Vegetation Composition in Hugumbirda-Gratkhassu National Forest Priority Area , South Tigray

The floristic composition and structure of the vegetation of Hugumbirda-Grat-Khassu forest, South Tigray, Northern Ethiopia is described and related to environmental factors. To analyze the vegetation and environmental data seventy-four relevè ( 20m x 20m) were used. For each species the cover/abundance value was estimated. Height and diameter at breast height (DBH) of all woody individuals taller than 2 m and thicker than 2 cm were measured. Importance Value Index was calculated for 24 tree/shrub species and the result helped to identify the five dominant tree/shrub species and to show the overall forest situation. The species and relevès were classified with the two-way indicator species analysis program TWINSPAN. We recorded 102 species belonging to 83 genera and 50 families. Five community types are described: Allophylus macrobotrys-Ficus sur, Nuxia congesta-Podocarpus falcatus, Acacia abyssinica-Olea europaea, Myrica salicifolia-Erica arborea-Maesa lanceolata and Acacia etbaicaDichrostachys cinerea-Acacia tortilis type. Of these, community type 2 has high species richness while community type 3 is poor in species richness. The general arrangement of all species was found to show high density at lower height and DBH classes. Based on the cumulative results of Importance Value Index, Juniperus procera, Olea europaea subsp. cuspidata, Nuxia congesta, Cassipourea mallosana and Olinia rochetiana were identified to be the most dominant tree species of the forest and they contributed 71.43% of the basal area. Analysis of community-environment relationships didn’t show significant differences except for altitude and slope.


Climate
Metrological data of monthly maximum and minimum temperature and monthly rainfall were taken from two stations (Alamata and Korem) for the period 1978 to 2001.One of the nearest metrological station is located in Alamata town at a distance of 5 km from Grat-Khassu. Here the mean annual minimum temperature is 14 o C and the mean annual maximum temperature is 29.8 o C. The hottest month is June with a mean maximum temperature of 34.3 o C, and the coldest is October with a mean minimum temperature of 10.7 o C. The mean annual precipitation is 750.9 mm.
The other nearest metrological station is in Korem town. Here the mean annual minimum temperature is 8.8 o C and the mean annual maximum is 22.1 o C. The hottest month is June with a mean maximum temperature of 24.3 o C, and the coldest is December with a mean temperature of 6.1 o C. The mean annual precipitation is 998.9mm.

Vegetation
Formerly, the area was covered with dense forest composed of different indigenous species.
According to Zenebe Gebre-Egziabher et al. (1998) and information obtained from local informants, the natural forest was exploited by Italian concessionary named Montu Doro who installed sawmills at Hugumbirda in 1950 with the permission of the then governor of Welo province.
The forest was officially put under the auspices of State Forestry Agency in 1965 (SFCDD, 1997). Then in 1981 the area was identified as one of the National Forest Priority Areas (NFPAs). Boundary demarcation, which is the basis for the current management of the forest, was undertaken in 1993. Based on this demarcation, the project covers a total area of 21, 654.24ha. Out of this about 532.75 ha is plantation forest whereas the rest contains disturbed natural high forest, bushes, shrubs, agricultural plots and settlement area.

Botanical composition survey
1. Rare-generally only one individual 2. Sporadic (few) -less than 5% cover of the total area 3. Abundant with less than 5% cover of the total area 4. Very abundant and less than 5% cover of the total area 5. 5-12% cover of the total area 6. 12.5-25% cover of the total area 7. 25-50% cover of the total area 8. 50-75% cover of the total area 9. 75-100% cover of the total area DBH (diameter at breast height) was measured at 1.3m using callipers for shrubs and trees that had a DBH equal or greater to 2cm (Martin, 1995;Tuxill and Nabhan, 2001). Height of trees and shrubs greater or equal to 2m was measured using hypsometer for taller trees and shrubs and measured pole marked at 0.5m intervals and 4m long (Tuxill and Nabhan, 2001;Bongers et al., 1988). The pole is held vertically at the base of the tree/shrub, enabling the observer standing far enough away from it to see the base and the top of the stem. The number of pole length was then counted to estimate the height of the plant. Those plant species having height and DBH less than two are counted by species.

Environmental data
The environmental variables measured for every stand (relevè) selected include altitude, slope, aspect (exposure), soil depth and estimate of grazing intensity and human impact.
Altitude was measured using an Everest Altimeter, slope was measured using clinometers, and Aspect was determined using Suunnto compass. As a possible indicator of total solar energy, aspect was codified according to Zerihun Woldu et al. (1989). Thus N=0; E=2; S=4; W=2.5; NE=1; SE=3; SW=3.3 and NW=1.3 From the corners and centre of each stand, soil depth was measured using a measured long metal marked at 10cm interval and 1.25m long. Average depth of these measurements was then taken for the stand soil depth. Then nature of the measured depth was determined following Parent (2000). The details are described below.
The state of human interference at each relevè was estimated following Gebremedhen Hadera (2000) and Kumlachew Yeshitla and Tamrat Bekele (2002) and with modifications. A 0-3 subjective scale was taken into consideration to record the presence or absence of stumps, logs and signs of fuelwood collection. Therefore, the magnitude of the impact was quantified as follows 0=nil; 1= low; 2= moderate; and 3=heavy

Vegetation data
Two-way indicator species analysis, which is a recommended technique for its robustness and effectiveness (Gauch and Whittaker, 1981), was used to classify the vegetation data.
TWINSPAN is a divisive polythetic method of vegetation classification; it classifies both relevès and species. The computer program used was the TWINSPAN program (Hill, 1979). The groups obtained were characterized as local plant community types, and described as "type", which were provisionally characterized by dominating and/or characteristic species, mainly trees and shrubs. A dominating species in this case is a species having a synoptic cover-abundance value (mean frequency x mean cover-abundance) (van der Maarel et al., 1978) of at least five, and a characteristic species having a high frequency in the type and a lower frequency in most other Leul, K, W., Tamrat, B and Sileshi, N (MEJS) Volume 2 (2): 2010 © CNCS Mekelle University 33 types. The community types identified were further characterized by means of environmental factors, which appeared to be correlated to the floristic composition of the type.

Structural data
The structure of the plants was described in terms of tree density, diameter and height. Tree density was computed by converting the count from the sample plot to a hectare basis. The diameter at breast height (DBH) was classified into 14 DBH classes and the percentage distribution of trees and shrubs in each class were computed. Height was classified into 9 height classes and the percentage distribution of the plants in each class was calculated.
The following parameters and index were calculated to determine the vegetation structure and the dominant species of the forest: for those individual species having DBH greater than 10cm and relative frequency greater than 0.60 %, relative density, relative frequency, relative dominance and importance value index (IVI) (Muller-Dombois and Ellenberg, 1974;Misra, 1974) were calculated using the following formulae:

Relative density = Number of individuals of a species/ Total number of individuals of all species x 100
Relative frequency =Frequency of a species/ Sum frequency of all species x 100 Relative dominance=Basal area of a single species/Total basal area of all species x 100 Importance Value Index (IVI) = Relative density + Relative frequency + Relative dominance.

Environmental data
To get a mean value for the various environmental parameters of each distinct plant community type, the values for all the relevè that make up the particular community type were added and averaged. One-way analysis of variance (ANOVA) was performed to detect variation among the community types with respect to any one environmental parameter. Tukey's tests were performed to detect significant differences among the different means of the environmental parameters of each community types. The correlation of the various environmental parameters among each other was evaluated by calculating Pearson's product moment correlation coefficient.
Leul, K, W., Tamrat, B and Sileshi, N (MEJS) Volume 2 (2): 2010 © CNCS Mekelle University 34 Meteorological data obtained from two stations of the study area were analyzed for mean monthly maximum and minimum temperature and mean monthly rainfall registered for 20 years.
Clima diagram was produced from these results.

Voucher Plant Specimen Collection and Identification
Every time a new species was encountered in the relevè, a specimen was collected in duplicate, numbered, dried and placed in a reference collection following standard Herbarium procedures (Bridson and Forman, 1992). The specimens were then identified by comparing them with already identified specimens in the National Herbarium of Ethiopia (ETH) and by referring to

Vegetation Classification
Five community types were obtained from the classification out put. Their description based on the dominant and characteristic species having synoptic cover-abundance values greater than one in at least one community type and their altitudinal distribution is as follows.

Allophylus macrobotrys -Ficus sur type
This community is found at altitude from 2215-2246. Allophylus macrobotrys is a dominant species and Ficus sur is characteristic species of the type. Associated species in the type include Ekebergia capensis, Teclea simplicifolia, Cassipourea malosana and Nuxia congesta.

Regenerating species of Grewia mollis, Maytenus undata, Pavetta oliveriana, Vernonia rueppelli
and Buddleja polystachya are abundant under the canopy layer. Woody climbers such as Leul, K, W., Tamrat, B and Sileshi, N (MEJS) Volume 2 (2) The results of Pearson's product-moment correlation of the environmental parameters show that some of the environmental parameters are correlated (Table 3). Altitude is positively correlated to soil depth and slope, and negatively correlated to human impact, species richness and aspect.
Species richness decreases with increase in altitude. Slope is negatively correlated to both grazing and human impact. Grazing and human impact are positively correlated to each other and are negatively correlated to soil depth.

Vegetation Structure
Height and DBH measurements were used to construct the density distribution for the various categories. The distribution of trees in different height classes is shown in Fig.2

Number of species
Correlation is significant at the 0.01 level **.
Correlation is significant at the 0.05 level *. DBH measurements reveal a trend similar to that of the height distribution (Fig.3). Most individuals have a diameter less than 20 cm (86%). About 36.4% belong to the lowest diameter class (<2 cm), and 19.1 % to the next higher class (2-5 cm). Only 0.4 % have DBH grater than 50 cm. The result confirms that the number of individuals decreases as the height and DBH of the individual's increase.    Leul, K, W., Tamrat, B and Sileshi, N (MEJS) Volume 2 (2) Relative density, relative frequency, relative dominance and importance value index were computed for 24 tree/shrub species with DBH greater than 10 cm and relative frequency greater than 0.60 %. The results of the analysis are presented in Table 4.

DISCUSSION
Five community types were identified from the classification strategies. These include community types 1, 2, 3, 4, and 5.
Community type 1, which is dominated by Allophylus macrobotrys and characterized by Ficus sur, is found in specialized habitats such as along river courses. The stands sampled in this type are located at the middle of the forest, which is less grazed by cattle and its human impact is found to be low. Regenerating species of Hagenia abyssinica and Podocarpus (Afrocarpus) falcatus are common here.
Community type 2 with good timber species: Juniperus procera, Olea europaea subsp. cuspidata and Podocarpus falcatus has experienced human interference in the form of selective cutting. Cattle interferences were also observed in some of its stands.
Community type 3 is rich in shrub layer species and woody climbers. The stands sampled in this community are located in an area having shallow soils with medium human interference in the form of firewood collection and selective cutting. This might be due to being near to the farmers settlement area. Although most area of this stands was highly affected before about 20 years being used as farming land, by now it is in good regeneration status. In most of its stands introduced exotic species of Cuppresus lusitanica and Pinus patula have been observed.
Community type 4 Myrica salicifolia-Erica arborea-Maesa lanceolata distributed at higher altitudes of the sampling site is unique in its own type having species different from other types. This community is surrounded by upper slope and bare rock area, thus little disturbance by Leul, K, W., Tamrat, B and Sileshi, N (MEJS) Volume 2 (2): 2010 © CNCS Mekelle University 41 human and cattle is encountered. Even though lianas are not recorded in this type it is having good species richness.
Community type 5 is highly influenced by people collecting firewood, charcoal making and grazing animals. This is due to its being nearby to Alamata town and having species of plants suitable for charcoal making and firewood. The stands of this community are from an area, which is at lower altitudes, having shallow soil and that receives lower amount of annual rainfall with higher temperature.
The community types don't show significant difference (P< 0.05) with respect to the environmental variables except for altitude and slope. The major discriminate among the community types is due to altitude. This is in line with the result observed by Kumlachew Yeshitla and Tamrat Bekele (2002) and Zerihun Woldu et al. (1989). The community types could be grouped into three based on altitude: group 1 with altitude 2250-2379 m (types 1, 2 and 3), group 2 community number 4 and group 3 community number 5 (with P value of 0.124). On the other hand, the communities were grouped in one in respect to aspect, grazing and number of species. All communities were homogenous with a P value of 0.651 in respect to aspect.
Density distribution at different height classes in the study area showed that 33.3 % of the individuals have a height less than 2 meters and 24.9 % fall between 2-5 meters range. Figure 2 shows that 78.8 % of all individuals are 9 meter or shorter. Individuals greater than 29 meters in height are rare (0.7 %). Thus the study confirms that the number of individuals decreased as the height of the individuals increased indicating long time disturbance (thus less mother trees).
Such result was also observed in Chilmo forest by Tamrat Bekele (1993).
Density distribution at different DBH classes also showed similar trend as that of height class distribution. It indicates dominance of small sized individuals. The pattern of such density can be an indicator for community dynamics in the forest. The total number of trees/shrubs in each DBH classes decreased with an increasing in DBH classes. This relationship also was observed in Dessa forest by Gebremedhin Hadera (2000); in Chilmo and Menagesha forests (Tamrat Bekele, 1993) and the tropical lowland rain forest of Los Tuxtlas, Mexico (Bongers et al., 1988).
The forest pattern is formed by the species structure with reversed J shaped in DBH class distribution. 36.4 % of the individuals in the forest have DBH less than 2 cm (Fig. 3).
Relative density distribution of the species showed Juniperus procera (14.47 %) followed by Olinia rochetania (9.63 %) and Olea europaea subsp. cuspidata (9.44 %) to have the highest Leul, K, W., Tamrat, B and Sileshi, N (MEJS) Volume 2 (2): 2010 © CNCS Mekelle University 42 relative density compared to the others. About twenty-nine per cent of the tree/shrub species have a relative density greater than five, while 70 % of the species have a relative density of less than 5 %. Analysis of the frequency distribution indicated that Juniperus procera, Rhus natalensis, Olea europaea subsp. cuspidata, Nuxia congesta, Rhus glutinosa and Olinia rochetania were found to be with the highest relative frequency indicating their good distribution throughout the forest (Table 4). Relative dominance, which is the basal area of a single species, divided by total basal area of the species ranges from 0.05 % (Dombeya torrida) to 30.34 % (Juniperus procera). Thus the contribution of each species to the basal area differs from one another. Species such as

CONCLUSION
The survey showed that the forest is dominated by small sized tree and shrub species in secondary stage of development, indicating that the forest was heavily exploited and affected in the previous periods, but good regeneration is in process at the present time. Therefore, to improve the natural diversity and structure of the forest, to minimize the influence of the surrounding communities and utilize the forest resources sustainably for present and future generation, the basic needs and traditional rights of the communities over the uses of forest resources should be recognized. The much-needed positive attitudes towards forest protection and development can only be obtained from the rural communities through the development of a genuine benefit sharing mechanism. Thus community participation is quite important.

ACKNOWLEDGMENT
We would like to thank Dr. Tesfaye Bekele, head of Forest Resources Conservation (MoA), who have provided us with a vehicle during reconnaissance survey. We are grateful to Ato Gebremedhin Hadera, head, and manager of woody biomass inventory and strategic planning project, and Ato G/Kidan Teklu, Natural Resources Conservation Department (MoA), for their material (Hypsometer, DBH caliper) and moral support. Thanks to National Metrological Service Agency for providing metrological data of the study area. We thank Swedish Agency for