Gut passage is an important but understudied component of animal-mediated seed dispersal that can impact germination and survival. Melastomataceae plants are abundant and ecologically important throughout the tropics, but studies have obtained contrasting results regarding effects of avian gut passage on melastome germination. We conducted a paired germination experiment to test how gut passage by four species of manakins—key avian dispersers of melastomes—affects germination of the pioneer melastome shrub Miconia rubescens. Manakin gut passage accelerated M. rubescens germination, with gut-passed seeds germinating an average of 5.1 days earlier than controls, and percent germination of gut-passed seeds was significantly higher at 20 and 40 days post-planting. Interestingly, manakin species varied in their gut passage effects, with L. velutina having stronger overall germination effects than M. manacus. Within species, greater body mass was correlated with higher percent germination and shorter times to first germination. Within M. manacus, seeds passed by females had significantly shorter times to first germination than seeds passed by males. Because control seeds were manually depulped in the experiment, the observed gut passage effects are likely attributable to minor scarification of the seed coat during gut transit. Our results suggest manakin gut passage can have biologically meaningful impacts on melastome germination, and the magnitude of these effects can vary based on manakin species, mass, and sex. This study refines our understanding of the ‘quality’ component of seed disperser effectiveness in an iconic dispersal mutualism, with implications for plant community composition and recovery of deforested neotropical ecosystems.

Study Area

This study took place in the Chocó biogeographic zone of northwestern Ecuador, a global hotspot for biodiversity experiencing extreme habitat loss and declining species abundance due to deforestation and other anthropogenic activity (Myers et al., 2000). Field work was conducted at Reserva FCAT (Fundación para la Conservación de los Andes Tropicales; 00◦ 23′ 28′′ N, 79◦ 41′ 05′′ W), a 700-ha private reserve within the Mache-Chindul Ecological Reserve in Esmeraldas Province. The average temperature in the area ranges between 23-25ºC and annual precipitation is approximately 2,500-3,000 mm, with the wet-season occurring from late December to June (Clark et al., 2006). At our site, agricultural crops (e.g., cacao, plantain) and pasturelands are prevalent and increasing, although intact and successional forests also remain relatively common (Van Der Hoek, 2017). However, heavy deforestation and fragmentation has occurred in the Mache Chindul Reserve, disrupting connectivity and species interactions (Van Der Hoek, 2017).

Study System

Our study species, Miconia rubescens (Gamba and Almaeda, 2018; formerly M. neomicrantha (Judd & Skean, 1991) and Octopleura rubescens (Triana, 1872)), is a woody shrub producing fruits that are small (mean height x width: 6.04 x 2.99 mm; mean weight = 0.11 g), round, watery, and white when ripe (mean Brix = 4.88%). This species is the most geographically widespread species in the Octopleura clade, occurring in 13 countries over Central and South America (Gamba & Almeda, 2014, 2018). Fruit produced by M. rubescens are consumed by numerous neotropical frugivorous birds, particularly manakins, tanagers, thrushes, and brush-finches (Kessler-Rios & Kattan, 2012; Loiselle & Blake, 1999; Stiles & Rosselli, 1993; Wheelwright et al., 1984).

Manakins are small passerines that constitute some of the most abundant frugivores in South and Central American rainforests, making them prominent agents of endozoochorous seed dispersal in these habitats (Cestari & Pizo 2012, 2013). In this study, the following manakin species were used in gut passage trials: Manacus manacus, Lepidothrix velutina, Ceratopipra mentalis, and Cryptopipo litae.  Manakins at our field site are regularly observed eating whole fruits of M. rubescens. This is especially true of M. manacus, which tend to situate their lek sites in disturbed secondary forest and along forest edges where M. rubescens is relatively abundant.

Gut passage trials

Gut passage trials were performed on wild-caught birds over the course of eight weeks, from early June to late July 2022. Passive mist netting took place in regenerating forest surrounding known lek sites of M. manacus using 2–7 nets, each 12m in length, that remained open in the morning hours until manakins were captured; we paused netting for subsequent stages of the trials. Birds were fasted for 45 minutes after capture to ensure that the majority of previously consumed seeds had left the bird’s digestive tract (Levey, 1986; Worthington, 1989). During the fasting period, manakins were extracted from nets, weighed, processed, and banded for identification. Once processing was complete, they were placed back into a fabric mist netting bag in the shade until the fasting period was complete.

Ripe fruits (determined by color and softness) were collected from the field < 24 hr prior to use in gut passage trials and stored in dry and dark conditions. Each paired trial in the experiment corresponded to fruits collected from a different M. rubescens individual (n = 41), which were marked to avoid resampling. Following the fasting period, birds were hand fed 1–4 whole M. rubescens fruits that had been previously collected from a single plant. The number of fruits fed to a given bird varied depending on the individual’s cooperation and condition. Immediately after feeding, birds were placed into a small pet carrier with a perch, covered for 45–60 minutes, then released. Fecal samples containing gut-passed seeds were collected from a piece of white printer paper at the bottom of the carrier, placed in sterilized marked containers, and processed for the experimental germination procedure later the same day (below).

Germination experiment

 We compared percent germination of gut-passed and non-digested (control) seeds using a paired experimental design. For a given experimental trial, gut-passed and control fruits were sourced from the same plant to account for potential individual variation in germination rate or success among plants. To minimize handling effects during extraction from feces (gut-passed treatment) or fruits (non-digested treatment), seeds were suspended in a small amount of sterile water and gently removed with a sterile metal spatula. Extracted seeds were then placed on sterile germination paper inside plastic petri dishes. For each paired trial, the same number of seeds were placed in control and gut-passed petri dishes. However, due to variation in the number of seeds that were recovered following gut passage, the number of seeds obtained varied among trials (mean: 31.83 ± 1.72 SEM, range = 6–42). Petri dish pairs were placed immediately adjacent to one another indoors where temperatures ranged from 13–22 ºC  and situated near a window to allow exposure to indirect natural sunlight on a natural 12 hr light/12 hr dark cycle. Germination paper was kept damp as needed using a spray bottle filled with sterile water. Control seeds received the same amount of water as paired gut-passed seeds and were watered at the same times. All petri dishes were monitored together at least 1–2 times per week to record the number of germinated seeds.

Statistical Analysis

To compare the effect of gut passage on germination times across all seeds, we fit a linear mixed-effects model using the nlme package in R (Pinheiro et al., 2017). The model included time elapsed until germination as a response variable (square root transformed to improve residual normality), treatment (i.e., gut-passed vs. non-digested) as a predictor variable, and dish within trial as a nested random effect. This nested random effect structure accounted for the fact that seeds belonging to the same trial (i.e., paired gut-passed and non-digested seeds) were derived from the same M. rubescens plant, and seeds in the same dish (in the case of the gut-passed treatment) passed through the same manakin. To test whether gut passage influenced variation in seed germination time, we calculated the within-dish variance (σ²) in time to germination for all gut-passed and non-digested seeds in each trial separately. We then conducted paired t-tests to compare the variances in germination times between treatments for all manakins, for L. velutina only, and for M. manacus only; variable transformations were applied as necessary to meet normality assumptions. To test whether gut passage influenced percent germination, we fit a model in lme4 (Bates et al., 2015) with final percent germination at the end of data collection as a response variable, treatment as a fixed effect, and bird ID and species as random intercepts.

To examine germination success at various timepoints throughout the experiment, we conducted Wilcoxon signed-rank tests comparing the percent of gut-passed (GP) versus non-digested (ND) seeds that germinated at 20, 40, 60, and 80 days after planting. Because seeds were not checked for germination on a daily basis, we used the percent germination recorded closest to the desired 20-day interval for analyses (20-day timepoint mean ± SD: 20.17 ± 1.14 days; 40-day timepoint: 40.32 ± 1.60 days; 60-day timepoint: 60.80 ± 2.01 days; 80-day timepoint: 80.20 ± 2.69 days). All seeds were checked for germination on the same days, so the distance from a given 20-day interval did not differ within an experimental-control pair. We compared percent germination of gut-passed seeds relative to paired controls for each of the following groups: all manakin species together, L. velutina only, and M. manacus only. Gut passage by C. mentalis and C. litae was not analyzed separately due to small sample sizes.

We fit additional linear mixed-effects models in nlme to investigate how bird mass and sex influenced germination parameters. Due to the small sample sizes of C. mentalis and C. litae, only L. velutina and M. manacus were included in these analyses, and only trials that reached 80 days post-planting (n = 30) were considered. To investigate the effects of mass, sex, and species on germination success, we fit a model that included the difference in percent germination between gut-passed and control seeds (D percent germination) as a response variable; mass, sex, and species as fixed effects; and individual bird ID as a random intercept. To investigate the effects of bird mass, sex, and species on germination rate, we fit a second model with the same fixed and random effects as the first, but a response variable of D days until first germination (i.e., the difference between gut-passed and control treatments in days elapsed prior to the first occurrence of germination). This model also included a fixed variance structure to account for heteroskedasticity. Finally, within M. manacus (the species for which we had the greatest number of known-sex individuals), we used a Wilcoxon test to compare D days until first germination between known males and females. Statistical tests and figures were conducted and generated in RStudio Version 4.4.1 (RStudio Team, 2024) and JMP Version 17.00 (JMP, 2021). All descriptive statistics are presented as means ± 1 SEM.