RELATIONSHIPS BETWEEN FRUGIVOROUS- BIRDS AND THEIR HABITATS IN FRAGMENTED FOREST AREAS NEAR TO VILLAVICENCIO CITY, VILLAVICENCIO, META, COLOMBIA

The degradation of the Llanos foothill ecosystems is one of the biggest ecological problems in the areas surrounding the city of Villavicencio, Meta. The generation of fragments produces new conditions for the establishment of bird guilds. This study was carried out during 2017 2018 in areas of the campus Loma Linda at the Santo Tomas University. It was evaluated whether the forest interior, forest border and grassland areas constitute habitats that support the establishment of frugivorous birds. The key factors that affect the richness, diversity and abundance of bird species were identified, from which recommendations were made for the establishment of conservation measures. The richness of birds and their abundance was positively correlated with intermediate values of plant richness and diversity and with the presence and coverage of species of palms and lianas. The abundance and diversity of frugivorous birds were higher in border areas and lower in grasslands and forest interiors. Apparently the diversity of the habitat in border areas of fragmented lands increases the diversity of frugivorous birds through an increase in the abundance values. Artificial enrichment is recommended with different tree, shrubs and palms species planted in the edge and interior areas of the forest, and the maintenance of remaining trees and shrubs in the areas of shrub grasslands.


INTRODUCTION
Habitat quality models that relate the occurrence of bird assemblies with the characteristics of the occupied habitat have been frequently referenced in the literature (KENNEDY et al., 2011). Such models are influenced by the spatial scale that is used which includes not only the hierarchical structure or interdependence of the different environmental variables that are considered, but also the hierarchical responses or interdependence of the birds towards the habitat they are occupying (SMITH et al., 2011;KENNEDY et al., 2010). However, there are still information gaps related mainly to the predictive capacity of the models when trying to associate particular species of birds with particular types of occupied habitats at different spatial scales (KENNEDY et al., 2010).
Although different studies of bird-habitat relationships have used different hierarchical approaches using regional scales, landscape level or plot scale, with greater emphasis on the latter (LEVEAU, 2019;SHOFFNER et al., 2018;FREEMAN et al., 2015), there are few studies that have attempted to relate the influence of habitat distinctive types on the occurrence and characteristics of the habitat by species of frugivorous birds. In the eastern plains of Colombia and particularly in the vicinity of the city of Villavicencio, piedmont ecosystems are susceptible FLORESTA to human disturbances such as deforestation, erosion and landslides, associated with an increase in urban pressure on these areas. Although there is no documented evidence of the effects of these activities on the structure and functioning of these ecosystems, it is assumed that such activities have an impact on resident bird populations, including frugivorous. Importantly, the remaining habitats can act as a refuge for a number of species of frugivorous birds. Environmental factors and the type of habitat or biotope occupied also play an important role in the conformation and persistence of bird assemblies (KRYNSKIA; GOLAWSKI, 2019;IMAI et al., 2017;FREEMAN et al., 2015;ESCRIBANO-AVILA et al.,2014).
Santo Tomas University at its Loma Linda campus has a forested relict of 8 ha in the process of natural regeneration that includes fragmented areas, grassland areas and a road that crosses part of the campus. However, no studies have been conducted on the composition and distribution of frugivorous bird species that may potentially favor the natural regeneration of the site, nor have the characteristics of the habitat occupied that may increase habitat availability for potentially useful species for restoration been identified. The succession of the vegetation depends to a large extent both on the seed rain dispersed by birds and from the buried seed bank in the ground (VELOSA, et al., 2018;VALOIS -CUESTA, et al., 2016). This study relates the variation in the composition and distribution of frugivorous bird species with the characteristics of the habitat occupied in three (3) piedmont biotopes (forest interior, forest edge and weed grassland). Habitat types play an important role in the formation and maintenance of bird guilds in modified environments (KRYNSKI; GOLAWSKI, 2019). Recommendations are also made for the management of such habitats that could be useful for the conservation of frugivorous birds in fragmented foothills.

MATERIALS AND METHODS
The study area is part of the green belt of the municipality of Villavicencio within the sub-humid forest biome and is located on the Loma Linda campus of Santo Tomas University, (4 ° 06'38.83``N; 73 ° 39'26.41 ' 'O), eastern slope of the eastern mountain range, at a height of 467 m above sea level, on the right bank of the road that leads from Villavicencio to the municipality of Acacias, Meta ( Figure 1 ; the soils are characterized by a light to moderately steep relief, with slopes 25-75%, poor in nutrients, strong reaction to extremely acid, and moderately thick to thin textures.

Establishment of sampling transects
In each of the biotopes, except in the open areas of the campus, 50 m x 2 m transects were established in line to determine the floristic composition and structure (vertical and horizontal) of the vegetation at the rate of: five (5) transects in the interior of the forest, two (2) transects at the edge of the forest and two (2) transects in the weed grassland. The transects were separated by a distance of 50 -100 m between their central points to ensure independence in bird censuses.

Bird censuses
For the censuses of birds made from the center of each transect, the method of counting at fixed radius points was used. At the midpoint of each transect, a period of 10 minutes was used, registering the species and the distance from the detection point to the center of the transect. The birds detected were recorded within a radius of no more than 50 m from the observation point of each transect and those that escaped as the observer approached the transect. Each fixed observation point was censored 15 times during a study period of 7 months in two different observation periods, namely: from 6:30 a.m. to 9:00 a.m., and from 3:00 p.m. to 5:30 p.m. Frugivorous birds detected outside the transect zone were also recorded, but not included in the habitat utilization analysis.

Habitat variables
Vegetation structure: The floristic structure and composition was determined based on all individuals with CAP> 4.0 cm. For each individual registered within the plot, the following information was established: Circumference at breast height -CAP (cm), Total height (m), Shaft height (m), Larger Diameter -Dm (m) and Minor Diameter -Dn (m) of the crown. Likewise, the coordinates of the start and end point of the transect were recorded, together with the height s.n.m. of each point. In each transect the number of species, the coverage by plant species in percentage (discriminated for tree, shrub, herbaceous, palms and liana species) and the presence of plant species in percentage (discriminated for tree, shrub, herbaceous, palms and liana species) were recorded. The midpoint of each coverage class and presence defined was averaged in order to obtain values of the two variables for each transect.
Floristic composition and diversity: No specimens were collected and their taxonomic determination was carried out by the dendrologist Francisco Castro. As an expression of alpha diversity, Fisher's alpha diversity index was calculated using the Past vers program. 3.0. As an expression of the homogeneity of the floristic composition between transects, the Bray-Curtis index was used to generate the similarity matrix and the classification dendrogram was developed.

Statistical analysis
An analysis of variance (ANOVA) was carried out to test the null hypothesis that there are no differences in the composition/abundance of the avian community among the biotopes considered. The Tukey multiple comparison test (p = 0.05) was used to establish the differences between the biotopes. Starting from the matrix of similarity generated, a dendrogram was constructed to group the bird species by means of the technique of average grouping into groups. The multivariate technique called "factor analysis" (F.A.) was used with a solution of main components to group bird species according to their relative position within the habitat of the study area. To obtain the coordinates corresponding to each of the observations (species), a regression analysis of the factors generated from the analysis of factors on each of the bird species was performed. The resulting regression coefficients were used as an ordination axes, taking two factors at the same time to graphically represent the position of each species within the study habitat.
Although the Factors Analysis provides information on the bird-habitat relationships of individual species, its main objective is to determine the relative effects of the habitat on the community or bird guild included as a whole. For this reason, Generalized Linear Models (GLM) were used to determine the key habitat conditions that determine the variations in composition and diversity of frugivorous bird species among the biotopes considered. A separate GLM was used for each of the dependent variables included: bird abundance, bird richness and bird diversity according to Shannon index. According to the distribution of the dependent variables, the generalized linear model with the Poisson distribution function was used for both the abundance, richness and diversity of birds according to Shannon index. The independent variables included in the models were the values of Factors 1, 2 and 3 derived from the Factors Analysis, and the following habitat variables: Presence (in percentage) of tree, shrub, herbaceous, liana and palms species; average coverage (in square meters) of tree, shrub, herbaceous, liana and palms species; number of species and plant diversity index (Fisher's alpha). Only results with a probability α ≤ 0.002 were considered statistically significant. Statistical analyses were performed with Past version 3.2.

RESULTS
The censuses of frugivorous birds included 2,103 individuals belonging to 30 species. Of the total, 1,128 individuals (53.6%) were located in forest edge, 861 (40.9%) in grassland areas and only 114 (5.4%) individuals in forest interior areas (Table 1). ANOVA rejected the null hypothesis that there are no differences in the frugivorous avian community between the habitat types considered (kw = 37.74, df = 2, p = 5.1 x10ˉ ⁹). The results of the multiple comparison test indicate that only the forest and grassland edge habitats were not significantly different (kw = 2,364, df = 2, p = 0.1239) with respect to the composition of the frugivorous avian community. The rest of the comparisons between habitat types showed significant differences in the composition of the avian community (p ≤ 5.3 x 10ˉ⁶).
Habitat overlap dendrograms are useful for visualizing relationships within the bird community. In the elaborated dendrogram, three groupings were defined (Figure 2). One of them, made up of the species Trogon viridis (L.), Ortalis guttata Spix, Catharus minimus Frederic de Lafresnaye and Momotus momota (L.) was characterized by the absence of records of these species in the grassland biotope. The second grouping consisting of the largest number of frugivorous species was characterized by its high abundance in the forest edge biotope. Among these species, the following stand out: The factor analysis performed produced 3 factors which explained 100% of the original variance. Factor 1 was interpreted as a tree-shrub gradient which was related to the abundance of tree-shrub species, the cover of tree species, the richness of plant species and the Fisher's alpha diversity index.
Factor 2 was interpreted as an herbaceous gradient resulting from the combination of the abundance and coverage of herbaceous species and the abundance of palm species. Factor 3 represented a gradient of abundance and coverage of liana species associated with the coverage of palm species.
The graphs of the regression coefficients of the species of frugivorous birds taken two factors at the same time illustrate the "location" of the species within the studied habitat. Figure 3 illustrates the position of the centroids of bird species in relation to the tree-shrub gradient (Factor 1) and the herbaceous gradient (Factor 2). The highest values of richness and diversity of species were recorded in the mature forest areas within the forest. Weed grassland areas had the lowest values of presence and coverage of treeshrub species, and richness and diversity, but they were the areas with the greatest coverage of herbaceous species and a high number of seedlings of palm species.
The richness and diversity of plants (Fisher's alpha) had a significant effect on the number (GLM; G = 9,456, p = 0.002 and G = 10,938, p ≤ 0.001 respectively) and abundance of frugivorous bird species (GLM; G = 40.46, p ≤ 0.001 and G = 44.74, p ≤ 0.001 respectively). The richness, abundance and diversity of frugivorous bird species was lower in forest interior areas compared to those with forest edge and weed grassland (Figure 4). The richness and abundance of birds was also dependent on the coverage -and presence of palm species (GLM; G = 10.12, p ≤ 0.001 and G = 55.1, p ≤ 0.001 respectively) and the coverage -and presence of liana species (G = -10.39, p = 0.008 and G = 33.29, p ≤ 0.001). Above all, the abundance and diversity of frugivorous bird species was higher in forest edge areas and lower in weed grasslands. Although not significantly, the type of vegetation covers most related to the abundance of bird species was recorded inside the forest (MANOVA, F 1.43 = 3.516, p = 0.06) ( Figure 5). The diversity of frugivorous birds (Shannon H') was not significantly related to any of the measured habitat variables. Habitat variables grouped in Factors 1, 2 and 3 of the Factors Analysis did not significantly affect the richness, abundance and diversity (Shannon H´) of frugivorous birds (Table 3).

DISCUSSION
Although only a small number of habitat types were included, the results of ANOVA and the multiple comparison test show significant differences in two (2) of the three (3) comparisons made with respect to the composition of frugivorous bird species. In the study area, vegetation types were considered distributed along different gradients with respect to the habitat of frugivorous birds. In one gradient, a type of vegetation is dominated by the presence and coverage of tree-shrub species and by the richness and diversity of plants ( 1) and in the other, by the presence and coverage of herbaceous species and by the presence of palm species (Factor 2). An additional gradient (factor 3) was associated with the abundance and coverage of liana species and the coverage of palm species.
The reported evidence regarding variations in the composition of frugivorous bird species indicates that significant differences can be found between specific units of differentiated local vegetation such as biotopes, which represent portions of important habitats for birds. In this study and in an area of 8 ha. a total of 30 species of frugivorous birds corresponding to 30% of the total bird species reported for the area were recorded. It is remarkable the occurrence of 11 species of birds of granivorous-frugivorous habits, including the following with low abundance in the study area: Trogon viridis, Brotogeris jugularis,, Pteroglossus inscriptus and Psarocolius angustifrons. Abundance of species such as Ortalis guttata, Columba cayannensis, Manacus manacus, and Cyanocorax violaceus were higher in edge forest areas in comparison with those of adjacent weed grasslands. This suggests that forest edges may be important for the conservation of frugivorous birds in fragmented areas of piedmont since they can be occupied by multiple species, including some with low population abundance.
The definition of habitat factors that explain the total variability in the occurrence of frugivorous birds in fragmented areas is difficult. Although in this study the variability explained by the first two factors of the Factor Analysis was 89.3%, not all the variables that could have a significant effect on the variability of the composition of frugivorous birds were included. Factors such as the analysis of food habits of the reported species and the distribution and abundance of fruits were not included. However, the dynamics of changes in plant cover due to the fragmentation effect in the study area and the definition of three (3) specific groupings of bird species associated with such changes suggest a composition of frugivorous birds without much rotation, which would lead to the establishment of relatively stable bird guilds.
The factors that significantly determine the variability of the species according to the type of biotope they occupy, may reflect the bird's preferences for specific habitat types. In this study, the occurrence of Ortalis guttata, Trogon viridis, and Catharus minimus was positively correlated with high values of both richness and diversity of plant species, tree and shrubs cover and palms species. These species recorded the highest abundance values in forest edge areas compared to areas of weedy grassland and forest interior, which reflects the preference of these birds for forest edge. In contrast, the presence of species such as Brotogeris jugularis, Psarocolius angustifrons and Psarocolius decumanus was positively correlated with low values of richness and diversity of plant species and high values of coverage of herbaceous species.
The role played by habitat diversity in determining the richness of bird species in fragmented and urban or semi-urban areas has been emphasized by different authors (PAKER, et al., 2014). In addition to this factor, the coverage of treeshrub species and palms, apparently plays an important role in maintaining high levels of richness of frugivorous bird species in fragmented areas of foothills llanos. In the study area, despite its small size, seven (7) species of palms (Arecaceae) were registered that provide food for the birds of the site.
Discriminated by biotope, the richness and abundance of frugivorous bird species was greater in forestedge areas compared to areas of weedy grassland and forest interior. Presumably, habitat diversity in forest-edge areas of fragmented areas of llanos piedmont increases the diversity of frugivorous birds through an increase in bird abundance. Given the differences in the composition of the guild of frugivorous birds between biotopes, it could be inferred that their distribution is at the same level at which the bird species discriminate. However, this statement requires for its verification of studies on the differential preponderance of the species of frugivorous birds in a greater number of biotopes.

CONCLUSIONS
• The forest edge areas constitute an important refuge for different species of frugivorous birds.
• The type of habitat occupied affected the richness and abundance of frugivorous bird species.
• The maintenance of remaining trees and shrubs in the weedy grassland areas as catalytic nuclei of the succession process, the artificial enrichment with various tree -shrub and palm species of the edge and interior areas of the forest, can contribute to the retention and maintenance of valuable habitats for frugivorous birds.