The present study was carried out in the Piedmont forest, the lower altitudinal belt of vegetation in the Southern Andean Yungas forest, in Argentina (Cabrera 1976). Southern Andean Yungas forests hold an important conservation value due to their high biodiversity, which includes several endemic and threatened flora and fauna species (Pidgeon et al. 2015). The Piedmont forest is located between the Yungas mountain forests to the west and the dry forests of the Chaco to the east (Brown et al. 2001). The Piedmont forest is between 400 and 900 m a.s.l and has marked rainfall seasonality with concentrated precipitation in the summer (December-March) (Brown et al. 2001; Mogni et al. 2015), and the year can be divided into wet (October to March) and dry (April to September) seasons. This sector has a high density of woody plant species endemism (30%) and approximately 70% of the tree species lose their leaves during the dry season (Brown et al. 2009). Dominant tree species are Calycophyllum multiflorum and Phyllostylon rhamnoides, which together with Anadenanthera colubrina, Myracrodruon urundeuva, Handroanthus impetiginosus and Myroxylon peruiferum constitute the tree canopy of 25 to 35 m height (Brown et al. 2009).

The Piedmont forest has been severely transformed and about 90% of its original surface has been converted into urban, agricultural and pasture areas, amongst others (Brown et al. 2009). Furthermore, the remaining forests of this ecosystem (approximately 908,000 ha) are mainly exploited by the selective extraction of timber species (Brown et al. 2009). The sites selected for this study are located in the Province of Jujuy, north-western Argentina and are separated by at least 30 kilometres (Finca Yuchán: 23°56'S, 64°54'W; Parque Nacional Calilegua: 23°38'S, 64°35'W and Reserva Privada Ecoportal de Piedra: 24°5'S, 64°23'W, Fig. 1). These sites have not been subjected to anthropogenic disturbance for at least the past 45 years.

Figure 1. 

Location of the study area in the Province of Jujuy, Argentina. The black square represents the study area. The red circles represent each study site. The Yungas forests are marked in green. Modified figure from Fundación Proyungas (2007).

A total of 30 phenology transects of 100 m × 6 m (10 in each site) were established to determine the variation in resource availability (Chapman et al. 1994). Transects were monitored by the same observer during the second week of each month, from May 2014 to June 2016, recording all seeding and fruiting trees with a diameter at breast height (DBH) > 10 cm. For each fruiting tree, DBH and estimated fruit abundance for each site and each month was measured from the sum of DBH and the total number of seeding and/or fruiting trees (Renton 2001). Only parrot food-plant species were included in the analysis at the stage of ripeness when consumed by parrots.

The Green-cheeked Parakeet diet was determined by direct observations of its feeding activity for two consecutive years (May 2014 to June 2016). Feeding observations were performed in the same area and at the same time as the phenology transects. At each of the three sites, a 100 ha grid was established where 60 randomly-located 300 m line transects were delimited. Monthly, the line transects were traversed from sunrise to 12 am, and from 4 pm to sunset, the time of greatest feeding activity of parrot species in general (Renton 2001). When parakeets were detected foraging, date, time, number of individuals, plant species and part of the plant consumed (seed, fruit, leaf, flower and nectar) were registered. We considered the total consumption of the fruit as a whole fruit, i.e. seeds and pulp. Seed consumption was considered when parrots discarded the exocarp and/or endocarp of the fruit to eat only the seed (Jordano 2014). Each event was taken as a single feeding episode. If a parakeet or a group of parakeets changed to another feeding source, either from the same or a different plant species, it was registered as a new feeding event (Altman 1974; Galetti 1993).

The relative frequency of different food items in the diet (seed, fruit, flower and others) was computed as the total number of records of item i (e.g. seed) divided by the total number of records of all items (seed, fruit, flowers and others). A contingency Chi-squared test was performed to observe the differences in the number of feeding events and the type of items consumed between two seasons, wet (October to March) and dry (April to September).

To estimate the trophic niche breadth, we used Levin’s measure of standardised niche breadth (Colwell and Futuyma 1971), calculated as BA = (Y²/ΣN²)-1/n-1, where Y is the total number of parrots feeding, N is the number of parrots associated with each plant species and n is the total number of food plant species. A value close to 0 indicates dietary specialisation, i.e. the species disproportionately selects some plant species, and a value close to 1 indicates a broad diet with a proportional use of several plant species (Krebs 1998). The overlap in parakeet diet between dry and wet seasons was evaluated using Morisita Index (Morisita 1959) that ranges between 0 (no overlap) and 1 (total overlap). Both indices were calculated for each season of the year taking into account the monthly data of both years.

Data from the phenology transects were used to determine food resource abundance and availability for the Green-cheeked Parakeet. We defined food resources available to the Green-cheeked Parakeet based on our field observations of foraging parakeet. Seasonal variation in food resource abundance was analysed by the U Mann-Whitney test on the sum of DBH and the total number of fruiting trees. Previous Kolmogorov–Smirnov analysis demonstrated a normal distribution for the variables of numbers of fruiting trees (D = 0.23; p < 0.12) and the sum of DBH of fruiting trees (D = 0.24; p < 0.10). To evaluate the relationship between food resource availability and the parakeet diet, the Spermank rank correlation was used and the significance of these correlations using the ‘cor.test’ function in R software (R Development Core Team 2020) was tested. Statistical analyses following Zar (1999) were carried out in R software version 4.0.1 (R Development Core Team 2020).

A total of 214 food plant trees of ten species from seven families were recorded in the phenology transects. Anadenanthera colubrina was the most abundant fruiting tree species (55% of fruiting trees). Piedmont forest showed no statistically significant differences for seasonal availability on food resource for both estimators (U = 156, N1 = N2 = 12, p < 0.68 for number of fruiting trees; U = 167, N1 = N2 = 12, p < 0.32 for sum of DBH). In the dry season, the food resource abundance was greater, where most species of trees have dry wind-dispersed fruit. At the end of the dry season, there was a decrease in the food resources availability, but there was a slight peak of flower production. In the wet season, there were peaks in fleshy fruit abundance.

Over the two years, the Green-cheeked Parakeet consumed 18 plant species from ten families (Table 1; Fig. 2), and only two of them were exotic (Morus nigra and Citrus sinensis). With a total of 220 feeding bouts, fruits represented 66.67% of the items consumed, followed by seeds (16.51%) and flowers (10.09%) and, to a lesser extent, leaves (4.59%) and nectar (1.83%). Some plant species were represented by more than one item (Table 1). The plant family with the highest number of species in the diet was Fabaceae (four species). Celtis iguanaea (30.73%; 67 feeding bouts) and Trema micrantha (22.01%; 48 feeding bouts) had the greatest number of feeding bouts. During the first year, 124 feeding bouts and the consumption of 15 plant species were registered, while, in the second year, the consumption of 11 species was detected. Citrus sinensis, Cedrela angustifolia, Jacaranda mimosifolia, Solanum riparium, Tessaria integrifolia and two unknown species were consumed only in the first year, while Maclura tinctora, Tipuana tipu and Vachelia aromo were consumed only in the second year.

Figure 2. 

The Green-cheeked Parakeet (Pyrrhura molinae) feeding on Citrus sinensis seed in the Yungas Piedmont in Argentina. Photo: Rodrigo Aráoz

The parakeet diet varied between seasons and there were statistically significant differences in the frequency of feeding bouts (X²₁ = 10.47; p < 0.001) and the frequency of item types consumed (X²₄ = 79.83; p < 0.001). The principal items consumed during the dry season were fruits of C. iguanaea (n = 18 feeding bouts) and seeds of C. sinensis (n = 11) and Parapiptadenia excelsa (n = 10). Only in this dry season, the consumption of leaves of Anadenanthera colubrina was observed. In the wet season, the most consumed items were fruits of C. iguanaea (n = 49) and T. micrantha (n = 40), along with flowers of Juglans australis (n = 14).

The niche breadth was variable amongst seasons and it was wider in the dry season (B = 0.69) than in the wet season (B = 0.28), with a high overlap in food items between seasons (Morisita Index = 0.59). The narrow diet niche breadth in the wet season indicates a tendency to concentrate food on only a few of the available resources. There were significant correlations between numbers of parakeet feeding and numbers of feeding bouts (r_s = 0.94, t₍₂₂₎ = 13.05, p < 0.001, Table 2, Fig. 3).

Figure 3. 

Linear relationship between the number of parakeets feeding and the number of feeding bouts as illustrated by the output of a Spearman’s rank correlation, in which the 95% confidence interval is represented by the grey shading.

Food resource availability for the Green-cheeked Parakeet did not demonstrate significant differences in this study, possibly due to the lack of food resource species in tree phenology samplings. In the study area, Benavídez (2018) found that the Piedmont forest has a marked seasonality with two different seasons, the dry (May-October) and the wet (November-April) seasons, where the plant community shows phenological behaviour, distinguishing flowering peaks in the dry season and two fruiting peaks, one in wet (fleshy fruit) and another in the middle of the dry season. Annual flowering and fruiting likely result from an influence of environmental factors. Several studies on dry forests have demonstrated that rainfall seasonality is the main factor that explains fruiting patterns in dry forests (Bullok and Solis-Magallanes 1990; Ragusa-Netto and Fecchio 2006; Valdez-Hernández et al. 2010). The phenology patterns of a vegetal community affect the availability of food items for the parrot assemblages, which may employ a combination of strategies of diet switching, habitat shifts and movements to track food resources (Renton et al. 2015).

The Green-cheeked Parakeet shows a broad diet in the study area, consuming a wide variety of plant species and items (from fruits to leaves) and varying throughout the year and seasons. These results are similar to the studies carried out in a forest in south-western Brazil, where the Green-cheeked Parakeet consuming between 16 to 20 plant species (Ragusa-Netto 2007; Nunes and Santos-Junior 2011) and similar other Pyrrhuras species (Roth 1984; Toyne et al. 1992; Kristoch and Marcondes-Machado 2001; Botero-Delgadillo et al. 2010, 2013; Buitrón-Jurado and Sanz 2016) were observed. One common feature of feeding habits in Pyrrhura genus is the extensive consumption of few, but abundant available food resources (Ragusa-Netto 2007; Botero-Delgadillo et al. 2010). Parrots are considered adaptable, consuming a great variety of plant species, as well as adapting to novel foods in modified environments, causing conflicts with farmers (Matuzak et al. 2008; Barbosa et al. 2021). However, the population decline of some parrot species has been linked to the decline of keystone plant resources (Berg et al. 2007; Renton et al. 2015). Our results showed a large consumption of Trema micrantha and Celtis iguanaea by the Green-cheeked Parakeet. Both native plant species could be a key food resource to the Parakeets in the Yungas forest and the potential reduced availability of these species, as a result of habitat fragmentation, may have consequences in the maintenance of wild parakeet populations.

In this study, fleshy fruit is the most consumed item and this fact was not unexpected given its high presence in the diet of Pyrrhura and other species of small and medium-size psittacids (Renton et al. 2015; Botero-Delgadillo et al. 2013; Benavídez et al. 2018). The high percentage of fruit in the diet of small parrots has been attributed to body size (Matuzak et al. 2008; Renton et al. 2015; Benavídez et al. 2018) and it is linked to energy requirements, due to the fact that small birds need more carbohydrates, because they display higher mass-specific metabolic rates (Suarez and Gass 2002). Furthermore, the high consumption of fruits, especially fruits with small seeds (e.g. Trema micrantha), might indicate the potential ecological role that this species might be playing in seed dispersal and, therefore, in the plant community structure (Blanco et al. 2018). Blanco et al. (2015) found that the Green-cheeked Parakeet in the dry tropical forest of the Bolivian inter-Andean valleys could act as a mutualist by dispersing seeds of native plants. Piedmont forest is currently one of the most threatened Andean forests in South America (Brown 2009) and, as a consequence, current functions and services provided by the Green-cheeked Parakeet and other parrot species in this forest requires further research, to have a better understanding of the Yungas ecosystem.

The Green-cheeked Parakeet might respond to changes in food availability, showing flexibility in the items consumed throughout the year and being mostly granivorous during the dry season, and more frugivorous/florivores during the wet season. Seasonal diet switching has been noted in other neotropical parrots (Galetti 1993; Renton 2001; Ragusa-Netto 2007; Botero-Delgadillo et al. 2010). We found that the Green-cheeked Parakeet exhibits a narrower dietary niche breadth in the wet season, which corresponds to a decrease in food resource availability. Fruit consumption occurs mainly during the wet season, when the availability of this resource increases in the Piedmont forest (Pacheco and Grau 1997; Brown et al. 2009), while seed and nectar are consumed mainly during the dry season. Consumption of flowers and nectar is important for some neotropical parakeet species (Galetti 1993; Cotton 2001; Ragusa-Netto 2005, 2007; Lee et al. 2014), especially in seasonal forests which provide an alternative food resource during the driest period of the year, when fruits and seeds are scarce (Ragusa-Netto 2005). The ability to exploit available resources may enable parrots to increase energy intake and fitness benefits in seasonal environments (Renton et al. 2015). Although we did not find any correlation between the parakeet diet and food availability, the expansion and contraction in the Levins Index and variation of food items consumed throughout the year and season demonstrate high flexibility in the diet and it might be closely related to the temporal variation in the availability of food resources. Furthermore, the absence of such correlation could also be associated with the absence of other food resources (e.g. understorey species) in the phenology transect. This is the case with Trema micrantha and Celtis iguanaea, two species that are relatively abundant in the understorey of the Piedmont forest and are the main plant species consumed by parakeets. The inclusion of understorey species in future studies of food resource availability needs to be taken into consideration.