This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Epiphytes are plants that use other plants as a habitat (Fig. 1). They are not considered parasites because they do not derive any kind of nutrition from their host plant. They derive water and nutrients both from the air and rain and from the dead bark and debris, accumulating around the epiphyte. Epiphytes are mostly found in the tropics and subtropics but especially mosses, liverworts and lichens can occur as epiphytes in almost every environment with trees. Epiphytes are photosynthetic and thus capable of producing their own energy. They use the plants they grow on only for physical support (Hanski & Gilpin 1997). The big advantage for epiphytes, besides the avoidance of competition with other plants, is the benefit of growing higher in the trees and as such receiving more light than terrestrial plants at the same location (Benzing 2004).
Because epiphytes occupy this special habitat, they are in general more dependent on the environmental conditions than terrestrial plants. The adaptation to life above the soil makes them more sensitive to environmental changes (Engwald et al. 2000; Padmawathe et al. 2004). In vascular epiphytes, the colonization and establishment of new seedlings is a rather slow process, making them more vulnerable for rapid environmental changes. They experience tree crowns as a mosaic of suitable and unsuitable habitats, where suitable habitats are species specific and, sometimes, only present in small and discrete patches (Hanski & Gilpin 1997).
Epiphytes are seen as an important component of biodiversity in tropical forest ecosystems. They can be very abundant in tropical forest canopies and can account for up to half of forest-plant richness (Benzing 1990; Engwald et al. 2000; Nieder et al. 2001). Besides their obvious biodiversity aspect, epiphytes have an important ecological function. They are part of ecosystem processes and have functions in nutrient and water cycling. Specifically in tropical forest ecosystem, a change in epiphyte composition can have cascading effects, affecting ecosystem services in general (Moorhead et al. 2009).
Epiphytes can be used as indicators of forest disturbance but they also provide resources and niche possibilities for canopy-dependent fauna (Benzing 1990; Cruz-Angon & Greenberg 2005). Since they are useful climatic indicators (Richter 1991), they can be used as a warning system for changing conditions in microclimate (Haro-Carrión et al. 2009) and even as indicators of global climatic change (Benzing 1998).
1.1.2. Epiphytic orchids
The Orchidaceae is currently believed to be the second largest family of flowering plants in the world with more than 22,000 species described. Although the family originated from terrestrial orchids, most of the species are epiphytic which can only be found in the tropics and subtropics (Benzing 2004). Epiphytes account for c. 70% of all the orchid species (Stevens 2001). Epiphytic orchids have no vascular connection with their host but the roots make sure the orchid is anchored to the host. The roots also function as nutrient and water storage and uptake mechanism. Cells in the roots contain chloroplasts for active photosynthesis. Some orchids have no leaves and fully depend on their roots for energy (Fig. 2).
Since they cannot rely on a regular supply of water via the soil, epiphytic orchids have some xerophytic adaptations. Their leaves are rather leathery, often being very succulent. The leaves have a thick cuticula that reduces water loss (Stevens 2001; Benzing 2004). A velamen develops in the roots (Fig. 3A). This is a tissue supporting one or more layers of dead cells. Special thickenings in the cell wall prevent the collapse of the cells and provide some protection to the roots against mechanical injury. Cells in the velamen can rapidly absorb water in wet conditions, even passively from the atmosphere (Benzing 2004). But in dry conditions, the cells only contain air and act as a barrier to prevent water loss via transpiration from the water conducting tissues in the inner part of the roots. The velamen reaches its maximum development in the roots that hang free in the air (Oliveira & Sajo 1999). Most epiphytic orchid species form enlarged stem segments called pseudobulbs, from which the leaf grows (Fig. 3B). Being succulent these pseudobulbs can store nutrients and water. The pseudobulbs swell or shrink, depending on the external moisture conditions. To minimize water loss, epiphytic orchids make use of the crassulacean acid metabolism (CAM), allowing for the uptake of CO2 during the night (Motomura et al. 2008). These adaptations are necessary for epiphytic orchids to overcome seasonal rainfall patterns, which sometimes results in months with alsmost no precipitation (Benzing 2004).
The distribution of epiphytic orchids in the forest is a subject which is rarely studied. Most orchid species occur clumped in a tree around the orchid mother plants. Orchids were shown to cluster more strongly than any other group of epiphytes (Hietz & Hietz-Seifert 2005). The reason for this phenomenon is that growing close to conspecific plants, chances are higher to form an association with mycorrhiza (Hietz & Hietz-Seifert 2005; Diez 2007). The seeds of epiphytic orchids are generally almost microscopic and are mainly dispersed by the wind. Orchids need microsites with specific levels of humidity, temperature and suitable substrate that fit their ecological requirements for germination and establishment (Winkler et al. 2005). Seeds are very numerous because chances to be dropped at the right microsite are small and a beneficial association with mycorrhiza is often necessary for germination (Winkler et al. 2009). Host tree species seems to be an important aspect of epiphytic orchid diversity (Haro-Carrión et al. 2009). The roughness of the bark, the water holding capacity of the bark and bark pH are important characteristics influencing epiphyte diversity (Patino & Gonzalez 2011). The establishment of orchid seeds and seedlings can be increased by a more corrugated bark or larger branches. Also the presence of other epiphytes, especially bryophytes can facilitate the establishment (Hietz et al. 2002). The extent of dispersal limitation acting on epiphytic orchids in fragmented forests remains largely unstudied (Wolf 2005).
1.1.3. Ecological importance
Epiphytes in general are an important part of the biodiversity in the rainforests and can contribute even more to the diversity by offering food and habitat to other species. Orchids in general are a broad and diverse group with many species, sensitive to ecosystem or environmental changes. The same is true for epiphytic orchids. However, not every species is affected in the same manner by human disturbance. Some orchid species are indicated to grow even better in more disturbed habitats (Solis-Montero et al. 2005; Werner et al. 2005; Hietz et al. 2006). Because of their vulnerable lifestyle and the broad differences between species, epiphytic orchids can be used as indicators of forest quality.
Orchid species do not grow random in the tree. The species composition shows a vertical stratification where different species occur in different layers in the canopy of the tree, depending on their microclimatic needs. Engwald et al. (2000) indicated that epiphytes of montane rainforests in particular are more vulnerable, compared to other forests, to changes in their environment. This is probably due to the importance of slowly growing, structural old trees in montane rainforests. Changes in the vertical distribution of the epiphyte species due to changes in microclimatic conditions can be used to indicate human disturbance (Padmawathe et al. 2004; Haro-Carrión et al. 2009). Identifying the species that are more sensitive to changes and those that are able to flourish in disturbed habitats can help to indicate forest disturbance.
Orchids in general and epiphytic orchids in specific are often used for conservation measures. Because of their spectacular flowers or because they are often rare or endemic, they can be used as flagship species for (sub)tropical forests. This helps to raise funding for research and conservation efforts, to gain the attention of the public and to enhance ecotourism. They can be of great importance for plant conservation and to protect certain orchid rich habitats, as such protecting whole ecosystems (Demissew et al. 2004).
1.2. Effects of forest fragmentation and management on epiphytic orchids
1.2.1. Forest fragmentation
Habitat fragmentation is considered as one of the three main causes behind the present biodiversity crisis (Young & Clarke 2000). Tropical forests are under considerably high human threat. Habitat fragmentation affects the ecology of tropical forests in many ways, and it consists of three major components: the direct loss of habitat, a reduction in the size of the remnant fragment and an increased isolation of the fragments (Andren 1994).
When large tracts of forests are dissected, organisms are exposed to different biotic and abiotic changes. Fragmentation changes the quality of the forest by altering its climatology (e.g. wind velocity and humidity), forcing species to cope with changes in microclimate and ecological functions. It also increases the amount of edge habitat with diverse ecological consequences. Epiphytes are believed to have great potential as indicators of forest edge effects (Esseen & Renhorn 1998).
Fragmentation usually reduces population size, making species more vulnerable to extinction due to stochasticity and the possible loss of genetic variation. Less specialized pollinators can be expected in small and isolated habitats (Roubik 2002; Honnay et al. 2005). This can reduce reproductive success of plants. Individuals from larger populations usually have higher fruit success than individuals from smaller ones (Leimu & Syrjanen 2002; Tremblay et al. 2005). The potential for a species to adapt to changes, whether environmental or climatic, decreases when its genetic diversity is reduced (Young & Clarke 2000). However, some tropical plants (e.g. some epiphytic orchids) occur naturally at low densities in tropical forests. These species can have certain aspects in their life history that maintain or even augment reproductive success after fragmentation took place. This can occur through long distance pollination (Murren 2002). This means the effect of fragmentation is species-specific.
Fragmentation can affect distribution and diversity of epiphytic orchids. Different aspects of the life history of epiphytic orchids suggest increased extinction risk when populations become fragmented. Because tropical orchids are often dependent on one or few specialized pollinators for successful fruit set (Ackerman 1996), it can be expected that they suffer from pollinator limitation in small fragments with small populations. The general lifestyle of epiphytic orchids also makes them more vulnerable to changes in climatologic conditions, especially decreased air humidity and precipitation (Murren 2002). These aspects put epiphytic orchids at an increased risk of local extinction after forest fragmentation.
1.2.2. Management changes
Conversion of tropical primary forest into anthropogenic habitats has consequences for the biodiversity of the forest. Epiphytes are one of the first life forms to be affected by changes in primary forests because they occupy forest canopies (Sodhi et al. 2008). When epiphyte diversity is compared between natural forests and anthropogenic habitats, most of the secondary habitats show a reduced diversity (Barthlott et al. 2001; Werner et al. 2005), though the extent of the changes in diversity can vary between study sites and habitats.
A number of studies have been investigating the human impact on epiphyte diversity and some of them do consider the habitat conversion to secondary forest types. Few studies were confined to single taxa such as bromeliads and/or orchids (Hietz et al. 2006). In general, changes in microclimatic conditions because of the selective logging of certain tree species, can strongly decrease epiphyte richness (Hietz-Seifert et al. 1996; Barthlott et al. 2001; Werner et al. 2005). Haro-Carrión et al. (2009) studied the contribution of shade cacao plantations to vascular epiphyte conservation and showed a reduced diversity of epiphytes in plantations relative to natural forest. But no difference was found in species richness of Orchidaceae.
Köster et al. (2009) showed a loss of epiphyte species in secondary forest with young secondary forests being less diverse than older secondary forests. Interestingly, this study showed no significant effect of spatial parameters such as fragment area, distance to edge or distance to primary forest on epiphyte diversity.
Barthlott et al. (2001) compared secondary vegetations with primary montane rainforest in the Andes. The study found a lower diversity of epiphytes in general and Orchidaceae in specific, for the secondary vegetation. This was not the case in a study of Moorhead et al. (2009). They found the orchid composition to be equally rich in polyculture coffee farms and in natural forests. However, when polyculture farms and natural forests were compared with monoculture coffee farms, a difference in richness was found. Other studies show that epiphyte diversity and abundance is lower in shaded coffee farms (Mexico) and home gardens (Ethiopia) than in nearby natural forests (Hietz 2005; Hylander & Nemomissa 2008). Some studies show that epiphyte diversity is positively correlated with increasing tree size (Hietz 2005; Moorhead et al. 2009).
Hietz et al. (2006) studied the abundance of epiphytic orchids and bromeliads in a montane forest in Mexico and concluded that disturbance does have complex species-specific effects, depending on many factors like host tree species. Thus it seems clear that not every orchid species reacts alike; drought-resistant species may benefit from the disturbance resulting in an increased occurrence in managed ecosystems (Larrea & Werner 2010). Especially species that require more shade and/or high humidity will decline or go extinct due to fragmentation and increased forest management (Hietz 2005; Werner et al. 2005; Wolf 2005). How much these species are negatively affected mainly depends on the degree of disturbance, the age of the secondary forest and the size and species composition of the remnant trees (Padmawathe et al. 2004). How epiphytes on the remaining trees react or how colonization on younger trees occurs, remains more elusive. Haro-Carrión et al. (2009) studied the vertical distributions of vascular epiphytes in shade cacao plantations relative to natural forest. They observed a downward shift of epiphytes on the remnant trees in shade cacao plantations.
Solis-Montero et al. (2005) studied the population structure of certain epiphytic orchids in a shade-coffee plantation in Mexico. They concluded that it is possible for orchid species to survive and reproduce in coffee plantations when the right microclimate conditions are present. Although plantations are not able to replace the original conditions of the primary forest.
To summarize, results of previous studies confirm that the composition and diversity as well as the vertical distribution of epiphytes can be used as an indicator of human-induced disturbance in a forest landscape (Hietz et al. 2006; Hylander & Nemomissa 2008; Haro-Carrión et al. 2009). To minimize the loss in epiphyte biodiversity, the maintenance of large, forest-like trees in managed plantations could help to conserve epiphyte diversity, not only in the canopy but also in the understory (Haro-Carrión et al. 2009). The crop itself can also contribute to the overall species richness. It can increase the habitat area for epiphytes and improve the microclimatic conditions (Hylander & Nemomissa 2008).
1.3. Epiphytic orchids in the coffee forests of Southwest Ethiopia
There are some major driving forces that increase pressure on the coffee forest in SW-Ethiopia. The most important is the rising population pressure causing deforestation for new settlements and agricultural land, as well as general overexploitation of the remaining forest. The others are road construction, making the forests more accessible, a poor policy and legislation with almost no control on illegal deforestation and the rising demand for coffee worldwide (Gove et al. 2008; ICO 2011).
Wild coffee, Coffea arabica L., is native to Ethiopia and grows as an understory shrub species in Afromontane rainforests between 1000 and 2000 m above sea level (asl) (Vavilov 1951). The local communities in Ethiopia have developed a long tradition of managing the forest for coffee production which is nowadays the main export product of Ethiopia (FAO & WFP 2009). This traditional management has only minor effects on the structure and biodiversity of the forest. Even today an important part of Ethiopian coffee beans (c. 35%) is produced in traditional coffee production forests, but with the rising demand for coffee and its higher prices (ICO 2011), the management intensifies, aiming for higher yields (Aerts et al. 2011).
Coffee yields are much higher in semi-plantation coffee systems (SPC) and semi-forest coffee systems (SFC) than in forest coffee systems (FC) because of forest management (Schmitt et al. 2009). Therefore, in forests in SW Ethiopia that are cultivated for coffee, the undergrowth is removed at least once a year to avoid competition with non-coffee shrubs. New coffee seedlings are planted whereas older shrubs are pruned to increase growth efficiency (Aerts et al. 2011). The tree layer is managed as well: both tree species that are less efficient in shading and slow growing species are cut, resulting in 30 % less canopy trees in SFC (Schmitt et al. 2009). Even the trees that are used as shade trees are managed, often for wood extraction. These modifications to the forest have led to a uniform, species poor tree canopy and a single-species (coffee-)shrub layer (Fig. 4) (Schmitt et al. 2009; Aerts et al. 2011). In more intensely managed coffee forests, there is no intermediate layer anymore. In the long term this will have serious implications for the regeneration of the forest when the mature trees reach a post-productive stadium (Aerts et al. 2011).
Some of the forest relicts, for example on more steep slopes, have been conversed to coffee forests. These relicts, which can vary greatly in size, are situated in an agricultural landscape matrix (Fig. 5). Depending on the quality of the surrounding matrix, it can be expected that this process of fragmentation and intensified management have caused significant changes in abundance and diversity of forest plant species. Comparisons of epiphytes in forests and coffee agroecosystems show that forests in SW Ethiopia generally maintain higher richness and abundance than coffee agroecosystems (Hylander & Nemomissa 2008, 2009).
Orchidaceae is, with around 167 species in 37 genera, the fifth largest family in the Flora of Ethiopia and Eritrea. Around 15 % of the region's orchids are endemic of which 22 species are terrestrial and only six are epiphytic. The orchid flora shows a decline in species richness from the southwest to the northeast, quite similar to the decreasing rainfall patterns (Demissew et al. 2004). Only 16% of the orchids are epiphytic or lithophytic (growing on rocks). Of them, the majority can be found in the more forested areas of west and SW Ethiopia. Epiphytic orchids in this more remote part of Ethiopia are poorly represented in reference collections and little is known about their distribution, physiology, ecology and conservation status. For a lot of epiphytic orchids, Ethiopia is the northern-most extent of the range of the species. This can in part explain why the share of epiphytic orchids in the total Ethiopian orchid family is rather low (worldwide around 70% of the Orchidaceae are epiphytic). It is suspected that many epiphytic orchids are threatened in Ethiopia because of massive deforestation. Although the conservation status for a lot of epiphytic orchid species in Ethiopia is not known, many species appear to be rare and/or endangered (Demissew et al. 2004).
Polystachya, a pantropical genus with its diversity centre in tropical Africa, is the genus that is best represented in the list of epiphytic orchids in Ethiopia. Twelve species are found, mostly in more wet forested areas although some can survive dry conditions by losing their leaves in the dry season. Some of the species in this genus, e.g. P. cultriformis (Fig. 6A), are widespread and common in tropical Africa whereas others, e.g. P. caduca (Fig. 6B), are endemic for Ethiopia (Demissew et al. 2004).
1.4. Problem statement and objectives
Changes in richness and abundance of different taxa have often been attributed to responses to disturbances. Deforestation in combination with changes in forest management has caused great changes in the pristine forest landscape of SW Ethiopia, leading to significant effects on the biodiversity of the area. Many studies have been studying the effect of human disturbance on biodiversity in different kind of landscapes. In the past, most studies focused on terrestrial plants for their research because they are easier to sample and the metapopulation structure is better known (Avila-Diaz & Oyama 2007). Currently, the focus lies on studying the epiphyte biodiversity in general because epiphytes are more sensitive to changes (Padmawathe et al.2004; Hietz 2005; Köster et al. 2009).
Previous studies have shown the negative effect of forest fragmentation and human disturbance on the diversity of epiphytes. This research has focused on an important group of epiphytes to investigate the effect on the composition and diversity of epiphytic orchids in the cultivated coffee forests around Jimma, Ethiopia. We aim to compare the orchid species diversity in semi-coffee plantations, semi-forest coffee systems and more natural coffee forest. In general, we expect the diversity and abundance of orchids to be lower in the cultivated and fragmented forest plots.
The study area consisted of three sites within 70 km west-northwest of Jimma. Jimma is a city situated in the Oromiya region in the highlands of Southwest Ethiopia (Fig. 7). The Oromiya regional state is the biggest of the nine federal states of Ethiopia. The forest index in the southwestern part of the country is 18 %, accounting for more than half of the remaining forest area in the entire country; forests account for less than 3% of the surface of Ethiopia (Gole 2003).
The study area is situated between 1800 and 2100 m asl. There is a humid, subtropical climate, with a yearly rainfall of about 1500 mm or more, a short dry season and relatively high cloud cover. A peak in rainfall occurs between July and September (long rainy season) and a smaller peak occurs between March and April (short rainy season). Differences in temperature throughout the year are small with a mean minimum and maximum annual temperature of 11.9 and 26.4 Â°C (Schmitt, 2006).
Species composition and altitude suggest that the forest in the study area is probably best classified as Afromontane rainforest (Van Mechelen 2009; Aerts et al. 2011). This forest occurs in the southwestern part of Ethiopia with an annual rainfall between 750 and 1500 mm (Friis, 1992). This forest occurs in Ethiopia only in the highlands (elevation between 1500 and 2600m). The canopy in the drier part of the rainforest is dominated by Afrocarpus falcatus but as rainfall increases, Pouteria adolfi-friederici becomes more prominent (Demissew et al. 2004). The latter was the dominant tree in the continuous forest where we have done part of the sampling. Below these emergent trees (>25-30m) an almost continuous canopy exists of medium-sized trees with species such as Ilex mitis, Prunus africana, Albizia sp. and Olea sp., Polyscias fulva, Sapium ellipticum and Syzigium guineese subsp. afromontanum (For a full list of sampled tree species, see appendix II). Lianas and epiphytes are widespread in the trees and understory of the forest (Demissew et al. 2004). In canopy gaps, the ground cover is rich in grasses and herb species (Gole et al
2.1.1. Study sites
Garuke: The site of Garuke is located c. 10km northwest of Jimma. The site is named after the village of Garuke (7Â°44' N, 36Â°44' E; elevation 2000-2100m) and can be reached on an asphalt road by car or public transport within 20 minutes from the city of Jimma. The landscape is heavily fragmented and there is a high degree of human disturbance. In this area, forest fragments of different size and management intensity lie within a matrix of grazing land, Eucalypt plantations, small villages or settlements and crop fields, mainly with Maize and Teff (Aerts et al. 2011).
The forest fragments here have been cultivated for quite some time by farmers as coffee plantations. Fragments are owned by different coffee farmers. Each owner works with a different management intensity in both the tree layer and the shrub layer. Management in the tree layer consists of thinning the canopy to create optimal shading conditions for the coffee shrubs. Managing the shrub layer is done by the removal of competing shrubs (Fig. 8A) and old or unproductive coffee. Moreover, planting of new coffee seedlings and pruning of older coffee shrubs (Fig. 8B) for higher productivity are common practice (Aerts et al. 2011). Thinning of the canopy has resulted in the disappearance of almost all emergent trees (e.g. the climax species Pouteria adolfi-friederici) because the shade of these species is too deep for maximum coffee growth, and because the wood is of high economic value (Gole et al. 2003). The trees that currently dominate in this forest type are pioneer or secondary species such as Albizia gummifera, which are typical for secondary or disturbed forest because they have a relative high growth rate in gaps created by coffee management (Chapman et al. 2002). The average size of forest fragments in this region is 4 ha (Van Mechelen 2009). We sampled 18 fragments in Garuke (Fig. 7), ranging in size from 0.25 ha up to 24 ha. The forest fragments in Garuke are considered as semi-plantation coffee production systems (SPC).
Fetche: The site of Fetche (7Â°42' N, 36Â°46' E) is situated less than 10km west-northwest from Jimma but is further from the asphalt road than the Garuke site. The remnant forest here is less fragmented resulting in 1 large cultivated "fragment" of approximately 100 ha surrounded by agricultural landscape. In this fragment we sampled 49 plots. Due to the proximity of Jimma city, this forest is of high interest for the production of coffee beans. Since the forest fragment has many different owners, the management intensity varies from very intensive to intermediate. Besides disturbance due to coffee productivity, the forest is also used for other functions such as firewood collection and charcoal production. The Fetche forest is considered to be a semi-forest coffee production system (SFC).
Gera: The third site we sampled is situated in the Gera sector of the Belete-Gera National Forest Priority Area; a large area of continuous forest with less disturbance and management than the previous sites. This site is situated c. 60km west of Jimma. The road between Jimma and the site is only partially surfaced with asphalt. We visited two areas within this site: Afalo and Qacho. Due to recent deforestation, Qacho (7Â°46' N, 36Â°17' E) is no longer connected to the larger forest complex of the Belete-Gera forest. In Qacho, wild coffee beans are collected and big trees are removed for construction and firewood, especially at the forest edge. At some sites in the forest, the canopy is partially removed and young shrubs were planted (Fig. 9). Deeper in the forest, the wild coffee shrubs grow almost unmanaged in what can be classified as a less disturbed Afro-montane rainforest. Around the small village of Afalo (7Â°38' N, 36Â°13' E) the tree layer was less disturbed, but the shrub layer was often managed with clearance of non-coffee shrubs and even sowing of coffee seedlings. The forest in Qacho and Afalo is considered to be wild coffee forest (FC systems).
2.1.2. Sampling sites
In our study area, we selected forest fragments of different size and different level of management intensity in Garuke and Fetche, and we also sampled the large continuous forest in Gera. Within each forest fragment we sampled a number of plots, depending on the size of the fragment. These plots consisted of one mature tree in the centre of a ground surface plot of 10mÂ·10m. In smaller fragments we sampled less plots than in larger fragments (Table 1). We used a random design and a density of approximately one plot per 0.5 to 1.5 ha. Within forest fragments, sample plots were located >25m apart to guarantee sample independence. Not every fragment was sampled equally because of sampling difficulties related to heavy rainfall, safety conditions or extremely time consuming plots. Nevertheless, a clear positive relation (RÂ²=0.714) remains between the sample size and the size of the fragment. In total, 339 plots were sampled over the three sites. All plots are situated between 1800 and 2100m asl. In Garuke, we sampled 151 plots, in Fetche 49 and in Gera forest 139 plots. The sampling was executed over a period from mid August till mid November 2010 (12/08/2010 - 11/11/2010) with 34 days of active sampling.
2.2. Sampling method
2.2.1. Sampling material
Sampling a tree in a tropical rainforest requires a full search from the base of the tree to the outer canopy. To achieve this target we ascended the trees, whenever we thought necessary, by using the single rope technique (next paragraph). This climbing technique was not injurious to trees and was safe for the researcher (Perry 1978; Jepson 2003).
When the tree was considered strong enough to climb, the rope was placed in the tree with the help of a catapult. With the catapult a throw bag was shot, with a small, light rope attached to it, around a strong branch. The weight of the bag helped to position the rope and to make sure the rope returned to the ground. The actual climbing rope, which was too heavy to shoot directly into the canopy, was then used to replace the small rope. On this rope we climbed using the single rope technique; shunts were used that can be pushed up the rope, when they do not experience a downward force. When a downward force is applied, the shunt locks, thus holding the climber's weight. Via the use of a foot loop, working on the same principle, the climber was able to climb up the rope. For the descent the climber changed the climbing technique while hanging on a lifeline. Then a combination was used of a reverso with a shunt (Fig. 10A).
This technique made it possible to climb almost any tree of any height (Fig. 10B). The material was easy to install and to carry, saving time and man force. But this method also had some disadvantages; there was a restriction considering the supportive capacity of selected trees (Perry 1978). Branches needed to be chosen and tested carefully before the climber could safely begin. Special training was required before actual use and the climber was more or less confined to a stationary lifeline, narrowing the sampling possibilities.
Only the writer of this thesis has climbed and sampled the trees. This has the advantage that the sampling of all the trees has been done by the same person ensuring less variation in counting method. In most cases, the trees were sampled from one point in the tree but sometimes it was necessary to move to other parts in the tree. This was possible by using the same rope or by shooting a second rope at another branch in the same tree. When a new species was found, samples were taken, whenever possible, for identification and for a digital herbarium. The researchers also used binoculars, a Kite Bonelli (10x42) and Kite Petrel (10x42), to observe orchids while hanging in the tree or sampling from the ground.
2.2.2. Sampling of the trees
Mature canopy trees were sampled in each fragment to measure epiphytic orchid diversity. Tree distance to the forest edge (DTE, in m) was measured to assess the effect of forest edge on distribution of epiphytic orchids. For trees located more than 100m from the edge, DTE was recorded as >100m. Since trees in close vicinity of each other tend to have a similar epiphyte flora due to the clumped distribution of many epiphyte species, trees standing well apart, separated by at least 25 meters and with crowns not overlapping, were selected (Gradstein et al. 2003). The height of the tree (H, in m) was defined by measuring the length of the climbing rope and/or through visual estimation.
We preferably selected older and larger trees to maximize the information on orchid diversity. They are usually richer in epiphytic orchids since the orchids had more time to colonize, and the crowns of the trees are more diverse causing a larger gradient in microclimatic conditions (Gradstein et al. 2003; Krömer 2003). Also, bark and canopy structure can have a strong influence on species diversity and composition of epiphytes. Therefore, we tried to sample tree species of different genera or families and we tried to maximize tree species diversity in every sampled fragment (Krömer 2007).
To document the habitat of the epiphytic orchids, the following characteristics of the host tree were measured:
1) Tree height (H)
2) Tree circumference at breast height (CBH, in cm) or height above buttresses
3) Vertical tree zone according to Johansson (1974) (see below and Fig. 11)
4) Estimation of fern species, found as epiphytes in the tree (FERNS)
Each sampled tree was schematically divided into five height zones for sampling (Johansson 1974, Fig. 11): zone 1, which ranges from 0 to 1.5m above ground; Zone 2 from 1.5m above ground to the first major ramification; Z3 from the first to the second ramification; Z4 the middle crown and Z5 the outer crown. Species richness and abundance of epiphytic orchids was determined for each tree zone.
2.2.3. Sampling of the shrubs
Around the randomly chosen tree, a plot of 10m by 10m was established. Within this plot the understory shrubs and treelets (<10m in height and >5cm in CBH) were sampled (Gradstein et al. 2003). The number of shrubs (nÂ° of shrubs) per plot was noted and used as an environmental variable. The CBH of every shrub was measured, the species name was noted and the trunks were surveyed for the presence of epiphytic orchids. Species number and the number of stands for every orchid were recorded. In total, almost 10 000 shrubs were measured and inspected.
2.3. Orchids species and stands
It was difficult to count the individual orchid plants on trees to measure the actual abundance of orchid species. It was sometimes hard, especially in the larger trees, to determine where one individual epiphyte ended and the other began. Many epiphytic orchids formed mats of pseudobulbs connected by, sometimes, long rhizomes. This resulted in large masses of orchids. For this reason we used the "number of stands" in a tree as a measure of abundance. We considered a stand as "a collection of individual stems and/or plants spatially separated from another group of the same species either by an area on the tree devoid of orchids or occupied by another species" (Sanford 1968). Whenever an intermingling of more than one species occurred in the same area on the tree, one stand was counted for each species present (Sanford 1968). Of course this is still open to interpretation and for that reason always the same person counted the orchids in the trees.
For the correct identification of the orchid species, it is important and often necessary to inspect the flowers of the species. Without the flowers some species can be difficult to distinguish from each other (see Results). When correct identification was not possible, we identified up to genus level (Table 2). Species were identified using the Field guide to Ethiopian orchids (Demissew et al 2004) and species identity was confirmed on digital photographs by P. Cribb of the Royal Botanical Garden at Kew. Epiphytic orchids were recorded by codes (A, B, C, etc.) in the field to avoid confusion and facilitate the sampling. This combination of richness values and abundance values gives a good estimation of the orchid communities present at the sampling sites.
2.4. Data analysis
For several reasons we decided to add the observations of the Fetche site to the Garuke site. This allows us to make a direct comparison between two forest types: the fragmented forests (sites Garuke and Fetche, 200 plots) vs. the continuous forests (site Gera, 139 plots).
2.4.1. Summary statistics
First, data were analysed with basic statistics. To identify differences in environmental variables (H, CBH, DTE and nÂ° of shrubs) and epiphytic orchids between tree and shrub layer and between forest types, we calculated simple means with standard deviations (SD). Nonparametric Mann-Whitney U tests were used to statistically test for these differences. We calculated indices for species diversity for every shrub and tree plot and analysed differences in diversity of orchids between forest types with Mann-Whitney U tests. We compared alfa (Î±, mean number of species per plot), Chao (mean richness estimator among runs), Fisher's alfa (parameter of a fitted logarithmic series distribution), Shannon Mean (Shannon diversity index), Simpson Mean (Simpson (inverse) diversity index) and Jack Mean (First-order Jackknife richness estimator) (Colwell 2009).
2.4.2. Community analysis
Before starting the community analysis we tested for outliers in the data using Outlier analysis. Plots, more than two standard deviations away from the mean, were removed from our dataset. We analysed the abundance data of the plots with nonmetric multidimensional scaling (NMS). NMS was used to explore (dis)simmilarities in the abundance data and to investigate indirect gradients influencing species distribution (Aerts et al. 2006). For every NMS ordination, we used Sørensen (Bray-Curtis) as a distance measure, six starting dimensions, 40 iterations to evaluate stability and an instability criterion of 10-5 (McCune & Mefford 2006). NMS dimensions were calculated for both the shrub plots and the tree plots. The dimensions were tested for differences between forest types with nonparametric Mann-Whitney U tests. We calculated Spearman rank correlations between environmental variables and NMS dimensions. After Bonferroni correction, providing a corrected level of significance for multiple tests, these coefficients were evaluated. With mixed model anova's we were able to correct for non independence of the fragments, fragment was here used as a random factor. With Indicator species analysis (ISA) we calculated indicator values (IV) for each species and the overall average p-value. We used the variable 'forest type' as a grouping variable, so for every species, the IV for the different types was calculated. The IV ranges from zero (no indication) to one (perfect indication).
After these analyses we divided the data in two datasets: one, containing the plots of the fragmented forests and one, containing the plots of the continuous forests. Again we used NMS ordination and calculated Spearman rank correlations between environmental variables and NMS dimensions. One environmental variable (fragment size, 'Area', in ha) was added for the correlations with the NMS dimensions of the fragmented forests.
With Cluster analysis (CA), data were clustered into groups using Sørensen (Bray-Curtis) as a distance measure and a flexible beta of -0.25 as group linkage method (Aerts et al. 2006). To determine the optimal number of groups in the Cluster analysis, we used ISA on each grouping variable, which is output from the CA. Indicator values for each species and the overall average p-value were calculated. The last cluster step that adds >0.05 significance to the average p-value was selected as the most informative number of clusters. Nonparametric Kruskal-Wallis ANOVA and pair-wise comparison were used to test for differences between clusters and variables lacking homogeneity of variance (Aerts et al. 2006).
The nonparametric multiresponse permutation procedure (MRPP) test is used for testing multivariate differences among pre-defined groups. We tested for differences in community composition between the groups (clusters) in our ISA. Again we used Sørensen (Bray-Curtis) as a distance measure. The group weighting factor was n/sum(n) (with n, the number of sample plots in each group) (Aerts et al. 2006). The test statistic (T) describes the separation between groups while the chance-corrected within group agreement (A) describes within-group homogeneity compared to random expectation. A=1-(observed delta/expected delta). When all items are identical within groups then A equals 1. If heterogeneity within groups equals expectation by chance, then A=0. If heterogeneity within groups exceeds expectation by chance then A<0. If there is more homogeneity within groups than expected by chance, then 1>A>0. In community ecology values for A are commonly below 0.1 (McCune and Mefford 2006).
2.4.3. Correlation of tree species with orchid richness
We used one-way ANOVA to compare diversity between tree species. Because not every tree was sampled equal times, we applied post-hoc Tukey's HSD (Honestly Significant Difference) test. This is a single-step multiple procedure, comparing all possible pairs of means, in conjunction with the ANOVA to find which trees are significantly different from each other (Linton & Harder 2007).
2.4.4. Vertical distribution of orchids
We used nonparametric Mann-Whitney U tests to compare orchid abundance in the shrub layer between forest types. We used t-tests, independent by groups, to compare the relative orchid abundance for each tree zone between forest types.
Outlier analysis, clustering, ISA and NMS ordination were perfomed in PC-ORD (Version 5.0 for Windows, McCune & Mefford 2006 ). For statistical tests, we used Statistica (Version 8.0 for Windows), except for the mixed models, which were run in SPSS (Version 18.0 for Windows, IBM, SPSS Inc., Chicago, IL). Diversity indices were calculated with EstimateS (Version 8.2.0, Colwell 2009).