Myiarchus flycatchers are the primary seed dispersers of Bursera longipes in a Mexican dry forest

Study sites

We conducted the study at three different successional stages of TDF that have been mostly unmanaged for varying periods of time since their last major disturbance (i.e., clear-cutting or burning). The three stages are described as follows: (1) The early successional stage (last disturbed ca. 20 y ago) consisted of vegetation regrowth in a plot once used for cattle ranching and, to a lesser extent, seasonal agriculture; (2) The intermediate successional stage (last disturbed ca. 35 y ago) corresponded with a transitional phase between a mature forest and fragmented areas with a matrix of pasture and seasonal corn and bean fields, yet was once dedicated to seasonal corn production and cattle ranching. Structural and floristic elements had developed that mirrored the original dry forest vegetation, to a large extent; (3) The mature successional stage was characterized by a closed canopy and presence of tree cover typical of mature dry forest (i.e., dominance of the Bursera spp.). This stage has not experienced a large scale disturbance for more than 50 y.

The successional stages were found in patches with areas of 97 ha (early stage), 45.3 ha (intermediate stage) and 24 ha (mature stage). The mean distance between successional stages was ca. 1 km (Fig. 1).

Bursera longipes

The genus Bursera is a distinctive component of TDF in Mesoamerica and is composed of ca. 107 species (De-Nova et al., 2012). Its distribution spans from northern Mexico to the northern region of South America (Becerra et al., 2012). The diversification of this genus is related to the southward expansion of TDF in response to the elevation of the Sierra Madre del Sur and the Mexican Volcanic Belt (De-Nova et al., 2012). Bursera evolutionary history indicates that a large portion of the biological richness of Mesoamerican TDF was derived from accelerated rates of speciation in habitats, from the early Miocene to the Pliocene, during pronounced arid periods (Becerra, 2005; Dick & Wright, 2005). This scenario matches other hypotheses that these lineages were mostly restricted to dry environments of Mexico and evolved during long periods of isolation (Valiente-Banuet et al., 2004).

In particular, Bursera longipes is endemic to TDF in the states of Mexico, Morelos, Puebla, Guerrero and Oaxaca in the Balsas basin (Fig. 1; Rzedowski, Medina Lemos & Calderón de Rzedowski, 2005). It is a deciduous species with trivalvate fruits that turn red at maturity; seeds have a slightly orange pseudoaril (Guízar & Sánchez, 1991). Fruits are 1.3 ± 0.02 cm (mean ± SE) in length and 0.87 ± 0.04 cm in width, with a fresh weight of 0.62 ± 0.01 g (N = 100 fruits). Flowering season begins with the onset of the rainy season (May or June), and fruiting occurs in early June or from May–July. Most fruits ripen between November and May.

Seed dispersal effectiveness

In this study we used the quantitative and qualitative components proposed by Schupp (1993) and Schupp, Jordano & Gómez (2010) to estimate the effectiveness of B. longipes seed dispersal in each successional stage. The quantitative component included the frequency of visits to B. longipes tree and the average number of fruits removed per visit by frugivorous birds. The qualitative component was based on the percentage of germination after seeds passed through the digestive system of birds, the probability of seed deposition under a nurse plant (adult plants that positively influence the recruitment of young seedlings) and the possible contribution of bird species to the establishment of B. longipes after seed deposition under nurse plants (Schupp, Jordano & Gómez, 2010). Seed dispersal effectiveness of each frugivore is calculated as the product of the subcomponents of quantity and quality, according to the following expression (Schupp, 1993; Schupp, Jordano & Gómez, 2010): Effectiveness = frequency of visits × average number of removed fruits per visit × proportion of seed germination × seed deposition probability in potential suitable microhabitats × contribution of birds to site of establishment.

Frequency of visits and average amount of removed fruit

Frequency of visits was determined by focal observations using binoculars (8 × 40 mm). Observations were focused on seven B. longipes individuals with ripe fruits at each successional stage and performed during January–May 2011 and March–May 2012 in the morning (0700–1130 h) and afternoon (1600–1830 h), when bird activity is higher. Each of the seven trees was sampled in each successional stage during both years. A total of 70 h of observation was recorded for each successional stage (10 h/tree), for a total of 210 h in all three stages.

Each tree was observed at a distance of ∼20–30 m for an observational period of 30 min, during which the frugivore species and number of visits were recorded, in addition to number of individuals, total time of visit (from arrival to departure) and number of fruits consumed per visit. The frequency of visits was analyzed in a χ² contingency table to determine differences among successional stages. In this case, the null hypothesis would indicate the existence of an equal number of visits between successional stages. The number of removed fruits was compared among stages with an unbalanced one-way ANOVA. For this analysis, data were transformed (log x + 1) to meet assumptions of normality and homogeneity of variance.

Seed germination

Seeds obtained from the faeces of birds captured by nine 12 m mist nets in each successional stage were used to determine the effect of passage of seeds through bird digestive system on the proportion of germination. Mist nets were placed during the months with greatest availability of mature B. longipes fruits (May and December 2010, January–May and December 2011 and March–May 2012). During each period, sampling was performed for 15 days. Mist nets remained open from 07:00 to 18:00 h, resulting in a total of 1,485 net-hours per stage and 4,455 net-hours for all successional stages.

Captured birds were placed in individual cages (40 × 40 cm) lined with mosquito netting and fed ad libitum with ripe B. longipes fruits for a day after capture. Retention time of seeds was estimated from the moment of fruit consumption until defecation or regurgitation of seeds. Birds placed in cages were monitored every 10 min, with the least possible disturbance, in order to confirm their consumption of seeds with pseudoarils. Defecation times were recorded at 10 min inspection intervals. Retention time of seeds is therefore an approximation, according to these intervals.

The premise in this portion of the study is that longer retention times would likely result in seeds being spread farther from the mother plant (Westcott & Graham, 2000). After evacuation, feces were collected, and birds were released. Since the techniques used to collect regurgitated seeds and feces were non-invasive, special authorizations were not required.

Seed viability was tested via flotation tests, in which floating seeds were considered nonviable due to lack of embryonic development (Thompson, Grime & Mason, 1997). Viable seeds were washed with 10% sodium hypochlorite, planted in cotton on petri dishes at ambient temperature and moistened daily with distilled water. This procedure was performed with seeds obtained from different sources and with different treatments as follows: (a) control group 1: seeds with pseudoaril obtained directly from the trees, (b) control group 2: seeds without pseudoaril obtained from trees and (c) seeds that passed through the digestive system of birds. For the final treatment, the germination experiment was only performed with bird species from which the largest number of seeds was obtained: Myiarchus nuttingi (N = 67), Myiodynastes luteiventris (N = 58), Myiarchus cinerascens (N = 33), Melanerpes chrysogenys (N = 29) and Myiarchus tyrannulus (N = 27). A total of 50 seeds per successional stage were used for each of the controls (fruits obtained from trees).

In the first phase of the germination experiment, we mixed the seeds of the control groups from all three stages and assigned a random number to each control treatment. Each control group had a total of five replicates with 30 seeds each. The objective of this portion of the experiment was to evaluate the effect of deinhibition and scarification (Robertson et al., 2006).

Germination experiments were performed directly in the field, where boxes with B. longipes seeds were placed under the canopy of nurse plants Mimosa polyantha and Senna wislizenni, which are commonly used by disperser birds for perching. The boxes were protected with mesh mosquito netting and boric acid was poured around the perimeter to avoid predation by ants. Over the course of 20 days, boxes were checked daily to count the number of germinated seeds. The emergence of a radicle indicated germination.

The estimated retention time that seeds remained in the digestive system of birds was compared between treatments with a one-way ANOVA, following a prior analysis of normality and homoscedasticity. The null hypothesis was that time of retention would be the same for all treatments. Multiple comparisons were analyzed with a Tukey HSD. The percentage of germinated seeds was analyzed using a generalized linear model (GLM) with a binomial distribution error and a logit link function (Crawley, 2012) to determine significant differences between treatments. Post hoc pairwise t-test comparisons were carried out for each germination treatment. To evaluate the effect of passage time through the bird gut on seeds on germination percentage, we performed a linear regression (Traveset, 1998). In addition, we also analyzed the relationship between the total body length of captured birds and the percentage of germinated seeds. The null hypothesis of this test was that a positive relationship would exist between these variables. We measured body length using a digital calibrator, according to the specifications given by Ralph et al. (1996). Analyses were performed with SPSS 17.0 software (SPSS, Inc., Chicago, IL, USA).

Potential suitable sites for recruitment

Frugivorous birds deposit faeces under their perch trees (Vasconcellos-Neto, Barbosa de Albuquerque & Rodrigues Silva, 2009), but only the canopy of certain shrubs and trees provides suitable conditions for recruitment of seeds in arid environments (Godínez-Álvarez, Valiente-Banuet & Rojas-Martínez, 2002). The deposition of seeds under the cover of trees or shrubs (potential nurse plants) was estimated by focal observations to record the number of visits to these perching sites by birds after fruit consumption. To facilitate the monitoring of birds after they finished eating and departed to fly in another direction or roost on another plant, one person was dedicated to post-consumer observations.

The number of visits of frugivorous birds to each of the following categories of perch plants was recorded: (1) conspecific, indicating that the individual remained in the same plant species (i.e., B. longipes) where they ate fruit; (2) Fabaceae species, including trees and shrubs of the Caesalpinoideae, Faboideae and Mimosoideae subfamilies, which have been identified as potential nurse plants in semi-arid environments (Godínez-Álvarez & Valiente-Banuet, 1998) or (3) other tree or shrub plants, including species of Opuntia or columnar Cactaceae. Focal observations ended when eye contact with observed birds was lost. A contingency table of χ² was used to compare the number of bird visits to each category of perch plant. The null hypothesis would be indicated by an equal number of bird visits among all perch categories across the three successional stages. Standardized residuals were used to evaluate the preferential use of certain perching sites by birds (Valiente-Banuet et al., 1991; Godínez-Álvarez, Valiente-Banuet & Rojas-Martínez, 2002). These residuals are distributed around a mean of 0 with a standard deviation of 1, and therefore, any resulting value ≥ 2 (approximately 5% of the normal distribution) was considered to be a significant deviation.

The probability that seeds were deposited in potentially suitable sites (under Fabaceae) was determined by the proportion of frugivore visits to these perch plants in relationship to the total number of recorded visits. Fabaceae plants have been shown to provide appropriate conditions for the recruitment of seedlings (Godínez-Álvarez & Valiente-Banuet, 1998).

Contribution of birds to seedling establishment in different successional stages

Two plots with a radius of 30 m (2,828 m² per plot) were randomly chosen in each successional stage. In each plot seedlings and young individuals of B. longipes (height < 50 cm) were counted under trees or shrubs used by birds to roost after ingesting B. longipes fruits. Recruited individuals were classified by the previously mentioned categories of nurse plants; these categories have been successfully used in others studies in semiarid forests (Valiente-Banuet et al., 1991; Godínez-Álvarez, Valiente-Banuet & Rojas-Martínez, 2002). The number of young B. longipes plants observed under nurse plants was compared to the number of individuals expected to be recruited at random, derived from examining a proportional and comparable area and counting B. longipes underneath all plants with a DBH ≥ 10 cm (Valiente-Banuet et al., 1991). The null hypothesis would indicate a proportional number of seedlings between the comparative plots, given the coverage of woody plants. Standardized residuals were calculated to analyze the significance. Plant cover was determined in a previous study corresponding to the study sites (Almazán-Núñez et al., 2012).

Finally, each bird species was assigned a value of 0–1 according to their contribution towards the establishment of B. longipes. This value was estimated from individual observations of bird species after feeding on B. longipes drupes, their flight destination and number of visits to other plants. The maximum value was assigned to birds with the highest number of flights to nurse plants under which the largest number of seedlings or young B. longipes individuals had been observed with respect what would be expected by chance, according to the standardized residuals for each plot.

Frequency of visits and number of removed fruits

A total of 20 bird species were recorded eating B. longipes fruits (Table 1). Frequency of visits to remove fruit varied between stages (X² = 54.78, df = 38, p < 0.05). Myiarchus tyrannulus and Tyrannus verticalis were the most frequent visitors to the early successional stage (Table 1), T. vociferans and T. verticalis to the intermediate stage and M. cinerascens to the mature stage. Spinus psaltria removed the greatest number of fruit at the early (5.00 ± 1.58) and intermediate (4.40 ± 0.51; Table 1) stages and Eupsittula canicularis at the mature stage (11.00 ± 4.00).

Overall, 17.9% of the fruits consumed in the three stages were removed at the early stage, 42.2% at the intermediate stage and 39.9% at the mature stage (n = 825), although no significant differences were found among stages (F_2,275 = 1.57, p = 0.210). The flycatcher T. verticalis remained for the longest time in the trees of the early stage (6.78 ± 1.30 min), T. vociferans in the intermediate stage (6.33 ± 1.13 min) and E. canicularis in the mature stage (8.00 ± 4.00 min; Table 1).

Seed germination

The shortest average seed retention time from fruit intake until evacuation was recorded for Myiarchus nuttingi and the highest for M. tyrannulus (Table 2). The latter had the widest range in seed evacuation time (minimum = 10 min and maximum = 230 min). The shortest average timeframe corresponded to Myiodynastes luteiventris (minimum = 12 min and maximum = 155 min; Table 2), although differences in retention time were not significant (F_4,122 = 0.98, p = 0.420). Body size of the frugivorous birds was positively correlated with time of passage of B. longipes seeds (R² = 0.79, F = 11.68, p = 0.04).

None of the seeds with intact pseudoaril (control group 1) germinated (Fig. 2). Seeds without pseudoaril (control group 2) had a germination rate of 10%. The seeds that passed through the gut of Myiarchus cinerascens had the highest germination percentage (27%, n = 33), followed by Myiarchus tyrannulus (26%, n = 27), Melanerpes chrysogenys (24%, n = 29), Myiarchus nuttingi (15%, n = 67) and Myiodynates luteiventris (12%, n = 58) (Fig. 2). According to GLM, seed germination varied among treatments (X² = 21.73, p = 0.001). Post hoc pairwise t-test comparisons indicated that the three bird species associated with the highest percentage of germination (M. cinerascens, M. tyrannulus and M. chrysogenys) significantly differed in comparison to seeds without pseudoaril (control group 2, p < 0.01) (Fig. 3). However, significant differences were not found in germination between seeds without pseudoaril (control group 2) and seeds eaten by M. nuttingi and M. luteiventris. The passage time of seeds through the gut and the germination percentage were marginally significant (R² = 0.71, F = 7.69, p = 0.06). Body size was not significantly correlated with germination percentage (p > 0.05).

Potential suitable sites for recruitment

After consuming fruits, birds perched in three categories of plants (Fig. 4). The preference was for Fabaceae species at all three stages (X² = 22.98, df = 12, p < 0.05; X² = 55.33, df = 20, p < 0.05; X² = 54.98, df = 20, p < 0.05, for the early, intermediate and mature stages, respectively) (Figs. 4A–4C). At the intermediate and mature stages, flycatchers M. nuttingi and M. tuberculifer remained for the longest period of time in Acacia and Mimosa plants following feeding episodes, while Tyrannus verticalis and Vireo gilvus spent more time in the same trees where they feeded on fruits. Thus, flycatchers of the Myiarchus genus were the most likely species to deposit B. longipes seeds beneath Mimosa and Acacia trees and shrubs across the three successional stages (Table 4).

Contribution of birds to sites of seedling establishment

The lowest density of B. longipes seedlings and non-reproductive individuals was found at the early stage (0.002 ind/m²). The average height of plants was 54.07 ± 7.90 cm. At the intermediate and mature stages, densities of seedlings and non-reproductive individuals were 0.007 ind/m² and 0.008 ind/m², with an average height of 50.93 ± 3.90 cm and 53.19 ± 3.80 cm, respectively, and did not differ significantly in density (F_2,5 = 0.89, p = 0.50) or in average height (F_2,104 = 0.12, p = 0.89).

At the early stage, the number of seedlings and young B. longipes individuals was significantly higher under Tecoma stans, Plocosperma buxifolium and Mimosa polyantha plants (Table 3). At the intermediate stage, the largest number of seedlings was found under Mimosa polyantha and Calliandra eryophylla, and at the mature stage, under Eysenhardtia polystachya, Senna wislizeni, Sebastiana pavoniana and Acacia cochliacantha (Table 3). Acacia subangulata was the only legume that presented a lower number of observed seedlings than expected by chance (Table 3).

The largest contribution to the establishment of B. longipes seedlings, calculated based on the number of flights to nurse plants with the largest number of observed seedlings with respect what would be expected by chance, was attributed to M. cinerascens at the early stage and to M. nuttingi at the intermediate and mature stages (Table 4).

Seed dispersal effectiveness

The effectiveness of seed dispersal was estimated for five bird species whose number of visits allowed for a reliable estimation, which varied depending on the stage (Table 4). For other species, dispersion was not determined due to lack of defecated seeds or other subcomponents that would allow for this assessment.

At all stages the best dispersers belonged to the genus Myiarchus. At the early stage, only M. cinerascens contributed to seed dispersion (Table 4). At the intermediate stage, M. nuttingi was the largest contributor to seed dispersion. In the mature stage, five species participated in seed dispersion, and M. cinerascens had the highest effectiveness (Table 4).