Study Site

The study was conducted at TBR, Sarapiquí, Heredia, Costa Rica (10°25′N; 84°47′W). The mean annual temperature and rainfall are respectively 24.3℃ and 3,777 mm. The area encompasses two life zones: tropical wet forest and tropical wet premontane forest basal transition, following Holdridge, Grenke, Hatheway, Liang, and Tosi (1971). TBR protects 345 ha of mature lowland rainforests with most of the reserve covered by primary forest. Most of the area surrounding TBR has been deforested. The Sarapiquí River borders the west limit of the reserve for an approximate distance of 2.5 km, while the Tirimbina River crosses the reserve across 3 km, entering the southeast corner and leaving TBR through their northwest-most portion. The area is relatively flat (180–220 m) and several small creeks and hillsides crisscross the reserve (Figure 1). A trail network of 10 km provides ample access to the reserve. Selective logging was done occasionally between 1962 and 1990. More than 1,200 plant species have been identified at TBR, of which 27 are native palms belonging to 14 different genera (Ley-López & Morales, 2015). Although palms represent only 2.25% of the species, they contribute significantly to forest structure (Aguilar-Barquero & Jiménez-Hernández, 2009; Lieberman & Lieberman, 1987). According to Kohlmann et al. (2010), TBR includes one of the most diverse areas for palms in Costa Rica.

Figure 1.

Map of Tirimbina Biological Reserve, Heredia, Costa Rica, and some of its main environments. (a) Gallery forest, (b) Forest edge, (c) Flat area, and (d) Hillside.

Palm Abundance, Fruit Collection, and Seed Germination

Palm abundance was estimated from 15 linear transects of 100 × 4 m following the layout of the trails (Figure 1). Transects were stratified covering all the main forest environments found in TBR, including gallery forests, flat areas, hillsides, forest edges, and forest interior (Figure 1). In each transect, we recorded and identified all adult palms located within 2 m on each side of the trail. We considered adult palms as those that reached the minimum size for which we have seen reproductive individuals in each species. Locating palms close to the trails is the simplest way to collect seeds for propagation studies. Palm abundance (Table 1) was defined according to these categories: (a) rare (R), species with less than 4 individuals; (b) uncommon (UC), species with 5 to 10 individuals; (c) common (CO), species with 11 to 20 individuals; (d) very common (VC), species with 21 to 50 individuals; and (e) abundant (AB), species with more than 50 individuals. Here, our main objective was to estimate the feasibility of finding reproductive individuals according to their abundance along the trails.

Table 1.

Study Species, Distribution, Forest Strata, Palm Abundance, Number of Parent Individuals per Species (N), Total Number of Seeds Sown, Month of Collection of Ripe Fruits, Germination Percentage (%G), Mean Length of Germination in Weeks (MLG), First and Last Week in Which a Seed Germinated, and Germination Time in Weeks in Which 50% of the Seeds Germinated (GT₅₀).

From October 2013 to November 2014, we walked the TBR trails in random order twice a week looking for ripe fruits (Figure 2). During each survey, we recorded the species with mature fruits. Our ultimate goal was not to measure the phenological patterns of fruit production per se but to determine the availability of seeds for planting purposes in the nursery. Seeds were collected from the first individual of each species that presented mature fruits in good condition. Attempts were made to collect seeds from at least 10 different individuals per species. For rare species, with few individuals and with highly localized distribution, we carried out specific monthly walks, which were maintained until January 2015.

Figure 2.

Examples of palm species found at Tirimbina Biological Reserve, Heredia, Costa Rica. Bactris gracilior (a), Chamaedorea deckeriana (b), Desmoncus moorei (c), Synechanthus warscewiczianus (d), Euterpe precatoria (e), Geonoma interrupta (f), and Reinhardtia gracilis (g). Source: Emmanuel Rojas and JML.

Seed collection, viability, cleaning, sowing, and germination conditions were implemented following the methods of Meerow (1991). For understory palms, seeds were extracted by hand from fully ripe fruits, whereas for subcanopy species, we used an extension pole to reach the fruits. We also collected seeds from recently fallen, fully ripe, and undamaged fruits. Fruit maturity was evident because of changes in fruit size and color. Seeds were taken to the greenhouse, separated completely from the mesocarp and other fruit parts, and cleaned by hand with a damp cloth to eliminate germination inhibitors that could have been present in the endocarp. Because our main goal was to document germination under standard, simple nursery conditions, we did not use chemical or physical scarification. We submerged seeds in water and discarded seeds that floated or showed evident physical or herbivory damage. A seed subsample was cut open to check for the presence of viable endosperm; if the endocarp was firm and the embryo filled the chamber, we assumed that the seeds were viable (Meerow, 1991).

Seeds were sown within 48 hr after field collection and were placed in germination trays filled with peat moss soil. We spaced seeds at 1 cm intervals and planted them at a depth of 1 cm. The trays were maintained in the greenhouse under a plastic, transparent roof, which permitted exposure to natural light. The greenhouse had an ambient temperature of 26℃ and 30% full sunlight. Trays were watered manually twice a week to keep the soil sufficiently humid but not saturated or waterlogged.

In each seed cohort, germination was checked once a week for 57 weeks. Germination was defined as the emergence of the first cataphyll from the soil surface. Defining germination as the emergence of the shoot apex from the soil surface could overestimate the actual time of germination in palms with remote germination, for which the shoot develops slowly belowground (F. M. Henderson, 2006). However, this condition did not affect our estimation of the germination time as all of our study species present adjacent germination (F. M. Henderson, 2006). At the end of 57 weeks, we inspected nongerminating seeds for viability by submerging them in water or cutting them in half and checking for the presence and consistency of the endosperm. Nonviable seeds were clearly empty or rotten. For all palm species, we recorded the month of seed collection, the final percentage of germination (%G = number of germinated seeds/number of seeds sown), the mean length of germination in weeks (MLG), which corresponded to the time difference between the first and the last germination week, and the number of weeks required to attain 50% of %G (GT₅₀). GT₅₀ was measured using a linear interpolation of the two germination values closest to the median germination time.

Statistical Analyses

Temporal germination patterns were evaluated using survival analysis, which is a more robust technique than other traditional analysis protocols (i.e., germination indexes or regression analysis) according to McNair, Sunkara, and Frobish (2012). Survival analysis takes into account several properties of the germination process, such as the delay in germination initiation, germination percentage, germination rate, and germination time change. In germination studies, the presence of right-censored observations violates the assumptions of classical statistical methods; alternatively, survival analysis concentrates on the distribution of the germination times of individual seeds instead of the cumulative germination, comparing the homogeneity in the temporary patterns of germination between the groups (McNair et al., 2012).

We estimated the survival functions using the nonparametric product limit method (Kaplan–Meier) for census groups. To test for the homogeneity of survival functions across genera and species, we used the nonparametric log-rank test statistic (χ²) following the recommendations of McNair et al. (2012). Multiple comparisons were corrected with a sequential Bonferroni protocol (Holm–Bonferroni method). All statistical analyses were carried out using R Statistical software ( http://www.r-project.org) and the functions survfit and survdiff from the package survival (Therneau, 2014).

Palm Abundance

We recorded 519 individuals belonging to 21 out of the 27 species recorded at TBR (Table 1). Four species were considered abundant (Asterogyne martiana, Geonoma congesta, Geonoma cuneata, and Geonoma deversa), comprising more than 60% of the individuals (Table 1). The high abundance of these species is reflected in their presence in almost all of TBR as well as their high density in some transects. Twelve species were considered uncommon to very common. These species were found in most of the sites but in low abundance (i.e., Bactris hondurensis and Synechanthus warscewiczianus) or were habitat specific. Chamaedorea tepejilote was very common but found only in riparian habitats, while Geonoma interrupta and Prestoea decurrens were found only close to creeks (Table 1). Eleven species were considered rare, six of which were not found on transects. All of these are very rare and only a few individuals have been located at specific places in TBR. Chamaedorea deckeriana has only been observed in an area of less than 1 ha in the northwest boundary of the reserve, whereas Astrocaryum alatum has been previously reported in the reserve but was not observed in this study. Rare species found along transects grow in very low densities with no apparent relationship to habitat type (i.e., Astrocaryum confertum, Chamaedorea lucidifrons, and Desmoncus moorei) or in a few specific places (i.e., Reinhardtia gracilis). Most palms (446) belong to understory species, whereas only 72 were canopy or subcanopy species. We found only one individual of the climbing species D. moorei (Table 1).

Fruit Availability and Seed Collection

We found fruits in 21 of the 27 native palms of TBR. As expected, species abundance and fruit phenology determined seed collection. All species in which no fruits were found were classified as rare. Some rare species, like C. deckeriana, had short and synchronous fruiting periods and grew in patches, but the rest of rare species had very few individuals so that characterizing the pattern of fruiting phenology was difficult. For rare species, our approach was to collect seeds in an opportunistic and punctual manner. It is likely that more intensive sampling covering difficult access sites in TBR would have increased the probability of finding more individuals and therefore, obtaining seeds of more rare species.

Continuous fruit availability throughout the year facilitated seed collection in the majority of the species cataloged as uncommon to abundant. Seed collection was limited by low fruit production (B. hondurensis), asynchronous fruit ripening (Chamaedorea spp.), rapid fruit consumption by the local fauna (S. warsecwiczianus and C. tepejilote), and in a few instances, seed predation (we observed larvae of curculionid beetles inside the seeds of G. deversa, G. interrupta, Geonoma longevaginata, and Iriartea deltoidea). At TBR, ripe fruits were available year-round with an average of 8 fruiting species per month. October and November had the highest number of fruiting species (15 and 13 species, respectively), while February and March had the lowest number of fruiting palms (5 and 3, respectively). The rest of the year was relatively constant with 6 to 11 palm species bearing ripe fruits (Table 1). In most species, high fruit yield facilitated seed collection, which is particularly important for the nursery propagation of uncommon and rare species.

However, these data should be analyzed with caution as the absence of fruits in a species may imply irregular or complex fruiting patterns associated with low population densities rather than the total absence of fruits. In rare species, the low abundance of individuals does not allow to infer clear phenological patterns of fruit production, whereas the fruiting period is likely to be longer than our results suggest in common to abundant species. For example, the 10 species in which we found fruits for more than six months showed irregular fruiting patterns. In these species, it is possible that fruit production is continuous for most of the year, but with a low number of individuals producing fruits in each period. Most species showing fruits for more than six months belong to Bactris and Geonoma (Table 1).

Germination Patterns

Our results indicate that germination was not the main obstacle to propagate native palm species. Average percentage germination was 65.6%, with only two species (B. longiseta and G. interrupta) showing germination values lower than 50% (Table 1). In most species, there was high intraspecific variation in %G meaning that seed quality could affect seedling production. For instance, while most species had individuals reaching germination percentages close to 100%, a few individuals of Bactris spp., Geonoma spp., and S. warsecwiczianus showed low germination (0%–15%).

In addition, seed dormancy and germination speed posed no major obstacles to propagate this palm flora. In all species, germination started before 10 weeks, and 90% of all seeds germinated between weeks 7 and 25 (Figures 3 and 4). MLG and GT50 were 16.4 and 14 weeks, respectively, meaning that obtaining a high number of seedlings relatively fast is possible (Table 1). The period needed to complete the whole germination response was 40 weeks, at which time 99% of all seeds completed germination. The exception was Bactris and Geonoma palms, in which only a small percentage (0.5%) of the nongerminating seeds was still viable after week 57 (Figures 3 and 4).

Figure 3.

Frequency distribution of germination over 57 weeks for the 21 palm species studied at Tirimbina Biological Reserve, Heredia, Costa Rica, from October 2013 until January 2016.

Figure 4.

Kaplan–Meier estimates of survivor functions for 16 palm species over 57 weeks at Tirimbina Biological Reserve, Heredia, Costa Rica. Upper and lower lines from the same color represent pointwise 95% confidence intervals. Bactris spp. (a), Chamaedorea spp. (b), Geonoma spp. (c), and all other species: Asterogyne martiana, Euterpe precatoria, Synechanthus warscewiczianus, Socratea exorrhiza, and Welfia regia (d).

There were differences in germination trajectories across genera and species (Figure 4; Bactris spp. χ² = 11.9, df = 2, p-Holm = .0076; Chamaedorea spp. χ² = 82.8, df = 2, p-Holm = .001; Geonoma χ² = 41.2, df = 4, p-Holm = .001). Bactris and Geonoma were characterized by long and gradual trajectories whereas other species showed shorter trajectories that leveled off around 20 weeks (Figure 4). Congruent with differences in germination trajectories, species diverged in synchrony and average germination time. Species with higher germination percentages (A. martiana, Chamaedorea spp., Euterpe precatoria, and P. decurrens) consistently showed shorter fruiting periods and lower MLG and GT₅₀. Euterpe precatoria showed the fastest germination rate (GT₅₀ = 7 weeks) with one individual starting and ending germination in weeks 5 and 7, respectively. The longest germination time was recorded in the genus Geonoma (Table 1, Figure 4). Here, one individual of G. congesta and another of G. deversa took 24 and 27 weeks, respectively, to begin germination, and did not complete it until after 57 weeks. Species in the genus Bactris also had a high value of GT₅₀. In summary, it takes 20 weeks to germinate most of species in this palm flora and obtain a high number of seedlings, but in the genera Bactris and Geonoma, it will take 40 weeks if the goal is to cover the entire germination response.

Palm Abundance, Fruit Collection, and Seed Germination

The use of native species in rainforest restoration is limited by a multiplicity of factors, including the lack of appropriate seed collection methods considering the reproductive phenology, as well as the abundance, dispersion patterns, and habitat preferences of adult plants (Hoffmann, Blum, Velazco, Gill, & Borgo, 2015). Securing access to seeds of native species is critical to initiate successful restoration processes, especially when the goal is to restore the functionality of complex and diverse systems such as tropical rainforests. This could pose significant challenges for landowners and restoration managers who have limited resources and decide to use the most abundant and better known species.

As many other tropical areas, TBR is characterized by a high palm diversity and abundance with continuous fruit production. To our knowledge, our work is the first effort to link these two factors with palm propagation for restoration purposes, even though information on tropical palm composition and palm phenology is already available (De Steven, Windsor, Putz, & León, 1987; A. Henderson, Fischer, Scariot, Whitaker-Pacheco, & Pardini, 2000; Peres, 1994). Palm propagation at TBR is more limited by the low abundance of rare species than by germination capacity or by ecological constraints such as supra-annual flowering, fruit predation, or seed predation.

Obtaining seeds from several species is crucial for restoration projects (Rodrigues, Lima, Gandolfi, & Nave, 2009), although most tropical species grow at very low densities, which clearly represents an obstacle for successful propagation. At TBR, more than half of the palm species are rare, uncommon, or habitat specific, and usually known from a few specific places in the reserve. Constraints on the propagation of low abundance tropical species have been discussed mainly for highly threatened species (Hoffmann et al., 2015), including potential negative effects on genetic diversity and plant performance caused by using seeds from small populations (Mijnsbrugge et al., 2010). For this reason, extensive monitoring and focal searches need to be in place to secure finding sufficient viable seeds (Wang et al., 2016). In the case of TBR, more intensive searches would allow finding more individuals of rare species, but this would make sampling more difficult for collection purposes in general. Therefore, initial efforts should focus on species with an additional conservation value (i.e., endemic or threatened) and in which it is subsequently possible to find sufficient individuals for their propagation. In the case of the rare palms at TBR, abundant fruiting could be advantageous as it allows the production of a large numbers of seedlings even when only a handful of mature individuals are found.

In addition to low population density, fruiting seasonality represents a second challenge to obtain sufficient seeds from different species (Blakesley et al., 2002). Phenological behavior is a fundamental aspect to consider when planning seed collection protocols, both at the species and community level (Morellato et al., 2016). In our case, we did not mark individuals, and several species were very scarce, making it difficult to draw complete conclusions regarding the overall pattern of fruiting phenology of palms at TBR. The characterization of the phenological behavior is affected by sampling size (Morellato et al., 2010). However, our data facilitated the characterization of the basic behavior of most species and its implications for seed collection. For example, data show that species of Bactris present fruits most of the year, whereas those of Chamaedorea are mainly seasonal.

According to Newstrom, Frankie, and Baker (1994), the phenological behaviors of tropical plants could be categorized as continual, subannual, annual, and supra-annual.

At TBR, almost half of the species showed continual to subannual patterns producing fruits all year round in large quantities (or at least for more than 6 months a year), which clearly represents an advantage for seed collection purposes. Nonetheless, for these, fruit production could be the result of the reproduction of low proportion of fertile individuals so that at a specific time, it is still difficult to obtain large seed quantities. To secure greater genetic diversity, several collections should be done throughout the year. On the other hand, a second group of species showed annual flowering (i.e., A. martiana, Chamaedorea spp.), which limits the seed collection period. For the TBR palms, fruit production coincided with the peak and end of the rainy season, which was the best time to collect seeds to increase species diversity. Finally, supra-annual flowering has been considered as another significant obstacle for obtaining sufficient seeds in tropical trees (Holl, 2002), but this does not seem to represent a significant problem in tropical palms in which very few show supra-annual fruiting (e.g., some canopy species, A. Henderson, 2002). The only species with this type of fruiting behavior at TBR is R. gracilis (J. M. L., personal observation).

Following fruiting seasonality, germination potential is the third key aspect for palm propagation. Few studies have addressed the germination behavior of palm floras. Our results show that germination percentages were similar or higher than those reported for tropical rainforest tree species (Garwood, 1983; Sautu et al., 2006).

Methodological, anatomical, and ecological factors explain our germination results. First, fruit collection was done weekly for a whole year, and seeds were sown immediately, which reduced the probability of observing lower germination percentages as many species have short viability and cannot resist prolonged storage (A. Henderson, 2002). Several studies reporting low palm germination concentrated on species from seasonal habitats. Hard seed coats, fibrous mesocarps and endocarps, and remote germination are characteristic of palms adapted to xeric conditions. Nonetheless, all palms in our study show adjacent germination, and most of them (with the exception of Bactris spp. and D. moorei) have thin endocarps, both traits being associated with fast germination (A. Henderson, 2002). These conditions provide an advantage in the humid climatic conditions of tropical rainforests which facilitate high and fast germination (Vázquez-Yanes & Orozco-Segovia, 1993), helping seeds and seedlings to escape predators and pathogens (Daws, Garwood, & Pritchard, 2005).

In tropical rainforests, germination has been considered slow when it takes more than 4 (Garwood, 1983) or 12 weeks (Blakesley et al., 2002). According to our data, most seeds germinated in 25 weeks, but 10% of them required more than 40 weeks. In addition, germination curves were long in some species, and there was significant variation in the percentage and timing of germination within and among species. This poses challenges for nursery management in which the goal is to obtain a rapid and homogenous germination. A common practice is to select plants with specific characteristics or harvest seedlings from seeds with faster germination. However, this should be avoided if possible for restoration purposes where the aim should be to represent the genetic and ecological base of the species under consideration (Thomas et al., 2014), and is particularly important in palms with ample germination responses (Ataroff & Schwarzkopf, 1992; Harms & Dalling, 1995). A simple strategy to cover a higher germination spectrum would be to increase the number of individual palms as well as to explore pre-germination treatments that accelerate germination (Potvin, Cansari, Hutton, Caisamo, & Pacheco, 2003).