Floristic overview of the epiphytic bryophytes of terra firme forests across the Amazon basin

Epiphytic bryophyte communities in terra firme forests of the Amazon region were investigated for the first time through standardized sampling across the Amazon basin. The sampling was carried out at nine localities, where bryophytes were collected in five height zones, from the forest floor to the canopy of eight canopy trees per locality. The sampling generated 3104 records, identifying 222 species and 39 morphospecies, within 29 families. The most common families were Lejeuneaceae (in 55%), Calymperaceae (in 8%), Leucobryaceae (in 4%) and Sematophyllaceae (in 4%). Richness and species composition did not show any geographical gradient. The bryoflora was significantly richer in the localities of Saul, in French Guiana, and Tiputini, in Ecuador, than in the other localities, probably due to differences in local climatic conditions. Among the 155 species recorded for more than one locality, 57 were classified as specialists. A total of 29 species (among which 3 were unidentified) were sampled only in the canopy, which reinforces the importance of canopy sampling for the study of epiphytic bryophytes in the Amazon.


Introduction
It is estimated that approximately 800 bryophyte species (mosses, liverworts and hornworts) occur in the Amazon region (Gradstein et al. 2001).In the Brazilian Amazon, the most updated species count indicates a total of 561 taxa (Costa & Luizi-Ponzo 2010).Only in the last decade, several known species have been cited as new records for the region (Moraes & Lisboa 2006;Alvarenga et al. 2007;Reiner-Drehwald & Schäfer-Verwimp 2008).In addition, species new to science have been collected and described in the region (Zartman & Ackerman 2002).
The present knowledge of the Amazonian bryoflora is primarily disseminated in publications of local inventories.Most of the bryophyte sampling has been carried out in the eastern Amazon, more specifically in the areas surrounding the Brazilian localities of Belém, Caxiuanã and Ilha do Marajó (Lisboa 1984(Lisboa , 1985;;Lisboa & Maciel 1994;Lisboa & Ilkiu-Borges 1997;Lisboa et al. 1999).Inventories conducted in scattered localities in the south of the state of Pará and the Serra dos Carajás have summarized the information available for the southeastern Amazon (Ilkiu-Borges et al. 2004;Moraes & Lisboa 2006).In central Amazonia, studies have mostly been concentrated in terra firme (non-flooded) forest and campinarana (white-sand forest) sites, within approximately 100 km of the city of Manaus (Lisboa 1976;Griffin III 1979;Zartman & Ilkiu-Borges 2007).Further sampling has been conducted across the states of Roraima (Yano 1992;Santiago 1997), Rondônia (Lisboa 1993) and Acre (Costa 2003).
The Amazon occupies an area of approximately 6 million km2 and is subject to a range of climatic conditions.Total annual rainfall ranges from 1500 mm (in northern Bolivia) to over 3000 mm (in the Upper Rio Negro region, the Colombian Amazon and northern Peru), the lowest monthly rainfall (in the three driest months) ranging from 0 mm to 447 mm, respectively, in those same regions.Moreover, environmental characteristics such as soil type, topography and catchment drainage vary at local scale, producing a mosaic of habitats and landscapes.Locally and across the region, different sources of variation are associated with diversity and compositional patterns in several plant groups (Londoño & Alvarez 1997;Costa et al. 2005;ter Steege et al. 2006;Drucker et al. 2008;Zuquim et al. 2009).
Recently, standardized and robust scientific data from extensive regions such as the Amazon have become a valuable requisite for conservation and governance policies.Large scale biological projects, funded either by the Brazilian government or by non-governmental organizations, have been establishing sampling protocols, a necessary step towards the primary biodiversity data needed to inform policy-makers.Despite the large amount of information Floristic overview of the epiphytic bryophytes of terra firme forests across the Amazon basin Sylvia Mota de Oliveira on bryophytes in the literature reviewed above, there is a need for a pre-established experimental or sampling design.In addition, few studies of bryophytes have explored the canopy, despite the fact that the importance of canopy sampling has previously been made clear by authors such as Cornelissen and Gradstein (1990), who observed that 50% of the listed species in a forest in Guyana were restricted to the canopy.
In order to provide an ecological, as well as a floristic, overview of the epiphytic bryophytes in terra firme forests of the Amazon, we sampled bryophyte communities at nine localities across the Amazon basin, using a standard protocol, from the forest floor to the canopy of eight trees per locality.In this paper, we provide a floristic overview of these communities, in which our major questions are as follows: What are the most abundant families and species of epiphytic bryophytes occurring in terra firme forests of the Amazon?What are the family and species rankings across the basin and along the vertical gradient on the host trees?Do these rankings change according to the geographical locality or to the height zone on the host tree?Are there species with a statistically significant preference for a given height zone on the host trees?In short, this is the first quantitative description of the epiphytic bryoflora across the Amazon region, conducted with the objective of providing the necessary background information for future large scale research on the ecology and biogeography of the group.

Sampling procedure and composition assessment
In order to sample communities of epiphytic bryophytes, we selected nine localities across the Amazon basin, along an east-west transect (Fig. 1).At each locality, epiphytic bryophyte communities were sampled from the bottom to the top of eight selected canopy trees growing on nonflooded plateaus (terra firme forest).The trees were divided into height zones, used as a surrogate for the microclimatic gradient found from the forest floor to the canopy.The height zones were established as follows (Mota de Oliveira et al. 2009): from the base of the tree up to 1.5 m (zone 1); the lower and upper trunk spaces (zones 2 and 3); between the upper trunk and the base of the crown (zone 4); and finally the outer sun-lit twigs (zone 6).The bryophyte communities were pooled samples of 4 patches of approximately 10 × 10 cm 2 each, per height zone.The collections from the sites Saül (French Guiana), Mabura Mora (Guyana) and Mabura Wallaba (Guyana) were taken from more than ten trees and included bryophyte samples taken from the middle canopy of the host trees (zone 5).In order to balance the sampling strategy, we used a sub-set of the data of these three sites, which consisted of samples from eight randomly chosen trees per locality and five height zones (zones 1-4 and zone 6).
The specimens present in the samples were identified using keys, monographs (Reese 1993;Gradstein 1994;Dauphin 2003) and available floras (Gradstein et al. 2001;Gradstein & Costa 2003).Material was deposited at the Herbaria of the Museu Paraense Emilio Goeldi (acronym, MG; Brazil), Instituto Nacional de Pesquisas da Amazônia (acronym, INPA; Brazil) and Nationaal Herbarium Nederland, Leiden University branch (acronym, L; the Netherlands).Material from the Guianas (sites 2-4) was re-identified to ensure synchrony in the taxonomy among the plots.Specimens that could not be identified down to the species level were classified as morphospecies, which allowed the analysis to run at species level.We called species "restricted" when, in our dataset, they were found in only one locality; we call species "typically epiphylls" when their main habit is epiphyllous, based on its description in current literature (Gradstein & Costa 2003;Zartman & Ilkiu-Borges 2007).

Data analysis
Each pooled sample (40 cm 2 ) was considered a "plot" and generated a species list.We considered the use of an abundance measure inappropriate, due to the impossibility of separating individuals for most of the species and to the intrinsic variation in plant size.To quantify community structure-species abundance distribution for the complete dataset and accumulation curves per locality-we used frequency as a surrogate for abundance, summing the number of plots per locality in which each species was recorded.This value ranged from 1 to 40, the maximum possible number of plots.In order to classify species as specialists (those with a tendency to occur in a given height zone) or generalists (those without such a tendency), we performed indicator species analysis (McCune & Grace 2002), with the software PCORD 5, for species recorded in at least three localities.This analysis also provides an indicator value for those species that show a significant preference for a given height zone.The indicator value ranges from 0 to 100, according to the strength of the preference.We calculated the weighted average height zone for the specialists, as additional and more straightforward information on species occurrence.

Results
A total of 351 plots, sampled from 72 trees, yielded 3104 occurrences of bryophytes, representing 29 families, 97 genera and 261 species or morphospecies (Appendix 1).As shown in Tab. 1, species richness of the localities varied from 51 for Reserva Ducke, Brazil (central Amazon), to 127 for Tiputini, Ecuador (western Amazon).Species richness was significantly higher for two of the nine localities: Tiputini; and Saül (in French Guiana) (Fig. 2).In terms of species richness, there were no significant differences between or among the other seven localities, which had an average of 64 species in eight host trees.On average, nine species were recorded per plot, ranging from 5-6 species in the eastern and central Amazon to a peak of 17 species per plot in Tiputini, Ecuador.Tiputini was also the locality where we found the highest proportion of restricted species-37.8%(sampled exclusively at one locality)-and the highest proportion of facultative epiphylls-16.5%-thelatter ranging from 5.7% to 10.8% among the other localities.
Of the 261 species identified, 155 were recorded in at least two localities.All localities investigated showed at least one locally abundant species that was not among the top 10% of abundant species in the complete dataset (Tab.1).
The family ranking over height zones 1-4 was rather consistent with the general ranking described above for the localities.It differed, however, in height zone 6, where Lejeuneaceae, again the most abundant family, was followed by Jubulaceae, Pterobryaceae, Macromitriaceae and Calymperaceae.
Species richness and the number of records were comparable across the different height zones (Tab.3).Species composition, however, was partially related to height zone.Out of 155 species recorded in at least two localities, 57 were significant indicator species for a particular zone.Zone 6 had the highest number of indicator species (30), followed by zone 1 (19), zone 4 (7) and zone 2 (1) (Tabs.3 and 4).Some species, although widespread across the basin, were completely restricted to zone 6, which was the case for Vitalianthus urubuensis, Colura greig-smithii and Caudalejeunea lehmanniana, as well as for all five Diplasiolejeunea species recorded (with the exception of 1 record of the 46, for D. cavifolia in zone 4).The two most abundant indicator species of zone 6, however, were found in at least three other height zones.Accordingly, the five most abundant indicator species of zone 1 were also found in other height zones, as were other zone 1 indicator species.Indicator species of zone 1 belonged to 11 different families and only 7 of the 26 species belonged to Lejeuneaceae.In zone 6, however, the indicator species belonged mainly to Lejeuneaceae, followed by Pterobryaceae and Jubulaceae.Accordingly, when we performed the same indicator analysis at the family level, we found that the number of families classified as specialists was higher in the understory-zones 1 and 2-than in the canopy-zones 4 and 6 (Tab.5).

Floristics
Is there a general Amazonian bryoflora?We found that the most abundant families (Lejeuneaceae, Calymperaceae, Sematophyllaceae, Leucobryaceae and Plagiochilaceae) kept their rank, in terms of abundance, across the basin.This result supports a general description of the Amazonian bryoflora (Gradstein et al. 2001).The family rank of epiphytic bryophytes only changed when the assemblages of the different height zones on the host tree were analyzed separately: the family composition of the outer canopy was distinct, Lejeuneaceae being followed by Jubulaceae, Pterobryaceae, Macromitriaceae and Calymperaceae.
It has long been known that bryophyte species composition differs between the understory and the canopy, both in terms of species composition and in terms of life forms (Richards 1984;Cornelissen & ter Steege 1989;Mota de Oliveira et al. 2009;Sporn et al. 2010).The restriction of some bryophyte species to the canopy has also previously been observed in the Amazon rain forest (Gradstein et al. 1990), as has the shift of the occurrence of some bryophyte species to lower height zones in a tree, as a result of opening of the canopy (Acebey et al. 2003).Based on such observations, epiphytic species were traditionally classified as "shade epiphytes" when restricted to the bottom part of the host tree, "generalists" when occurring in many height zones, and "sun epiphytes" when restricted to the canopy (Richards 1984).We suggest that the mechanisms leading Acta bot.bras.27(2): 347-363.2013.
to the restriction/preference of a given species to one of the habitats are not only related to light intensity, as suggested by the denomination, but also to population dynamics.While the occurrence of some species in the canopy is clearly related to the requirement of high light levels for development, such as reported for Frullania species (Romero et al. 2006), we hypothesize that our observations of typical epiphylls in the canopy branches, such as Colura greig-smithii, Odontolejeunea rhomalea and Leptolejeunea elliptica, are more related to their ability to rapidly colonize recent substrates than to light availability.This working hypothesis certainly deserves further testing with a proper experimental design.

Species richness
Is bryophyte species richness across the Amazon comparable to tree species diversity?According to our results, species richness was highest in two localities.The high richness in geographically separate localities such as Tiputini and Saül, as well as the unchanged family ranking across the basin, clearly differed from the gradients in diversity and composition of canopy trees in the Amazon (ter Steege et al. 2003;ter Steege et al. 2006).In canopy trees, alpha diversity increases along the east-west axis and the composition shows a clear change in family ranking along the axis running from the Guiana Shield, in the northeast part of the basin, to Bolivia, in the southwest.In the epiphytic bryophytes of the region, however, the sharpest compositional gradient is apparently established locally, rather than across geographic distances, at least in the Guianas region (Mota de Oliveira et al. 2009).We therefore believe that Tiputini (Ecuador) and Saül (French Guiana) showed high species richness due to specific microclimatic conditions.In biomes adjacent to the Amazon, such as the Andes and the Atlantic Forest of Brazil (Wolf 1993;Kessler 2000;Costa & Lima 2005), the richness gradient in bryophyte assemblages is set by altitude and is usually attributed to the correlated gradient in relative humidity and temperature.Saül and Tiputini are also at higher elevations than are the other localities studied.Elements of sub-montane and midmontane forest, such as the genus Porotrichum, occurred in these two localities and nowhere else.The high species richness of Saül has been related to the high and constant levels of relative humidity and regular fog in comparison with other Amazonian terra firme sites (Gradstein 2006).Tiputini may offer similar conditions.In our data, the percentage of typically epiphyllous species growing on bark in Tiputini was significantly higher than in any other locality, which may again indicate favorable conditions of relative humidity.For instance, Odontolejeunea rhomalea and O. lunulata, Colura greig-smithii, Radula mammosa, found on the bark of the host trees only in Tiputini, are typical epiphylls.Published records of those same species indicate their presence in Central and Eastern Amazonia (so there is no dispersal barrier), albeit growing on leaves (Zartman & Ilkiu-Borges 2007).Apparently, the more favorable microclimatic conditions in Tiputini and Saül simply allow the establishment of a greater number of species.In Tiputini, the proximity with the Andes may also play a role as an extra source of species.The greater relative importance of the family Neckeraceae, for instance, can be related to the fact that this is an important Andean family in terms of abundance (Churchill 2009); and the higher number of Plagiochila species, a well known feature of sub-montane and montane forests, is probably caused by the abundance of Plagiochila in the Andes.

Understory specialists and canopy specialists
The indicator species of zone 1 belonged to 11 different families, whereas those of zone 6 belonged to only three families: Lejeuneaceae, Pterobryaceae and Jubulaceae.This result suggests that the canopy habitat may be a recent "acquisition" in evolutionary time, as has been demonstrated for an epiphyllous habitat (Wilson et al. 2007) and for ferns (Schneider et al. 2004;Schuettpelz & Pryer 2009).However, the attempt to verify habitat specialization at the family level, as we have shown in our results, should be carefully discussed.The fact that Lejeuneaceae, Calymperaceae, Sematophyllaceae and Leucobryaceae presented a statistically significant preference for a given zone is not entirely consistent, because these families have indicator species in both the understory zones and the canopy zones along the gradient.On the basis of our results, the families that can be consistently classified as indicators are Calypogeiaceae, Leucophanaceae, Hookeriaceae, Leucomiaceae, Thuidiaceae and Fissidentaceae for the lower (understory) height zones; and Jubulaceae, Pterobryaceae and Macromitriaceae for the upper (canopy) height zones.These can be designated indicator families because most of theie species are "restricted" to their respective height zones.

Sampling considerations and future research
The results of our study clearly indicate the need to include the canopy of trees in standard protocols for bryophyte inventories in the Amazon, which has not been the case.Interestingly, species recently described, such as Vitalianthus urubuensis (Zartman & Ackerman 2002) and Cheilolejeunea neblinensis (Ilkiu-Borges & Gradstein 2008) were recorded in several of our localities and thus proved not to be rare.That means that we can still expect to find even common new bryophyte species in the Amazon.After describing Vitalianthus urubuensis, Zartman and Ackerman (2002) posed the question of whether the new species was restricted to Manaus region or was undersampled.The answer is now clear: the canopy was undersampled.That species was first collected by the authors exactly in a secondary forest, where canopy opening allows some canopy specialists to occur in lower height zones.
Regarding the geographic distribution of bryophytes in the Amazon, we propose the investigation of the local conditions at Tiputini and Saül, followed by the identification and analysis of climatically comparable areas in order to test whether bryophyte species richness is climate related.

Figure 1 .
Figure 1.Map of the study area, showing the sampling localities from east to west in the Amazon basin.

Figure 3 .
Figure 3. Species abundance distribution based on the complete dataset.Axis x: species ranked by abundance (number of records); axis y: number of records.

Figure 2 .
Figure 2. Species accumulation curves of epiphytic bryophytes measured on eight host trees, in each of the localities sampled.Axis x: number of host trees; axis y: number of species recorded.

Table 2 .
The 11 most abundant families in the dataset and the proportional distribution of records among the nine sampled localities.
*Correspond to observations in the text.

Table 3 .
Data related to indicator species, extracted from the complete dataset, per height zone (Z): total number of species (S), total number of records (N), average number of species per plot (P), number of indicator species calculated for species occurring in at least two localities (155 sp) according to ISA (p<0.05)(IndS), most abundant indicator species calculated for the complete dataset according to ISA (p<0.05).
ContinuesActa bot.bras.27(2):347-363.2013.HZ -height zone; WA -weighted average (height zone, from all records); IV -indicator value; IVR -indicator value for randomized groups; IVRSD -standard deviation of the indicator value for randomized groups.*p<0.05, by indicator species analysis, for all; **where the WA is a natural number, the species occurred only in that height zone.

Table 5 .
Families to which the species classified as specialists belong.* HZ -height zone; WA -weighted average (height zone, from all records); IV -indicator value; IVR -indicator value for randomized groups; IVRSD -standard deviation of the indicator value for randomized groups.*p<0.05,by indicator species analysis, for all.Acta bot.bras.27(2): 347-363.2013.