Two thrush species as dispersers of Miconia prasina ( Sw . ) DC . ( Melastomataceae ) : an experimental approach

We carried out a seed germination experiment using two thrush species in captivity. We compared the number of germinated seeds and germination time of control seeds (manually removed from fruits) and ingested seeds of Miconia prasina by two bird species, Turdus albicollis and T. amaurochalinus, and also compared retention times of seeds by both thrush species. Control seeds germinated more frequently than those ingested for one species, T. albicollis. The germination time of ingested seeds by T. amaurochalinus was similar to the control seeds but seeds ingested by T. albicollis took longer to germinate than the controls. Both thrush species had a similar seed defecation pattern. The cumulative number of defecated seeds increased by 2 hours after fruit ingestion. At the end of the first 30 minutes both species had already defecated approximately 50% of the seeds ingested Our results suggest that both species could act as disperser agents of M. prasina.


Introduction
Seed dispersal depends on many aspects of the plant-disperser interaction (Herrera, 1995).For example, retention time, manipulation tactic, number of seeds, and place where seeds are deposited are all important for the qualitative and quantitative importance of the seed dispersal agent (Barnea et al., 1992).Classically, pulp removal has been suggested as advantageous since it may enhance seed germination (Ridley, 1930), because the pulp may inhibit germination or be infected by patho-gens (Traveset, 1998).Different bird species may be able to treat pulp and seeds differently (Levey, 1986).Small seeds, such as those of Miconia (Melastomataceae), are more slowly separated from pulp than large seeds.Therefore, small seeds are retained longer in the gut (when compared with regurgitated seeds), which might reduce their viability (Murray et al., 1994) but, in general, seeds defecated by frugivores are viable (see Smith, 1975;Levey, 1986).in different areas of Atlantic forest in Brazil (Talora and Morellato, 2000;Oliveira, 2002;Antonini and Freitas, 2004) mostly in forest edges and in regenerating areas, and their fruits are an important food resource for species of different bird families (Correia, 1997).

Study bird species
The White-necked Thrush, Turdus albicollis Vieillot 1818, inhabits mainly the forest (Sick, 1997), and it has been seen feeding on fruits of Miconia species at the forest edge (R.D. Antonini, unpublished data) and the Creamy-bellied Thrush, T. amaurochalinus Cabanis 1851 is found in open habitats, such as the border of the forests (Sick, 1997), secondary forests (M.A.S. Alves unpublished data) and restingas (Alves et al., 2004).It also feeds on fruits and arthropods (Gomes, 2006).

Germination studies
Fieldwork was carried out in the months of August and September 1999 and 2001, during the fruiting period of M. prasina.For the germination experiment, three individuals of T. albicollis and four of T. amaurochalinus were captured and kept in individual cages (53 x 46 x 30 cm) and tested with M. prasina fruits, all taken from the study population site.Only completely ripe and not damaged fruits were collected on the day of the experiment.Birds were maintained for three days for acclimatization and fed with a standard diet (following Denslow et al., 1987) and water ad libidum.On the morning of the fourth day, 10 ripe berries of M. prasina (from seven individuals) were offered to each bird and feces were collected every 10 minutes (after the first fruit was eaten) within the first 30 minutes, and after that every 30 minutes during 4 hours and 30 minutes, totaling 12 samples collected for Many studies on the effects of seed germination after ingestion by birds showed that germination is variable (Smith, 1975;Sorensen, 1981;Izhaki and Safriel, 1990;Barnea et al., 1991;Ellison et al., 1993;Castiglioni et al., 1995;Traveset, 1998;Figueroa and Castro, 2002), specially those of germination of seeds collected from feces and seeds extracted manually from pulp (Ellison et al., 1993;Samuel and Levey, 2005).Experimental studies are important as the results can give us clues to comprehend the process of seed dispersal in nature.
Trees and shrubs of the genus Miconia (Melastomataceae) are conspicuous elements in the secondary vegetation (Talora and Morellato, 2000;Manhães, 2003;Oliveira, 2002) producing small berries containing a large number of small seeds (Snow, 1965).Melastome fruits are consistently mentioned as being among the most important resource for many frugivorous birds in the tropics (Snow, 1965;Snow, 1981;Loiselle and Blake, 1990;Stiles and Rosselli, 1993;Galetti and Stotz, 1996), providing fruits at times when fruits in general are scarce (Levey, 1990).
The goals of the present study were: 1) to test whether two thrush species impede or improve germination of Miconia prasina (Sw.)DC. after ingestion; 2) to test whether the time until germination is reduced or not in defecated seeds compared to the control, and 3) to compare seed retention time after ingestion by both thrush species, to determine whether the two thrush species are equal as dispersers agents of M. prasina.

Study site
The study was carried out at a secondary Atlantic forest area near Vila Dois Rios (23° 11' S and 44° 12' W), in Ilha Grande (Figure 1), the largest island (approximately 19,000 ha) on the coast of Rio de Janeiro State, southeastern Brazil.Ilha Grande is covered by Atlantic forest in different successional stages because of prior agricultural activities, with some areas of undisturbed forest being present in the centre of the island.The climate is warm and humid, with mean annual temperature of 23 °C and total annual rainfall of 2200 mm (data obtained from the Central Nuclear Almirante Álvaro Alberto meteorological center -CNAAA).

Study plant species
Miconia prasina is a typical early successional species with small seeds (mean ± sd) (81.4 ± 5.7 seeds per fruit; n = 11) inside purple berries (4.5 ± 0.3 mm lenght; 6.1 ± 0.6 mm width; 0.12 ± 0.03 g; n = 37).This plant occurs Both Turdus species had a similar pattern of seeds defecation (Figure 2) with the exponential phase between the first hour and the stationary phase reaching after 3 hours (T.albicollis) or 4 hours (T.amaurochalinus).The greatest number of seeds recorded in a single interval was at 30 minutes after ingestion for both species.Through this time interval T. albicollis had defecated 50% (n = 823) of the ingested seeds and T. amaurochalinus 54% (n = 1123) (Figure 2).The retention time of the seeds in the digestive tract did not differ between the studied species (U = 1191.0;p = 0.805).

Discussion
The closely related thrush species studied in the present work differed in terms of seed germination of M. prasina, as the seeds defecated by T. amaurochalinus germinated faster than those ingested by T. albicollis.
Passage through digestive tracts of birds increases seed germination (Izhaki and Safriel, 1990).Ingestion may have consistent influence on seed germination (Lieberman andLieberman, 1986, Barnea et al., 1991;Mallet-Rodrigues, 2001), since it can cause changes at external structure of the seeds (thermic, chemical and mechanical action) and consequently dormancy removal (Agami and Waisel, 1988;Izhaki and Safriel, 1990;Barnea et al., 1991), or on seed viability (Figueroa and Castro, 2002).Birds had an effect on the rate of germination (55% of the studies), accelerating the germination of ingested seeds more commonly than delaying (Traveset, 1998).In the present study seed ingestion for both thrushes decreased the number, and for T. albicollis delayed the time of seeds germination.Seed germination should be favored after pulp removal, if the pulp has germination inhibitors or is infected by fungi or other pathogens (Traveset, 1998) more easily than seeds separated from the pulp (Traveset, 1998).But only few studies used intact fruits (with seeds left in the fruits) as control (Traveset, 1998;Samuel and Levey, 2005).The results every bird.After the trial, birds were released in the area where they had been captured.
Seeds found in feces were washed with distilled water and sown in petri dishes with filter paper (both autoclaved to prevent contamination).Control seeds were taken manually from intact fruits and washed with distilled water.In each experimental time interval, ten defecated seeds and ten control were placed in different petri dishes for germination comparisons.If fewer than 10 seeds were defecated then the total number of seeds (from two to nine) was used along with the same number of control seeds.
All seeds were wetted every two or three days with distilled water and the Petri dishes were kept in the laboratory under continuous white light and the same conditions of temperature and humidity with no artificial control of them.The seeds were checked weekly (every two or three days) for 50 days.The emergence of the radicle indicated germination.The day of the first seed germinated in each Petri dish was used to estimate the germination of ingested and control seeds.
To compare germination efficiency among treatments (control and experiment) Chi-square test was used.The germination time among ingested and control seeds were compared using a Mann-Whitney test.The difference between the speed of the germination curves between seeds ingested and control was tested by Kolmogorov-Smirnov test.Tests followed Zar (1999), using the Statistical Software Systat 8.0.

Results
Overall, the ingested seeds germinated less than the control seeds (Table 1), even though there was significant difference only for T. albicollis.The number of germinated seeds did not differ significantly after ingested by both thrushes (χ 2 = 1.36, p > 0.05).
Seeds ingested by T. albicollis took significantly more time to germinate than the control seeds (Table 2).The difference between the speed of the germination curves between seeds ingested and control for T. albicollis was marginally significant (Kolmogorov-Smirnov test, KS = 0.41, p = 0.071) while no significant difference was found for T. amaurochalinus (Kolmogorov-Smirnov test, KS = 0.23, p = 0.476).Although we do not have much information on the foraging tactics of both Turdus species, our results suggest that both bird species act as dispersers agents of M. prasina.Further studies can elucidate the role of time delay on seeds germination by dispersers, such as T. albicollis, compared to the control.
of the present study suggest that the passage through the digestive tract of T. amaurochalinus seems not to have a negative effect on seeds germination as observed for the seeds ingested by T. albicollis which germinated significantly less than the controls.However, both thrush species may favor Miconia seeds germination after pulp removal.
Longer retention times of seeds in digestive systems might cause a stronger effect of seed coat abrasion and therefore better germination (Barnea et al., 1991).However it does not explain the difference found in the present study for ingested and control seeds for T. albicollis in terms of seed germination, since there was no evidence of different times of seed retention between the thrush species.
Turdus albicollis was the most captured bird in the understory of a forest less disturbed by man (Alves, 2001), feeds on fruits and arthropods, and preliminary data suggest it is an opportunistic species in terms of its diet (R.D. Antonini and V.C. Tomaz unpublished data).This species can potentially transport seeds from the interior to the edge of the forests and vice-versa.On the other hand, T. amaurochalinus is found in open habitats, feeding also on fruits and arthropods (Gomes 2006).This species is migrant and its role on seeds dispersion deserves further investigation, as migrant bird species have been reported as important seed dispersers of different species in arid Mediterranean habitats (Barnea et al., 1992) and tropical ones (Mallet-Rodrigues, 2001).The process of forest regeneration does not depend exclusively on specialist frugivorous, but also on a large number of generalist species of secondary forests, with diet generally based on fruits and insects, including species of Muscicapidae (see Jordano et al., 2006) family of the Turdus studied.
Individuals of Turdus albicollis and T. amaurochalinus are gulpers (eating the whole fruits and ingesting all seeds) (Stiles and Rosseli, 1993;pers. obs.), and are expected to transport, defecate away from the parental plant (Stiles and Rosselli, 1993).As the Atlantic forest is nowadays distributed in many small fragments, the opportunistic feeding behavior of both bird species may increase the probability of M. prasina seeds to be dispersed between forest patches and colonization of new areas.In the same study area a previous study on T. albicollis using capture-marking-recapture in two sampled areas 1 km far from each other, crossed by a river, three individuals captured in one location were recaptured in the other one (Ritter 2000).This indicates that T. albicollis move between habitats.

Table 1 .
Absolute number and percentage of seed germination of control and ingested seeds of Miconia prasina by Turdus albicollis (White-necked Thrush) and T. amaurochalinus (Creamy-bellied Thrush).