Unraveling algae and cyanobacteria biodiversity in bromeliad phytotelmata in different vegetation formations in Bahia State , Northeastern Brazil

Knowledge of algal and cyanobacterial diversity of phytotelmata remains poorly-known, especially for bromeliads from different vegetation formations. We investigated the microalgae communities of four species of tank bromeliads from different vegetation formations in Bahia State, Northeast Brazil, highlighting the composition, richness and diversity of taxa. Sampling of water stored in bromeliads was carried out quarterly between 2014 and 2016, and abiotic variables and morphometric attributes of bromeliads were measured. A total of 89 taxa of algae and cyanobacteria were recorded for the four bromeliad species studied. The microalgae communities of the phytotelmata varied among vegetation formations, with one tank bromeliad, Alcantarea nahoumii, with more complex architecture (higher number of leaves and thus more cavities), being distinguished by its high species richness (73 taxa). The bromeliads exhibited little similarity in species composition, with only one species (Phacus polytrophos) occurring in all four species. Throughout the entire sampling period, classes with higher species richness, especially due to A. nahoumii, were Zygnematophyceae, Cyanophyceae and Chlorophyceae, which accounted for about 80 % of all species inventoried. Our results contribute to the knowledge of microalga communities of bromeliad phytotelmata in Brazil with regard to species richness and composition, as well as significant environmental characteristics.


Introduction
The term phytotelm (φυτόν, phyton = plant; τέλμα, telm = pool) refers to the small amounts of water that accumulate in plant structures, such as leaves, flowers, or tree trunk and maintain an associated biota (Varga 1928;Maguire 1971).Phytotelmata occur in almost all regions of the world, although they are most common and most diversified in the tropical region, mainly due to the high rainfall there and the large numbers of plants capable of accumulating water (Fish 1983).
Little is currently known about the diversity, distribution patterns, and compositions of phytotelmata microalgae communities.Among the few published studies are those of Gebühr et al. (2006) with allochthonous populations of Sarracenia purpurea (Sarraceniaceae) in Germany, and Hernández-Rodríguez et al. (2014) with Tillandsia multicaulis (Bromeliaceae) in Mexico.Studies of phytotelmata algae began in Brazil in the 1970s, but have only recently become more frequent, especially in Bahia State, where several taxa have been recorded for the first time for that country, including a number of species new to science (Ramos et al. 2017a;b;c;d;2018a;b).
Most phytotelmata algal studies have taken floristic approaches (morphospecies), usually addressing specific groups (Lyra 1971;Sophia 1999;Ramos et al. 2011), or ecological approaches (Brouard et al. 2011;Carrias et al. 2014), although they have often been published without reliable taxonomic support at the species level.Studies comparing microalgae communities from bromeliad phytotelmata found in different vegetation formations, however, have not yet been undertaken.
We investigated the algae and cyanobacteria communities present in four species of tank bromeliads in different vegetation formations in northeastern Brazil, emphasizing the composition, richness, and diversity of the species in those environments in order to: (1) evaluate the influence of morphometric attributes of the bromeliad tanks and the abiotic variables of the water retained in them on the richness of the algal and cyanobacterial communities; and, (2) determine the species richness and similarities of the algal and cyanobacterial communities in four tank-bromeliad species.

Sampling
Study material was gathered from the water retained in the cavities (tanks) formed by the leaves of the bromeliads (four species, Figs.1B-E) in the four different vegetation formations.The accumulated water was collected using a plastic hose coupled to a 50 ml syringe.To ensure greater efficiency in collecting all biological material, that procedure was repeated several times in each tank.The water in some bromeliads was available in two kinds of phytotelmata (central and lateral), but the same collection procedures were used in both situations.Water from distinct tanks (if present), but from the same bromeliad specimen, were mixed so that each bromeliad represented a single sampling unit.
Quarterly excursions were carried out in the four study areas during 14 months, between 2014 and 2016, totaling 320 sampling units (80 bromeliads from each study area).Water from 20 tank-bromeliads were collected during each sampling period; the bromeliads sampled were randomly chosen based on the throw of a die (1 = North; 2 = East; 3 = West; 4 = South; 5 and 6 =, irrelevant), indicating the direction to follow to the next bromeliad.The distance between a sampled bromeliad and the next was always at least 5 m.
The collected material was held in properly labeled plastic containers (50 ml) and transported in a portable cooler to the Phycology Laboratory (UEFS); all collected materials were subsequently fixed in Transeau solution (Bicudo & Menezes 2017).

Environmental variables
Abiotic information concerning the tank water, such as temperature, pH, electrical conductivity, and total dissolved solids (TDS) was measured using a Hanna multiparameter probe; dissolved oxygen was measured using a portable digital Instrutherm (MO-910).All limnological variables were measured immediately after harvesting the bromeliad tank water samples.The morphometric characteristics of the bromeliads, such as height, width, and numbers of leaves were also recorded.

Identification of the microalgae and the attributes analyzed
All of the microalgae material was examined under an Olympus LX35 Optical Microscope.Taxonomic identifications were carried out to the infrageneric level whenever possible, and were mainly based on the morphological and metric characteristics of the population, consulting the specialized literature (Huber-Pestalozzi 1955;Prescott et al. 1975;Komárek & Fott 1983;Komárek & Anagnostidis 1998;2005;John et al. 2011;Wołowski 2011;Komárek 2013;Carty 2014).
Taxonomic richness was calculated based on the total number of species sampled in each bromeliad tank (Brower et al. 1998).The frequency of occurrence of each algae species (in each bromeliad species) was calculated considering the number of samples in which the taxa occurred in relation to the total number of samples collected.Categories of frequency followed Matteucci & Colma (1982): > 70 % (quite frequent, VF); ≤ 70 % and > 40 % (frequent, F); ≤ 40 % and > 10 % (occasional, O); and ≤ 10 % (rare, R).

Statistical analyses
One-way analysis of variance (ANOVA) was used to detect differences in the morphological characteristics of the bromeliads, the abiotic variables of water, and microalgae richness (dependent variables), among the different bromeliad species (independent factor).We also tested for pairwise differences, employing the Tukey post-hoc test using the multcomp package (Hothorn et al. 2008).For this analysis, the data were log10 (x + 1) transformed to fit the assumptions of normality and homoscedasticity.
Principal Component Analysis (PCA) was performed using a variance-covariance matrix; the data was transformed by Z-score to reduce the dimensionality of the morphological and abiotic water parameters of the different bromeliads tanks sampled.All analysis were performed in R environment (R Core Team 2017).The similarities of the algal and cyanobacterial compositions between the four bromeliad species studied were determined by using the Sørensen similarity index (Muller-Dombois, 1981): 2c/2c + A + B × 100, where A and B represent the number of species in areas A and B; and c corresponds to the number of species held in common in both areas.
A Venn diagram was prepared to illustrate the distribution of algal and cyanobacterial species richness among the different bromeliads, using software available at the Bioinformatics & Evolutionary Genomics (2017) site.

Morphological attributes of the bromeliads, limnological conditions, and microalgae richness
In terms of the characteristics of the bromeliads sampled, Hohenbergia stellata had high mean diameter and height values while Hohenbergia littoralis had low mean values for those parameters; the other two bromeliad species showed similar measures of those same morphological characteristics.Significant differences were found in the mean numbers of leaves among the bromeliad species, especially between Alcantarea nahoumii (37.8 leaves) and H. littoralis (14.9 leaves) (Tab.2).The waters in the bromeliad tanks of the four species were predominantly acidic, but with other significant differences between them.Aechmea cf.lingulatoides and A. nahoumii showed high mean electrical conductivities, and their total dissolved solids were two to three times greater than those observed in H. littoralis and H. stellata.The greatest mean dissolved oxygen concentration was found in A. nahoumii (7.8 mg L -1 ), followed by H. littoralis (6.8 mg L -1 ).That parameter displayed considerable variation among the different bromeliad species.
The bromeliads studied here were organized according to their morphological characteristics and abiotic water variables in the PCA, with the first two axes explaining 55.8 % of the data variability (eigenvalues: axis 1 = 2.35, axis 2 = 2.03, permutation test = 0.009).Most individuals of H. stellata were grouped on the negative side of axis 1, and were mainly correlated with higher plant diameter (r = -0.76)and height (r = -0.61)values; most individuals of A. nahoumii were also grouped there, correlated to pH (r = -0.65)and numbers of leaves (r = -0.58).Individuals of H. littoralis were grouped on the positive side of axis 1, correlated with high water temperature (r = 0.52).Most individuals of A. nahoumii and some of A. cf.lingulatoides (mainly the ones sampled in Jan/2015) were grouped on the negative side of axis 2, and were correlated with the highest TDS (r = -0.82)and conductivity values (r = -0.81)(Fig. 2).
The phytotelmata algae and cyanobacteria communities were represented by a total of 89 taxa in the four bromeliads species studied, distributed among 54 genera  The taxonomic richness of microalgae and cyanobacteria was highest in Alcantarea nahoumii (Serra da Jiboia -rock outcrops) with 73 taxa, followed by Hohenbergia littoralis (Parque das Dunas -restinga) (nine), Aechmea cf.lingulatoides (Fazenda Itaberaba -caatinga) (seven), and Hohenbergia stellata (Reserva Sapiranga -Atlantic Forest) (six).In terms of the species richness in the different bromeliads in the four vegetation formations studied, those on the rock outcrops containing 69 exclusive taxa, followed by restinga and caatinga vegetation (six each) and Atlantic Forest (four) (Fig. 3).Only one species, Phacus polytrophos Pochmann, occurred in all four bromeliad species studied.The most representative genus was Cosmarium (nine taxa), and it was the only desmid occurring in more than one bromeliad species.
In terms of the entire sampling period, the microalgae classes with the highest species richness in A. nahoumii were Zygnematophyceae, Cyanophyceae, and Chlorophyceae, which contributed approximately 80 % of the total number of species inventoried.Among the bromeliads species with the lowest richness, A. cf.lingulatoides contained mainly by diatoms, whereas the most representative class in H. littoralis was Euglenophyceae (Fig. 4).
The species compositions of the microalgae communities showed low similarities among the different bromeliad species (Tab.4).The species composition in A. nahoumii was only 3 % similar to that in A. cf.lingulatoides, 5 % similar to that in H. stellata, and 7 % similar to that in H. littoralis.The species composition in A. cf.lingulatoides was most similar to that in H. littoralis (13 %) and H. stellata (15 %).
Chlorophyta -Green algae were one of the most diverse groups occurring in phytotelmata in Bahia State, being represented in the present study by two classes: Chlorophyceae (19 taxa) and Trebouxiophyceae (six taxa), which occurred in all four areas, especially Serra da Jiboia (20 taxa); Monoraphidium was the most representative genus (four species).Some taxa, such as Enallax costatus, were widely distributed in the bromeliad tanks of A. nahoumii; Oedogonium pulchrum was the only species consistently encountered, often forming large and dense populations in H. littoralis tanks.
Acta Botanica Brasilica -32(4): 567-577.October-December 2018  Cyanobacteria -Cyanobacteria displayed the greatest species richness in the phytotelmata at Serra da Jiboia (21 taxa), especially Hapalosiphon stuhlmannii, which was the principal species forming gelatinous masses located in the central rosette of the bromeliads.Those masses contained numerous cyanobacteria representatives including coccoid and colonial, but mainly filamentous organisms.
Bacillariophyta -Diatoms were not very common in the phytotelmata studied, with only four taxa occurring in a small number of bromeliads in Serra da Jiboia and Fazenda Itaberaba.
Dinophyta -Dinoflagellates were represented by a single species (Parvodinium umbonatum) that was restricted to bromeliads at Serra da Jiboia.Nonetheless, that species showed the greatest frequency of occurrence (F = 79 %) among all taxa encountered at Serra da Jiboia.
Euglenophyta -Euglenophytes were represented by 11 taxa; Euglena mutabilis was the most frequent species (F=56 %) of that group in the bromeliads studied.
Table 3. List of microalgae taxa and their classification by size class, life form, and frequency of occurrence in each bromeliad species.

Microalgae richness and the influence of environmental factors
The high taxonomic richness of algae and cyanobacteria in the bromeliads at Serra da Jiboia was apparently related to the following factors: (1) plant architecture -Alcantarea nahoumii generally produces many (over 30) wide leaves that form large numbers of cavities (greater complexity) and consequently more places for water to accumulate; those leaves are also held at open angles that allow higher solar illumination; (2) the regional climate -the bromeliads grow on mountain tops, with forests on their slopes (distinct from the other three areas) -which favors the sites being constantly humid, so that the bromeliad tanks would not usually experience abrupt variations in water volumes (allowing greater community stability); (3) environmental data -the tank water was slightly acidic to neutral, with moderate conductivity and high TDS (indicative of high concentration of dissolved salts and nutrients) as compared to the other bromeliads; the usually high dissolved oxygen content probably contributed to a greater microalgae diversity.
In terms of plant architecture, Carrias et al. (2014) reported that greater bromeliad complexity (larger numbers of leaves -and consequently more sub-reservoirs) was associated with lower algal richness.The opposite was observed during the present study, however, mainly in A. nahoumii -the bromeliad species with the greatest number of leaves and the highest algal and cyanobacterial richness when compared to less complex bromeliads.Water volume may also directly influenced algal and cyanobacterial species richness and diversity, as the bromeliads at Serra da Jiboia usually contained large amounts of water well-distributed among the leaves in all four sampling periods.
Overall, the algal and cyanobacterial community similarities among the different bromeliads species were considered quite low (Tab.4), and probably linked to the distinct environmental conditions in each of the four study areas.Additionally, different bromeliad species tend to host distinct algal communities (Carrias et al. 2014).
Recent studies have demonstrated that changes in rainfall distribution can reduce chlorophyll-a concentrations in bromeliad tanks and therefore significantly affect microalgae dominance (Pires et al. 2017).Although we did not measure chlorophyll-a concentrations in the water accumulated in each bromeliad tank, it was evident that the plants from Serra da Jiboia (A.Variations in the algal and cyanobacterial communities in phytotelmata would be quite natural, with some wellestablished taxa and others more temporary.When comparing the desmids in A. nahoumii tanks (Serra da Jiboia) identified during this study with the 16 taxa previously identified for the same area (Ramos et al. 2011), nine continued to inhabit the tanks, whereas the others had disappeared.The nine taxa that persisted are widely distributed among bromeliads in the area, and well-adapted to local conditions, usually forming large populations, even with considerable external impacts (such as fires).
In terms of H. stellata (Reserva Sapiranga), only 13 out of 80 bromeliads sampled contained representatives of algae or cyanobacteria, and the numbers of species per sample were very low.Some contributing factors to that situation probably included: (1) luminosity -the bromeliads were largely shaded, but many of plants exposed to high sunlight (open areas in the forest) showed low diversity.Even in open areas, leaves from surrounding trees will fall into the tanks, often almost completely covering them; and, (2) dissolved oxygen -DO concentrations were commonly quite low (mean of 4.8 mg L -1 ; near 2.5 mg L -1 in some samples).According to Laessle (1961), algae are usually abundant in bromeliads that have high DO concentrations -a condition rarely observed in H. stellata in the present study.Additionally, many of those bromeliads held only small amounts of water (or were completely dry during some sampling periods).Among the main components found in H. stellata tanks were fungi, pollen grains, protozoa and organic matter debris.

Taxonomic groups -richness and main representatives
Desmids are one of the main microalgae groups occurring in bromeliad phytotelmata in Brazil (Sophia 1999;Sophia et al. 2004;Ramos et al. 2011).Water conditions, such as low pH and low conductivity, are typical of phytotelmata environments and favor desmid development (Sophia et al. 2004).A number of desmid species were recently described for the first time from bromeliads in Bahia State, suggesting that those environments as important centers of desmid biodiversity (Ramos et al. 2018a).
In terms of the main Chlorophyta taxa, we highlight the genus Rhopalosolen, which was recently reported for the first time for Brazil (Ramos et al. 2017c).In the bromeliads with a predominance of filamentous green algae (such as the genus Oedogonium), common predators such as rotifers, cladocerans, and copepods could be influencing the low observed algal richness.
The cyanobacteria comprise one of the most important groups found in phytotelmata in terms of taxonomic diversity and their ecological roles (Bermudes & Benzing 1991).An interesting feature of the cyanobacteria identified in the bromeliad phytotelmata was their normal association with rivers, waterfalls, tree barks, rocks, even hot springs (Komárek & Anagnostidis 1998;2005;Komárek 2013) -so that bromeliad phytotelmata are important environments for cyanobacteria diversity and bring together in just one place species otherwise known from very diverse habitats.
The low diversity of diatoms encountered in bromeliad tanks was not surprising.In a study of diatoms in bromeliads in Rio de Janeiro State, however, Lyra (1971) reported reduced numbers of those microalgae and attributed that result to their oligotrophic tank water conditions, as well as other environmental factors such as light intensity and temperature.According to that author, mineral element concentrations in those environments could considerably interfere with frustule development -although Pinnularia is one of the most common diatom genera inhabiting those microhabitats, and has been reported in a number of different bromeliad species (Lyra 1971;1976).
Parvodonium umbonatum (the only dinoflagellate identified in the present study) have been reported as occurring in bromeliads in Rio de Janeiro State (as Peridinium umbonatum; Sophia 1999).During dry periods, spherical, reddish-brown cysts of P. umbonatum were often encountered in the bromeliads at Serra da Jiboiapossibly representing a reproductive strategy in response to unfavorable environmental conditions.That species was consistently observed forming large populations during all of the sampling periods (Ramos et al. 2016).
In terms of Euglenophytes, Carrias et al. (2014) reported that heterotrophic microalgae species were dominant in areas exposed to direct sunlight -but that tendency was precisely the opposite among bromeliads in Bahia -with euglenophytes appearing predominantly in shaded (or partially shaded) bromeliads.Colorless euglenophytes were encountered at Reserva Sapiranga, but chlorophyllous taxa were much more common in bromeliads exposed to direct sunlight, especially Euglena mutabilis, which commonly forms large populations (Ramos et al. 2017d) Although a total of nine microalgae classes were encountered in the four bromeliad species, other microalgae groups may occur in those phytotelmata, such as Xanthophyceae (Sophia 1999) and Cryptophyceae (Hernandez-Rodriguez et al. 2014), although little is currently known about the diversity of those groups in those microhabitats, and what environmental conditions are ideal for their development.

Conclusion
The algae and cyanobacteria communities found in the phytotelmata of bromeliads growing in distinct vegetation formations were all very different, with one bromeliad species (Alcantarea nahoumii) being especially distinguished by its high taxonomic richness.We also observed low microalgae species similarity among the bromeliads studied, with only one species (Phacus polytrophos) occurring in all four bromeliad species.
Overall, the following conditions were found to be favorable to high algae and cyanobacteria richness in bromeliad phytotelmata: slightly acidic water at high temperatures, moderate conductivity, large numbers of leaves (with large amounts of water), high dissolved oxygen levels, and high environmental light intensities (Laessle 1961;Sophia et al. 2004;Brouard et al. 2011;Marino et al. 2011), high rainfall (Pires et al. 2017), and available nutrients (Laessle 1961;Marino et al. 2011).
Combined analyses of limnological variables, bromeliad position (shade or sun), and plant morphology are important to understanding the patterns of microalgae communities in phytotelmata environments, and more detailed studies will be needed to increase our knowledge of algae and cyanobacteria distributions and their ecological relationships within that interesting microhabitat.

Figure 2 .
Figure 2. Principal component analysis based on morphological characteristics and abiotic water variables of four bromeliad species (Diam: Diameter, Heig: Height, Leav: number of leaves, T: water temperature, pH, EC: electrical conductivity, TDS: total dissolved solids, DO: dissolved oxygen).

Figure 4 .
Figure 4. Relative species richness of the taxonomic classes of the microalgae communities in four bromeliad species.

Figure. 3 -
Figure. 3 -Venn diagram showing the numbers of microalgae taxa found in the phytotelmata of different bromeliad species in four vegetal formations in Bahia, Brazil.
nahoumii) and Parque das Dunas (H.littoralis) contained greater volumes of water than those at Reserva Sapiranga (H.stellata) and Fazenda Itaberaba (A. cf.lingulatoides) -and the bromeliads in the former two areas showed quite different richnesses.The unevenly distributed water storage by H. littoralis leaves (with large amounts of water accumulated in the central tank and considerably smaller volumes in the lateral tanks) may affect the low species richness in that bromeliad in Parque das Dunas.The leaves forming the central tank are usually long and perpendicular to the ground, which allows less light into the bottom of the central tank; the leaves forming the lateral tanks, however, are held at more open angles, permitting greater light penetration.Additionally, the very acidic waters and lower conductivity and TDS values measured in H. littoralis may depress algal and cyanobacterial richness.

Table 1 .
Rainfall data of the four study areas (by municipality; INMET 2018) between December/2014 and February/2016 and air temperature during the samplings.Values are represented by mean.

Table 2 .
Mean and standard deviation values (n = 80) of the morphological characteristics, abiotic water conditions, and algae richness sampled in four different bromeliads species (Superscript letter represents the Tukey test results: different letters indicate significant differences, p < 0.05).

Table 4 .
Sorensen's similarity index (expressed as %) applied to presence-absence matrix of microalgae species in water tanks of different bromeliad species.