Seasonality and landscape characteristics impact species community structure and temporal dynamics of East African butterflies

Abstract

Species community structures respond strongly to habitat changes. These are either driven by nature or human activities. The biota of East African drylands responds highly sensitively to natural and anthropogenic impacts. Thus, seasonality strongly influences resource availability in a cyclic manner during the year, with cyclic appearance of the different developmental stages of invertebrates, while man-made landscape transformations profoundly and permanently modify habitat structures and, as a consequence, species communities. Butterflies are an excellent model group for the study of the effects of seasonality, and to test for biodiversity responses to anthropogenic activities such as habitat modification, degradation and destruction. We performed transect counts of adult butterflies in riparian forests and their adjoining areas, either dry savannahs with occasional pasturing (i.e. near-natural status) or farmland areas with fields, gardens and settlements (i.e. highly degraded status with lack of original vegetation). Transects were set along the river beds as well as at 250 m and 500 m distances parallel to these rivers, with eight transects per distance class and site (i.e. 48 transects in total). We recorded habitat structures for each transect. Counts were conducted during the dry and the rainy season, with 16 repetitions for each single transect, i.e. eight per season and transect. We compiled trait data on morphology, geographic distribution, ecology, behaviour, and life-history for all butterfly species encountered. Our results show higher species richness and numbers of individuals in farmland transects compared with the savannah region. Seasonal fluctuations of the detectable species abundances between the rainy and dry season were severe. These fluctuations were much more pronounced for the savannah than the farmland area, i.e. was buffered by human activities. Farmland and savannah support two distinct butterfly communities, with generalist species being more common in the farmland communities. Strict habitat associations were comparatively weak and typical dry savannah and riparian forest species were not clearly restricted to the near natural landscape.

Figure 1

Location of our study region in Kenya (star in small inlaid map), enlarged map of our study region around Kitui, with the two study sites, Nzeeu River and Kainaini River, including all transects (aerial images).

Figure 2

Species richness (a), and number of individuals (b), along the Nzeeu (orange) and Kainaini (yellow) rivers; given is the mean per transect. (c) Average scores of the dominant (explained variance: 91.5%; red: dry season, violet: rainy season) and subdominant (explained variance: 3.5%; green: dry season, brown: rainy season) PCoA eigenvectors across transects. (d) Bray–Curtis similarities of the intermediate (I) and far (F) transects to the near river transects (N), as well as average Bray–Curtis similarity within the eight near river transects N_W (green bars; only calculated for the rainy season; data insufficient for dry season) and the average similarities across the eight near river transects N_B (blue bars; data insufficient for Kainaini during the dry season). Error bars in (a), (b), and (c) denote standard errors from eight sample points in each transect. Error bars in (d) are based on 1000 bootstrap samples. Abbreviations: N—near river transects; I—transects at intermediate distance from river (i.e. 250 m); F—transects far from river (i.e. 500 m); N_W—average Bray–Curtis similarity within the eight walks in the near river transects; N_B—average similarities across the eight near river transects.

Figure 3

Average abundance-based butterfly ecological traits in transects along the Nzeeu (orange) and Kainaini (yellow) rivers near Kitui city in southern Kenia. Error bars denote standard errors. Traits used are (a) geographic distribution (4 categories), (b) forest index (5 categories), (c) habitat specialisation (3 categories), (d) savannah index (5 categories), (e) tree index (3 categories), (f) wetness index (3 categories), (g) larval foodplant specialisation (3 categories), (h) larval food plant type (dicotyledonous, monocotyledonous), and (i) hemeroby index (4 categories). Categories (apart from larval food plant type) are always in increasing order. Abbreviations: N—near river transects; I—transects at intermediate distance from river (i.e. 250 m); F—transects far from river (i.e. 500 m).