SEARCH

SEARCH BY CITATION

Keywords:

  • Africa;
  • agroecosystem;
  • avifauna;
  • matrix habitat;
  • remnant trees

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Coffee cultivation plays a role in biodiversity alteration and conservation in much of the tropics. This is particularly so in Ethiopia, where coffee is an indigenous shrub and a major commodity in national and local trade. In southwestern Ethiopia, coffee (Coffea arabica, “highland coffee”) is harvested from both forests (its natural habitat) and from within farmland where it is grown in small patches under isolated shade trees. We investigated the effects of management practices on bird assemblages in each of these systems. We found that bird assemblages in the forest remnants were distinct from those in the farmland even if many species were regularly found in both habitats. Coffee cultivation in open farmlands promoted bird species diversity through the retention of forest trees, while coffee cultivation in forests reduced bird diversity. Forest coffee management may, however, ensure that forest remnants are not converted to other forms of more open agriculture. Certification standards for “ecologically friendly” coffee in Ethiopia need to take this complexity into account.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Coffee is the world's second-most traded commodity and therefore has the potential to influence biodiversity conservation in much of the tropics where it is cultivated. It is also Ethiopia's largest export and most important cash crop (Petit 2007). Many studies have demonstrated that shade coffee systems contain higher levels of forest biodiversity than coffee grown in more intensely cultivated sun (nonshaded) coffee plantations (Perfecto et al. 1996; Greenberg et al. 1997; Moguel & Toledo 1999; Hietz 2005). Such findings have lead, most notably in the New World, to the certification and marketing of shade coffee cultivation, which is least detrimental to biodiversity. Such certification defines standards for tree density, vertical structure of tree strata, and species density of trees (Mas & Dietsch 2004; Philpott et al. 2007). However, it has also been suggested that promotion of “shade coffee” could lead to higher rates of human encroachment and degradation of forest remnants, and thus be counter-productive (Rappole et al. 2003; see also Moegenburg & Levey 2002 for another crop).

Highland coffee is indigenous to southwestern Ethiopian forests, and therefore, conservation of the last remnants of montane rain forests that contain a genetic reservoir for coffee is of high priority (Silvestrini et al. 2007). Wild coffee and its natural densities are difficult to define due to the long history of forest coffee cultivation. However, coffee naturally occurs sparsely, while its density increases with more intense management within forests. Such management may include removal of competing vegetation, and planting of coffee seedlings (Schmitt 2006). Besides growing under dense natural shade in forests, coffee is also cultivated in small mixed-farming systems, with small patches of several coffee plants shaded by individual trees embedded among other crops including corn (Figure 1). Land conversion of forests to larger unshaded or partly shaded coffee farms is also an increasing threat (Wakjira 2007). These current configurations of coffee/tree systems therefore represent important management strategies throughout a large part of the landscape. As tree retention appears fundamental to restoring functionality to open, degraded landscapes (Hughes et al. 2002; Gove et al. 2005; Harvey et al. 2006; Sekercioglu et al. 2007), promotion of farmland coffee cultivated under individual shade trees could be an important tool in maintaining the ecological integrity of these landscapes. Hence, the landscape mosaics of southwest Ethiopia may approximate the best of conservation strategies, in which remnant natural vegetation remains relatively intact while a high-quality human-modified matrix is also maintained (Lindenmayer & Franklin 2002; Donald & Evans 2006; Fischer et al. 2006). Ecological studies aimed at guiding conservation policy or biodiversity-focused certification schemes in complex landscape mosaics may be most effective if both forest patches and the agricultural matrix are considered simultaneously to grasp comprehensively the threats to biodiversity and consider most appropriate levels of management (Aerts et al. 2008; Harvey et al. 2008).

image

Figure 1. The southwestern Ethiopian landscape. (A) One of the largest forest remnants in the study area. (B) A relatively undisturbed forest interior. (C) Forest interior heavily utilized for coffee cultivation. (D) Typical agricultural landscape. (E) Isolated forest tree with under-story coffee-cultivation in agricultural matrix.

Download figure to PowerPoint

Here we analyze the dilemma that coffee cultivation may present for habitat conservation in the Ethiopian landscape; coffee cultivation may introduce or maintain trees in open farmland, while at the same time modify forest remnants. We consider how these two effects of coffee cultivation may be best considered simultaneously using birds as focal organisms.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Study region and sites

We conducted the study in a 40 × 35 km area surrounding the town of Bonga (36° 14′ E, 7° 16′ N), 450 km southwest of Addis Ababa, Ethiopia (Figures 1 and 2). The area's climate is warm tropical with the main rainy season occurring from March to October. Annual average rainfall is approximately 1700 mm/year and dominant vegetation is Afromontane rainforest (Friis et al. 1982) with species such as Sapium elipticum (Euphorbiaceae), Schefflera abyssinica (Araliaceae), and Millettia ferruginea (Fabaceae). The landscape is a mosaic of forests and agricultural areas with both large (> 1 km2) and small (< 1 ha) patches of forest and agriculture (Figure 2). We carried out all fieldwork during June–August 2007, during the bird breeding season, when identifying canopy birds from song was easiest (A. Shimelis, Personal observation).

image

Figure 2. Map of study site. Gray shading represents forest and white, agriculture. Closed circles represent forest sites, black and white circles agricultural sites, and open circles major townships. Inset: 2-dimensional ordination of sites, based on bird assemblage composition, using nonmetric Multidimensional Scaling and Sørensen's similarity index. Closed diamonds represent forest, open circles agricultural sites.

Download figure to PowerPoint

We chose 19 farm sites and 19 forest remnant sites, widely dispersed across the focal landscape (Figures 1 and 2). The sites were chosen from tri-decadal global Landsat images produced in 2000, with 15-m resolution. We separated our sites into farmland and forest as the landscape is clearly defined by these two habitat types, both superficially and by local communities responsible for their management. Forest plots included a range of coffee densities (Table 1), which, along with tree densities and tree species richness, correspond to a gradient between "Primary" forest and "Rustic" coffee, based on the schemes of Philpott et al. (2008) and Moguel & Toledo (1999). This is in contrast to coffee cultivation on Ethiopian farms, which does not resemble any of such schemes due to the sparse and scattered distribution of coffee plants. All forest plots would likely pass current shade coffee standards (Mas & Dietsch 2004). Both farm and forest sites were chosen to cover a variation in density of surrounding forest and were separated by at least 2 km within each habitat category. Altitude ranged from 1,700 to 2,350 m above sea level. We were led to each site using GPS and at each site we established a 100 × 200 m plot.

Table 1.  Vegetation cover, altitude, and surrounding forest cover in farm and forest plots
 FarmlandForest
MeanRangeMeanRange
  1. N= 19.

  2. aExcluding coffee. Most dominant food plants in forest were Piper capense and Syzgium guinense ssp. afromonatanum and in farmland Corn (Zea mays).

Cover of food plantsa (%)5110–82173–35
Tree density (stems > 15 cm/ha)101–30104 50–232
Tree species9.8 spp/2 ha3–1511.9 spp/0.2 ha7–20
Cover of coffee (%) 20–8110–54
Number of subplots with coffee (max 50)   7.80–22  
Forest cover within 1 km (%)350.4–775321–88
Altitude (m asl)1,864   1,492–2,3401,854   1,542–2,340

Vegetation and bird surveys

At each farm site, the 100 × 200 m plot, was divided into fifty 20 × 20 m subplots in which we recorded density and diameter of all trees (diameter at breast height > 15 cm) and percent cover of coffee. In each forest patch, a 100 × 200 m plot was established at least 25 m from the nearest forest edge (little more was possible due to the small area of some of the forest patches). As measures of habitat modification which could be quantified simply and efficiently, we focused on tree density and species richness, and the density of food plants and coffee. These variables have previously been associated with a reduction in habitat complexity and increased management intensity (Senbeta & Denich 2006; Philpott et al. 2008). Due to higher tree densities in forests, we established five 20 × 20 m subplots within each 100 × 200 m plot in which tree and coffee cover was quantified and tree species inventoried. The mean from the subplots was used to represent the whole plot. Subplot location within the plot was randomized once, and this scheme was used for every plot. Cover of coffee was estimated visually as percent foliage cover in each subplot in farms, while in the forests, cover was estimated visually as percent foliage coverage across four tape measures of 20 m length within each subplot. Where tree species could not be identified in the field, specimens were taken to the National Herbarium of Ethiopia, Department of Biology, Addis Ababa University for identification. These specimens were deposited in the Herbarium collection.

At each farm and forest site, we used 22 eight-minute point counts evenly distributed within the 100 × 200 m plot. Given the small size and high heterogeneity of many farms and forest patches, we employed a relatively dense array of point counts. These were located in five transects running along the longest axis (three transect of four points, staggered with two transect of five points) with points separated by 50 m along each transect. Each point was sampled once. We recorded all bird species seen and heard within a 25-m radius. On several occasions we did not complete all 22 point counts (at least 18; usually due to rain). Each site survey was carried out in one day soon after sunrise and lasted approximately 3–4 hours.

Data analysis

Based on the information on main habitat requirements in Sinclair & Ryan (2003), bird species were categorized into four response groups: (1) open habitat species (using fields and grazing areas), (2) shrubland species (using shrubby areas and edge habitat), (3) woodland species (using wooded savanna and parks), and (4) forest species (using dense closed forests) (see Table S1).

We used ArcGIS 9.1 to calculate the percentage of forest cover present within a 1-km radius of each sample site. We used data on tree assemblages found in each of our forest patch sites as a reference to quantify the relative habitat value of each farm-scale site; for each agriculture site, we calculated the percent of forest tree species that were found in each farm plot (see Table S2). We used Sørensen's similarity to compare the bird assemblages of the forests and agriculture sites and a nonmetric multidimensional scaling ordination to illustrate this relationship, while testing it with analysis of similarity (ANOSIM) with 999 random permutations (Clarke & Green 1988). As we had uneven sample sizes (between 18 and 22 point counts in each site), we rarefied the samples to 18 samples using the Mau Tau bootstrapping method in EstimateS (Colwell 2005).

We tested for bird-habitat relationships separately for the farmlands and the forest patches. The relationship between tree and coffee presence within each of the 50 quadrats within a farm plot was assessed using a general linear model with binomial distribution and logit link function. Relationships between continuous variables were assessed using linear models. Rarefied species density (e.g., the number of species found in a site) or rarefied species density of bird species classified by habitat (see above) were the dependent variables considered. The independent variables considered for model inclusion were: tree density, tree species density, coffee cover, cover of food crops, altitude, and remnant forest within 1 km. As we sampled 50 quadrats in each farm, we were able to express the intensity of coffee cultivation as the number of quadrats that contained coffee plants in each farm. In the forests, with only five quadrats per site, we were confined to expressing coffee density as percent cover. When the tree assemblage response (species density, tree density, or similarity to all forest plot assemblages) was the dependent variable, the independent variables were coffee cover, other food crop cover, altitude, and remnant forest within 1 km. The most parsimonious models were selected using a best-subsets approach based on the Bayesian information criterion (Miller 2002). All analyses were performed in R2.6.1 (R Development Core Team 2007).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We recorded 106 species of birds during the study. Cumulatively, 94 species were sampled from agriculture sites (of these, seven were forest species and 22 woodland species), and 72 species in the forest patches (eight forest species and 24 woodland species). The species densities of both open- and shrubland birds were, respectively, 18 and 2 times higher in the farmlands than in the forests (F1,37= 70.3, P < 0.001, R2= 0.65; F1,37= 41.0, P < 0.001, R2= 0.52), while species densities of woodland and forest birds were, respectively, 1.7 and 9.7 times higher in the forest patches (F1,37= 16.3, P= < 0.001, R2= 0.29; F1,37= 50.2, P < 0.001, R2= 0.57). Fifty-eight percent of bird species were shared between the most similar agriculture and forest plots, suggesting that despite some value to forest species of the farm plots, they are still distinct from forest habitats (Figure 2, inset; ANOSIM, R= 0.769, P < 0.001). In farms, woodland and shrubland species densities were positively associated with tree species density (Figure 3A, Table 2), while forest bird species density was positively associated with tree density and with the amount of forest within 1 km (Table 2).

image

Figure 3. On farms, species density of tree-dependent bird species is correlated with the number of tree species (A). The level of farmland coffee cultivation is associated with (B) the species density of farmland trees, and (C) the similarity of the farmland tree assemblage and the forest tree assemblages. See Table 2 for whole model summaries.

Download figure to PowerPoint

Table 2.  Statistical analysis of associations between farmland biodiversity (tree and bird assemblages) and habitat variables
Dependent variableNumber of variables in best modelVariables included in modelR2P
  1. The best model was defined as that with the lowest BIC value.

Trees
 Tree density No sign. model  
 Species density10.28 (Coffee)0.200.033
 Similarity to forest10.60 (Coffee)0.51<0.001 
Birds
 Total species density10.80 (Tree spp.)0.340.005
 Forest species density20.03 (Tree density) + 0.0001 (Forest < 1 km)0.56<0.001 
 Woodland species density10.37 (Tree species)0.220.025
 Shrubland species density10.40 (Tree species)0.350.004
 Open species density1No sign. model  

Although coffee presence in a subplot was positively associated with tree presence within that subplot (Z= 4.12, N= 450, P < 0.001), the level of coffee cultivation in individual farms was not correlated with tree density (Table 2). At this scale, coffee frequency was, however, positively correlated with tree species density (Figure 3B) and to the proportion of forest trees found in the farm (Figure 3C). (If we were to use percent coffee cover in quadrats, as in the case of the forest patches, rather than coffee occurrence, the relationships were similar, but with slightly poorer model fit.)

In forest patches, bird species density was negatively associated with coffee density (Figure 4A, Table 3). This pattern was similar for all tree-dependent bird groups. There was no evidence of a relationship between coffee coverage and the density or species density of trees (Table 3). Bird species density was also negatively correlated with tree density (Figure 4), and together with coffee density a three-variable model could explain 66% of the variation in total bird species density in forests.

image

Figure 4. Within forest patches, coffee cultivation is related to a decrease in species density of tree-dependent birds. See Table 3 for whole model summaries.

Download figure to PowerPoint

Table 3.  Statistical analysis of associations between forest biodiversity (tree and bird assemblages) and habitat variables
Dependent variableNumber of variables in best modelVariables included in modelR2P
  1. The best linear model was defined as that with the lowest BIC value.

Trees
 Tree density No sign. model  
 Species density No sign. model  
Birds
 Total species density3–1.58 (Coffee) + 0.64 (Tree spp) – 0.04 (Tree density)0.66<0.001 
 Forest species density2–0.42 (Coffee) + 0.003 (Altitude)0.520.001
 Woodland species density2–0.60(Coffee) – 0.01 (Tree density)0.470.003
 Shrubland species density3–0.46 (Coffee) + 0.38 (Tree spp) – 0.02 (Tree density)0.360.021
 Open species density No sign. model  

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

It is clear from our study that the primarily structural drivers of the bird assemblage are local habitat variables such as tree species density, tree density, and extent of coffee cover, all of which are largely determined by local management practices. These practices, which tend to simplify habitat structure in forests while augmenting it within agriculture, thus largely determine the fate of bird biodiversity conservation in the current landscape of southwestern Ethiopia. Traditionally, conservation has focused on formal reserve establishment and strict protection, which to be effective requires committed governance and a positive response from local communities. Here, we focus our discussion instead on how local and realistic management decisions, independent of a reserve system, could affect local biodiversity conservation.

Habitat management can have varied results, as seen in the forests where bird diversity responds negatively to coffee cultivation, most likely by simplification of habitat structure (Senbeta & Denich 2006; Philpott et al. 2008), while reduced tree densities (likely due to selective logging) had a positive effect on some bird groups, probably due to the diversity of micro-habitats created.

Conversely, coffee cultivation on farms, often under single isolated trees, increased habitat complexity, and hence diversity of birds found within farmland. The promotion of isolated farm trees through coffee cultivation could lead to increased quality of matrix habitat as has been found in other regions. Isolated trees promote conservation of invertebrates (Dunn 2000; Gove et al. 2005), and epiphytes (Hietz-Seifert et al. 1995), and are used as movement stepping stones in otherwise inhospitable habitats (Fischer & Lindenmayer 2002; van de Ree et al. 2003). Upon abandonment, these trees can also act as important nuclei of habitat restoration (e.g., Guevara et al. 1986; Williams-Linera et al. 1997). While identified as important in other studies (Anand et al. 2008), there was evidence for a minor role of the extent of surrounding forest in maintenance of forest bird species on farms. Its lesser role in our study may be due to the high-quality matrix habitat of current farmlands, and the overall abundance of forest habitat within the landscape.

Ethiopia's unique situation, with coffee as an indigenous forest shrub, implies that coffee cultivation in degraded landscapes could be considered a form of landscape rehabilitation and should therefore be promoted in regions already cleared of native vegetation (Hylander & Nemomissa, in press). Although this type of coffee cultivation found within the landscape matrix could not be certified as "shade coffee" under current biodiversity-focused certification schemes (Mas & Dietsch 2004), its value in terms of increasing conservation potential of a heavily modified landscape is clear. A coffee certification program introduced to Ethiopia could emphasize the value that coffee cultivation under shade trees in the agricultural matrix has in improving deforested and degraded landscapes.

Our results also suggest that increases in forest coffee density can be detrimental to bird diversity. Current certification schemes emphasize the maintenance of a complex canopy structure (Mas & Dietsch 2004); Shrub-level structure and the density of coffee plants are not a focus. Certification of shade coffee in Ethiopia may lead to further degradation of remaining forest remnants through an increase in coffee densities. However, as the conservation of native forest remnants is dependent upon the positive attitudes of the surrounding land-users, endowing local forests with a readily harvested and profitable resource may be the best way to ensure the forests longevity, albeit in a partially degraded state (Stellmacher & Gatzweiler 2005; Brooks et al. 2006). Ensuring that coffee farmers receive a reasonable price for the commodity is perhaps most important. Tesfaye & Bierschenk (2004) have indicated that without sufficiently high coffee prices, Ethiopian farmers anticipate further forest clearing and production of cereals or crops such as the narcotic khat (Catha edulis)—crops of higher value than coffee. Land conversion to larger government-sponsored unshaded coffee farms is also a threat (Wakjira 2007).

In our study, reductions in tree density (e.g., from selective logging) may have some positive effects on forest bird assemblages, probably due to an increase in small-scale habitat diversity. But this effect is not apparent for specialist forest birds, illustrating that different types of human disturbances can have profoundly different effects on forest biota and that there is a great need for research on management practices which can be considered sustainable.

Human livelihoods play an important role in maintaining biodiversity values (Wakjira 2007). Ethiopia may be able to gain relative high coffee prices by offering authentic "forest coffee." However, defining "true" forest coffee is a notoriously difficult task and hence, Wiersum et al. (2007) suggest an "area-based" certification approach for Ethiopia which focuses on the sustainable management of overall landscapes, including retention and management of forest patches combined with sustainable and habitat-promoting farming practices, and operates at a scale most conducive to certification of many small-hold farmers. To make certification accessible to such farmers in Mexico, Ávalos-Sartorio et al. (2006) also suggest an area-based certification scheme.

Finding sustainable multiple usages of forest resources may be the most important way of retaining or increasing the forest cover of southwestern Ethiopia. Forest coffee management certainly has an important role to play. Nonetheless, it should be recognized that coffee cultivation within forests simplifies their structure (Senbeta & Denich 2006), leading to a decrease in habitat value (demonstrated here for birds), while coffee cultivation on farms could actually improve the habitat value of these degraded habitats. We suggest that Ethiopia has a unique set of qualities, particularly the indigenous nature of coffee, and hence represents a unique set of management dilemmas not well suited to a one-size-fits-all universal style of coffee and environmental certification.

Editor : Dr. Andreas Baldi

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We thank R. Dunn, J. Ehrlén, and three anonymous referees for very helpful comments on the manuscript and Teshome Legesse and Woldeyohannes Enkossa for field and herbarium assistance. The study was supported by a grant from Swedish International Development Cooperation Agency (SIDA) to K.H.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
CONL_033_sm_TableS1-S2.doc258KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.