and coffee agroecosystems varying in vegetation complexity. All field sites are located between 600 and 1400 m elevation, approximately 40 km north of Tapachula between 15.202N, 92.383W (NW corner) and 15.144N, 92.297W (SE corner) (Fig. 1). The geology of this area of Chiapas can be generally characterized as comprised of volcanic rocks and sediments derived from volcanic rocks, which are quite prone to

Plot scale analyses

Individual vegetation variables and the vegetation complexity index varied reflecting a general decline in vegetation structure and diversity along the management intensification gradient (Table 1). We found minimal vegetation damage as a result of the hurricane, and no small landslides within the study plots. We counted a total of one fallen tree (in a shade monoculture farm) and three defoliated shade trees (two in a shade monoculture farm, and one in a sun coffee farm). We recorded three

Discussion

The results from our study demonstrate that both land use and topographic features influence certain types of environmental damage sustained by hurricanes, and that the observed impacts vary with the scale at which impacts are measured. Most of the damage from Hurricane Stan seemed to stem from excessive rainfall, rather than wind damage, particularly evident from the lack of damage to foliar vegetation. The majority of the damage, such as fruit drop, damage to buildings and infrastructure, and

Conclusion

In sum, we demonstrate two important factors relating to susceptibility of agricultural landscapes to hurricanes using a case study of Hurricane Stan in the Soconusco region of Chiapas. First, for our case, the scale at which the impact of the hurricane was measured was important to understanding which landscape factor most strongly influenced landslide incidence and severity. We found no measurable effects of coffee management at the smallest spatial scale in terms of vegetation damage or

Acknowledgements

We thank G. Lopez Bautista, G. Dominguez, J.A. Garcia-Ballinas, and B. E. Chilel for assistance in collecting field data. G. Ibarra Nunez from El Colegio de la Frontera Sur provided logistical support. We thank all of the farm owners for access to their property and for providing information on economic losses sustained during the hurricane event. A. Rodriguez Camacho and the members of the Unión de Café Tacaná generously provided access to their damage reports. Thanks to C. Taylor, P. Bichier,

References (35)

  • V. Dale

    Ecosystem management in the context of large, infrequent disturbances

    Ecosystems

    (1998)
  • K. Emanuel

    Increasing destructiveness of tropical cyclones over the past 30 years

    Nature

    (2005)
  • D.R. Foster et al.

    Patterns of forest damage resulting from catastrophic wind in central New England, USA

    Journal of Ecology

    (1992)
  • J.L. Frangi et al.

    Hurricane damage to a flood plain forest in the Luquillo mountains of Puerto Rico

    Biotropica

    (1991)
  • C.A. Gresham et al.

    Hurricane Hugo wind damage to southeastern U.S. coastal forest tree species

    Biotropica

    (1991)
  • Laing, A., 2003. Tropical cyclone floods in Florida: geographical influences and community preparedness. Joint Session...
  • M.C. Larsen et al.

    How wide is a road? The association of roads and mass-wasting in a forested montane environment

    Earth Surface Processes and Landforms

    (1997)
  • Cited by (129)

    • Long-term trends in functional crop diversity across Swedish farms

      2023, Agriculture, Ecosystems and Environment
    • Climate change adaptation through agroforestry: opportunities and gaps

      2023, Current Opinion in Environmental Sustainability
    View all citing articles on Scopus
    1

    Tel.: +1 212 854 8253.

    View full text