Study site

Our study was conducted in the TMCF of the FFR, in the central cordillera of western Panama. Our site was located in the Quebrada Honda section of the reserve (8°45’40” N, 82°14’22” W, 1,150 m asl), where forest structure and dynamics have been studied extensively by J. Dalling and coworkers (Andersen, Turner, & Dalling, 2010; Heineman et al., 2015; Heineman, Turner, & Dalling, 2016). During the study period, the annual precipitation was c. 5,700 mm yr⁻¹ according to our own recordings (HOBO Data logging rain gauge RG3, Onset Computer Corp., Cape Cod, MA, USA) set up c. 20 m outside the forest near the research station from March 2014 to April 2016.

The bedrock in Quebrada Honda is Rhyolitic tuff, which is acidic and poor in nutrients, and the topsoil is organic, with an acidic pH < 5.0 (Andersen et al., 2010). Emergent forest trees include Oreomunnea mexicana and Quercus sp. Other important genera of trees are Hedyosmum, Faramea, Trichilia, Guettarda, Ardisia, and Inga. The subcanopy and understory vegetation consists of trees and shrubs from the Rubiaceae, Malvaceae, Melastomataceae, Myrtaceae, Meliaceae, and others (C. Rodríguez, unpublished. data). Data are available on stem DBH of all trees >5 cm DBH in a 1-ha permanent forest plot (Prada et al., unpublished data) in the vicinity of our study area, which allowed us to scale up from our census to a plot-level estimation (see below). Stem density was 1,597 (trees >5 cm DBH) and 787 (trees > 10 cm DBH), and the total basal area was 46.3 m² ha⁻¹ (Prada et al., unpublished data).

Study trees and sampling methods

We censused all epiphytic vascular plants in the 22 selected trees, dividing the trees vertically into six height zones (0–5, 5–10, 10–15, 15–20, 20–25, 25–27 m). The uppermost zones were not present in all trees, the smallest sampled tree being only 9 m tall.

Individual abundance was quantified followed the “stand method” (Sanford, 1968), in which a “stand” is a group of individual shoots belonging to the same species that is spatially isolated or otherwise clearly separated from other groups of plants from the same species. Not all mechanically dependent plants were included in our biodiversity census: We excluded vines (climbing plants that germinate in and stay connected to the ground throughout their lives) and mistletoes (parasitic plants). Juvenile epiphytes less than two thirds of the adult’s size were not considered either because of the difficulty of identification.

Epiphytic vascular plants were separated into three categories: vascular epiphytes (plants that germinate on other plants and never establish a connection to the ground), hemiepiphytes (plants that germinate on other plants but later establish contact with the ground via aerial roots), and nomadic vines (climbing plants that germinate on the ground and later may lose their connection to the ground as they ascend) following the terminology outlined by Zotz (2013a). Vascular plants were identified using taxonomic keys and critical cases were identified by specialists. Voucher specimens were deposited in the herbaria of the University of Panama and the University of Chiriqui.

To estimate species richness in the area based on our sampling, we calculated the Chao 1 index (1984), which recognizes the fact that the sampled diversity normally is less than the actual diversity present in an area.

Our sampling strategy for the quantification of epiphytic matter was meant to assure a reliable estimate of the amount, composition, and variability of epiphytic matter on different surface types (branches and trunks) and tree sizes.

The DBH of host trees is usually a good predictor of epiphytic matter across trees of a forest (Chen et al., 2010). We determined the epiphytic matter on 22 trees with DBH ranging between 9 and 75 cm in order to obtain a wide representation of the available tree sizes to establish the quantitative relationship between epiphyte matter and DBH (Hsu et al., 2002; Wolf et al., 2009). The 22 trees ranged from 9 to 27 m in height with their crown branches starting from 8 to 14 m above the ground. The smallest tree was a palm lacking branches that was included because of the importance of palms in this forest and their inclusion in the tree survey used for extrapolations.

We sampled epiphytes on trunk and branch and additionally on surrounding vegetation up to a height of 2 m in 2 × 2 -m plots around the selected trees. Trees were climbed using single-rope techniques, or ladders were used in the case of smaller trees and lower sections of all trees. The epiphytic matter, that is, all living vascular and nonvascular epiphyte biomass and dead organic matter (DOM), of the 22 trees was sampled from one half of the tree trunk (180° of each trunk).

For estimating epiphytic matter in the tree crowns, we sampled all epiphytes from 28 branches (2–11 cm basal diameter) from 15 trees after cutting and lowering the branches to the ground. This allowed the determination of the relationship between branch diameter and epiphytic matter. We then estimated the diameters of all the primary branches of 19 of the 22 selected trees. Consistent with the approach used by Köhler et al. (2007), the remaining three trees were not taken into account because one of them was a palm without epiphytes in the crown and the other two were small with a strong crown deformation. The numbers and sizes of all branches were used to estimate epiphytic matter for the crown of each of the 19 trees.

For the understory, all epiphytic matter below 2 m in height inside the 2 × 2-m plots, with the exception of the focal tree, was collected.

In the lab, the epiphytic matter was separated into five categories: DOM, lichens, bryophytes, vascular plants, hemiepiphytes, and nomadic vines. Hemiepiphytes and nomadic vines were lumped for biomass because it was not always clear to which group they belonged. Each group was weighed within 2 hr of collection in order to obtain fresh weight. Subsamples were then oven dried at 80℃ for 48 hr (for most of the living tissues) and 96 hr for some Araceae and Bromeliaceae (depending on their size and level of succulence) and DOM. After drying, they were reweighed to obtain dry weight.

Total epiphytic matter per tree was regressed against DBH. The same was done for biomass/epiphytic matter by category. The resulting functions were used to estimate total matter per category on each of the trees in a nearby 1-ha forest plot, where tree DBH data were available (Prada et al., unpublished data). Two estimates of epiphytic matter for the whole forest (in kg ha⁻¹) were calculated, one based on regressions using the total epiphytic matter and one on the sum of the separate categories. All analyses were performed in R version 3.2 (R CORE TEAM 2015).

Epiphyte species richness and floristic structure

Each of the 22 trees hosted numerous bryophytes and at least one vascular epiphyte. We recorded a total of 2,423 stands, belonging to 155 species of vascular epiphytes, 84 of which could be identified to species level (Table A1). From the 155 species, we recorded 47 genera and 18 families. Of these, 133 species (85%) were epiphytes, 9 species (6%) were hemiepiphytes, 6 species (4%) were nomadic vines, and the remaining 7 species (4%) could not unambiguously be assigned to hemiepiphytes or nomadic vines (Table A1). Five species could not be identified even to family level because reproductive organs were lacking. In our analysis, those species not identified to species or genus level were included as morphospecies.

The species accumulation curve for vascular epiphytes in the sampled trees (Figure 1) does not reach an asymptote which indicates that not all species in the area were included in our sample. The Chao 1 estimation of the total number in the study area was 189 ± 16 (SE) with a 95% confidence interval of [169, 239] species.

Figure 1.

Epiphyte species accumulation curve (155 species, 22 trees) in a montane cloud forest in western Panama (Quebrada Honda, Fortuna). The shaded area indicates the 95% confidence interval.

Ferns and fern-allies were represented by 51 species in 17 genera and nine families. The most constant (widespread among trees) epiphyte species was an unidentified Elaphoglossum (Dryopteridaceae), which was found on 15 of the 22 sampled trees (Table 2). The most abundant epiphyte species was a filmy fern with 188 stands (Hymenophyllum axillare Sw., Hymenophyllaceae), followed by Elaphoglossum nigrescens (Hook.) T. Moore ex Diels (Dryopteridaceae) with 115 stands (Table 2). Although the dominant species in terms of constancy and abundance were ferns, angiosperms were more species rich, with 104 species in 53 genera and nine families, representing five eudicotyledonous and four monocotyledonous families (Table A1). The vascular epiphyte assemblages were dominated by Orchidaceae with 38 species, contributing 25% (Figure 2) to epiphyte species diversity, and 494 stands, representing 20% of total abundance (Figure 3). Furthermore, orchids were found on 19 trees with Pleurothallis as the most species-rich genus of the family occurring in nine trees (Table 2). Pleurothallis dentipetala was the most abundant orchid species, with 57 stands (Table 2). Other species-rich families were Araceae with 23 species (15%), also documented on 19 trees, Hymenophyllaceae (9%), Dryopteridaceae and Polypodiaceae (each 8%), and Bromeliaceae (7%, Figure 2).

Figure 2.

Species richness of epiphyte families recorded on 22 trees in western Panama (Quebrada Honda, Fortuna). For a complete species list, see Table A1.

Figure 3.

Proportional abundance of epiphytes by family (from a total of 2,423 individuals) recorded on 22 trees in western Panama (Quebrada Honda, Fortuna).

Table 2.

Most Important Epiphyte Species Sorted by Constancy.

Anthurium (Araceae) and Elaphoglossum (Dryopteridaceae) were the most species-rich genera (each 10% of the total), followed by Pleurothallis (Orchidaceae, 6%), Guzmania (Bromeliaceae), Peperomia (Piperaceae), and Trichomanes (Hymenophyllaceae; each 5%).

Epiphytic biomass

A total of 28 kg of epiphytic matter was removed from the 22 sampled tree trunks. From the 28 branches in the crowns, 25 kg of epiphytic matter was removed and 12 kg from the understory. Taken together, 62% of epiphytic matter was DOM and the remaining 38% were vascular and nonvascular epiphytes (Table 3). Vascular epiphyte biomass was dominated by bromeliads (73% of total vascular plants), with a lower contribution of ferns (12%) and orchids (8%). The remaining vascular epiphytes, mostly Araceae and Piperaceae, contributed just 7% to the total.

Table 3.

Epiphytic Matter Components in the Tropical Montane Forest (1,150 m asl) of Quebrada Honda, Fortuna Forest Reserve, western Panama (kg cm⁻²).

Epiphytic matter increased exponentially with DBH as well as with branch diameter (R² = .65; p < .001; Figure A1). The total epiphytic matter per tree was estimated by adding the calculated epiphytic matter on all branches per tree to the measured epiphytic matter on the trunks for each of the 22 sampled trees. This epiphytic matter per tree (on trunk plus branches) increased exponentially with tree size (DBH) (linear regression analysis with log [epiphyte matter]: R² = .53; p < .01; Figure 4). Similar relationships were found for individual categories: DOM (R² = .59; p < .001; Figure A2), lichens (linear relationship, R² = .60; p < .001; Figure A3), bryophytes (R² = .71; p < .001; Figure A4), and vascular epiphytes (R² = .53; p < .001; Figure A5). For hemiepiphytes, the relationship was not significant (R² = .26; p > .001; Figure A6).

Figure 4.

Relationship of epiphytic matter and trunk diameter at breast height (DBH; cm), based on epiphytes on the branches and trunks of 22 trees in western Panama: Quebrada Honda, Fortuna (log 10 epiphyte matter) = 1.26 + 0.04 DBH, R² = .53; p < .001).

Based on these regression equations (or the mean in the case of hemiepiphytes) for each epiphyte group and the DBH of all trees in a nearby 1-ha forest plot (Prada et al. unpublished data), plus an extrapolation of the overall mean biomass in the understory in the 2 × 2-m plots, we estimated the total epiphytic matter in this forest to be 16,439 kg ha⁻¹ (Table 3). We obtained a somewhat higher estimate (18,974 kg ha⁻¹) based on the direct regression of total epiphytic matter per tree with DBH, shown in Figure 4.

Epiphyte species richness

The epiphytic community in the studied forest is rich in species and comparable to other primary tropical lower montane forests between 1,000 and 2,000 m asl (Table 4). It is characterized by a high proportion of rare species, so that the species abundance curve did not saturate. The real number of species in the forest is expected to be considerably higher than the observed number (189 vs. 155) according to our Chao 1 index estimate. Interestingly, the estimated number of vascular epiphyte species roughly equals the observed number of tree species (>5 DBH cm) in the 1-ha census used for our extrapolation (190 species; Prada et al. unpublished data).

Table 4.

Number of Species of Vascular Epiphytes Recorded in Tropical Montane Forests.

For dominant epiphyte families, the registered species represented only 9% (Orchidaceae) or 15% (Bromeliaceae) of the estimated number of species in the FFR. This indicates that apart from the high α-diversity found in this study, the β-diversity is also likely very high in this heterogeneous mountain reserve, which ranges from 700 to over 2,000 m asl, stressing its conservation value. Many of the orchid and bromeliad species occurred in a small number of trees. This high proportion of uncommon species may also explain why the 22 sampled trees hosted quite a low percentage of the species reported in the reserve (McPherson unpublished data, Meisner & Zotz, 2012).

The representation of angiosperms families in this study agrees very well with general patterns of epiphytes diversity in the Neotropics, the most common taxon being Orchidaceae, followed by Araceae and Bromeliaceae (Gentry & Dodson, 1987). However, taking into account all species, some fern families were more species rich than the Bromeliaceae. According to previous studies, Araceae are the second-most species rich group after orchids in tropical lowland forests, while in montane forests, ferns tend to become more important than Araceae (Kreft, Köster, Küper, Nieder, & Barthlott, 2004; Krömer, Kessler, Gradstein, & Acebey, 2005). However, in our study area, both groups show similar species richness. The Neotropical epiphytic eudicotyledonous flora is more diverse in families but much less diverse in species compared to the monocots (Zotz, 2016). The largest eudicot epiphyte family globally is the Piperaceae, which also contributed 6% (nine species) to the total species count in this study, occupying the first place of eudicotyledonous families, followed by Ericaceae and Gesneriaceae. The large contribution of the Orchidaceae to epiphyte species richness confirms patterns found in forests from both the tropical lowlands and montane primary and secondary forests (Barthlott, Schmit-Neuerburg, Nieder, & Engwald, 2001; Kreft et al., 2004; Küper, Kreft, Nieder, Köster, & Barthlott, 2004). On a global scale, orchids account for ca. 68% of all species and 59% of all genera of epiphytes (Zotz, 2013b).

The epiphyte species and families were not randomly distributed in the forest, but most showed clear preferences for height zones. The lower strata were preferred by aroids and several fern families, especially the filmy ferns (Hymenophyllaceae). Eight of the 14 species of Hymenophyllaceae were exclusively recorded in the lower zone (0–5 m), six were found in more than one zone, but just one, a Trichomanes species, was found in the upper forest canopy. In contrast, the Orchidaceae as a family preferred the higher strata of the forest, whereby this general preference differed considerably between species (Table A1). Orchids were less common on understory shrubs than on the lower 2 m of the trunks. One explanation could be that they prefer thick branches because these offer a different substrate, another that trunks have been available for colonization for a longer (Hietz & Hietz-Seifert, 1995).

Epiphytic matter

Because of differences in the estimation methods in previous studies, comparisons are not straightforward. For example, the 2-fold difference in the epiphytic matter estimates of Köhler et al. (2007) and Nadkarni et al. (2004) for a montane cloud forest of Monte Verde, Costa Rica (16,215 kg ha⁻¹ vs. 33,100 kg ha⁻¹) may be explained by the different number and different limits of DBH classes, different stratification of the trees, and different sampling methods (Table 1). Alternatively, the results may reflect real differences between plots or may be explained by a mixture of biological and methodological reasons. The epiphytic matter estimate of the present study was 16,439 kg ha⁻¹ and coincides with that of Köhler et al. (2007). Taken together, informed tentative estimate of typical epiphytic matter in this type of forest would be in the range of c. 20,000 kg ha⁻¹.

The highest estimation of epiphytic loads so far is from an upper montane forest in Colombia (Hofstede et al., 1993) with an estimation of 44,000 kg ha⁻¹ of epiphytic matter. This estimate was based on sampling data from three trees, chosen as representative of three canopy strata. This approach may easily yield an overestimate, but epiphyte loads are indeed very high in the wet upper montane forest sampled, with large (c. 1 -m diameter) balls of bryophytes covering the trees (M.Y. Bader, personal observation). In comparison, the moss layer in our study area was modest, though still considerable. In the montane forest, low temperatures probably favor both bryophyte growth and reduce decomposition rates (Conant et al., 2011), which may explain the high fraction of DOM (63%). Similarly high estimates of DOM are reported by Nadkarni et al. (2004) and Kanzaki & Sri-Ngernyuang (2012), which also coincidence with the values of our study (62%).

The estimated 6,214 kg ha⁻¹ epiphytic biomass (all epiphytic matter excl. DOM) contributes about 2% to the total aboveground biomass in the Quebrada Honda forest, which is estimated at 360,000 kg ha⁻¹ based on tree survey used for our extrapolations (K. Heineman, unpublished data). Considering total forest biomass, that is, aboveground and belowground biomass, this relative contribution is even lower and represents only a very modest fraction. Only few studies report direct estimations of both tree leaves and epiphytes foliage from tropical forests (Kanzaki & Sri-Ngernyuang, 2012; Socher, Vellozo Roderjan, & Galvão, 2008) with c. 13%. For a lowland rain forest in Panama, in turn, the contribution to green biomass was estimated to be much lower with only 0.02% (Golley, McGinnis, Clements, Child, & Duever, 1969). This difference, even after taking possible methodological biases into account, illustrates the great importance of epiphytes in tropical montane forests compared to lowland forests.

Reported epiphyte biomass values for tropical montane forests in the literature cover a very wide range, and our estimate lies roughly in the middle of this range. Some of the previous estimates may be considered unreliable because of the very limited sampling (e.g., three trees only) and simple extrapolation methods (Hofstede et al., 1993; Nadkarni, 1984). In spite of the uncertainties related to the epiphyte biomass versus DBH relationships in our study, the extensive sampling and the detailed information about forest structure used for the extrapolation makes our estimate relatively robust. Our data show that epiphytic matter increased exponentially with tree size (DBH, linear regression analysis) consistent with other reports Díaz et al. (2010). For future surveys, we would suggest using this tree-size based method with some modifications. Studying a range of tree sizes is necessary, as there is a strong relationship between epiphyte load and DBH. However, as the smaller trees were more similar in their biomass, fewer trees in the smaller size classes would need to be sampled. On the other hand, as large trees have high but also very variable epiphyte loads, we recommend including more large trees to improve the regression model in this influential size range. Sampling the understory is not necessary for estimating the biomass per hectare, as this part of the forest contributed only about 2% to total epiphytic matter, which is much less than the uncertainty associated with the biomass-size model. This recommendation does not, however, hold for biodiversity studies, as conditions specific for the understory might support specific species not found in higher strata (Krömer et al., 2007; Zotz, 2007).

Even though bryophytes were found in all trees, they contributed relatively little to the total epiphytic matter, at variance with some other montane forests (e.g., bryophytes contributed nearly 75% of green epiphyte biomass in Hofstede et al., 1993). However, their role in the ecosystem may be much larger than suggested by this low contribution to biomass. For instance, their contribution to DOM is important since most of the DOM is derived from bryophytes (Turetsky, 2003). They may also contribute to the maintenance of high epiphyte biomass by facilitating vascular-epiphyte germination, establishment, and survival (Zamfir, 2000). Additionally, their poikilohydric nature and large water-holding capacities are probably very important in their regulatory potential for water fluxes and microclimate in the forest (Cornelissen, Lang, Soudzilovskaia, & During, 2007).

Epiphytes and climate change

Because of their existence at the interface of vegetation and atmosphere, epiphytes have been called particularly vulnerable to climate change. In contrast, at least in seasonal climates, epiphytes are well adapted to survive irregular access to water and appear relatively insensitive to, for example, increased drought in lowland forests (Zotz & Bader, 2009). In montane cloud forests, however, epiphytes thrive because of the humid conditions and may indeed be quite susceptible to drought. As a result, changes in microclimate due to forest disturbance have been linked to a floristic turnover in the epiphyte vegetation of tropical montane forests (Larrea & Werner, 2010). Climatic changes, especially changes in atmospheric water inputs, are therefore expected to have detrimental consequences for cloud-forest epiphytes.

To date, information on biomass or biodiversity of epiphytes is lacking for the great majority of tropical montane forests. Thus, any less than catastrophic change may easily go unnoticed. With this study, we created a baseline for a diverse montane forest and contribute to the still-small body of epiphytic matter estimates in tropical forest ecosystems. More studies are needed to understand and quantify the current and future role that epiphytic plants play in these forests.