Study area and site selection
The study was conducted in Southern Ethiopia, Silte zone, Kibet town agroforestry system (Fig. 1). The town is located 148 km from Addis Ababa on the road between Butajira to Arabaminch. It is 173 km far from Hawassa and 26 km from Werabe the capital towns of the Region and Zone, respectively. Its geographical location is approximately at 07°56′ N and 38°14′ E. Its mean annual temperature is 18.8 °C and its annual rainfall pattern is 1200.5 mm which is the climatic data of the closest town of Butajira since there is no available data for the district [15].
The agroforestry system in the study area mainly consists of the chat (Catha edulis), coffee (Coffea arabica L.), and Enset (Ensete ventricosum). Fruit trees such as Sweet orange (Citrus sinensis L.), Lime (Citrus aurantifolia), Avocado (Persea Americana Mill), Mango (Mangifera indica L.), and Banana (Musa paradisiacal L) are common. Unlike chat and coffee, fruits and vegetables are produced for consumption and local market. There are also several multipurpose tree varieties such as Tid (Juniperus procera), Zigba (Podocarpus gracillor), Weira (Olea Africana), Sholla (Fiscus gur), Bisana (Croton macrostachyus), Wanza (Cordia africana), and different Acacia spp. provide various services including fuelwood.
The vegetation cover of the field within the survey area was mapped during the bird survey. Fields were classified into four categories in terms of their field vegetation cover. These are home gardens, live fences, parkland, and fast-growing eucalyptus woodlots.
Home gardens (Hg) are composed of a high diversity of plants and an important source of household subsistence and cash needs which are characterized by being near the residence. The dominant components of the home gardens are coffee, enset, pepper, various annual crops, and numerous kinds of vegetables.
Live fences (Lf) are barriers of closely spaced trees or shrubs to protect crops or structures against livestock and human interference. It may be established all around the farm, but it is commonly established around the homesteads and gardens. Plant species like Koshim (Dovyalis abyssinica) are common native tree species to be promoted for this purpose other such as Qentaffa (Pterollobium stellatum), Quelqwal (Euphorbia abyssinia), and Acacia spp.
Eucalyptus woodlot (Ew) where woody perennial is planted and managed over time to produce fuelwood, and poles. The most common eucalyptus is the Eucalyptus globulus.
Parkland (Pl) involves the growth of individual trees and shrubs in wide spaces in croplands. Some good examples of this practice are Cordia africana and Acacia albida intercropping with maize.
Study design
The study was conducted from December 2019 to early October 2020 encompassing both wet and dry seasons. We assigned four study sites that reflected the impact of changing agroforestry type on bird diversity in the Kibet town agricultural landscapes. Fifteen sampling sites were established encompassing four study sites (habitat types): Home gardens (Hg, n = 5), Live fences (Lf n = 4), Eucalyptus woodlot (Ew, n = 3), and Parkland (Pl n = 3). We established a line transect in each sampling site with 500-700 m in length [8]. Each line transect was 100 m far away from the roadside to avoid edge effect and 300 m far away from each other to avoid double counting of the same individual of a species following the work of [6].
Feeding guilds were classified based on direct observations and available literature on feed bioecology [16].
Sampling method
Data were collected from December 2019 to April for the dry season and from late May to early October 2020 for the wet season. Four field visits (two visits in dry season and two visits in wet season) were conducted to observe the composition and diversity of bird species. Line transects were surveyed in each sampling site and pooled the data for analysis. The transect line was walked at a constant pace for approximately 30 min. Species observation mostly took place through vision and also through acoustic.
Bird survey
In all transects walk at a steady pace (30 min) between 7:00 am and 9:00 am in the morning and 4:30 pm to 6:00 pm in the late afternoon when most birds are active [6]. In the home garden area footpaths were used. To minimize disturbance during the count, a waiting period of 3 to 5 min before counting was applied [40]. No count was conducted in the presence of passing vehicles [28]. Species observation mostly took place through vision by using 8 × 17 binoculars and eyes, but also through sound. Overflying species were not included in the count.
Identification was visual except in some rare cases when the voice will be used if the bird cannot be seen. Identification and categorization of birds to their respective taxonomic groups’ done following field guide books [4, 46, 47]. By considering all the recorded species during the study period, a baseline of bird checklist (database) was prepared.
Data analysis
Different samples of each habitat were pooled before analysis. Data were analyzed using the Shannon-Weaver Index to determine the species diversity and evenness in the study area [35]. A similarity analysis (SIMPER) was carried out to calculate the percentage contribution of each bird species to different study sites (i.e., species that are characteristic of each habitat type) along with the agroforestry system. But only the three most contributing species were considered for dissimilarity comparison. Differences in feeding guilds (in terms of feeding functional group) similarity between habitat types were also described by SIMPER and a one-Way Analysis of Similarity (ANOSIM). ANOSIM is a non-parametric permutation test that uses similarity matrices or, in this case, the Bray-Curtis index [13]. When there are no group differences global R is centered at zero, or R ≤ 0, whereas if there are high group differences R = 1 [37]. Additionally, differences in species composition among sites were analyzed by a non-metric multidimensional scaling (NMDS), using the Bray-Curtis index. The species data can be collapsed into two dimensions using this ordination technique [30]. A plot of NMDS was applied as a visual aid to interpreting how species composition differed among habitat types [23]. The ANOSIM and SIMPER test was carried out with PAST 4.03 [20].