Summer feeding ecology of Great Pampa-finches, Embernagra platensis at Laguna de Guaminí, Buenos Aires, Argentina

The aim of this study was to assemble data on the summer feeding ecology of the Great Pampa-finch, Embernagra platensis at the Laguna de Guaminí, Buenos Aires, Argentina, and to explore the differences related to the dietary patterns for each sex between winter and summer when possible. The stomach contents of 43 birds were analyzed. The animal fraction was composed of Hymenoptera (45.1%), Coleoptera (32.4%), Lepidoptera (6.0%), Araneae (5%) and Orthoptera (3.2%). The application of the index of relative importance (IRI) resulted in 1490.4 for Coleoptera, 428.5 for Hymenoptera and 162.5 for Lepidoptera caterpillars. The vegetal fraction consisted of Triticum aestivum (26.9%), Cyperaceous (25%), Poaceae (Gramineae) (19.3%) and Panicum sp. (11.2%). The IRI values were 893.8 for Triticum aestivum, 174.5 for Gramineae, 126.5 for Panicum sp. and 112.8 for Scirpus sp. The food niche width was 0.33 for both sexes; the diversity index resulted in 1.06 for females and 1.33 for males and specific diversity ranged from 1.87 to 2.84. A canonical component analysis (CCA) was performed on environmental and morphometric variables, and a Monte Carlo test confirmed the canonical correlations. A t-test showed that some birds harmonized with a logarithmic model and some with a geometric curve. During the summer, Embernagra platensis ingests Hymenoptera and Coleoptera more often than seeds, suggesting that two biological mechanisms could be taking place in this bird.


INTRODUCTION
The Great Pampa-finch, Embernagra platensis, belongs to the Emberizinae finches, distributed from the southeast Brazil (north of Minas Gerais to Espírito Santo), through most of Paraguay (local in the dry Chaco), northwest Bolivia (north to Beni), and Uruguay, to central and south Argentina (Misiones, Formosa, Chaco, Santa Fé, La Pampa and Río Negro (Short 1975, Ridgely andTudor 1989)).
It inhabits open grasslands, often with scattered shrubs, and can fairly frequently be seen flying along and over roadsides, although some prefer damp places.They form pairs or small groups, are territorial and con-Previous papers on the Great Pampa-finch tered mainly arthropods, mostly insects as its po preys (Marelli 1919, Aravena 1928, Zotta 19321940, Montalti et al. 2005).Beltzer (1990) report the diet of Embernagra platensis in the middle va the Río Paraná (Province of Santa Fé, Argentin dominated by an animal fraction (16 taxa), follow a plant fraction (8 taxa).Current feeding studi ried out at the Laguna de Guaminí during the registered arthropods (40% biomass), mostly Fo dae and Coleoptera, and typical seeds of the P (60% biomass) as part of its diet (Montalti et al.The previously known facts about the dietary co tion of the Great Pampa-finch were incomplete, "main" -2010/8/6 -20:54 -page 664 -#2

664
LAURA M. FERMAN and DIEGO MONTALTI the summer, compare its diet between seasons and sexes at the Laguna de Guaminí, Province of Buenos Aires, Argentina.

STUDY AREA
This research was carried out in the area of the Laguna de Guaminí (37  Merrit and Cummins (1978) and Kusnezov (1978).The systematic determination of plant species was made following Cronquist (1993).
The minimum sample was calculated using the collected stomachs (represented by each individual) and taxa (prey items).

STATISTICAL ANALYSIS
In this study, the relationship between prey item vari-morphometric measurements (environmental variables) of birds were also analyzed.All data were combined in a matrix.The associations between sets of two variables were investigated performing a canonical correlation analysis (CCA) using the Canoco statistical software.
In order to compare diet patterns through the diversity index among groups from different seasons, statistical tests were carried out using SYSTAT 7.0.When running the program, we operated with groups of birds collected in each season as "different communities", and prey species items were named by taxonomical category.We worked with four sets of data named from 'C1' to 'C4', and with the abundances of prey items.In the analysis, 'C1' refers to summer males, 'C2' to summer females, 'C3' to winter males, and 'C4' to winter females (Montalti et al. 2005).
The abundance range hypothetical model method was used to determine whether the real data fit the hypothetical model curves of richness-abundance.The model curves tried were logarithmic, geometric and broken stick shaped.The analysis evaluates and compares biodiversity community rates.We worked with 'C1' to 'C4' and proposed hypothetical curves for each community using SYSTAT 7.0.
The food niche breadth (FNB) was calculated using Levins' index (Levins 1968, Krebs 1989), as 1/( p i 2), where p i is the proportion of prey i in the diet.A standardized food niche breadth value (FNBst) was calculated, as (F N B − 1)/(n − 1), where n is the total number of prey categories (Colwell and Futuyma 1971).Lower values of Levins' index indicate relatively more specialized species, whereas higher values indicate more generalized ones (Krebs 1989).
The importance relative index (IRI) was used to establish prey contribution to diet composition.The H' index (Shannon and Weaver 1949) was calculated between seasons and sexes, and a t-test was also used in order to compare the differences in diversity indices among the communities mentioned above.
Insect length ranged from 5.5 mm in Curculionidae and Formicidae to 30 mm in Lepidoptera caterpillars.The resultant IRI for the arthropod fraction was 1490.4 (149%) for Coleoptera, 428.5 (42.85%) for Hymenoptera, and 162.5 (16.25%) was represented by Lepidoptera caterpillars.
The trophic spectrum based on the identification of 507 items was integrated by 41 taxa (9 plants and 32 insects) and the standard FNB resulted in 0.33 for the summer period.Gastroliths (5 samples) ranged between 1 and 2 mm.
The mean stomach sample was 0.57g (SD±0.21)and it ranged from 0.16 g to 1.36 g, representing ca.1.36% of the body mass.The mean body mass was 41.7 g, and the mean number of items was 12 per stomach per bird for the summer.
The H' index was 2.24 for males and 1.94 for females during the summer.The diversity index was 1.08 for males and 1.5 for females during the winter, and there was no difference between these seasons.
Table III shows in the CCA analysis that the second components contemplate about 71% of the information among variables, meaning that the samples (birds) corre-   Tables IV and V show the mean values of diet items for the winter and summer, based on 38 and 43 stomach contents, respectively.

DISCUSSION
In order to discuss the feeding ecology of the Great Pampa-finch during the summer, we made some comparisons concerning the proportions and the diversity of prey items ingested between the two seasons when possible.The specimens collected had consumed 17% of seeds (biomass) in the summer and 60% in the winter.Plant seed families were Chenopodiaceae, Asteraceae, Cyperaceae, and Poaceae for the winter, and they differed from the summer in certain species, namely Helianthus annuus, Setaria ssp., Zea mays, and Eleusine tristachia (Montalti et al. 2005).The summer plant species diverged considerably from the winter species (Table II), but they were all local components, both cultivated and ornamental crops.We assume that the undetermined vegetal fraction was obtained as part of indigestible seed husks and grass that were picked while birds were eating fallen seeds.Wheat and corn were obtained from leftover crops at the end of August-November.The harvesting activity spreads native and exotic seeds, but also grain storage offers "free" nourishment, especially for pest insects commonly preyed upon seedeating birds.Even though seeds are important components of the winter diet, during the nesting season most of the birds feed on insects.Energy and protein requirements increase in birds during breeding (Klasing 1998).Arthropod ingestion provides more nutrients than fruit for most passerine birds (Izhaki 1998).During the summer, the amount of seeds and gastroliths decreased in comparison with the winter (Montalti et al. 2005).
"main" -2010/8/6 -20:54 -page 667 -#5 FEEDING ECOLOGY OF GREAT PAMPA-FINCHES    Blattidae, Coleoptera, and Hymenoptera, compared to the winter even though they had eaten a larger quantity of preys.The presence of Araneae, Lycosidae, Carabidae, Curculionidae and Tenebrionidae may suggest that birds are digging the ground, removing leaves, tree barks and tissues, and exploring soil surfaces.Dermaptera (Forcicula sp.) and Heteroptera (Periplaneta sp.) are not abundant, but we suggest that they were trapped during flying displays or while scavenging the soil stratum.A similar dietary diversity was registered in South African passerine birds also composed by beetles, termites, berries, seeds and ants (Kopij 2003(Kopij , 2005)).Aquatic arthropods were represented by larval stages of Odonata (dragonflies) and Hemiptera such as Belostoma sp., that were found in six different samples (Beltzer 1990).We found several brownish and greenish caterpillars.A diversity measurement gave us the trophic spectrum, which was composed of 41 taxa (9 plants, 32 arthropods) for the summer and 21 (9 plants, 12 arthropods) for the winter (Montalti et al. 2005).
During the summer, the mean seed weight of the content (0.57g, SD±0.21) (0.21g) was half of the mean weight encountered for the winter (Tables IV and V), which was 1.6% and 3% of the body mass, respectively.Other summer values resulted in a minor proportion in comparison to the values of the winter.The items consumed throughout the summer were also half in quantity (abundance) in comparison with the winter period.This behavior could be related to the feeding period of chicks.However, variety in the trophic spectrum was higher during the summer, whereas the mean body weight was similar for both seasons (Tables IV and V).
Concerning the animal fraction, we would like to emphasize the proportion of ants in the diet of Embernagra platensis, which is notorious and very abundant when compared to other insect proportions.Beltzer (1990) found Atta sp. and Acromyrmex sp., the same genus found in this study for both seasons.These "leafcutter ants" can be universally found in a variety of ecosystems.44.1% of ants (out of a total of 83% of insect biomass for the summer) and 39% (out of a total showing a behaviour indulged in by birds called "anting".Many bird species have been observed picking up ants and rubbing them on their plumage, a behavior which is displayed frequently (Clark and Clark 1990).Most anting reports involve passerines, especially of the Corvidae, Sturnidae and Emberizidae families (Wenny 1998) but, in fact, though anting is not well understood yet, birds seem to acquire the defensive secretions of ants due to their fungicidal and bactericidal properties.Video-recording could be a helpful tool in case of laboratory experiments, but what happens during the anting behavior is not known.These finches could be performing a defensive treatment by using ants or stimulating the formic acid discharge before eating them because of their taste.In this sense, the Great Pampa-finch could naturally be displaying some kind of pest biological control in local agriculture, feeding on harmful insects, ants and beetles (e.g.Curculionidae).This could be an auspicious hint of an opportunity to study some of the feeding mechanisms that Embernagra platensis uses.

ACKNOWLEDGMENTS
This work was part of an integral research program carried out by many groups of investigators researching in the same Laguna de Guaminí area.We want to thank the Dirección Provincial de Recursos Naturales de la Provincia de Buenos Aires (Direction of Natural Resources of the Province of Buenos Aires) for the permission granted according to federal laws to do research in this area in December 1997.We also want to thank Magdalena Ponce for manuscript corrections.

*
Mean values were based on N = 38 stomach contents collected during the winter.TW = total stomach content weight, GN = number of gastroliths, GW = weight of gastroliths, SN = seed number, SW = seed weight, IN = insect number, IW = insect weight, ItN = number of items.

TABLE II Plant species number (N) and frequency (F) of occurrence consumed by Embernagra platensis during the summer at the Laguna de Guaminí.
*Undeterm.= undetermined fraction due to deterioration, material that was weighed and contributed to the diet but that couldn't be classified as a plant gender or species.Total number of plants = 223. of about 0.016 (F-ratio = 42.287;P-value = 0.002).The test of significance of all canonical axes (trace) was 0.041 (F-ratio = 139.205;P-value = 0.002).The first two canonical correlations were significant, representing a good association between the two sets of variables.

TABLE III CCA analysis summarizing the correlation among variables and the information explained by axes.
All four eigen-values reported above are canonical and correspond to axes that are constrained by the environmental variables.

TABLE V Mean summer dietary items values of Embernagra platensis at the Laguna de Guaminí during December 1997.
# Mean values were based on 43 stomach contents collected during the summer.TW = total stomach content weight, GN = number of gastroliths, GW = weight of gastroliths, SN = seed number, SW = seed weight, IN = insect number, IW = insect weight, ItN = number of items.andColeoptera.The summer dietary components were terrestrial insects, such as the 44.1% of Formicidae (bio-were trapped during short flights.We also foun nopsis saevissima, Acromyrmex lundi and Acrom