Effects of body size on the diet of Rivulus haraldsiolii ( Aplocheiloidei : Rivulidae ) in a coastal Atlantic Rainforest island stream , southern Brazil

The effect of body size on the feeding habits of the little-known killifi sh Rivulus haraldsiolii, collected in a coastal Atlantic Rainforest island stream, was investigated. Samples were collected during a study trip on Biotemas, 23 (4): 59-64, dezembro de 2010 ISSN 0103 – 1643 doi: 10.5007/2175-7925.2010v23n4p59


Introduction
The Neotropical killifish Rivulus haraldsiolii Berkenkamp, 1984 was recently redescribed, and recorded as being endemic to Atlantic Forest coastal drainages in southern Brazil, between Paraná and Santa Catarina states (Costa, 2007).The little-known species of Rivulidae are small-sized forest-dwelling fi sh that inhabit preferably the shallowest parts of small streams, temporary swamps, fl ooded areas and man-made drainage ditches (Santos-Filho, 1997).As for any other killifi shes, despite the remarkable life cycle (Costa, 2003, Mitcheson andLiu, 2008) and the occurrence in perennial aquatic biotopes and seasonal pools (Costa, 2006), there is scarce information about their natural feeding ecology.Available evidence suggests that the group might have a generalist diet (Taylor, 1992;Shibatta and Rocha, 2001;Shibatta and Bennemann, 2003;Laufer et al., 2009), and differences in the diet richness and prey size are related to variations in body sizes (Santos-Filho, 1997;Laufer et al., 2009).
We describe here the effect of body size on the feeding habits of individuals of R. haraldsiolii collected in an Atlantic Rainforest island brackish stream, based on data collected during fi eld studies conducted on 13 July 2009, and related to the fi eld ecology course promoted by the Ecology and Conservation Post Graduation Program of the Federal University of Paraná on Ilha do Mel Island.The study site is considered to be one of the most southern Atlantic Forest areas (Marques and Oliveira, 2004) where freshwater and brackish streams fl ow directly into the Atlantic Ocean, belonging to the coastal drainages of eastern Brazil province.

Material and Methods
The fi eld work was carried out in a 100m stretch of a brackish stream located in the northern part of the Island (25 o 31'S 48 o 18'W) (Figure 1).The study site comprises remaining areas of the coastal Atlantic Body size effects on R. haraldsiolii diet Rainforest, with fl oating vegetation and herbaceous riparian vegetation (Figure 2a).All specimens were caught under dead tree branches and leaves in slow-fl owing water, with sand on the bottom and a great amount of vegetable debris.Specimens were collected using manual sieves (5mm mesh) and 1.50 x 5.00m seine nets (5mm mesh) (Figure 2b).Captured specimens were fi xed in the fi eld in 10% formalin solution and brought to the laboratory of the Centro de Estudos do Mar (CEM-UFPR).Voucher specimens were deposited in the fi sh collection of the Museu de História Natural Capão da Imbuia (MHNCI 12391).
Fishes were collected with IBAMA (Brazilian Institute of Environment and Renewable Natural Resources) authorization 10320.
Because of the lack of information about their life history (size at fi rst maturity, growth, and reproduction), specimens were classified as juveniles and adults according to the size captured, coloration pattern and fi n confi guration (Santos-Filho, 1997;Costa, 2007).Digestive tracts were removed and their contents analyzed.The food items were grouped in broad taxonomic or ecological categories: aquatic immature insects (Diptera pupae and Chironomidae), aquatic insects (Hemiptera -Vellidae), microcrustaceans (Cladocera), aquatic ticks (Acarina), gastropods (Pomacea sp.), terrestrial insects (Coleoptera, Formicidae and Isoptera), insect fragments, and plant fragments.
To evaluate the abundance of food items in the diet of each individual, an adaptation of the Hynes' (1950) method of points was performed: the abundance of each item was interpreted as the number of points (score) that the item covered on grid paper.Abundance data was then investigated using non-metric multidimensional scaling analysis (MDS) with cluster overlay, analysis of similarity (ANOSIM) and similarity percentage (SIMPER) methods performed by the Primer v6 software (Clarke and Gorley;2006).Signifi cant differences in the diet composition and abundance between juveniles and adults (factors) were assessed by ANOSIM investigation.The SIMPER analysis was used to identify which food item primarily accounted for observed differences between groups.A similarity matrix with the abundance values of food items was generated using the Bray-Curtis similarity coeffi cient.

Results
Eighteen specimens were measured and dissected, ranging from 21.4 to 54.9mm in total length.Five specimens were classifi ed as juveniles (21.4-29.3mm),and thirteen as adults (30.2-54.9mm).Considering the consumption of both animal (mainly aquatic immature dipterans and terrestrial insects) and vegetable (mainly plant fragments) material, the feeding habit of R. haraldsiolii seems to be omnivorous.In relation to the origin of the food items, among the autochthonous ones, the most frequent were aquatic ticks (50%) and Chironomidae (27.8%).Formicidae (50%), plant fragments (27.8%),Coleoptera (16.6%) and Isoptera (16.6%) were the most common allochthonous food items.
Diet composition demonstrated differences between juvenile and adult categories as indicated by the ordination resulting from MDS (Figure 3) and the one-way ANOSIM test (Global R=0.485; signifi cance level=1%).SIMPER results showed high dissimilarity between juvenile and adult diets (92.6%), with greater representativeness of Formicidae (63.9%), plant fragments (16.9%) and aquatic ticks (12.8%) for adult categories, and aquatic larvae of Chironomidae (67.6%) and aquatic ticks (27.85) for juveniles (Figure 4).The larger the grey circle (bubble), the greater the abundance value.
Body size effects on R. haraldsiolii diet

Discussion
Despite the fact that this study represents a shortterm survey on the feeding ecology of this poorly known killifi sh, our fi ndings suggest that plants and a variety of terrestrial and aquatic invertebrates form the bulk of Rivulus haraldsiolii diet.Davies et al. (1990) and Taylor (1992) studying Rivulus marmoratus in Florida mangrove marshes, and Shibatta and Bennemann (2003) studying Rivulus pictus in small lakes of Brazilian savanna, also noted that aquatic and terrestrial invertebrates (mostly insects) are considered to be one of the main food items consumed by rivulids.
The food habits changed ontogenetically, which may be a refl ection of changes in feeding capabilities (e.g.mouth gap, locomotion ability) or habit shifts (Wooton, 1999).Adults (large individuals) feed mainly on Formicidae, plant fragments and aquatic ticks, whereas juveniles (small individuals) feed basically on aquatic immature insects (Chironomidae) and aquatic ticks.A shift in food habits with size increase is well known in fi shes (Nikolsky, 1963;Amundsen et al., 2003;Rezsu and Specziár, 2006;Alcaraz and García-Berthou, 2007), and has already been registered for Rivulidae (Laufer et al., 2009), as well as for R. luelingi in a laboratory experiment (Santos-Filho, 1997).Generally, larger fi sh eat larger food and a greater variety of organisms, whereas smaller fi sh feed on less diverse and smaller sized food items.The feeding habits of R. haraldsiolii seem to follow this trend.
The presence of terrestrial insects and the anatomical characteristics of R. haraldsiolii, such as the fusiform body shape, large eyes dorsolaterally placed on the head, and upward turned mouth, with the upper and lower jaws moderately protractile (Figure 5), indicate surface picking behavior, by which the fi sh swims upstream and catches terrestrial items on the water surface (Sazima, 1986), mainly in the marginal areas with slow fl owing waters, probably due to a reduced ability for continuous swimming.According to Santos-Filho (1997), Rivulus is a visually oriented predator, strongly attracted by prey movement and items that fall into the water from the trees, in a similar way to that observed for several characin species in Atlantic Rainforest streams (Costa, 1987;Casatti and Castro, 1998;Aranha et al., 1998;Abilhoa et al., 2007;2009).
The relevance of aquatic immature stages of mosquitoes (Diptera), terrestrial insects (Formicidae) and plants to its natural diet highlights the importance of the marginal forest of the Atlantic Rainforest biome to the feeding habits of stream fi shes, a result which has also been recorded for several other stream fi shes (Sabino and Castro, 1990;Esteves and Lobón-Cerviá, 2001;Costa et al., 1987;Graciolli et al., 2003;Abilhoa et al., 2007;Vitule et al., 2008).In fact, most fi sh species use food items of allochthonous origin (Lowe-McConnell, 1999) and changes in the riparian vegetation can cause alteration in the feeding habits of freshwater fi shes, affecting many links of the trophic chain (Barrela et al., 2000).The present scientifi c note is the fi rst register on the feeding ecology of R. haraldsiolii, an important component of temporary pond communities and stream ecosystems.
Mar (CEM-UFPR) for laboratory facilities.Several papers on Rivulidae biology and ecology were kindly provided by Scott Taylor (Environmentally Endangered Lands Program -Florida).

FIGURE 1 :
FIGURE 1: Map of South America, showing the Paraná state coast (southern Brazil) and the location of the study site on Ilha do Mel Island (black square).

FIGURE 2 :
FIGURE 2: (a) The study site, a small brackish water stream in the northern part of Ilha do Mel Island, southern Brazil.(b) Rivulus haraldsiolii, total length 32.1mm, male.

FIGURE 3 :
FIGURE 3: Ordination resulting from Nonmetric Multidimensional Scaling Analysis (MDS) with cluster overlay (traces), based on the abundance of food items in juvenile and adult categories (numbers=total length).

FIGURE 4 :
FIGURE 4: Contribution of Formicidae, plant fragments, Chironomidae, and aquatic ticks (Acarina) to the Rivulus haraldsiolii diet, according to the ordination resulting from Nonmetric Multidimensional Scaling Analysis (MDS) with cluster overlay (traces).The larger the grey circle (bubble), the greater the abundance value.