Effects of Nutrient Limitation on Understory Communities in a Tropical Forest

Home , Alsophila erinacea, Austroblechnum, Blotiella, Lindsaeaceae, Stigmatopteris

EFFECTS OF NUTRIENT LIMITATION ON UNDERSTORY COMMUNITIES IN A TROPICAL FOREST

BY JÉSSICA LIRA VIANA

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Plant Biology in the Graduate College of the University of Illinois at Urbana-Champaign, 2020

Urbana, Illinois

Doctoral Committee: Professor James W. Dalling, Chair Associate Professor Anthony Yannarell Associate Professor Jennifer Fraterrigo Associate Professor Wendy Yang

ABSTRACT

Ferns are the second most diverse group of vascular plants whose diversity and abundance

peak in mid-elevation tropical forests. Soil nutrient limitation is an important factor influencing

plant communities and yet little is known about the factors influencing fern floristic composition

and functional traits in montane tropical forests. My dissertation compares composition variation

and decay-rates between terrestrial ferns and understory palms, since the extent to which their resource needs overlap has not been determined. Furthermore, ferns and palms are two essential components of forest understories that differ widely in physiology and reproductive traits, while competing for the same limited resources. I found that soil factors impacted compositional similarity of both ferns and palms, with fern compositional variation related to total soil nitrogen to phosphorus ratio (soil N:P) and light conditions (red:far read – RFR) and palm compositional variation related to bulk density. Distance–decay rates in compositional similarity were slightly higher for palms than ferns. In addition, the abundance of understory palms and tree ferns showed opposite trends with soil N:P and RFR compared to herbaceous ferns. I also analyzed the responses of functional composition and functional dispersion to soil and precipitation gradients, to determine to what degree environmental factors influence trait distribution and diversity of fern

communities across their phylogeny. The examination of functional composition and functional dispersion of fern communities along the soil fertility gradient demonstrated that functional composition was affected by soil variables and dry season rainfall. Leaf traits contributed to most of the variation among all trait variables evaluated. Functional dispersion decreased with soil N:P and manganese (Mn), and increased with dry season soil moisture. Trait and phylogenetic diversity showed clustered patterns. Due to field collections, the fern species list of Fortuna is expanded. I give floristic details of species occurrence and extend the discussion on habitat preference of

herbaceous ferns and tree ferns. Several indicator taxa associated with soil nutrient conditions

related to parent material type were identified. By contrasting compositional and abundance

patterns of similar groups with differing dispersal strategies in highly diverse ecosystems, this

dissertation uncovers how distinctly terrestrial ferns and understory palms respond to soil nutrient

limitation gradient. Incorporating edaphic gradients with the phylogenetic framework of the fern

communities, this research reinforces the importance of eco-evolutionary information to better understand the mechanisms of assembly that maintain diversity in tropical forests.

INDEX WORDS: tropical rainforest, nutrient limitation, environmental gradients, pteridophytes, ferns, understory palms, floristics, composition turnover, functional traits Fortuna Forest Reserve,

Panama

iii

Dedicated

with all my affection to my mother Maria Edileuza Lira,

and sisters Jamily Lira Viana and Carla Jesseany Lira Viana,

to my grandparents Pedro Vieira Lira, Maria Lina Vilar Lira;

in loving memory of my father Carlos José Rodrigues Viana, and my grandparents José da Costa Viana and Maria Rodrigues Viana.

I also dedicate this thesis with all my gratitude to Ernesto Machado.

ACKNOWLEDGEMENTS

Foremost, I would like to express my sincere gratitude to my advisor, Prof. James Dalling,

for his patience, guidance, inspiration, comprehension, enthusiasm, tremendous expertise, and for

the continuous support of my Ph.D. studies and research. Throughout the research and writing

process of this thesis, his advice helped me to thrive and become a skilled researcher. For my Ph.D.

thesis, I could not have imagined finding a better mentor.

In addition to my advisor, I would like to thank the research committee, Prof. Wendy Yang,

Prof. Anthony Yannarell, and Prof. Jennifer Fraterrigo, for their support, informative comments,

and challenging questions.

My sincere thanks go to Dr. Robbin Moran (New York Botanical Garden); Prof. Surangi

Punyasena, Dr. Carol Augspurger and Dr. Astrid Ferrer (Plant Biology Department – University of Illinois); Dr. Ben Turner (Smithsonian Tropical Research Institute); Dr. Alfonso García Piñeres

(Universidad de Costa Rica); Pedro Caballero (Universidad Autónoma de Chiriquí); and the staff from the Herbarium of the Universidad de Panama. Each of them has kindly helped me along my

PhD studies, either teaching me or supporting me, or helped me in a crucial step of my research.

I thank my labmates in the Dalling Lab: Adriana Corrales and Kate Heineman that received

me in the lab with enthusiasm, Jennifer Jones, Cecilia Prada, Zarluis Mijango, and Sierra Brown

for reviewing my talks and for sharing happy moments in the professional and personal life. I wish

to all my labmates, including current and new ones (Manuel Flores and Taylor McClellan), a

successful academic experience during their studies.

This research could have been more difficult if not for the help from Carlos Espinosa and

Evidelio Garcia during fieldwork. Thank you for walking with me countless times in muddy, steep

and treacherous conditions in Fortuna, only because I saw a beautiful fern. v

I feel extremely grateful for the funding I received from Brazil’s agency National Council for Scientific and Technological Development (CNPq), Center for Latin American and Caribbean

Studies, Torrey Botanical Society, Smithsonian Tropical Research Institute, Govindjee and Rajni

Govindjee, the School of Integrative Biology and the Plant Biology Department of the University of Illinois at Urbana-Champaign.

I appreciate the initial help from Dr. Maria Regina Barbosa (Universidade Federal da

Paraíba), my first postgraduate mentor and Dr. Wayt Thomas (New York Botanical Garden) for the support, and both for encouraging me to pursue this path.

I recognize the immeasurable value of all my friends Ingrid Romero, Dessireé Zerpa,

Georgia Seyfried and Cecilia Prada

I thank Ernesto Machado, for endless encouragement, patience and emotional support.

I have immense gratitude for my family, for allowing me time away to pursue my career.

TABLE OF CONTENTS

CHAPTER 1: AN OVERVIEW OF FERNS ...... 1

CHAPTER 2: COMPOSITIONAL VARIATION IN UNDERSTORY FERN AND PALM COMMUNITIES ALONG A SOIL FERTILITY AND RAINFALL GRADIENT IN A LOWER MONTANE TROPICAL FOREST ...... 22

CHAPTER 3: DIVERSITY AND ECOLOGY OF THE FERNS IN FORTUNA ...... 58

CHAPTER 4: TRAIT DISPERSION AND PHYLOGENETIC RELATEDNESS IN A GUILD OF TROPICAL TERRESTRIAL FERNS ...... 93

CHAPTER 5: CONCLUSIONS ...... 133

APPENDIX A: SUPPLEMENTARY MATERIAL FOR CHAPTER 2 ...... 138

APPENDIX B: SUPPLEMENTARY MATERIAL FOR CHAPTER 3 ...... 146

APPENDIX C: SUPPLEMENTARY MATERIAL FOR CHAPTER 4 ...... 153

PLATES ...... 164

vii

CHAPTER 1. AN OVERVIEW OF FERNS

Prominent ecological and evolutionary aspects among pteridophytes

Ferns and lycophytes, also known collectively as pteridophytes, are distinguished from other vascular plants by the absence of seeds, flowers and fruits. In addition, the identity of pteridophytes is also linked to their life cycle, in which the gametophyte is free from sporophyte regulation. Although sharing these features, lycophytes and ferns are two evolutionary lineages of spore bearing vascular plants organized in two distinct classes, Lycopodiopsida and

Polypodiopsida, respectively. Together, the two classes comprise approximately 12,000 species widely spread across the globe (PPG I, 2016). Richness and abundance are particularly high in tropical, subtropical, and southern temperate forested ecosystems where lycophytes and ferns coexist, whereas in northern temperate forests, terrestrial ferns from basal groups seem to be scarce

(Kessler et al., 2016; Moran, 2008).

Pteridophytes were a pervasive component of the terrestrial flora before the angiosperm diversification during the Cretaceous period, which is thought to be associated with a drop in pteridophyte diversity (Schneider et al., 2004). Although most of the extant ferns are restricted to herbaceous forms, fossil records indicate that pteridophytes were once comprised of a broad range of diverse life forms, including rhizomatous herbs, epiphytic herbs, shrubs, vines and trees of diverse growth architectures (Rothwell, 1996). Climate changes occurring between the Paleozoic and Mesozoic periods were likely to have caused a substantial decrease of specialized groups like arborescent forms (Scott and Galtier, 1985). The decline of pteridophyte diversity combined with the angiosperm diversification in the Cretaceous, subsequentially propelled herbaceous forms to diversify and generate the most derived clade of pteridophytes, polypod ferns, which comprised about of 80% of living fern species (Schneider et al., 2004).

Even though most ferns and angiosperms underwent diversification almost in parallel, the selective effects of evolution may have differed between these two plant groups given the distinctive reproductive apparatus in both lineages (Rathinasabapathi, 2006). While evolutionary pressure in angiosperms may have targeted specific traits associated with flowering, pollination and dispersal, innovative traits may have evolved in ferns to deal with new environmental stressors that were absent before the appearing of angiosperms. For example, Feild et al. (2004) suggests that the dominance of fast-growing angiosperms promoted a more competitive environment for ferns. In addition, the poor light availability in the forest understory, conditioned by a dense

canopy, posed a new challenge for fern species. The rise of an innovative photoreceptor in

polypods is thought to be an explicit response for the shaded understory conditions provided by

forested environments. The new Neochrome 1, formally phytochrome 3, discovered in the

gametophyte cells of Adiantum capillus-veneris L. species, was revealed to be responsible for the phototropism response under red light conditions (Kawai et al., 2003).

Despite of the emergence of an important innovation to adapt to the dimness of understories, the relationship of ancient fern assemblages with the environment, especially volcanic areas, has been suggested as important to explain evolution and environmental relationships in ferns (Scott and Galtier, 1985). For example, in a study conducted in Hawaii,

Amatangelo and Vitousek (2008) analyzed leaf stoichiometry of ferns species belonging to basal groups (non-polypods) and derived groups (polypods) across a natural gradient of nitrogen (N)

and phosphorus (P) availability ranging from a young (0.3 K years) N and P limited site relatively

high in cations (1 site); to more developed (20 and 150 K years) N and P rich, low cation sites (2

sites), and old sites (4100 K years) moderately rich in N and P and low in cations. They found that

although foliar nutrient of both non-polypods and polypods changed significantly with soil P and

macronutrients, non-polypod ferns had the highest leaf N at the lowest N site and the highest leaf

P at the lowest P site. On the other hand, polypod ferns had significantly higher foliar N and P than dicots and highest foliar K across all sites. Non-polypod ferns had lowest foliar Ca and Mg across

all sites. Investigating the ecological traits, such as leaf stoichiometry, of non-polypods and polypods ferns and their phylogenetic correlates can improve the understanding of the ecological role of these plant groups. In addition, leaf stoichiometry of ferns, in particular, might influence nutrient cycling in terrestrial ecosystems, especially where ferns are diverse and abundant

(Amatangelo and Vitousek, 2009; Funk and Amatangelo, 2013).

Ferns as a model to study biological processes

The greenness that once prevailed on the flowerless terrestrial Earth persists today, part of it as extent ferns. Ferns endured climatic changes and mass extinctions and are of considerable importance for evolutionary, taxonomic, genetic, physiological, ecological and environmental studies.

The sequencing and study of the genome of fern species has propelled researchers towards a better understanding of genome evolution across land plants (Sessa et al., 2014). However, it was not until 2018 that Azolla filiculoides Lam. and Salvinia cucullata Roxb. (Salviniales) had their nuclear genomes sequenced (Li et al., 2018), almost 20 years after the first angiosperm plant,

Arabidopsis thaliana (L.) Heynh. (The Arabidopsis Genome Initiative, 2000). Azolla filiculoides is an aquatic fern that was the one first candidate taxa to be considered for genome sequencing because it is economically important for the production of rice in Southeast Asia, due to the symbiotic relationship with the nitrogen fixer cyanobacterium Nostoc azollae (Perkins and Peters,

1993). Although the sequencing of the first fern species has been achieved (Li et al., 2018), broader comparative analyses of genome evolution are still missing. These comparative analyses could

reveal whether important genes for land plants share the same location in chromosomes across

embryophytes. In addition, genomic studies in ferns could enable the inference of the origin of

specific genes linked to insect resistance and symbiosis. For example, Li et al. (2018) suggest that

the scTM12 protein present in the Salvinia genome - a chitin-binding protein that confer insect

resistance – could be a product from bacteria or other microorganisms that horizontally transferred

the gene to ferns. Furthermore, it was found that genes associated with arbuscular mycorrhizal and

root nodule symbiosis are absent in Azolla and Salvinia, which might suggest that most of the common symbiosis pathways have been lost in both lineages. Thus, investigating the genomes for fern species might reveal beneficial biotechnological information to improve crops worldwide, and help answer essential questions about the evolution of land plants.

Earlier studies focusing on cytological studies have produced considerable data on chromosome attributes of living ferns (Britton, 1974; Mickel et al., 1966; Sorsa, 1968; Tryon et al., 1975). These studies have revealed that ferns belonging to homosporous lineages have on average three-fold more chromosomes than flowering plants (ferns n = 57.05; angiosperms n =

15.99) (Klekowski and Baker, 1966). Actually, Ophioglossum reticulatum L., a small (< 30 cm height) herbaceous fern is ranked as the organism with the highest chromosome number reported

(2n = 1440) (Khandelwal, 1990). In addition to chromosome numbers, the genomes of extant ferns are exceptionally large (Bennett and Leitch, 2001), even after accounting for chromosome doubling events (polyploidy) (Bainard et al., 2011). Nakazato et al. (2008) suggest that among land plants, homosporous ferns are the single lineage to exhibit positive correlation between chromosome number and DNA amount. Heterosporous ferns (distinct male and female spores – i.e. Lycophyte and Selaginella), on the contrary, exhibit chromosome numbers similar of those found in angiosperms, with a mean of 13.6. However, the disparity in the number of chromosomes

between homosporous and heterosporous (lycophytes and angiosperms) lineages of land plants is

yet to be explained (Barker and Wolf, 2010). The phylogenetic position of ferns combined with

the extensive morphological and reproductive differences between ferns and spermatophytes

provide opportunities to improve our understanding of plant evolution and to identify ancestral

versus derived developmental patterns (Rensing, 2017).

Another central query in plant sciences concerns leaf evolution and development. Ferns

have recently been considered a plant group of interest because of the phylogenetic circumscription

and high morphological diversity of their leaves (Vasco et al., 2013). These characteristics make

ferns a crucial plant lineage for comparative studies on the evolution of leaves in vascular plants.

Although there is molecular data that provides support for the homology between microphylls

(containing a single unbranched vein) and megaphylls including angiosperms and lycophytes

(sensu Shubin et al., 2009), how leaves evolved among euphyllophytes (Polypodiopsida and seed

plants) is still an open debate. Vasco et al. (2016) employed a comparative developmental approach

and phylogenetic analyses to study how the C3HDZs genes are expressed among ferns. C3HDZs

genes are responsible for regulating both vascular and meristem development, and for specifying adaxial leaf identity in angiosperms. Key results suggest that C3HDZs genes were duplicated and underwent neofunctionalization in the euphyllophyte ancestor, and that both ferns and seed plants share similar leaf developmental mechanisms. Across lycophytes and ferns, the expression of

C3HDZs genes is conserved, with a role that is related to the initiation of lateral primordia (leaves or sporangia). However, the expression of C3HDZz genes in the sporangia in lycophytes is maintained, whereas in ferns, an ancestral role in the growth of sporangium is suggested. Vasco et al. (2016) provide molecular support for the hypothesis that leaves in lycophytes derived from the sporangium, a change in the paradigm of lycophyte leaf evolution theories.

The Impact of ferns on the environment

In many ecosystems, ferns play an important role in influencing community assembly and

ecological processes. In eastern North American temperate forests, for example, ferns are able to

suppress the emergence of some canopy tree species by altering the light conditions and litter in

the understory (George and Bazzaz, 1999a). By negatively affecting the establishment of tree species, fern cover may change patterns of species composition, density and seedling spatial distribution (George and Bazzaz, 1999a). Furthermore, experiments manipulating fern understory cover in the same ecosystem showed that, when a fern understory was present, the growth and survival of most of the seedling species tested were reduced. Among the seedlings studied, only

Quercus could emerge from low light conditions created by the fern understory cover (George and

Bazzaz, 1999b). Therefore, the presence or absence of fern foliage can alter the

microenvironmental conditions for seeds and seeding survival, leading to a decrease in seedling

densities.

Tropical rainforest canopies are densely occupied by epiphytes, especially by fern species

in mid-elevation rainforests (Kluge and Kessler, 2011; Watkins and Cardelús, 2012). The bird’s

nest fern, Asplenium nidus L., is the dominant species among canopy epiphytes in Malaysian

tropical rainforests, and can reach very high densities (~180 per hectare) distributed across the full

canopy height (Fayle et al., 2009). Asplenium nidus appear to function as a critical key stone species providing a suitable microhabitat for a number small vertebrate (Scheffers et al., 2014) and invertebrate groups (Ellwood and Foster, 2004; Karasawa and Hijii, 2006). This is because bird's nest ferns are able to significantly reduce the daily temperature variation in comparison to canopy habitats where the fern is absent (Turner and Foster, 2006). The potential ability of epiphyte ferns to buffer microclimate variation, and the capacity to accumulate limiting resources like water and

litter in the canopy environment, turn them into remarkable “islands of biodiversity in the

rainforest canopy” (Fayle et al., 2008).

The impact of ferns on the environment has been studied for the capacity of certain species to remove contaminants from the ambience through accumulation, chelation, and detoxification mechanisms. Traits suitable for model candidates to be used in phytoremediation (the use of plants to eliminate pollutants) include fast growth, easy reproduction or propagation, and the ability to associate with nitrogen-fixing microorganisms (Drăghiceanu et al., 2014; Negri and Hinchman,

1996). Azolla and Salvinia are both aquatic fern species which have shown potential for removing heavy metals and organic compounds from the environment. Benaroya et al. (2004) detected lead precipitates in the vacuoles of the aquatic fern A. filiculoides when growing in 20 mg of lead(II) medium. Lead content in the whole plant dry weight was composed of 2.3% after four days of growth. Furthermore, there was a 100% increase in the leaf element content of phosphorus, sulfur, sodium and calcium, and 40 % decrease in magnesium and chloride compared to control plants.

Similarly, Olguín et al. (2002) found that Salvinia minima Baker accumulated lead, cadmium and chromium from liquid solutions, but higher or lower levels of these metals were depended on light conditions and pH levels. Among terrestrial fern species, Pteris and Pityrogramma (both

Pteridaceae) are the best-used species for removing several types heavy metals from the environment (Francesconi et al., 2002; Xie et al., 2009).

Environmental controls on ferns in tropical forests

Large scale patterns of fern diversity follow the prominent latitudinal diversity gradient pattern, with the highest diversity associated with tropical mid-elevation forests and oceanic island habitats (Moran, 2008). Mountains with elevations of 800-2000 m maintain more species and higher endemism (Kluge and Kessler, 2006) and differ in species composition from lowland

forests (Jones et al. 2011). The potential explanations for this patterning, from mid-domain effects or historical and geoclimatic factors, remain in dispute (Hawkins and Diniz-Fihlo, 2002; Watkins et al., 2006).

Independent of elevation gradients, the occurrence of fern species and the structure of their communities have long been associated with variation in nutrient availability (e.g. Tuomisto and

Ruokolainen, 1993; Young and León, 1989). Evidence for positive correlations between floristic patterns of ferns and soil attributes over landscape or regional scales have been accumulated, especially in western and central lowland Amazonian sites where soils are characterized by a gradient from nutrient-poor sandy soils to relatively fertile clay soils and minimal climatic variation (Costa et al., 2005; Poulsen et al., 2006; Ruokolainen et al., 1997; Tuomisto et al., 2002,

2003, 2014; Tuomisto and Poulsen, 1996; Tuomisto and Ruokolainen, 2005; Vormisto et al., 2000;

Zuquim et al., 2014, 2007). In lowland rainforests, ferns exhibit composition patterns that are similar to other plant groups despite their differences in dispersal strategies and phylogenetic lineage (Costa et al., 2005; Tuomisto et al., 2002; Vormisto et al., 2000). However, the responses

of terrestrial fern abundance to soil nutrient gradients have not yet been compared with other plant

groups. Furthermore, in comparison to lowland tropical rainforests, studies focusing on floristic

composition turnover in montane tropical forests are scarce (but see Jones et al., 2011).

Environmental factors such as soil chemistry, drainage and topographic position are determinants of local fern floristic variation, as well as other plant taxa in lowland forests, but the key environmental influences in montane rainforests, and their relative value, may be different due to, among many other factors, distinct geological processes of soil formation (Coates and Obando,

1996; Higgins et al., 2011), topography, and differing patterns of nutrient availability (Vitousek,

1984). Measuring the relative importance of stochastic and deterministic processes in structuring

8 terrestrial fern communities, and contrasting their occurrence patterns with other understory plant groups provides an opportunity to address how floristic composition and abundance of major understory plant groups relate with environmental factors in diverse ecosystems such as in lower montane neotropical rainforests.

Functional traits can either influence or respond to environmental changes (Nock et al.,

2016). The responses of species to environmental gradients, as well as the changes in plant strategies have shown to be mediated by functional traits (Diaz et al., 1998; Westoby et al., 2002).

In addition, trait and phylogenetic patterns can provide insights into ecological processes structuring communities (Webb, 2000). One study on functional traits of fern species along an elevational gradient in Costa Rica suggests that fern communities under mid-elevation tend to be phylogenetically overdispersed as a consequence of interspecific competition, whereas communities in high elevations tend to be phylogenetic clustered due to environmental filtering

(Kluge and Kessler, 2011). Sessa et al. (2018) analyzed the phylogenetic and functional diversity of fern community assembly at two spatial scales and demonstrated that environmental filtering played a larger role at the smaller scale. Other studies linking phylogenetic and functional composition to environmental conditions only found limited phylogenetic signal (Carvajal-

Hernández et al., 2018; Tanaka and Sato, 2015). As an important component of warm temperate rainforests from New Zealand, tree ferns have also been included in the study of functional traits of canopy tree across soil fertility gradients (Jager et al., 2015). However, most fern species are herbaceous, and their functional composition and phylogenetic diversity have not been fully explored, and less so along gradients of nutrient availability.

Overview

For my doctoral dissertation, I explore the effects of nutrient limitation on understory

communities in a tropical rainforest in western Panama. My research was conducted at the Fortuna

Forest Reserve and neighboring Palo Seco Protected Forest with altitudes that range from 700 to

1,500 meters and comprise a well-preserved primary forest ecosystem with exceptionally high

biodiversity. Soil and precipitation data have previously been determined in 12 one-hectare

permanent plots across the forests by earlier studies (Andersen et al., 2010; Prada et al., 2017).

Soils in Fortuna and Palo Seco forests are characterized by extreme contrasts in nutrient availability associated with the underlying parent material, which allows for the study of the effects of nutrient limitation on ecological communities. There is also variation in the amount and seasonality in rainfall and cloud cover across the plots, which range from aseasonal conditions in the Caribbean sites, to occasional dry periods (monthly rainfall <100 mm) in sites surrounding

Lake Fortuna, and a longer dry season on the plots on the Pacific slope sites.

For my second chapter, I focused on a comparison between understory palms and terrestrial ferns, which substantially differ in their dispersal ability and physiology despite occupying the same forest stratum. I analyzed the compositional patterns of understory palm and terrestrial fern communities along a gradient of nutrient availability. In addition, I tested whether the observed abundance patterns of these two plant groups are correlated to specific soil and light variables such as soil N:P, R:FR ratio of irradiance, and soil bulk density. I used ordination analyses and generalized linear regression models to answer the following questions: (a) Do environmental variables influence fern and palm distributions in a similar manner across a gradient in soil nutrient availability? (b) Does larger propagule size and lower reproductive output in palms result in greater spatial aggregation of populations and, consequently, greater distance–decay rates in community

10 similarity in palms than ferns? The results and conclusions of this study are published in the

Journal of Vegetation Science, and is coauthored by Dr. Benjamin Turner, and Dr. James Dalling.

For my third chapter, I describe the components of the lycophyte and fern flora of the

Fortuna Forest Reserve and neighboring Palo Seco Protected Forest combining an existing list of species and field collections. I also addressed the relationships of species with environmental variables, highlighting the habitat associations of herbaceous and tree ferns. Furthermore, I identified indicator species in each of the distinct communities based on soil, precipitation and parent material characteristics. This study is a chapter coauthored with Dr. James Dalling in the book “Fortuna Forest Reserve, Panama: Interacting Effects of Climate and Soils on the Biota of a Wet Premontane Tropical Forest” published as Smithsonian Contributions to Botany volume

112.

For my fourth chapter, I combined trait and phylogenetic data to analyze trait-dependency in species occurrence patterns in relation to environmental conditions, and to infer assembly mechanisms of terrestrial fern communities. I employed RLQ and fourth-corner analyses to characterize and test trait-environment relationships, and beta regression models to analyze functional dispersion responses to soil and precipitation factors. Additionally, I examined trait and phylogenetic diversity patterns and how traits of fern species are distributed in relation to the phylogeny. My hypothesis was that nutrient availability in our study site acts as a strong environmental filter, structuring traits and phylogenetic diversity in fern communities. This study will be submitted to the journal “Functional Ecology”, with Dr. Benjamin Turner, and Dr. James

Dalling as coauthors.

In the final chapter, I summarize the main results from each of the above-mentioned chapters and discuss the future directions for fern research.

REFERENCES

Amatangelo, K. L., and Vitousek, P. M. (2009). Contrasting predictors of fern versus angiosperm

decomposition in a common garden. Biotropica, 41(2), 154–161.

https://doi.org/10.1111/j.1744-7429.2008.00470.x

Andersen, K. M., Turner, B. L., and Dalling, J. W. (2010). Soil-based habitat partitioning in

understorey palms in lower montane tropical forests. Journal of Biogeography, 37(2), 278–

292. https://doi.org/10.1111/j.1365-2699.2009.02192.x

Bainard, J. D., Henry, T. A., Bainard, L. D., and Newmaster, S. G. (2011). DNA content

variation in monilophytes and lycophytes: Large genomes that are not endopolyploid.

Chromosome Research, 19(6), 763–775. https://doi.org/10.1007/s10577-011-9228-1

Barker, M. S., and Wolf, P. G. (2010). Unfurling Fern Biology in the Genomics Age.

BioScience, 60(3), 177185. https://doi.org/10.1525/bio.2010.60.3.4

Benaroya, R., Tzin, V., Tel-Or, E., and Zamski, E. (2004). Lead accumulation in the aquatic fern

Azolla filiculoides. Plant Physiology and Biochemistry, 42(7–8), 639–645.

https://doi.org/10.1016/j.plaphy.2004.03.010

Bennett, M. D., and Leitch, I. J. (2001). Nuclear DNA amounts in pteridophytes. Annals of

Botany, 87(3), 335–345. https://doi.org/10.1006/anbo.2000.1339

Britton, D. M. (1974). The Significance of Chromosome Numbers in Ferns Author(s): Annals of

the Missouri Botanical Garden, 61(2), 310–317. http://www.jstor.org/stable/2395059

Carvajal-Hernández, C. I., Gómez-Díaz, J. A., Kessler, M., and Krömer, T. (2018). Influence of

elevation and habitat disturbance on the functional diversity of ferns and lycophytes. Plant

Ecology and Diversity, 11(3), 335–347. https://doi.org/10.1080/17550874.2018.1484526

Coates, A. G., and Obando, J. A. (1996). Geological evolution of the Central American Isthmus.

In J. B. C. Jackson, A. F. Budd, and A. G. Coates (Eds.), Evolution and Environment in

Tropical America (pp. 21–25). University of Chicago Press.

Costa, F. R. C., Magnusson, W. E., and Luizao, R. C. (2005). Mesoscale distribution patterns of

Amazonian understorey herbs in relation to topography, soil and watersheds. Journal of

Ecology, 93, 863–878. https://doi.org/10.1111/j.1365-2745.2005.01020.x

Diaz, S., Cabido, M., and Casanoves, F. (1998). Plant functional traits and environmental filters

at a regional scale. Journal of Vegetation Science, 9(1), 113–122.

https://doi.org/10.2307/3237229

Drăghiceanu, O. A., Dobrescu, C. M., and Soare, L. C. (2014). Applications of pteridophytes in

phytoremediation. Current Trends in Natural Sciences, 3(6), 68–73.

Ellwood, M. D. F., and Foster, W. A. (2004). Doubling the estimate of invertebrate biomass in a

rainforest canopy. Nature, 429(6991), 549–551. https://doi.org/10.1038/nature02560

Fayle, T. M., Chung, A. Y. C., Dumbrell, A. J., Eggleton, P., and Foster, W. A. (2009). The

Effect of Rain Forest Canopy Architecture on the Distribution of Epiphytic Ferns

(Asplenium spp.) in Sabah, Malaysia. Biotropica, 41(6), 676–681.

https://doi.org/10.1111/j.1744-7429.2009.00523.x

Fayle, T. M., Ellwood, M. F., Turner, E. C., Snaddon, J. L., Yusah, K. M., and Foster, W. A.

(2008). Bird’s nest ferns: islands of biodiversity in the rainforest canopy. Antenna, 32, 34–

37. http://www.royensoc.co.uk/publications/Antenna.htm

Feild, T. S., Arens, N. C., Doyle, J. A., Dawson, T. E., and Donoghue, M. J. (2004). Dark and

disturbed: a new image of early angiosperm ecology. Paleobiology, 30(1), 82–107.

https://doi.org/10.1666/0094-8373(2004)030<0082:dadani>2.0.co;2

Francesconi, K., Visoottiviseth, P., Sridokchan, W., and Goessler, W. (2002). Arsenic species in

an arsenic hyperaccumulating fern, Pityrogramma calomelanos: A potential

phytoremediator of arsenic-contaminated soils. Science of the Total Environment, 284(1–3),

27–35. https://doi.org/10.1016/S0048-9697(01)00854-3

Funk, J. L., and Amatangelo, K. L. (2013). Physiological mechanisms drive differing foliar

calcium content in ferns and angiosperms. Oecologia, 173(1), 23–32.

https://doi.org/10.1007/s00442-013-2591-1

George, L. O., and Bazzaz, F. A. (1999a). The Fern Understory as an Ecological Filter :

Emergence and Establishment of Canopy-Tree Seedlings. Ecology, 80(3), 833–845.

https://doi.org/10.1890/0012-9658(1999)080[0833:TFUAAE]2.0.CO;2

George, L. O., and Bazzaz, F. A. (1999b). The fern understory as an ecological filter: Growth

and survival of canopy-tree seedlings. Ecology, 80(3), 846–856.

https://doi.org/10.1890/0012-9658(1999)080[0846:TFUAAE]2.0.CO;2

Hawkins, B. A., and Diniz-Fihlo, J. A. F. (2002). The mid-domain effect cannot explain the

diversity gradient of Nearctic birds. Global Ecology and Biogeography, 11, 419–426.

Higgins, M. A., Ruokolainen, K., Tuomisto, H., Llerena, N., Cardenas, G., Phillips, O. L.,

Vásquez, R., and Räsänen, M. (2011). Geological control of floristic composition in

Amazonian forests. Journal of Biogeography, 38(11), 2136–2149.

https://doi.org/10.1111/j.1365-2699.2011.02585.x

Jager, M. M., Richardson, S. J., Bellingham, P. J., Clearwater, M. J., and Laughlin, D. C. (2015).

Soil fertility induces coordinated responses of multiple independent functional traits.

Journal of Ecology, 103(2), 374–385. https://doi.org/10.1111/1365-2745.12366

Jones, M. M., Szyska, B., and Kessler, M. (2011). Microhabitat partitioning promotes plant

diversity in a tropical montane forest. Global Ecology and Biogeography, 20, 558–569.

https://doi.org/10.1111/j.1466-8238.2010.00627.x

Karasawa, S., and Hijii, N. (2006). Does the existence of bird’s nest ferns enhance the diversity

of oribatid (Acari: Oribatida) communities in a subtropical forest? Biodiversity and

Conservation, 15(14), 4533–4553. https://doi.org/10.1007/s10531-005-5443-4

Kawai, H., Kanegae, T., Christensen, S., Kiyosue, T., Sato, Y., Imaizumi, T., Kadota, A., and

Wada, M. (2003). Responses of ferns to red light are mediated by an unconventional

photoreceptor. Nature, 421(6920), 287–290. https://doi.org/10.1038/nature01310

Kessler, M., Karger, D. N., and Kluge, J. (2016). Elevational diversity patterns as an example for

evolutionary and ecological dynamics in ferns and lycophytes. Journal of Systematics and

Evolution, 54(6), 617–625. https://doi.org/10.1111/jse.12218

Khandelwal, S. (1990). Chromosome evolution in the genus Ophioglossum L. Botanical Journal

of the Linnean Society, 102, 205–217. https://doi-org/j.1095-8339.1990.tb01876.x

Klekowski, E., and Baker, H. (1966). Evolutionary Significance of Polyploidy in the

Pteridophyta. Science, 153(3733), 305–307. http://www.jstor.org/stable/1719509

Kluge, J., and Kessler, M. (2006). Fern endemism and its correlates: Contribution from an

elevational transect in Costa Rica. Diversity and Distributions, 12(5), 535–545.

https://doi.org/10.1111/j.1366-9516.2006.00231.x

Kluge, J., and Kessler, M. (2011). Phylogenetic diversity, trait diversity and niches: Species

assembly of ferns along a tropical elevational gradient. Journal of Biogeography, 38(2),

394–405. https://doi.org/10.1111/j.1365-2699.2010.02433.x

Li, F. W., Brouwer, P., Carretero-Paulet, L., Cheng, S., De Vries, J., Delaux, P. M., Eily, A.,

Koppers, N., Kuo, L. Y., Li, Z., Simenc, M., Small, I., Wafula, E., Angarita, S., Barker, M.

S., Bräutigam, A., Depamphilis, C., Gould, S., Hosmani, P. S., … Pryer, K. M. (2018). Fern

genomes elucidate land plant evolution and cyanobacterial symbioses. Nature Plants, 4(7).

https://doi.org/10.1038/s41477-018-0188-8

Mickel, J. T., Wagner, W. H., and Chen, K. L. (1966). Chromosome Observations on the Ferns

of Mexico. Caryologia, 19(1), 95–102. https://doi.org/10.1080/00087114.1966.10796208

Moran, R. C. (2008). Diversity, Biogeography, and floristics. In T. A. Ranker and C. H. Haufler

(Eds.), Biology and Evolution of Ferns and Lycophytes (pp. 367–394). Cambridge

University Press. https://doi.org/10.1017/CBO9780511541827.015

Nakazato, T., Barker, M., Rieseberg, L., and Gastony, G. (2008). Evolution of the nuclear

genome of ferns and lycophytes. In T. Ranker and C. Haufler (Eds.), Biology and Evolution

of Ferns and Lycophytes. Cambridge University Press.

Negri, M. C., and Hinchman, R. R. (1996). Plants that remove contaminants from the

environment. Laboratory Medicine, 27(1), 36–40. https://doi.org/10.1093/labmed/27.1.36

Nock, C. A., Vogt, R. J., and Beisner, B. E. (2016). Functional Traits. In eLS. John Wiley &

Sons, Ltd: Chichester. (pp. 1–8). https://doi.org/10.1002/9780470015902.a0026282

Olguín, E. J., Hernández, E., and Ramos, I. (2002). The effect of both different light conditions

and the pH value on the capacity of Salvinia minima Baker for removing cadmium, lead and

chromium. Acta Biotechnologica, 22(1–2), 121–131. https://doi.org/10.1002/1521-

3846(200205)22:1/2<121::AID-ABIO121>3.0.CO;2-F

Perkins, S. K., and Peters, G. A. (1993). The Azolla-Anabaena symbiosis: endophyte continuity

in the Azolla life-cycle is facilitated by epidermal trichomes. New Phytologist, 123(1), 53–

64. https://doi.org/10.1111/j.1469-8137.1993.tb04531.x

Poulsen, A. D., Tuomisto, H., and Balslev, H. (2006). Edaphic and floristic variation within a 1-

ha plot of lowland Amazonian rain forest. Biotropica, 38(4), 468–478.

https://doi.org/10.1111/j.1744-7429.2006.00168.x

PPG I. (2016). A community-derived classification for extant lycophytes and ferns. Journal of

Systematics and Evolution, 54(6), 563–603. https://doi.org/10.1111/jse.12229

Prada, C. M., Morris, A., Andersen, K. M., Turner, B. L., Caballero, P., and Dalling, J. W.

(2017). Soils and rainfall drive landscape-scale changes in the diversity and functional

composition of tree communities in premontane tropical forest. Journal of Vegetation

Science, 28(4), 859–870. https://doi.org/10.1111/jvs.12540

Rathinasabapathi, B. (2006). Ferns represent an untapped biodiversity for improving crops for

environmental stress tolerance. New Phytologist, 172(3), 385–390.

https://doi.org/10.1111/j.1469-8137.2006.01889.x

Rensing, S. A. (2017). Why we need more non-seed plant models. New Phytologist, 216(2),

355–360. https://doi.org/10.1111/nph.14464

Rothwell, G. W. (1996). Pteridophytic evolution: An often underappreciated phytological

success story. Review of Palaeobotany and Palynology, 90(3–4), 209–222.

https://doi.org/10.1016/0034-6667(95)00084-4

Ruokolainen, K., Linna, A., and Tuomisto, H. (1997). Use of Melastomataceae and

pteridophytes for revealing phytogeographical patterns in Amazonian rain forests. Journal

of Tropical Ecology, 13(2), 243–256. https://doi.org/10.1017/S0266467400010439

Scheffers, B. R., Phillips, B. L., and Shoo, L. P. (2014). Asplenium bird’s nest ferns in rainforest

canopies are climate-contingent refuges for frogs. Global Ecology and Conservation, 2, 37–

46. https://doi.org/10.1016/j.gecco.2014.06.004

Schneider, H., Schuettpelz, E., Pryer, K. M., Cranfill, R., Magallón, S., and Lupia, R. (2004).

Ferns diversified in the shadow of angiosperms. Nature, 428(6982), 553–557.

https://doi.org/10.1038/nature02361

Scott, A. C., and Galtier, J. (1985). Distribution and ecology of early ferns. Proceedings of the

Royal Society of Edinburgh. Section B. Biological Sciences, 86(B), 141–149.

https://doi.org/10.1017/S026972700000806X

Sessa, E. B., Banks, J. A., Barker, M. S., Der, J. P., Duffy, A. M., Graham, S. W., Hasebe, M.,

Langdale, J., Li, F. W., Marchant, D. B., Pryer, K. M., Rothfels, C. J., Roux, S. J., Salmi,

M. L., Sigel, E. M., Soltis, D. E., Soltis, P. S., Stevenson, D. W., and Wolf, P. G. (2014).

Between Two Fern Genomes. GigaScience, 3(1), 1–7. https://doi.org/10.1186/2047-217X-

3-15

Sessa, E. B., Chambers, S. M., Li, D., Trotta, L., Endara, L., Burleigh, J. G., and Baiser, B.

(2018). Community assembly of the ferns of Florida. American Journal of Botany, 105(3),

549–564. https://doi.org/10.1002/ajb2.1073

Shubin, N., Tabin, C., and Carroll, S. (2009). Deep homology and the origins of evolutionary

novelty. Nature, 457, 818–823. https://doi.org/10.1038/nature07891

Sorsa, V. (1968). Chromosome Studies on Puerto Rican Ferns (Gleicheniaceae). Caryologia,

21(2), 97–103. https://doi.org/10.1080/00087114.1968.10796287

Tanaka, T., and Sato, T. (2015). Taxonomic, phylogenetic and functional diversities of ferns and

lycophytes along an elevational gradient depend on taxonomic scales. Plant Ecology,

216(12), 1597–1609. https://doi.org/10.1007/s11258-015-0543-z

The Arabidopsis Genome Initiative. (2000). Analysis of the genome sequence of the flowering

plant Arabidopsis thaliana. Nature, 408, 796–815.

https://doi.org/10.1134/S1022795411020074

Tryon, A. F., Bautista, H. P., and Araújo, I. D. S. (1975). Chromosome studies of Brazilian ferns.

Acta Amazonica, 5(1), 35–43. https://doi.org/10.1590/1809-43921975051035

Tuomisto, H., and Poulsen, A. D. (1996). Influence of edaphic specialization on pteridophyte

distribution in neotropical rain forests. Journal of Biogeography, 23(3), 283–293.

http://www.jstor.org/stable/2845846

Tuomisto, H., and Ruokolainen, K. (1993). Distribution of Pteridophyta and Melastomataceae

along an edaphic gradient in an Amazonian rain forest. Journal of Vegetation Science, 4(1),

25–34. https://doi.org/10.2307/3235634

Tuomisto, H., and Ruokolainen, K. (2005). Environmental heterogeneity and the diversity of

pteridophytes and Melastomataceae in westen Amazonia. Biologist Skrifter, 55, 37–56.

Tuomisto, H., Ruokolainen, K., Aguilar, M., and Sarmiento, A. (2003). Floristic patterns along a

43-km long transect in an Amazonian rain forest. Journal of Ecology, 91(5), 743–756.

https://doi.org/10.1046/j.1365-2745.2003.00802.x

Tuomisto, H., Ruokolainen, K., Poulsen, A. D., Moran, R. C., Quintana, C., Cañas, G., and Celi,

J. (2002). Distribution and diversity of pteridophytes and Melastomataceae along edaphic

gradients in Yasuni National Park, Ecuadorian Amazonia. Biotropica, 34(4), 516–533.

Tuomisto, H., Zuquim, G., and Cárdenas, G. (2014). Species richness and diversity along

edaphic and climatic gradients in Amazonia. Ecography, 37, 1034–1046.

https://doi.org/10.1111/ecog.00770

Turner, E., and Foster, W. A. (2006). Assessing the Influence of Bird’s Nest Ferns (Asplenium

spp.) on the Local Microclimate across a Range of Habitat Disturbances in Sabah, Malaysia.

Selbayana, 27(2), 195–200. https://www.jstor.org/stable/41760282

Vasco, A., Moran, R. C., and Ambrose, B. a. (2013). The evolution, morphology, and

development of fern leaves. Frontiers in Plant Science, 4, 1–16.

https://doi.org/10.3389/fpls.2013.00345

Vasco, A., Smalls, T. L., Graham, S. W., Cooper, E. D., Wong, G. K.-S., Stevenson, D. W.,

Moran, R. C., and Ambrose, B. A. (2016). Challenging the paradigms of leaf evolution:

Class III HD-Zips in ferns and lycophytes. New Phytologist, 212(3), 745–758.

https://doi.org/10.1111/nph.14075

Vitousek, P. M. (1984). Litterfall , Nutrient Cycling , and Nutrient Limitation in Tropical

Forests. Ecology, 65(1), 285–298. https://doi.org/10.2307/1939481

Vormisto, J., Phillips, O. L., Ruokolainen, K., Tuomisto, H., and Vasquez, R. (2000). A

comparison of fine-scale distribution patterns of four plant groups in an Amazonian

rainforest. Ecography, 23(3), 349–359. https://doi.org/10.1111/j.1600-0587.2000.tb00291.x

Watkins, J. E., Cardelús, C., Colwell, R. K., and Moran, R. C. (2006). Species richness and

distribution of ferns along an elevational gradient in Costa Rica. American Journal of

Botany, 93(1), 73–83. https://doi.org/10.3732/ajb.93.1.73

Watkins, J. E., and Cardelús, C. L. (2012). Ferns in an angiosperm world: Cretaceous radiation

into the epiphytic niche and diversification on the forest floor. International Journal of

Plant Sciences, 173(6), 695–710. https://doi.org/10.1086/665974

Webb, C. O. (2000). Exploring the Phylogenetic Structure of Ecological Communities: An

Example for Rain Forest Trees. The American Naturalist, 156(2), 145–155.

https://doi.org/10.1086/303378

Westoby, M., Falster, D. S., Moles, A. T., Vesk, P. A., and Wright, I. J. (2002). Plant ecological

strategies: Some leading dimensions of variation between species. Annual Review of

Ecology and Systematics, 33, 125–159.

https://doi.org/10.1146/annurev.ecolsys.33.010802.150452

Xie, Q. E., Yan, X. L., Liao, X. Y., and Li, X. (2009). The arsenic hyperaccumulator fern Pteris

vittata L. Environmental Science and Technology, 43(22), 8488–8495.

https://doi.org/10.1021/es9014647

Young, K. R., and León, B. (1989). Pteridophyte species diversity in the central Peruvian

Amazon: Importance of edaphic specialization. Brittonia, 41(4), 388–395.

https://doi.org/10.2307/2807552

Zuquim, G., Costa, F. R. C., and Prado, J. (2007). Fatores que determinam a distribuição de

espécies de pteridófitas da Amazônia Central. Revista Brasileira de Biociências, 5(2), 360–

362.

Zuquim, G., Tuomisto, H., Jones, M. M., Prado, J., Figueiredo, F. O. G., Moulatlet, G. M.,

Costa, F. R. C., Quesada, C. A., and Emilio, T. (2014). Predicting environmental gradients

with fern species composition in Brazilian Amazonia. Journal of Vegetation Science, 25(5),

1195–1207. https://doi.org/10.1111/jvs.12174

CHAPTER 2: COMPOSITIONAL VARIATION IN UNDERSTORY FERN AND PALM COMMUNITIES ALONG A SOIL FERTILITY AND RAINFALL GRADIENT IN A LOWER MONTANE TROPICAL FOREST1

ABSTRACT

Ferns and palms are key components of tropical wet forest understories. The distributions

of these groups are influenced by variation in rainfall and soil properties, although the degree to

which their resource requirements overlap remains unclear. Palms and ferns also represent

extremes in a spectrum of reproductive traits, which may impact habitat requirements and

compositional turnover. We asked whether: (a) environmental variables influence fern and palm

distributions in a similar manner across a gradient in soil nutrient availability; and (b) larger

propagule size and lower reproductive output in palms result in greater spatial aggregation of

populations and, consequently, greater distance–decay rates in community similarity in palms than ferns. All ferns and palms were sampled in fifteen 5 m × 5 m subplots in each of 12 one-hectare

plots. The plots span a strong gradient in soil nutrient availability related to parent material. Light,

precipitation and soil variables were measured in each subplot. Species associations were analyzed

using ordination and regression, as well as Procrustes and Mantel tests. Parent material significantly impacted compositional similarity of both ferns and palms, with fern compositional variation related to total soil N:P ratio and red:far red (R:FR), and palms with bulk density (p <

0.05). Palm and fern abundances were not correlated; however, tree fern abundances showed an

opposing abundance relationship with total soil N:P and R:FR compared to herbaceous ferns and

palms. Distance–decay rates in compositional similarity were slightly higher for palms than ferns.

Ferns and palms showed similar patterns of compositional variation in response to resource

1 This article is published as: Viana, J. L., Turner, B. L., and Dalling, J. W. (2020). Compositional variation in understory fern and palm communities along a soil fertility and rainfall gradient in a lower montane tropical forest. Journal of Vegetation Science, jvs.12947. https://doi.org/10.1111/jvs.12947 22

availability but were related to distinct environmental factors. Soil fertility and light availability

emerged as key factors in predicting the abundance of plant groups, with clear partitioning of

environmental gradients only apparent when tree ferns were compared with other groups.

INTRODUCTION

A major challenge in ecology is to understand the processes responsible for shaping the abundance and distribution of species. While many hypotheses have been proposed to explain how plant communities assemble, most can be cast as conforming to one of two contrasting views, where communities are either structured primarily by the dispersal capacity of species (dispersal assembly) or by species requirements for particular sets of environmental conditions (niche assembly) (Hubbell, 2001; Silvertown, 2004). Although these views are not mutually exclusive, their relative importance may depend on the life history of the species groups under investigation.

Ferns are one of the earliest vascular plant lineages that persist today (Taylor et al., 2009).

They evolved in terrestrial landscapes, have spread across ecosystems worldwide, and remain a locally dominant plant group in many moist temperate and tropical montane forests (Brock et al.,

2016; Sharpe and Shiels, 2014). The variety of life forms found within ferns, including terrestrial herbs, aquatic floating plants, epiphytes, and tree ferns, has allowed them to fill most ecological niches. Furthermore, fern species contribute significantly to the diversity of tropical and temperate forests (Gentry and Dodson, 1987), and have practical use as indicators of both environmental heterogeneity and, potentially, environmental change (Bergeron and Pellerin, 2014; Silva et al.,

2018). In addition, the tiny spores produced by ferns suggest that they are less affected by dispersal limitation than most other vascular plant groups (Mehltreter et al., 2010; Tuomisto and Poulsen,

1996).

Some ferns have evolved unique anatomical and physiological traits to cope with sudden

changes in water availability allowing them to maintain populations that undergo periodic dry conditions (Holmlund et al., 2020). However, fern richness peaks in mesic mid-elevation forests

(Moran, 2008). To a certain extent, the narrow association of some fern taxa with mesic, low-light environments typical of forest understory might be explained by limits imposed by passive stomatal control, which contrasts with the active stomatal control found in other vascular plant groups (Brodribb and McAdam, 2011; Pittermann et al., 2011), and by low hydraulic conductivity relative to seed plants (Watkins et al., 2010). Analyses of fern distribution patterns at large spatial scales in lowland Amazonian forests with relatively homogenous climate and with long gradients of soil nutrient availability have also revealed that fern assemblages are sensitive to soil physical and chemical properties (Tuomisto et al., 2014; Tuomisto and Poulsen, 1996; Tuomisto and

Ruokolainen, 1993; Vormisto et al., 2000). Similarly, in lowland Central American rain forests, compositional similarity is correlated with soil pH, and exchangeable base cation, nitrogen (N), phosphorus (P) and carbon (C) concentrations (Jones et al., 2006, 2013). Studies of the drivers of distribution patterns of ferns in neotropical montane and lower montane forests, where abundance and diversity peaks, and where both soil and climate vary at short spatial scales, are relatively scarce (but see Jones et al., 2011).

Short–stature palms often occupy the same tropical forest understory stratum as ferns. As with terrestrial ferns, palms also exhibit habitat associations governed by soil properties (Andersen et al., 2010; Clark et al., 1995; Svenning, 1999; Vormisto et al., 2004a) . However, understory palms and terrestrial ferns represent opposite ends of a spectrum of propagule size and dispersal mechanism in vascular plants. While ferns produce small (25–60 μm) wind-dispersed spores

(Warren and Wagner, 1974), understory palms are dependent on vertebrate dispersal of their relatively large seeds. For example, Chamaedorea, the largest genus of understory palm in the

Americas, produces 3–25 mm diameter fruits (Henderson et al., 1995). The number of propagules

per plant is, therefore, also drastically different in these plant groups. Understory palms produce

tens to hundreds of fruits per individual in each reproductive episode, whereas annual spore

production in ferns varies from 0.5 to 100 million per individual depending on species (Conant,

1978). Consequently, the contrasting reproductive ecology of ferns and palms provides an

opportunity to examine how plant life-history traits influence the relative importance of dispersal

and niche assembly processes for co-occurring taxa. One prediction arising from the large

difference in reproductive output, and therefore source limitation to seed dispersal (sensu Clark et

al., 1998) is that species with lower dispersal ability will have more aggregated populations and

will therefore show greater rates of decline in compositional similarity with increasing distance

between communities (Qian, 2009; Soininen et al., 2007).

Previous work has compared the distribution of palms and ferns across sites varying in soil

texture, base cation availability and hydrological regime in the lowland Amazon (Tuomisto et al.,

2016; Vormisto et al., 2000), and between trees and ferns across a gradient of soil nutrient and

moisture in lowland forest in Panama (Jones et al., 2013). These studies, however, included canopy

palms and did not directly compare species with similar stature and forest understory requirements.

Montane forests also differ from lowland forests by having more heterogeneous habitat conditions at local scales resulting from altered temperature and moisture conditions associated with variation in aspect, exposure, and elevation, and from local variation in soil nutrient availability associated with parent material and duration of weathering (Dalling et al., 2016). Furthermore, whereas studies of plant community associations with soil variables in Amazonian forests have focused on base cation and aluminum concentrations, montane forests are often characterized by nitrogen limitation, or potentially co-limitation between nitrogen and phosphorus (Dalling et al., 2016;

Tanner et al., 1998). Here we asked whether (a) the distributions of understory palm species and

terrestrial fern species respond in similar ways across steep soil nutrient and moisture gradients in

a mid-elevational forest; (b) the larger propagule size and lower reproductive output in understory

palms result in stronger spatial aggregation of populations, and therefore greater compositional

beta-diversity and distance–decay rates in community similarity than ferns.

METHODS

Site of study

The study was carried out in two forest reserves: the 19,500 ha Fortuna Forest Reserve

(8°45′ N, 82°15′ W) and adjacent higher elevation areas of the Palo Seco Protected Forest (8°45′

N, 82°13′ W). Both reserves support humid to super-humid rain forest (Holdridge, 1947). For

brevity, the two sites are referred to collectively as Fortuna. Elevation ranges from 700 m to 1,500

m asl and includes both Pacific and Caribbean slopes as well as a central valley surrounding Lake

Fortuna. Mean annual rainfall ranges between 4,600 mm and 6,300 mm. Fog and cloud cover are

frequent, varying with elevation and rainfall. Mean annual temperature ranges from 19°C to 22°C

(Cavelier et al., 1996).

Plot description

Fern and palm surveys were carried out in twelve one-hectare plots arrayed across Fortuna

(Figure 2.1). The plots occupy a natural gradient in soil fertility, determined by the underlying parent material. Sites are classified as being underlain by one of five parent materials, ordered according to increasing fertility: (i) deep layers of rhyolitic tuff with a shallow organic layer

(Chorro A & B), or rhyolite with a deep organic horizon (Honda A & B); (ii) rhyolite to andesite

transition (Samudio); (iii) andesite, with clay-rich soils and shallow or no organic horizons (Bonita,

Palo Seco, Verrugosa A & B); (iv) dacite or undifferentiated volcanics with more organic rich

topsoil (Hornito, Alto Frio); and (v) basalt (Pinola) (Table S1,Prada et al., 2017). Soils developed

on these parent materials vary in soil pH, N, P, and cation exchange capacity (Table 2.1). Sites

also vary in the amount and seasonality in rainfall and cloud cover, ranging from aseasonal

conditions in Caribbean slope plots (Palo Seco, Verrugosa A & B), occasional dry periods

(monthly rainfall < 100 mm) in sites surrounding Lake Fortuna, and a longer dry season on the plot on the Pacific slope (Alto Frio; Table 2.1). Previous work has shown that both the overstorey tree community and understory palm community exhibit a high degree of compositional turnover

in response to these environmental variables (Andersen et al., 2010; Prada et al., 2017).

Plot environmental data

Soil physical and chemical properties and rainfall data were obtained from Prada et al.

(2017). Briefly, 0–10 cm deep soil samples collected in 13 locations in each plot were analyzed

for pH, bulk density, total soil C, N and P, extractable inorganic N (NH4 and NO3), resin

extractable P (resin P), extractable cations (Al, Ca, Fe, K, Mg, Zn), effective cation exchange

capacity (ECEC: Al, Ca, Fe, K, Mg, Mn, Na), and total exchangeable bases (TEB: Ca, Mg, K, Na).

Soil data for each plot were obtained by calculating the average over the 13 soil samples collected.

Rainfall data were collected biweekly over a period of seven years (2007–2009 and 2011–2014)

in a large gap adjacent to a subset of plots. For Bonita, Chorro B, and Verrugosa A and B, we used

rainfall data from the nearest plots (<3 km away and with comparable slope position; Prada et al.,

2017). Only rainfall data from 2013 and 2014 were available from Pinola and Alto Frio.

In addition, for this study, we measured gravimetric soil moisture, red:far red ratios (R:FR)

and canopy openness. These additional variables were measured in each plot across fifteen 5 m ×

5 m subplots 20 m apart along three 5-m wide transects each 30 m apart, which were previously

established by Andersen et al. (2010) for an understory palm survey. Here, we used the same

subplot set-up to also survey terrestrial ferns and resurvey understory palm communities.

Gravimetric soil moisture content was measured during the mid-dry season (Feb–Mar). For each 27

plot, one soil sample was collected in each of fifteen 5 m × 5 m subplots. Leaf litter was removed,

and mineral soil collected from the upper 5 cm of the profile. Light conditions were quantified by

measuring R:FR using an SKR 100 sensor (Skye Instruments Ltd, Llandrindod, UK), and canopy

openness using hemispherical photographs. R:FR is highly correlated with percent light

transmission under cloudy conditions across a range of canopy conditions (Capers and Chazdon,

2004). The SKR 100 sensor was placed at ground level, free from the shade of most bordering branches and leaves. Hemispherical photographs were taken simultaneously at a constant height

of 1 m. Both R:FR and canopy openness were measured at three points (0, 2.5, 5 m) along the

central north–south line of each subplot. Hemispherical photographs were taken using an Opteka

fisheye lens attached to a Nikon D3300 camera. All measurements were taken under uniformly

cloudy conditions. Canopy openness were analyzed with Gap Light Analyzer v. 2.0 (Frazer et al.,

2000). For each plot, we calculated the average over the 15 samples of gravimetric soil moisture and light variables to obtain the plot variables.

Community sampling

Understory palms and terrestrial ferns were surveyed within the same fifteen 5 m × 5 m subplots used for environmental measurements. All palms and ferns > 10 cm in height were counted and identified. Terrestrial ferns included herbs and tree ferns growing on the ground, or on small rocks. Because the objective was to compare habitat requirements of palms and ferns, all holo-epiphytic taxa (obligate epiphytes) were excluded, and primary hemi-epiphytes (where the life cycle starts as an epiphyte and roots grow downward later) were not considered. Potential ramets of ferns and palm species when spaced one meter from the nearest conspecific, and where no obvious connections could be observed, were counted as distinct individuals. Understory palm taxa included all species below the canopy layer (<15 m height). Nomenclature follows Henderson et al. (1995) for understory palms and PPG I (2016) for extant ferns. Fertile plants were collected

to verify identification and then used as representative voucher specimens deposited in the

University of Panama Herbarium (PMA).

Data analysis

Species accumulation curves were permuted to assess the adequacy of plant community

sampling. Fisher's alpha was used to quantify fern and palm diversity. A Principal Components

Analysis (PCA) was used to examine the environmental variation among twelve 1-ha plots,

including 19 soil variables, two rainfall variables, two light variables, and elevation (Table 2.1).

All environmental variables were standardized to have a mean of zero and unit variance before

performing the PCA.

To test whether fern and palm community composition respond similarly to environmental

gradients, we compared ordinations of the two groups of plants. First, we performed a Principal

Coordinates Analysis (PCoA) of fern and palm data separately, using square-rooted Bray–Curtis

distance of relative abundances. Then we tested the strength of the association of individual

environmental variables with the dissimilarity matrix of each of the plant groups by fitting

standardized (mean zero and unit variance) environmental variables a posteriori to the PCoA. We

used the envfit function from the vegan package, that uses multiple regression techniques where

the first two dimensions of the PCoA are the independent variables. The significance of fitted

environmental variables was assessed by 1,000 permutations. To account for multiple comparisons, we applied the Holm–Bonferroni method to adjust p-values (Holm, 1979). Second, we used Procrustes analysis, in which the ordination space configuration of the two plant groups is compared using a superimposition approach with configurations scaled to equal dispersions and then rotated until the sum of squared differences is maximally reduced (Peres-Neto and Jackson,

2001). The Procrustes analysis was permuted 1,000 times to test its significance.

In addition, we used Mantel tests to examine correlations between floristic, environmental,

and geographic distance matrices (Legendre and Legendre, 2012). Bray–Curtis distances were

calculated with species abundance to generate compositional dissimilarity matrices, and Euclidean

distances were calculated for geographic and standardized environmental data. Universal

Transverse Mercator (X and Y) coordinates from each plot were used for calculating Euclidean

distances. Mantel tests were conducted between floristic dissimilarity data and all environmental

variables together, and geographic distances. In addition, we also tested correlations between

floristic data and three subsets of the environmental variables: soil, precipitation, and light

variables. The soil data consisted of 19 variables, which were standardized and summarized using

PCA. The first three axes of the PCA of the soil subset data were used to compute a Euclidean

distance matrix. Rain and light subsets were composed of two variables each. Pure environmental

and pure spatial effects on floristic data were analyzed with partial Mantel tests with significance

based on 999 permutations. Because previous studies at our sites showed that canopy trees and

understory palm composition are significantly associated with parent material (Andersen et al.,

2010; Prada et al., 2017), we used ADONIS, a non-parametric multivariate analysis of variance,

to determine whether both terrestrial fern and understory palm compositions also differ across

parent materials.

To test whether fern and palm abundances are affected differently by the environmental

variables that structure their community composition, we used a generalized linear mixed model

(GLMM) of palm and fern abundance with Poisson errors and log-link function. We selected the

environmental factors which were most strongly related to species composition in the fern and palm PCoA analyses (total soil N:P, red:far red, bulk density) and used them as univariate predictor variables for herbaceous fern, tree fern and palm abundance, with random effects to account for the nesting of subplots within plots. We used the Akaike information criterion (AIC) to compare

each full model with an intercept-only model that included only the random terms. We also tested for the association between fern and palm abundances at the 1-ha plot level using the non- parametric Spearman's test.

To determine whether palms have greater compositional beta-diversity than ferns, resulting from either lower reproductive output or shorter dispersal distances, we fitted a generalized linear regression model for the relationship between the log of geographical distance and Bray–Curtis pairwise floristic similarity at the plot level for the two plant groups. To test whether the slopes and intercepts differed between palms and ferns, we used a generalized linear model with Gaussian errors with plant group as a categorical variable. For all statistical analyses, we used environmental factors data from the plots, including the GLMM analysis. All analyses were carried out in the R environment (R Core Team, 2019) using the vegan, and lme4 packages (Bates et al., 2015;

Oksanen et al., 2019).

RESULTS

Fern and palm compositional structure

In total, 4,528 ferns and 2,237 palm individuals were recorded across the 12 plots, giving a total density of 10,061 and 4,971 fern and palm individuals per hectare, respectively, within the total area surveyed (0.45 ha). The fern community was represented by 37 genera and 77 species and the understory palms by 10 genera and 30 species. The palms all belonged to the subfamily

Arecoideae Burnett. Species richness (S), diversity (α), and abundances (n) of ferns and palms varied among sites and parent material (Table 2.1). The most seasonal plot, Alto Frio, supported the lowest richness, diversity, and abundance of both ferns and palms (S = 6, α = 1.78, n = 50; S =

3, α = 0.76, n = 31, respectively). In contrast, plots on andesite lithology in relatively aseasonal sites supported the highest richness, diversity, and abundance of both ferns and palms. For ferns,

the Bonita plot had the highest richness (S = 25), Verrugosa A had the highest diversity (α = 7.8),

and Samudio the highest number of individuals (n = 865). For palms, andesite plots also had the

highest richness, diversity (Palo Seco; S = 16, α = 3.73), and abundance (Verrugosa A, n = 479).

Fern and palm individual-based accumulation curves failed to saturate for any sites and taxa except

for palms at Bonita (Figure A.1). No fern or palm species occurred in all plots. Didymochlaena

truncatula was the most abundant fern species, occurring in all sites except Bonita (andesite) and

Alto Frio (undifferentiated volcanics). Among the 17 fern families, Dryopteridaceae, Cyatheaceae,

Anthyriaceae, and Thelypteridaceae were the most species-rich, while Diplazium (Athyriaceae) was the richest genus with 11 species. Geonoma cuneata was the most abundant palm species, occurring in all plots except Alto Frio. The most species-rich palm genus was Chameadorea, with

13 species.

Environmental gradients and their effects on compositional variation

The PCA of all environmental variables (excluding floristic data) indicated that the environmental gradient at Fortuna is determined by soil variation, with sites distributed according to parent material, and to some extent by total and dry-season rainfall (Figure A.2). The first two dimensions of the PCA explained 65.6% of the variance in the environmental dataset and produced two main gradients. The first dimension (50.0% explained variance) was positively correlated with soil variables and negatively correlated with rainfall variables, extractable Al and soil moisture.

The soil variables ECEC, exchangeable Ca, TEB, and exchangeable Mg were most strongly correlated with the first ordination dimension. The second dimension (15.6% explained variance) was positively correlated with elevation, R:FR, and total soil N:P (Table A.1). Additional descriptions of the environmental conditions of the 12 sites can be found in Prada et al. (2017).

Distinct patterns of floristic similarity were detected in fern and palm communities based on PCoA. However, compositional shifts in these communities were strongly correlated with soil 32

nutrients for both groups, with plots grouped according to parent material (Figure 2.2a,b). PCoA

axis 1 and axis 2 combined explained 32% of the variation in ferns and 45% of the variation in

palms. After Holm–Bonferroni analysis, total soil N:P and R:FR were positively correlated with

compositional variation of ferns (r2 > 0.80; p < 0.05; Table A.2), whereas only bulk density was

positively correlated with compositional variation of palms (r2 = 0.87; p < 0.05; Table A.2). Fern

and palm PCoA ordinations showed a high degree of concordance based on Procrustes analysis

(m2 = 0.09; p = 0.004; Procrustes rotation correlation = 0.96), indicating that compositional

patterns of palms and ferns were significantly more similar to each other than expected by chance.

Environment and plant community correlations

The floristic distance matrices of both ferns and palms were significantly correlated with

the distance matrices (Figure 2.3) of: (1) all environmental factors excluding geographic data; (2)

the soil PCA distance matrix; and (3) geographic distances (0.28 > r2 < 0.5; p < 0.05). Distances

of all environmental factors were also positively correlated with geographic distances. From these,

the lowest coefficients were obtained for the relationship between floristics and geographic

distance (Table A.3). After controlling for geographic distance, only ferns remained significantly

correlated with the distance matrix of all environmental factors, and neither ferns nor palms were

correlated with geographic distances when controlling for environmental factors. Fern floristic

distances were also positive and significantly correlated with precipitation. Palm floristic distances

were unrelated to precipitation, and both plant group distances were unrelated to light distances.

Consistent with the Procrustes analysis, fern and palm floristic distances were significantly

correlated with one another and remained significantly correlated after controlling for

environmental and geographic effects. There were significant effects of parent material (classified

in Table 2.1) on compositional similarity of both ferns and palms (ADONIS: F = 1.62; p = 0.006

and F = 2.23; p = 0.01, respectively).

Fern and palm abundance relationships with environmental variables

At the plot level, palm and fern abundances were not significantly correlated (Spearman ρ

= −0.005, p = 0.95), although some plots were notably palm-dominated (Verrugosa A) or fern- dominated (Samudio and Hornito; Table 2.1). Contrasting patterns in fern and palm abundance responses to total soil N:P, R:FR, and bulk density were obtained with the GLMM testing. There was a significant interaction effect for each of the GLMMs testing total N:P, R:FR, and bulk density individually on the abundances of the plant groups. Palms showed a decrease in abundance with an increase in total soil N:P (z = −20.33; p < 0.001) and R:FR (z = −7.94; p < 0.001), while fern abundance showed an opposite trend for each of the environmental factors tested (Figure

2.4a,c; Table A.4). A similar but weaker relationship was observed for soil bulk density (Figure

A.3a). The interaction effect of plant group abundances and total soil N:P and R:FR was driven by differential responses of herbaceous ferns and tree ferns to these variables (Figure 2.4b,d; Table

A.5). While herbaceous ferns and palms responded similarly to total soil N:P and R:FR, tree ferns were most abundant in low-fertility sites with high total soil N:P (z = 27.01; p < 0.001) which were also sites with higher light penetration to the understory and therefore higher R:FR (z = 28.32; p <

0.001). The response of herbaceous ferns and tree ferns to bulk density had a similar pattern, with their abundance decreasing with an increase in bulk density (Figure A.3b). Models including one environmental variable performed better (based on the AIC) than models based only on random effects for palm and fern abundances, as well as for tree fern and herbaceous fern abundances

(Tables A.6 and A.7).

Life-history effects on distance–decay rates of compositional similarity

Compositional similarity declined linearly with the logarithm of inter-plot distance both for ferns and palms (t = −4.54; p < 0.0001; t = −4.34; p < 0.0001, respectively). When considering

all taxa, there was a slightly steeper decline in floristic similarity for palms than ferns (t = 2.057, p < 0.05), although variation in pairwise similarity was high across all distances (Figure 2.5).

DISCUSSION

Ferns and palms dominate the understory stratum of many lowland and montane wet tropical forests (Moran, 2008; Svenning, 1999). Here we explored whether ferns and palms share similar environmental requirements, and whether differences in propagule size and number result in differences in beta-diversity patterns. Overall, we found that fern and palm community composition is structured in broadly similar ways across the environmental gradient of soils and climate; however, the composition of palms was most strongly associated with variation in soil bulk density, whereas fern composition was more strongly associated with soil N:P and light quality (R:FR). Furthermore, we observed that the differences between these groups were largely driven by tree ferns, which in contrast to herbaceous ferns and palms, reached peak abundances on the most infertile soils with the most open forest canopy. While we predicted more clumped distributions for palms than ferns, we found that distance–decay patterns in community similarity were broadly similar between the two groups, with only slightly higher distance–decay rates in palms.

Environmental influences on community structure

Both ferns and palms showed a high degree of compositional variation and variation in abundance across gradients of soil nutrient availability and seasonality. For both groups, richness and diversity patterns peaked in sites with andesite-derived soils of intermediate fertility (Bonita,

Palo Seco, Verrugosa A) while their abundance patterns diverged. Palms were more abundant on continually wet sites (Bonita, Verrugosa A and B), while peak fern abundance occurred at the transition between low-fertility rhyolite and andesite-derived soil (Samudio) and in a more

seasonal higher fertility site on dacite (Hornito). Across the seasonality gradient, the four ever-wet

Caribbean slope sites (Bonita, Palo Seco, Verrugosa A and B) and the strongly seasonal Pacific slope site (Alto Frio) only shared one fern species and had no palm species in common. Rainfall seasonality, however, was not necessary for compositional turnover; the relatively high- and low- fertility aseasonal sites (Pinola and Chorro A) only 3 km apart, shared no fern species and only three palm species. Palm community composition is known to be correlated with climate and soil properties, with a focus in lowland Amazonian sites on exchangeable base concentration, aluminum concentration and soil texture (Cámara-Leret et al., 2017; Costa et al., 2009; Poulsen et al., 2006; Vormisto et al., 2000). In contrast, earlier work with understory palms at Fortuna highlighted the importance of inorganic N availability as the strongest environmental factor influencing their composition (Andersen et al., 2010). Here we found that palm community composition was more strongly correlated with bulk density, which was correlated with the second non-metric multidimensional scaling (NMDS) axis in Andersen et al. (2010). The difference between these two studies may be attributable to the inclusion of a greater range of soil types in the current study. The strength of soil effects on palm community composition is variable and depends on the spatial scale of the study area (Eiserhardt et al., 2011), ecosystem type, and sometimes even varies with habit (i.e., canopy versus understory species; Costa et al., 2009). Most previous analyses of palms have grouped overstory and understory taxa together (Tuomisto et al.,

2016; Vormisto et al., 2000, 2004b), which may account for weaker habitat associations of palms than ferns as some canopy palms in neotropical lowland forests are widespread and abundant habitat-generalists (Tuomisto et al., 2016). At Fortuna, the abundance of overstory and understory palms does not coincide. Relatively few understory palms occur on rhyolite at the Chorro sites

(Table 2.1), while the overstorey of these forest plots is dominated by three canopy palms

(Colpothrinax aphanopetala, Wettinia quinaria and Euterpe precatoria) that collectively account

for 21%–45% of the basal area (Prada et al., 2017).

Fern species distributions have also been studied extensively across environmental

gradients. In sites where climatic conditions are relatively uniform over large scales, as in lowland

Amazonian forests, soil variables play a significant role structuring fern distribution patterns at

local to regional scales (Costa et al., 2005; Tuomisto et al., 2002; Tuomisto, Poulsen, et al., 2003;

Tuomisto, Ruokolainen, Aguilar, et al., 2003; Tuomisto, Ruokolainen, and Yli-Halla, 2003;

Tuomisto and Poulsen, 1996; Zuquim et al., 2007). These studies indicated that changes in fern composition are linked to the variation in the availability of base cations and soil texture, however plant-available soil P, an important determinant of tree species distributions (Condit et al., 2013), has been infrequently measured (but see Jones et al., 2013). Studies of fern communities across habitats, including more fertile sites, have identified the potential influence of a broader range of soil variables, including total N and total P, as well as extractable bases (Jones et al., 2006, 2011,

2014). Analyses of fern response to environmental gradients may also be contingent on taxonomy in this group. Two phylogenetically distinct groups: ferns and lycophytes (collectively

“pteridophytes”), may have distinct habitat requirements, while here we highlight how the monophyletic clade of Cyatheaceae that includes tree ferns may also differ in ecological requirements from its herbaceous relatives.

Beyond community structure, previous studies have also revealed that some groups of fern taxa are consistently restricted to particular habitat types. For example, Lindsaea species diversity

peaks in highly infertile white-sand soils that are broadly distributed across the western Amazonian

basin and Guiana shield (Lehtonen and Tuomisto, 2007; Zuquim et al., 2014). In our study, two

Lindsaea species were likewise restricted to low-fertility rhyolite sites that develop on volcanically derived soils resembling white sands. An additional Lindsaea species that did not enter our survey

also occurs in N-poor sites in association with the ectomycorrhizal tree Oreomunnea mexicana, sometimes dominating the understory strata (J. Lira Viana, personal observation). Gradients in soil texture, cation content and inundation have also been found to underlie niche differentiation in

Polybotrya species in western Amazonian forests (Tuomisto, 2006). Furthermore, in lowland tropical forests, fern communities on rich soils have been found to contain more closely related lineages than those in poor soils (Lehtonen et al., 2015). Soil effects on ferns communities, therefore, go beyond species distribution, influencing fern species coexistence and diversification.

Environmental partitioning between palms, herbaceous ferns, and tree ferns

Fern and palm compositional patterns showed a high level of similarity to each other, reflected in a significant congruence in their PCoA ordination configuration based on Procrustes analysis, and by the significant positive matrix correlation between their dissimilarities based on a

Mantel test. Similarity in the structuring of ferns and palm communities is consistent with earlier studies in lowland tropical forest at fine spatial scales (Poulsen et al., 2006; Vormisto et al., 2000).

More broadly, ferns and other plant groups, especially small-seeded Melastomataceae shrubs, have also been found to share similar floristic distributional patterns across edaphic conditions at larger spatial scales in lowland Amazonian forests (Ruokolainen et al., 1997; Tuomisto et al., 2002;

Tuomisto and Ruokolainen, 1993). Consequently, multiple lineages, particularly of understory plants, appear to respond similarly to environmental variation across multiple ecosystem types.

Despite overall similarities in the way that environmental variables impact the compositional variation of ferns and palms, there were some distinct differences. Palm community composition was most strongly related to soil bulk density, while for ferns, composition was related to total soil N:P and light quality (R:FR). The influence of soil bulk density for palms in this study differs from what was found in previous studies conducted in Amazonian forests, where exchangeable base cation, clay content, and topography were related to density, composition, and 38

phylogenetic structure (Cámara-Leret et al., 2017; Costa et al., 2009; Muscarella et al., 2019;

Poulsen et al., 2006; Vormisto et al., 2004a). Soil physical properties such as effective soil depth and soil structure also predicted the basal area or arborescent palms in Amazonian forests (Emilio et al., 2014). In contrast, the composition of terrestrial ferns was strongly associated with R:FR, a measure of canopy light interception, and total soil N:P, a measure of soil fertility. In our study, terrestrial herbaceous ferns were more abundant in sites with low R:FR, indicating greater canopy light interception associated with a taller canopy and more phosphorus-rich soils, while tree ferns were more abundant in more open and infertile sites. The association of herbaceous ferns with low-light conditions may further reflect the presence of an unusual photoreceptor, phytochrome 3, in these taxa, responsible for leaf and chloroplast orientation and that is especially sensitive to low- light conditions therefore potentially providing an advantage in deep shade (Kawai et al., 2003).

Neither total precipitation nor dry-season precipitation were retained as significant correlates of community composition. This likely reflects the fact that the majority of the census plots are located in either aseasonal Caribbean slope forests, or at sites with only short or infrequent dry spells.

Fern–palm interactions

A feature of Fortuna is that the forest understory in watersheds only a few kilometers apart can be dominated either by ferns (e.g. Samudio) or by palms (e.g. Verrugosa A; Table 2.1). At the plot level, however, fern and palm abundances were not correlated. Differences in the relative abundance of these groups at more local scales could reflect resource partitioning based on physiological differences in resource requirements, stochastic processes, potential linkage to past disturbance, or biotic interactions, potentially including direct competitive interactions between palms and ferns. Determining which of these processes can account for variation in the abundance of these species groups would require an experimental approach.

Nonetheless, further analyses of abundance relationships indicated that while low total soil

N:P was generally associated with palm or herbaceous fern dominance, high N:P sites were dominated by tree ferns. At the highest N:P sites in the Chorro watershed sites, tree ferns were three times more abundant than herbaceous ferns. In contrast, at nearby Pinola, a low N:P site on

basalt parent material, herbaceous ferns were 19 times more abundant than tree ferns. Sensitivity

of tree fern abundance to N and P availability has been observed in previous studies although

relationships differ from our findings. In a cool-temperate rain forest in New Zealand, the highest

stem density and basal area of the widely distributed tree ferns Cyathea smithii and Dicksonia squarrosa were recorded in alluvial forests featuring higher total soil P (mean = 87 g m−2) in

comparison to terrace forest (mean = 11 g m−2; Coomes et al., 2005). Similarly, tree ferns are most

abundant on high P soils along chronosequences of soil and vegetation development at Franz Josef,

Haast and Arawhata in New Zealand (Turner et al., 2018). In contrast, in tropical forest sites in

Hawaii, cover of the tree fern Cibotium glaucum was lower in sites with either high N or P

concentrations and greater in sites with N and P co-limitation (Vitousek, 2004). Tree ferns deserve

further study as an independent functional group to determine the physiological basis underlying

their habitat associations.

Floristic similarity and life-history effects

Distance decay in floristic similarity is a consequence of characteristics of both organismal

traits and environmental characteristics. Comparisons of distance–decay patterns among taxa

along the same environmental gradient can, therefore, provide some insight into the importance of

dispersal in structuring communities (Soininen et al., 2007). We predicted that stronger dispersal

limitation arising from larger propagule size and lower reproductive output in palms would result

in faster distance–decay rates in community similarity (beta-diversity) for palms than ferns. We

found both higher initial dissimilarity among fern communities, consistent with higher alpha-

diversity, and consistent with previous studies comparing organisms with spore vs seed dispersal

(Qian, 2009; Soininen et al., 2007), and a significantly steeper decline in similarity with distance

for palms. Slopes of the distance–decay relationship, however, were quite similar, and variation in

pairwise similarity between plots along the distance axis was high, both for ferns and palms,

consistent with results from the Mantel tests where neither fern nor palm floristic distance matrices

were correlated with geographic distance after controlling for environmental factors. In montane

forests, characterized by highly heterogeneous habitat conditions, distance–decay relationships are

therefore likely to be primarily governed by environmental dissimilarity (Jones et al., 2011, 2014)

and potentially landscape configuration (Garcillán and Ezcurra, 2003) rather than by species life

history.

CONCLUSIONS

In wet tropical forests, ferns and palms can occupy much of the available space in the forest

understory. Our results further support the findings of earlier studies from lowland forests that

ferns, palms and other plant groups, including canopy trees, generally show similar compositional

responses to environmental gradients despite large differences in growth form and reproductive traits. However, tree ferns reached peak abundance in plots on different parent materials to palms and herbaceous ferns and therefore showed a contrasting response to the environmental variables most strongly associated with compositional variation in this forest. Resource partitioning may,

therefore, potentially contribute in part to the coexistence of these major understory groups.

lms at *

54.2 1135 617.5 7.52 4964 * 159 0.64 Pinola PIN Basalt 13 10 2.7 1.9 351 313 0.5 2.3 6.6 10.12 1.8 6.98 0.3 0.35 5.4 4.5

Frio

51.1 94 1100 4641 503 9.6 0.31 Alto ALT Undiff. volc.ꭞ 6 3 1.8 0.8 50 38 0.66 2.5 8.47 11.89 1.4 3.76 0.03 0.42 5.6 4.7

nd understory pa nd understory

35 203 1330 5164 280 10.38 0.67 Hornito HOR Dacite 19 4 3.5 1.2 847 31 0.26 1.2 4.9 13.65 2.2 1.8 0.48 0.34 5 2.8

erns f

28.7 445 850 6257 317 8 1.10 0.4 Verrugosa B VEB Andesite 20 10 4.3 1.9 465 327 11.44 0.67 0.3 0.31 0.2 5.1 2.2 0.85 1.17

errestrial errestrial t

6257 37.9 445 970 11.16 0.42 Verrugosa A VEA Andesite 22 15 7.4 2.9 138 479 0.92 241 10.05 0.98 0.5 0.4 0.49 4.6 2.7 1.6 1.7

bundance bundance a 6257 32.5 445 880 369.2 7.67 0.41 Palo Seco Palo PAS Andesite 16 16 3.8 3.7 248 268 1.02 0.8 0.4 7.42 4.4 2.2 0.8 1.1 1 0.41

5507 47.6 351 1300 351 10.75 0.34 Bonita BON Andesite 25 12 5.7 2.3 451 388 2.31 1.77 1.1 12.34 4.5 3.8 1.9 2.5 0.3 0.74 LE

–

TAB

4833 51.3 215 1215 270 13.76 0.4 Samudio SAM Rhyo And 20 7 3.7 1.6 865 136 1.04 1.17 0.4 11.09 4.2 3.6 0.6 1.5 1.1 0.4

6159 1240 332 127.7 40.9 0.17 Honda B HOB Rhyolite 20 8 4.7 2.9 327 42 0.84 0.43 1.9 16.1 9.68 3.8 2.3 1.78 1.2 0.1 0.32

richness, diversity, and total total richness, and diversity,

6255 1155 381 180.6 43.9 0.29 Honda A HOA Rhyolite 18 11 4.3 3.1 290 101 1.07 1.2 0.2 17.46 9.06 3.6 2.9 2.25 1.1 0.04 0.5 nd a

6159 1240 86.3 50.6 332 31.53 0.31 12 1.8 2.3 258 73 2.35 Chorro B Chorro CHB Rhyolite 9 11.79 4.7 1.17 2.7 0.01 1.3 2.4 1.2 0.83 eatures eatures f

5507 26.95 1100 3.7 0.2 1.8 0.1 0.8 0.8 57.2 29.6 0.5 3 2.7 1.9 238 41 351 2.98 Chorro A Chorro CHA Rhyolite 12 0.13 9.74 0.73

hectare plots

nvironmental nvironmental ) -3 e ) (%)

Ferns Palms Ferns Palms Ferns Palms -1

(g g

(g cm (mm) Plot

(m) he12 one-

(mm)

†

‡ 2.1.

‡ O) ‡ ‡ 2

Density

N -N -

3 4

§ † Al Rainfall Rainfall Total N:P Total C Ca Total P pH (H NH Abundance Elevation rainfall Dry Parent Material Richness Fisher’s Alpha Plot Code Canopy openness openness Canopy Total N R:FR NO Soil moisture Bulk Bulk Resin P Table of each

1.09 Pinola PIN Basalt 0.1 0.15 0.02 8.5 0.005 8

2.29 0.02 Alto Frio ALT Undiff. volc.ꭞ 0.12 0.42 11.4 0.005 10.9

1.01 0.03 Hornito HOR Dacite 0.18 0.71 7.4 0.005 6.2

0.15 Verrugosa B VEB Andesite 0.04 0.07 0.04 1.8 0.005 0.6

0.17 0.03 Andesite 0.06 0.04 2.5 Verrugosa A VEA 0.05 0.6

0.33 Andesite 0.06 0.19 0.04 2.8 Palo Seco Palo PAS 0.01 1.5

0.35 Andesite 0.11 0.03 0.03 3.4 Bonita BON 0.05 0.8

ꭞ undifferentiated volcanics. undifferentiated ꭞ

– ;

-1 L 0.32 Rhyo And 0.07 0.14 0.05 3.4 Samudio SAM 0.06 1.6 M

units C units § ; 0.06 0.01 Rhyolite 0.02 0.01 1.5 Honda B HOB 0.03 0.2 -3

0.04 0.01 Rhyolite 0.02 0.01 1.3 Honda A HOA 0.02 0.1 units mg cm †

;

-3

0.07 0.03 3.1 Chorro B Chorro CHB Rhyolite 0.07 0.04 0.04 1.7

units µg cm units

‡

0.1 0.28 1.8 0.9 Chorro A Chorro CHA Rhyolite 0.1 0.01 0.01 2014; 2014; .) ont

§ §

§ § Mg Mn Na ECEC TEB Plot Code Parent Material Fe K Table 2.1. - 2013 for only available *Data

FIGURES

Figure 2.1. Map of the twelve 1-ha plots in the site of study, Fortuna Forest Reserve and adjacent upland areas of the Palo Seco Protected Forest. Symbols indicate the location of the 12 1-ha plots established across soils with distinct parent material.

Figure 2.2. The first two dimensions of a Principal Coordinate Analysis of the floristic composition of (a) ferns, and (b) palms, where different symbols indicate parent material. Vectors indicate environmental variables that remained significantly correlated with the fern and palm communities (p ≤ 0.05) after the Holm-Bonferroni analysis for multiple comparisons.

Figure 2.3. Comparison of Pearson correlation coefficients obtained from Mantel and partial

Mantel tests between distance matrices derived from geographic and environmental variables

(Euclidean distances) and floristic similarity (Bray-Curtis index) for ferns (open bars) and palms

(filled bars). The x-axis represents the subsets of environmental variables tested: (i) Soil – includes

the first three principal component axes of soils variables; (ii) All Environ. - includes all

environmental variables (soils, light and precipitation); (iii) Geographic - includes geographic

coordinates; (iv) Env|Geo - environment controlling for geographic distance; (v) Precipitation -

includes only rainfall and dry season rainfall; (vi) Geo| Env - includes geographic distance

controlling for environment; (vii) Light - includes canopy openness and R:FR. Asterisks indicate

significant correlations at p ≤ 0.05.

Figure 2.4. Relationship between environmental factors and the total abundance of plant groups

(all ferns and palms), and total abundance of ferns (tree ferns and herbaceous ferns) from generalized linear mixed models. Palm and fern abundances in response to total soil N:P (a) and red: far-red (c). Tree fern and herbaceous fern abundance in response to total soil N:P (b) and red: far-red (d). C.I. = Confidence Intervals.

Figure 2.5. Relationship between pairwise floristic similarity and geographic distance in kilometers for 1-ha plots. Linear regression fit and 95% CI are shown.

REFERENCES

Andersen, K. M., Turner, B. L., and Dalling, J. W. (2010). Soil-based habitat partitioning in

understorey palms in lower montane tropical forests. Journal of Biogeography, 37(2), 278–

292. https://doi.org/10.1111/j.1365-2699.2009.02192.x

Bates, D., Mächler, M., Bolker, B., and Walker, S. (2015). Fitting linear mixed-effects models

using lme4. Journal of Statistical Software, 67(1), 1–48.

https://doi.org/10.18637/jss.v067.i01

Bergeron, A., and Pellerin, S. (2014). Pteridophytes as indicators of urban forest integrity.

Ecological Indicators, 38, 40–49. https://doi.org/10.1016/j.ecolind.2013.10.015

Brock, J. M. R., Perry, G. L. W., Lee, W. G., and Burns, B. R. (2016). Tree fern ecology in New

Zealand: A model for southern temperate rainforests. Forest Ecology and Management,

375, 112–126. https://doi.org/10.1016/j.foreco.2016.05.030

Brodribb, T. J., and McAdam, S. A. M. (2011). Passive origins of stomatal control. Science,

331(February), 582–585. https://doi.org/10.1126/science.1197985

Cámara-Leret, R., Tuomisto, H., Ruokolainen, K., Balslev, H., and Kristiansen, S. M. (2017).

Modelling responses of western Amazonian palms to soil nutrients. Journal of Ecology,

105(2), 367–381. https://doi.org/10.1111/1365-2745.12708

Cavelier, J., Solis, D., and Jaramillo, M. A. (1996). Fog interception in montane forests across

the Central Cordillera of Panamá. Journal of Tropical Ecology, 12, 357–369.

https://doi.org/10.1017/S026646740000955X

Clark, D. A., Clark, D. B., Sandoval, R. M., and Castro, M. V. C. (1995). Edaphic and human

effects on landscape-scale distributions of tropical rain forest palms. Ecology, 76(8), 2581–

2594. https://doi.org/10.2307/2265829

Clark, J. S., Macklin, E., and Wood, L. (1998). Stages and spatial scales of recruitment limitation

in southern appalachian forests. Ecological Monographs, 68(2), 213–235.

https://doi.org/10.1890/0012-9615(1998)068[0213:SASSOR]2.0.CO;2

Conant, D. S. (1978). A radioisotope technique to measure spore dispersal of the tree fern

Cyathea arborea. Pollen et Spores, 20(4), 583–593.

Condit, R., Engelbrecht, B. M. J., Pino, D., Perez, R., and Turner, B. L. (2013). Species

distributions in response to individual soil nutrients and seasonal drought across a

community of tropical trees. Proceedings of the National Academy of Sciences, 110(13),

5064–5068. https://doi.org/10.1073/pnas.1218042110

Coomes, D. A., Allen, R. ober. B., Bentley, W. arre. A., Burrows, L. E., Canham, C. D., Fagan,

L., Forsyth, D. M., Gaxiola-Alcantar, A., Parfitt, R. L., Ruscoe, W. A., Wardle, D. A.,

Wilson, D. J., and Wright, E. F. (2005). The hare, the tortoise and the crocodile: the ecology

of angiosperm dominance, conifer persistence and fern filtering. Journal of Ecology, 93(5),

918–935. https://doi.org/10.1111/j.1365-2745.2005.01012.x

Costa, F. R. C., Guillaumet, J. L., Lima, A. P., and Pereira, O. S. (2009). Gradients within

gradients: the mesoscale distribution patterns of palms in a central amazonian forest.

Journal of Vegetation Science, 20(1), 69–78. https://doi.org/10.1111/j.1654-

1103.2009.05314.x

Costa, F. R. C., Magnusson, W. E., and Luizao, R. C. (2005). Mesoscale distribution patterns of

Amazonian understorey herbs in relation to topography, soil and watersheds. Journal of

Ecology, 93, 863–878. https://doi.org/10.1111/j.1365-2745.2005.01020.x

Dalling, J. W., Heineman, K., González, G., and Ostertag, R. (2016). Geographic, environmental

and biotic sources of variation in the nutrient relations of tropical montane forests. Journal

of Tropical Ecology, 32(5), 368–383. https://doi.org/10.1017/S0266467415000619

Eiserhardt, W. L., Svenning, J. C., Kissling, W. D., and Balslev, H. (2011). Geographical

ecology of the palms (Arecaceae): Determinants of diversity and distributions across spatial

scales. Annals of Botany, 108, 1391–1416. https://doi.org/10.1093/aob/mcr146

Emilio, T., Quesada, C. A., Costa, F. R. C., Magnusson, W. E., Schietti, J., Feldpausch, T. R.,

Brienen, R. J. W., Baker, T. R., Chave, J., Álvarez, E., Araújo, A., Bánki, O., Castilho, C.

V., Honorio C, E. N., Killeen, T. J., Malhi, Y., Oblitas Mendoza, E. M., Monteagudo, A.,

Neill, D., … Phillips, O. L. (2014). Soil physical conditions limit palm and tree basal area in

Amazonian forests. Plant Ecology and Diversity, 7(1–2), 215–229.

https://doi.org/10.1080/17550874.2013.772257

Frazer, G. W., Canham, C. D., and Lertzman, K. P. (2000). Gap light analyzer (GLA) Version

2.0: Image-processing software to analyze true-color hemispherical canopy photographs.

Bulletin of the Ecological Society of America, 81(3), 191–197.

Garcillán, P. P., and Ezcurra, E. (2003). Biogeographic regions and β-diversity of woody dryland

legumes in the Baja California peninsula. Journal of Vegetation Science, 14(6), 859–868.

https://doi.org/10.1111/j.1654-1103.2003.tb02219.x

Gentry, A. H., and Dodson, C. H. (1987). Diversity and biogeography of neotropical vascular

epiphytes. Annals of the Missouri Botanical Garden, 74(2), 205.

https://doi.org/10.2307/2399395

Henderson, A., Galeano, G., and Bernal, R. (1995). Field guide to the palms of the Americas.

Princeton University Press.

Holdridge, L. (1947). Determination of world plant formations from simple climatic data.

Science, 105, 367–368. http://science.sciencemag.org/content/105/2727/367.short

Holm, S. (1979). A Simple Sequentially Rejective Multiple Test Procedure. Scandinavian

Journal of Statistics, 6(2), 65–70. https://doi.org/https://www.jstor.org/stable/4615733

Holmlund, H. I., Davis, S. D., Ewers, F. W., Aguirre, N. M., Sapes, G., Sala, A., and Pittermann,

J. (2020). Positive root pressure is critical for whole-plant desiccation recovery in two

species of terrestrial resurrection ferns. Journal of Experimental Biology, 71(3), 1139–1150.

https://doi.org/10.1111/nph.16067

Hubbell, S. P. (2001). The unified neutral theory of biodervisity. Princeton University Press.

Jones, M. M., Cicuzza, D., van Straaten, O., Veldkamp, E., and Kessler, M. (2014).

Determinants of fern and angiosperm herb community structure in lower montane rainforest

in Indonesia. Journal of Vegetation Science, 25, 1216–1224.

https://doi.org/10.1111/jvs.12181

Jones, M. M., Ferrier, S., Condit, R., Manion, G., Aguilar, S., and Erez, R. P. (2013). Strong

congruence in tree and fern community turnover in response to soils and climate in central

Panama. Journal of Ecology, 101, 506–516. https://doi.org/10.1111/1365-2745.12053

Jones, M. M., Szyska, B., and Kessler, M. (2011). Microhabitat partitioning promotes plant

diversity in a tropical montane forest. Global Ecology and Biogeography, 20, 558–569.

https://doi.org/10.1111/j.1466-8238.2010.00627.x

Jones, M. M., Tuomisto, H., Clark, D. B., and Olivas, P. (2006). Effects of mesoscale

environmental heterogeneity and dispersal limitation on floristic variation in rain forest

ferns. Journal of Ecology, 94, 181–195. https://doi.org/10.1111/j.1365-2745.2005.01071.x

Kawai, H., Kanegae, T., Christensen, S., Kiyosue, T., Sato, Y., Imaizumi, T., Kadota, A., and

Wada, M. (2003). Responses of ferns to red light are mediated by an unconventional

photoreceptor. Nature, 421(6920), 287–290. https://doi.org/10.1038/nature01310

Legendre, P., and Legendre, L. (2012). Numerical ecology (3rd English ed.). Elsevier.

https://vufind.carli.illinois.edu/vf-uiu/Record/uiu_7260036

Lehtonen, S., Jones, M. M., Zuquim, G., Prado, J., and Tuomisto, H. (2015). Phylogenetic

relatedness within neotropical fern communities increases with soil fertility. Global Ecology

and Biogeography, 24, 695–705. https://doi.org/10.1111/geb.12294

Lehtonen, S., and Tuomisto, H. (2007). Ecological and genetic notes on Lindsaea digitata

(Lindsaeaceae), a new fern species from western Amazonia. Nordic Journal of Botany,

25(3–4), 145–151. https://doi.org/10.1111/j.0107-055X.2007.00109.x

Mehltreter, K., Walker, L. R., and Sharpe, J. M. (2010). Fern Ecology. Cambridge University

Press. https://doi.org/10.1017/CBO9780511844898

Moran, R. C. (2008). Diversity, Biogeography, and floristics. In T. A. Ranker and C. H. Haufler

(Eds.), Biology and Evolution of Ferns and Lycophytes (pp. 367–394). Cambridge

University Press. https://doi.org/10.1017/CBO9780511541827.015

Muscarella, R., Bacon, C. D., Faurby, S., Antonelli, A., Kristiansen, S. M., Svenning, J. C., and

Balslev, H. (2019). Soil fertility and flood regime are correlated with phylogenetic structure

of Amazonian palm communities. Annals of Botany, 123(4), 641–655.

https://doi.org/10.1093/aob/mcy196

Oksanen, J., Blanchet, F. G., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., Minchin, P. R.,

O’Hara, R. B., Simpson, G. L., Solymos, P., Stevens, M. Henry H. Szoecs, E., and Wagner,

H. (2019). vegan: Community Ecology Package. R package version 2.5-4. Https://CRAN.R-

Project.Org/Package=vegan.

Peres-Neto, P. R., and Jackson, D. A. (2001). How well do multivariate data sets match? The

advantages of a Procrustean superimposition approach over the Mantel test. Oecologia,

129(2), 169–178. https://doi.org/10.1007/s004420100720

Pittermann, J., Limm, E., Rico, C., and Christman, M. A. (2011). Structure-function constraints

of tracheid-based xylem: a comparison of conifers and ferns. New Phytologist, 192(2), 449–

461. https://doi.org/10.1111/j.1469-8137.2011.03817.x

Poulsen, A. D., Tuomisto, H., and Balslev, H. (2006). Edaphic and floristic variation within a 1-

ha plot of lowland Amazonian rain forest. Biotropica, 38(4), 468–478.

https://doi.org/10.1111/j.1744-7429.2006.00168.x

PPG I. (2016). A community-derived classification for extant lycophytes and ferns. Journal of

Systematics and Evolution, 54(6), 563–603. https://doi.org/10.1111/jse.12229

Prada, C. M., Morris, A., Andersen, K. M., Turner, B. L., Caballero, P., and Dalling, J. W.

(2017). Soils and rainfall drive landscape-scale changes in the diversity and functional

composition of tree communities in premontane tropical forest. Journal of Vegetation

Science, 28(4), 859–870. https://doi.org/10.1111/jvs.12540

Qian, H. (2009). Beta diversity in relation to dispersal ability for vascular plants in North

America. Global Ecology and Biogeography, 18(3), 327–332.

https://doi.org/10.1111/j.1466-8238.2009.00450.x

R Core Team. (2019). R: A language and environment for statistical computing. R Foundation

for Statistical Computing. https://www.r-project.org/.

Ruokolainen, K., Linna, A., and Tuomisto, H. (1997). Use of Melastomataceae and

pteridophytes for revealing phytogeographical patterns in Amazonian rain forests. Journal

of Tropical Ecology, 13(2), 243–256. https://doi.org/10.1017/S0266467400010439

Sharpe, J. M., and Shiels, A. B. (2014). Understory fern community structure, growth and spore

production responses to a large-scale hurricane experiment in a Puerto Rico rainforest.

Forest Ecology and Management, 332, 75–86. https://doi.org/10.1016/j.foreco.2014.01.023

Silva, V. L., Mehltreter, K., and Schmitt, J. L. (2018). Ferns as potential ecological indicators of

edge effects in two types of Mexican forests. Ecological Indicators, 93, 669–676.

https://doi.org/10.1016/J.ECOLIND.2018.05.029

Silvertown, J. (2004). Plant coexistence and the niche. Trends in Ecology and Evolution, 19(11),

605–611. https://doi.org/10.1016/J.TREE.2004.09.003

Soininen, J., McDonald, R., and Hillebrand, H. (2007). The distance decay of similarity in

ecological communities. Ecography, 30(1), 3–12. https://doi.org/10.1111/j.2006.0906-

7590.04817.x

Svenning, J.-C. (1999). Microhabitat specialization in a species-rich palm community in

Amazonian Ecuador. Journal of Ecology, 87(1), 55–65. https://doi.org/10.1046/j.1365-

2745.1999.00329.x

Tanner, E. V. J., Vitousek, P. M., and Cuevas, E. (1998). Experimental investigation of nutrient

limitation of forest growth on wet tropical mountains. Ecology, 79(1), 10–22.

https://doi.org/10.1890/0012-9658(1998)079[0010:EIONLO]2.0.CO;2

Tuomisto, H. (2006). Edaphic niche differentiation among Polybotrya ferns in western

Amazonia: implications for coexistence and speciation. Ecography, 29(3), 273–284.

https://doi.org/10.1111/j.2006.0906-7590.04390.x

Tuomisto, H., Moulatlet, G. M., Balslev, H., Emilio, T., Figueiredo, F. O. G., Pedersen, D., and

Ruokolainen, K. (2016). A compositional turnover zone of biogeographical magnitude

within lowland Amazonia. Journal of Biogeography, 43(12), 2400–2411.

https://doi.org/10.1111/jbi.12864

Tuomisto, H., and Poulsen, A. D. (1996). Influence of edaphic specialization on pteridophyte

distribution in neotropical rain forests. Journal of Biogeography, 23(3), 283–293.

http://www.jstor.org/stable/2845846

Tuomisto, H., Poulsen, A. D., Ruokolainen, K., Moran, R. C., Quintana, C., Celi, J., and Cañas,

G. (2003). Linking floristic patterns with soil heterogeneity and satellite imagery in

Ecuadorian Amazonia. Ecological Applications, 13(2), 352–371.

https://doi.org/10.1890/1051-0761(2003)013[0352:LFPWSH]2.0.CO;2

Tuomisto, H., and Ruokolainen, K. (1993). Distribution of Pteridophyta and Melastomataceae

along an edaphic gradient in an Amazonian rain forest. Journal of Vegetation Science, 4(1),

25–34. https://doi.org/10.2307/3235634

Tuomisto, H., Ruokolainen, K., Aguilar, M., and Sarmiento, A. (2003). Floristic patterns along a

43-km long transect in an Amazonian rain forest. Journal of Ecology, 91(5), 743–756.

https://doi.org/10.1046/j.1365-2745.2003.00802.x

Tuomisto, H., Ruokolainen, K., Poulsen, A. D., Moran, R. C., Quintana, C., Cañas, G., and Celi,

J. (2002). Distribution and diversity of pteridophytes and Melastomataceae along edaphic

gradients in Yasuni National Park, Ecuadorian Amazonia. Biotropica, 34(4), 516–533.

Tuomisto, H., Ruokolainen, K., and Yli-Halla, M. (2003). Dispersal, environment, and floristic

variation of western Amazonian forests. Science, 299(5604), 241–244.

http://science.sciencemag.org/content/299/5604/241.short

Tuomisto, H., Zuquim, G., and Cárdenas, G. (2014). Species richness and diversity along

edaphic and climatic gradients in Amazonia. Ecography, 37, 1034–1046.

https://doi.org/10.1111/ecog.00770

Turner, B. L., Wells, A., Andersen, K. M., and Condron, L. M. (2018). Consequences of the

physical nature of the parent material for pedogenesis, nutrient availability, and succession

in temperate rainforests. Plant and Soil, 423(1–2), 533–548. https://doi.org/10.1007/s11104-

017-3514-4

Vitousek, P. M. (2004). Nutrient Cycling and Limitation: Hawai’i as a Model System. Princeton

University Press.

Vormisto, J., Phillips, O. L., Ruokolainen, K., Tuomisto, H., and Vasquez, R. (2000). A

comparison of fine-scale distribution patterns of four plant groups in an Amazonian

rainforest. Ecography, 23(3), 349–359. https://doi.org/10.1111/j.1600-0587.2000.tb00291.x

Vormisto, J., Tuomisto, H., and Oksanen, J. (2004). Palm distribution patterns in Amazonian

rainforests: What is the role of topographic variation? Journal of Vegetation Science, 15(4),

485–494. https://doi.org/10.1111/j.1654-1103.2004.tb02287.x

Warren, H., and Wagner, J. (1974). Structure of spores in relation to fern phylogeny. Annals of

the Missouri Botanical Graden, 61(2), 332–353. https://doi.org/10.2307/2395061

Watkins, J. E., Holbrook, N. M., and Zwieniecki, M. A. (2010). Hydraulic properties of fern

sporophytes: Consequences for ecological and evolutionary diversification. American

Journal of Botany, 97(12), 2007–2019. https://doi.org/10.3732/ajb.1000124

Zuquim, G., Costa, F. R. C., and Prado, J. (2007). Fatores que determinam a distribuição de

espécies de pteridófitas da Amazônia Central. Revista Brasileira de Biociências, 5(2), 360–

362.

Zuquim, G., Tuomisto, H., Jones, M. M., Prado, J., Figueiredo, F. O. G., Moulatlet, G. M.,

Costa, F. R. C., Quesada, C. A., and Emilio, T. (2014). Predicting environmental gradients

with fern species composition in Brazilian Amazonia. Journal of Vegetation Science, 25(5),

1195–1207. https://doi.org/10.1111/jvs.12174

CHAPTER 3: DIVERSITY AND ECOLOGY OF THE FERNS IN FORTUNA2

ABSTRACT

Ferns are the second most diverse plant group, exhibiting a broad variety of habits from 1

mm floating herbs to 20 m tall tree ferns. Habitats with continually wet and warm environments,

which are associated with mid-elevation tropical rainforests, harbor the highest fern diversity in

the Neotropics. Here we describe the fern and lycophyte flora of the lower montane tropical

rainforests of Fortuna Forest Reserve and adjacent Palo Seco Protected Forest, and discuss the associations of their component species with environmental variables. As a result of extensive field collections, the current fern species list of Fortuna is extended, including two new distribution records for Panama. Floristic similarity across sites at Fortuna reflects similarities in underlying

parent material and associated soil fertility, with several indicator taxa associated with particular

soil nutrient conditions related to parent material type. Compositional variation among herbaceous

ferns and tree ferns was remarkably similar with respect to parent material and soil variables.

However, the abundance patterns of tree ferns diverged from those of herbaceous ferns in response

to soil nutrient and light availability.

INTRODUCTION

The record of ferns on Earth is remarkably long. In the Late Devonian through Permian

periods (~400 million years [Ma] ago), the Coenopteridales are believed to have exhibited the first

fern-like morphology and to have given rise to the earliest-emerging extant filicalean families,

Gleicheniaceae and Osmundaceae (Phillips, 1974). They were characterized by laminar fronds,

2 This study is a chapter coauthored with Dr. James Dalling in the book “Fortuna Forest Reserve, Panama: Interacting Effects of Climate and Soils on the Biota of a Wet Premontane Tropical Forest” published as Smithsonian Contributions to Botany volume b112.

circinate venation and foliar borne annulate sporangia, morphological traits that characterize

modern fern taxa. While the relatively invariant morphology prevalent in ferns seems to have failed

to generate the evolutionary novelties of angiosperms, ferns have nevertheless succeeded by

preserving their leaf as a functional photosynthetic and reproductive organ. As a bifunctional

organ, leaves in ferns have also undergone a diversification in shape and form reflecting the

ecological requirements of the species (Vasco et al., 2013; Watkins et al., 2016).

Phylogenetic studies employing chloroplast markers of living taxa for evolutionary

inferences (Schneider et al., 2004; Schuettpelz and Pryer, 2009, 2007) suggest that ferns are a

monophyletic lineage and the closest relative to the seed plants, with lycophytes being a sister

group to ferns and seed plants. These studies imply a radiation event in the leptosporangiate ferns

(i.e. ferns in which sporangia emerge from a single epidermal cell) during the Late Cretaceous

period. More recently, Testo and Sundue (2016) used chloroplast markers and fossil data

calibration to suggest that the origin of all ferns – including both Equisetaceae and Psilotaceae

families – took place during the Silurian period (~ 440 Ma ago), and leptosporangiate ferns, during

the Carboniferous period. This is much earlier than previously hypothesized suggesting that the

most modern taxa arose in the Cretaceous period and diversified during the Cenozoic era

(Schneider et al., 2004; Schuettpelz and Pryer, 2009). One concern is that these studies (i.e.

Schneider et al., 2004; Schuettpelz and Pryer, 2009) are based on plastid gene markers, which

provide incomplete information on ancestry given their single linkage group and uniparental origin

(Wolf et al., 2018). Furthermore, hybridization events, significant and recurrent in ferns, are

missed in plastid phylogenetics. New approaches point toward the use of low-copy nuclear sequence data for acquiring multilocus nuclear-encoded genes, which should provide a better detection of signatures of complex evolutionary histories such as those of ferns.

With no major significant changes in their reproductive biology over millions of years, ferns preserve a remarkable and singular means of reproduction given their free-living gametophyte stage. Three types of sexual breeding systems are undertaken by ferns. Outcrossing and sporophytic selfing breeding types need eggs and sperm from different gametophytes, but for outcrossing types, gametophytes need to come from differing sporophyte parents, whereas in sporophytic selfing types, gametophytes can be derived from the same sporophyte individual.

These breeding systems are equivalent to those of seed plants. The third type is gametophytic selfing, in which eggs and sperm derive from a single, bisexual gametophyte, a unique breeding system in homosporous fern lineages that enable their capacity for extreme inbreeding (Sessa et al., 2016). In the short term, gametophytic selfing reduces genetic variation resulting in progeny that have identical pairs of genes for any given pair of hereditary characteristics. However, gametophytic selfing also increases the recruitment potential of homosporous ferns by enabling colonization following long-distance dispersal (Klekowski, 1979).

In addition to possessing these diverse breeding systems, ferns are set apart from bryophytes and seed plants by having sporophyte and gametophyte stages independent of one another. This singular dichotomy, found only in the life cycle of ferns, poses not only stimulating and demanding questions to future field-based ecological studies, but also a concern about how to approach ferns when considering the effects of one distinct life stage on another. While the ecological distribution of sporophytes may be dependent on the capacity of gametophytes to acquire resources for growth and reproduction, gametophytes can also increase species ranges by occurring in areas where its sporophyte stage is absent (Greer and McCarthy, 1999; Nitta et al.,

2016). Thus, it is imperative that upcoming ecological studies involving ferns consider how the

60 biology of gametophytes might impact the fate of sporophytes, considering the advantages and disadvantages of having a flexible breeding system as ferns do (Page, 2002; Sessa et al., 2016).

FERN DIVERSITY AND ENVIRONMENTAL GRADIENTS

After flowering plants, ferns are the most diverse vascular plant group in the world (Kreft et al., 2010). The highest diversity is associated with tropical mid-elevation forests and oceanic island habitats (Moran, 2008). Mountains with elevations of 800 to 2000 m maintain the highest species richness and levels of endemism (Kluge and Kessler, 2006). Tropical forests featuring mid- elevation slopes usually experience higher precipitation rates and reduced seasonality when compared to lowland tropical forests, which coupled with frequent cloud cover provides the humid conditions for ferns to complete their life cycle. Although it is thought that water is essential for ferns, there are species with physiological and reproductive adaptations that enable them to cope with dry conditions (Holmlund et al., 2020). Furthermore, local habitat heterogeneity associated with slope, aspect and soil variation on mountains may also contribute to the observed higher diversity at these sites. In an exploration of species microhabitat distributions and their elevational ranges, Jones et al. (2011) suggested that microhabitat requirements of mid-elevation fern species may differ from those of lowland forests.

Fortuna Forest Reserve, and the adjacent southeastern part of the Palo Seco Protected

Forest (henceforth ‘Fortuna’), are lower montane humid to super-humid rainforests (Holdridge,

1947) located in western Panama with altitude varying from 700 to 2000 m above sea level (asl)

Twelve one-hectare plots were established across the reserves to sample variation in climate and underlying parent material and soil characteristics (Figure 3.1, Dalling et al., n.d.; Prada et al.,

2017). In this area, with elevations from 850 to 1,300 m, mean annual rainfall ranges from 4,400 to 6,600 mm with increasing seasonality from the Caribbean to the Pacific slope. Both parent

material and associated soil conditions and precipitation have been found to strongly influence

canopy tree species and understory palm species distributions (Andersen et al., 2010; Prada et al.,

2017). Here we focus on (1) describing the major floristic components of ferns and lycophytes in

Fortuna and (2) characterizing habitat associations of terrestrial herbaceous ferns and tree ferns.

To examine the regional diversity of ferns in Fortuna we considered the most prominent

macroenvironmental features. These include (1) parent material, which consists of multiple distinct

volcanically derived rocks: rhyolite, mafic-volcanic, dacite, basalt and undifferentiated volcanics

(Silva et al., in press) that give rise to soils differing in nutrient availability (Turner and Dalling, in press); (2) annual and seasonal precipitation; (3) light availability, given differences in canopy structure across Fortuna and the high frequency of low-level cloud cover in these lower montane

forests (Dalling et al., in press). The analysis of the fern community at Fortuna included an existing

list of taxa compiled by McPherson et al. (2010) at the MOBOT projects webpage and in situ field

collections. Field collections were focused mostly on terrestrial species at the 12 one-hectare plots.

Specimens were also collected along the trails that lead to those plots. All specimens collected were identified and deposited in the Universidad de Panama Herbarium (PMA) and the

Universidad Autónoma de Chiriquí (UNACHI). The Taxonomic Name Resolution Service

(TNRS) was used to obtain standardized scientific names and to resolve synonyms (TNRS iPlant

Collaborative, version 4.0, http://tnrs.iplantcollaborative.org).

FORTUNA FERN AND LYCOPHYTE COMPOSITIONAL DIVERSITY

From the species list retrieved from McPherson et al., (2010), a total of 265 species names were listed, including epiphytes, terrestrial herbaceous ferns, and tree fern taxa, of which 241 species names were accepted, 23 were synonyms and one name was unresolved. We obtained 102 species during in situ field collections (Table B.1), of which 77 came from the 12 one-hectare plot

surveys and 25 from surrounding access trails. These collections contributed an additional 48 taxa

(of which 39 are terrestrial) to the McPherson list, resulting in a total of 289 fern and lycophyte

species. These taxa represent 25 families and 81 genera. Dryopteridaceae (53 spp.) was the richest

family (Figure 3.2). The genus Elaphoglossum, nested in this family, contributed more than half of the species (29 spp.), representing a large fraction of the epiphytic component of the flora of

Fortuna. Elaphoglossum has also been reported as the most species-rich genus in a fern flora of a mid-elevation montane forest in Peru (Jones et al., 2011). Other families important at Fortuna

include Polypodiaceae (39 spp.), containing mostly epiphytic species, and Hymenophyllaceae (22

spp.) with both epiphytic and terrestrial species. Diplazium (Athyriaceae) and Asplenium

(Aspleniaceae) were the fourth-ranked genera with 19 species each. Diplazium is comprised

exclusively of terrestrial species, whereas Asplenium exhibits varied habits. The genus Cyathea

(15 spp.) and the subfamily Thelypteridaceae (18 spp.) are also important components of the

Fortuna fern flora. Thelypteridaceae are mostly terrestrial species, while Cyathea are tree ferns.

Only 11% of the genera were represented by a single species. Lycophytes contributed an additional

21 species, with Palhinhae and Selaginella the most species-rich genera (9 spp. each). Terrestrial

ferns represented 15.1% of the flora of Fortuna when considering only the vascular taxa censused

in the plots.

An examination of different growth forms among fern species was conducted by consulting

the online version of Flora Mesoamericana (http://legacy.tropicos.org/Project/FM) which provides detailed descriptions of species. Seven categories of growth form were listed for fern species: terrestrial; epiphyte, tree fern, hemiepiphyte (plants that spend part of their life cycle as an epiphyte), climbers, epiphyte/terrestrial/rupicolous (rupicolous: plants that grow on rock or stone)

(ETR) and epiphyte/terrestrial (ET). Fern species were distributed across growth-form as follows:

52.9% terrestrials, 35% epiphytes, 7.6% tree ferns, 2.1% ETR, 1.7% hemiepiphytes, 0.4% climbers

(represented by a single species, Salpichlaena volubilis), and 0.4% ET indicating that only a single

species, Asplenium harpeodes, can adopt both growth forms. The low percentage of taxa assigned

to the hemiepiphyte growth-form might reflect the difficulty in documenting this habit in natural

environments, although hemiepiphytism occurs in a diversity of families including Blechnaceae,

Hymenophyllaceae, Lomariopsidaceae and Tectariaceae (Kramer and Green, 1990) which occur

in Fortuna. Terrestrial and epiphytic ferns were the two most frequently encountered growth forms

(Figure 3.3). The higher proportion of terrestrial ferns may not reflect actual richness differences

at Fortuna, but instead can be explained by field collectors who were focused primarily on

terrestrial species.

Data on species collections from the 12 permanent plots (Table B.1) revealed that the

overall abundance of terrestrial ferns peaks in sites with dacite and rhyolite-derived soils.

Furthermore, when abundance is partitioned by growth-form, tree ferns are more abundant in

rhyolite sites that have soils characterized by a high total nitrogen to total phosphorus ratio (soil

N:P >16) and low pH (<5). In contrast, terrestrial herbaceous ferns are more abundant in dacite- derived soils associated with low soil N:P and high effective cation exchange capacity (ECEC)

(Viana et al., 2020). In contrast to abundance, the highest fern diversity and richness were obtained for two sites with mafic-volcanic-derived soils (Verrugosa A: Fisher’s alpha diversity = 7.4;

Bonita: richness = 25 species; Figure 3.4). Therefore, these patterns suggest that at the regional scale, richness and diversity of ferns appears to be greatest on sites with intermediate fertility, whereas abundance patterns might depend on growth form, with the highest abundance of tree ferns at low fertility sites and with the highest abundance of herbaceous ferns at high-fertility sites.

LOCAL HABITAT ASSOCIATIONS OF TERRESTRIAL SPECIES

Fern biogeography in the Neotropics is well understood across precipitation and elevational

gradients (Jones et al., 2013; Watkins et al., 2006). Beyond these broad-scale climatic effects on fern distributions, more localized habitat conditions, characterized by variation in soil and parent material, likely also play an important role. In lowland Amazonian forests where climatic conditions are relatively uniform, base cation concentration, soil clay content, pH, soil carbon to nitrogen ratio (C:N), and total phosphorus (P) and calcium (Ca) explain most of the variation in fern species composition (Tuomisto and Poulsen 1996; Jones et al., 2006; 2014; 2016; Zuquim et

al., 2014). Steep gradients in the availability of key nutrients over short spatial scales, as occurs at

Fortuna, could therefore provide new insights into the influence of soil variables on fern

distribution and growth.

In this section, we describe the compositional associations of terrestrial ferns with environmental predictors by performing a nonmetric multidimensional scaling (NMDS) analysis.

We used raw species abundances and standardized environmental data collected at 12 one-hectare

plots established across distinct parent materials (Dalling et al., in press, Table 1.1). Environmental

data consisted of 19 soil variables, rainfall, and red to far-red light ratios (R:FR) measured at each

plot. The compositional patterns of tree ferns and terrestrial herb fern species were analyzed

together and then separately given the difference in habit and observed patterns in abundance

across plots. A Procrustes analysis was performed to determine whether tree fern communities and

herbaceous fern communities have similar or distinct configurations of ordination space. The significance of the Procrustes analysis was tested in 999 permutations. These ordination analyses are complementary to results presented in Viana et al (2020) which contrasted all fern taxa and

palm taxa at Fortuna. In addition, we report species associations with parent material and soil

nutrient status based on an indicator species analysis (De Cáceres et al., 2010), which analyzes

species associations with combinations of sites. Eleven 1-ha plots were selected and combined in

three groups according to parent material: low fertility rhyolite sites (Chorro A and B, Honda A

and B); intermediate fertility mafic-volcanic sites (Bonita, Palo Seco, Verrugosa A and B); high

fertility basalt/dacite/undifferentiated-volcanic sites (Pinola, Hornito, Alto Frio). Samudio site was

excluded because it contains both rhyolite and mafic volcanic microsites that are not mapped

within the plot. Indicator species were identified using the multipatt function implemented using

the package indicspecies in R (De Cáceres, 2013), which calculates an indicator value (IV) from

the product of relative abundance and relative frequency of the species in the groups (Dufrêne and

Legendre, 1997). We also repeated the analysis separating two rhyolite groups: the most infertile

and deep rhyolite deposits (the two Chorro plots) and shallower rhyolite deposits (the two Honda

plots). Significance was assessed by 999 random permutations with p < 0.05. In addition, species

occurring in at least six plots, three site groups, and having greater than 150 individuals were

classified as generalist taxa.

Herbaceous fern habitat associations

Fern species composition was similar among plots within the same parent material group

(Figure 3.5). Two distinct clusters formed in the NMDS space, one with all Rhyolite plots, and

another with all mafic-volcanic plots on the Caribbean slope (Palo Seco, Bonita, Verrugosa A and

B). Plots with high soil fertility (Pinola, Alto Frio and Hornito) did not cluster, but were correlated

with positive coordinates of the NMDS first dimension (NMDS1). Soil and light variables were significantly correlated with fern composition (Table B.2), notably moisture (r2 = 0.83, p < 0.01),

soil N:P (r2 = 0.74, p < 0.01) and total P (r2 = 0.73, p < 0.01). Herbaceous and tree fern NMDS

ordinations showed a high degree of concordance based on Procrustes analysis (m2 = 0.22; p

<0.001; Procrustes rotation correlation = 0.88), indicating that species composition-environment

relationships of herbaceous and tree ferns were significantly more similar to each other than

expected at random.

Twelve species were identified as indicator taxa based on preferences for the three classes

of soil fertility (Table 3.1). Low-fertility rhyolite sites provided seven species, of which three were

restricted to Honda sites, two were restricted to Chorro sites, and two occurred on both Chorro and

Honda sites. Intermediate fertility mafic-volcanic sites provided three species, and high-fertility

basalt/dacite sites provided a single species. All seven species restricted to low fertility sites

belonged to three families: Lindsaeaceae (1 sp.), Cyatheaceae (4 spp.) and Blechnaceae (2 spp.).

The four species from the intermediate-fertility sites were distributed into three families:

Dryopteridaceae (2 spp.), Cyatheaceae (1 sp.), and Marattiaceae (1 spp.). The single species restricted to high fertility sites belongs to the Athyriaceae family.

Numerous taxa that were not selected by the indicator analysis also had constrained distributions. Desmophlebium lechleri and Thelypteris frigida were relatively frequent at both

Chorro A and B, while Lindsaea imrayana and Danaea sp. were restricted to Chorro B but were not abundant. Parablechnum schiedeanum§ (§ symbol denotes taxa recorded at the site but not

included in the plot survey) was also observed only at Chorro. In the Honda plots we often found

the generalist species Olfersia cervina, and numerous individuals of Cyathea nigripes. Scattered

aggregations of Megalastrum biseriale, and M. pulverulentum were restricted to Honda sites. At

Samudio, which was excluded from the indicator species analysis, Diplazium macrophyllum dominates the understory where the canopy is open. Its occurrence was restricted to this site.

Dicksonia sellowiana was also found only in Samudio despite its broad distribution across Central and South America. It is an endangered species and is reportedly found in area with high rates of

canopy change (Alfonso-Moreno et al., 2011). The presence of these two taxa at Samudio may reflect a large number of treefalls in the plot that occurred in 2005 to 2006.

All plots located on the Caribbean slope of the Continental Divide (Bonita, Palo Seco,

Verrugosa A and B) are classified as having mafic-volcanic parent material, with soil N:P ranging

from 7.7 to 11.1, pH ranging from 4.5 to 5.1, and relative clay-rich soils lacking a developed

organic layer (Andersen et al., 2010; Turner and Dalling, in press). The filmy fern Trichomanes

elegans, displaying iridescent blue leaves, was recorded in all mafic volcanic sites except Palo

Seco, although it was observed in shady and sloping microsites outside the plot. Tuomisto and

Ruokolainen (1993) reported the restriction of T. elegans to clay soils in a rain forest in the

Peruvian Amazonia. Dracoglossum plantagineum and Mickelia hemiotis occurred only at the

Verrugosa A and B sites, and Thelypteris gigantea occurred only at Verrugosa A.

The two plots on the southern side of Fortuna, Hornito and Alto Frio, are the most fertile

plots in terms of extractable cations and P availability. Both sites are markedly seasonal, with

lower mean annual precipitation at Alto Frio (Dalling et al., in press). Hornito maintains several

species-rich genera including Asplenium, Thelypteris, and Tectaria, while Alto Frio has only a

species-poor assemblage. Species in common with other plots were Adiantum tetraphyllum and

Pteris altissima. Blechnum occidentale and Ctenitis submarginalis were the most abundant

species. Phanerophlebia jugladifolia§ and P. quadriaurita were found only at Alto Frio. Another

high-fertility site, Pinola, with high rainfall and basalt parent material where soil total P is 618 µg

cm-3, had three taxa restricted to it: Lastreopsis killipii, Diplazium striatastrum, and Thelypteris

eggersii.

While many taxa showed strong substrate or site preferences, we also found taxa that

showed no evidence of an association with habitat. A key species that fits in this pattern was

Didymochlaena truncatula, which occurred in most plots with high rainfall and well-drained soils.

This species has a pantropical distribution (Garcia and Salino, 2008) and was relatively abundant at Fortuna. Farias et al. (2015) tested whether phenological patterns in D. truncatula were linked to temperature and rainfall in a lower montane forest in Brazil. No correlation was found between fertility or leaf production and environmental factors, only leaf mortality and leaf growth per month were negatively correlated with rainfall. If phenological responses of D. truncatula are not correlated to specific environmental factors, further functional traits such as those regarding its gametophyte (e.g., longevity, drought resistance or light requirements) may provide novel insights.

Other species that exhibit no clear soil habitat association were Megalastrum atrogriseum,

Diplazium urtifolium, both terrestrial herbs, and the tree ferns Alsophila erinacea and Cyathea multiflora.

Based on regression trees and indicator species analysis, Zuquim et al. (2014) found that

Adiantum and Pteris species at a lowland forest located in Brazilian Amazonia were both

associated with high cation sites, in common with observations at Fortuna, where the only species

of Adiantum occurred in dacite and basalt sites. In the same study, Lindsaea species were indicators

of the poorer soils. Interestingly, distinct species from Lindsaeaceae were separated by soil type

in Fortuna. Lindsaea arcuata and L. imrayana occurred exclusively on the infertile rhyolite plots,

while L. klotzschiana§, was observed only in a fertile site with dacite parent material. Furthermore, at both dacite and rhyolite plots, Lindsaea species were frequently found in association with

Oreomunnea mexicana trees, one of the few ectomycorrhizal trees at Fortuna, and a species that

forms monodominant stands characterized by a reduced availability of inorganic nitrogen

(Corrales et al., 2016).

Overall, the observed species occurrences at Fortuna provide strong support for species

edaphic preferences across the soil nutrient gradient. Determining the response of ferns to soil

nutrients and other soil characteristics, such as particle size from plot-based plant and soil surveys

is useful because small-scale soil variation affecting plants is often absent from large-scale soil

surveys, (Zuquim et al., 2014). Furthermore, our knowledge of fern species and their soil

associations provides a platform for future work to elucidate the functional basis of soil affinities

of ferns, possibly focusing on mycorrhizal associations and/or physiological traits associated with

distinct root morphologies. Seeking an explanation for how ferns have survived for more than 400

Ma might improve our understanding of other plant groups and their relationship with soils as well.

In this section, we have illustrated the predictive power of soil conditions over the

composition of terrestrial fern species. The association of some epiphytic taxa with specific soil

and site conditions has been already demonstrated (Tuomisto, 2006). At the community level,

although it is anticipated that terrestrial ferns show greater habitat specialization than epiphytic

ones, this prediction is yet to be tested.

Tree fern habitat associations

Tree ferns are distinct from herbaceous ferns because of the development of a caudex, or

trunk-like structure. This morphological feature likely gave tree ferns a competitive advantage

over herbaceous plant groups in the light-limited understory. However, the character is not unique to the clade Cyatheaceae to which most tree ferns belong. Neither do all ferns in Cyatheaceae have a trunk-like habit (Large and Braggins, 2004). Cyatheaceae consists of 643 species (PPG 1, 2016), making them the most speciose family of ferns behind the Cyatheales. The family Cyatheaceae radiated during the late Jurassic (Korall and Pryer, 2014) and expanded its range across tropical, subtropical and temperate regions.

With a broad latitudinal range (23°N to 50°S latitude), tree ferns are present in most

rainforests around the world as an important compositional and structural unit (Brock et al. 2016).

Most tree ferns are associated with tropical montane and pre-montane cloud forests, that provide a combination of topographic and climatic features, including persistent orographic rainfall.

Temperature is also a key factor explaining tree fern distribution. With the exception of a few New

Zealand species, most tree ferns do not occur where temperatures drop below freezing (Brock et al., 2016).

The floristic survey of the 12 one-hectare plots yielded 814 individuals: 12 species and 1 morpho species in the Cyatheaceae and a single species in the Dicksoniaceae family. Cyathea rojasiana accounted for 32% of all tree fern individuals. Only two plots (Chorro A and B; rhyolite parent material) accounted for all individuals of this species. In fact, from the 258 fern individuals recorded in Chorro B, 79% were tree ferns. In contrast, Cyathea multiflora, ranked as the second most abundant tree fern, occurred in all but rhyolite sites and Alto Frio. Cyathea divergens and

Alsophila erinacea were the third- and fourth-most abundant taxa.

Procrustes analysis of the NMDS ordination of herbaceous ferns and tree ferns showed that floristic distances among taxa in these groups were significantly more similar to each other than expected by chance. Environmental factors significantly correlated with tree fern floristic composition overlapped with those of herbaceous ferns. Moisture and soil N:P obtained the highest correlation values for both tree ferns and herbaceous ferns ( Table B.2). Exchangeable iron, the R:FR ratio, and total P obtained higher correlation values for tree fern composition, while total

P, dry season rainfall, exchangeable calcium, total extractable base cations and effective cation exchange capacity obtained higher correlation values for herbaceous fern composition. To compare visually the floristic patterns of herbaceous and tree ferns, we fitted soil fertility as a

categorical variable into the NMDS ordination. Sites were assigned one of the three levels of soil

fertility variable: low (four rhyolite sites), intermediate (four mafic volcanics sites), and high (three

sites: basalt, dacite and undifferentiated volcanics). Samudio was excluded from this analysis. Soil

fertility was significantly correlated with both herbaceous and tree ferns (r2 = 0.73 , p = 0.001 and

r2 = 0.60, p = 0.023). Soil fertility levels clustered similarly in both herbaceous and tree ferns

(Figure 3.6).

Although herbaceous and tree ferns showed similar floristic association patterns, the two groups showed opposing abundance patterns in relation to soil N:P and R:FR (Viana et al., 2020).

Tree fern abundance was associated with low soil fertility sites and significantly positively

correlated with soil N:P (and therefore associated with rhyolite sites), whereas herbaceous fern

abundance was negatively correlated with high soil N:P (and therefore associated with the dacite site). A similar response of these two groups was also obtained with R:FR. Differences in the abundance of herbaceous ferns and tree ferns across parent materials suggest that coexistence of these groups in the understory might in part reflect resource partitioning.

Interestingly, the direction of tree fern species associations with P availability may be ecosystem dependent. At Fortuna, high tree fern abundance is associated with relatively infertile, low soil P sites. Similarly, in Hawaiian tropical forests, tree fern cover is lower in sites with either high N or high P concentrations (Vitousek, 2004). In contrast, in cool-temperate rainforest systems in New Zealand, tree fern abundance is higher on alluvial terrain and the early stages of chronosequences characterized by N limitation of productivity (Coomes et al., 2013; Turner et al.,

2018). Soil moisture conditions have also been associated with tree fern distributional patterns

(Brock et al., 2016). However, there is still much to reveal about the ecology of tree ferns. For

72 instance, we do not know whether having a free-living gametophyte stage influences tree fern response to distinct soil conditions.

NOTES ON SPECIES OCCURRENCE

Patterns of endemism in tropical ferns and lycophytes are similar to those of richness

(Moran, 2008) characterized by a unimodal relationship with elevation. However, endemism peaks at a higher elevation than the peak species richness (Kessler, 2010). According to Moran and Riba

(1995), 15% of Costa Rican and Panamanian ferns and lycophytes are endemic to the two countries. Binational endemism reflects the occurrence of many fern taxa in the La Amistad

International Park (PILA), a protected area that straddles the border and encompasses half of the

Talamanca mountain range. Of the five endemic fern species found at PILA, only one,

Trichomanes consanguineum, is found at Fortuna. Among lycophytes, Selaginalla corraea

Valdespino is the only species endemic to Fortuna (Valdespino, 1993). It is described as a very small plant, much like a moss in appearance, and with an epipetric or terrestrial habit (Figure 3.7).

Selaginella corraea is associated with rhyolitic rocks; with soils on steep, forested slopes; and an elevation of 1,100 to 1,200 m (I. Valdespino, personal communication, 15, November 2018), an environmental description that fits the Honda and Chorro watersheds. The occurrence of Cyathea rojasiana, one of the few tree fern species endemic to Panama, is also restricted to very wet cloud forests. The type specimen used to describe the species was collected in the vicinity of Quebrada

Bonito, one of the wettest parts of the Fortuna reserve.

Extensive field collections at Fortuna are likely to yield additional taxa. Based on recent field collections, we report new records for Panama of two fern species. The first is Bolbitis hastata, which had a previously known range that extended from Mexico to Costa Rica. This species was found in the Hornito plot in ephemeral streams beds. No other population of this

73 species has been observed in the Fortuna reserve. The second is Thelypteris hatchii, which ranges from southern Mexico to northern Costa Rica. However, T. hatchii is common under high canopy forest and gentle slopes in Hornito plot, accounting for 22% of fern individuals in that plot. We also report the occurrence of Diplazium x verapax in this area, a rare rather than endemic hybrid species resulting from the interaction between D. plantaginifolium and D. werckleanum, its likely co-occurring parents (Testo et al., 2017). Diplazium x verapax is a peculiar-looking fern; it shows an intermediate appearance between its parental progenitors (Figure 3.8). Diplazium plantaginifolium has simple leaves and D. werckleanum, pinnate leaves. This hybrid could be easily be confused with D. riedelianum, although a recent treatment of these species restricted D. riedelianum to Argentina, Brazil, and Paraguay (200-800 m), whereas D. x verapax ranges from

Mexico to Panama (650–1,690 m). Because hybrids are usually sterile, the only way to reproduce is by the means of buds located in the lamina, apparently the reason this species is reported to be rare (Testo et al., 2017). Indeed, in Fortuna, D. x verapax is found only in the Hornito area.

SUMMARY

Fortuna forests are characterized by diverse fern and lycophyte taxa. The examination of a previous list of the fern flora combined with in situ collections resulted in a 19% increase in species number. A major floristic contribution comes from the family Dryopteridaceae, comprised of epiphytic and terrestrial genera, in which the epiphytic genus Elaphoglossum – the richest in species – is included. Polypodiaceae, the second-most species-rich family, contributes exclusively epiphytic taxa. Prominent diverse terrestrial genera are the herbaceous taxa Diplazium and

Thelypteris and tree fern genus Cyathea, considered a key component of the vegetation of mid- and high-elevation cloud forests in the Neotropics.

Fern species show clear habitat association across the gradient in soil fertility and parent material at Fortuna, with 12 indicator species significantly associated with groups of sites based on parent material. Rhyolite sites provided the highest number of indicator species. Herbaceous and tree ferns showed similar compositional patterns, but their relationship with soil N:P was distinct: tree ferns were related to infertile soils (high soil N:P), while herbaceous ferns were related to fertile soils (low soil N:P).

TABLE

Table 3.1. Indicator species analysis for terrestrial ferns in 11 one-hectare plots. Plots are divided into three groups based on parent material soil fertility: rhyolite plots (Chorro A and B, Honda A and B), mafic-volcanic plots (Bonita, Palo Seco, Verrugosa A and B), and high-fertility plots

(Pinola, Hornito, and Alto Frio). Generalist taxa are ordered by abundance. Observed IV = observed indicator value; P = significance values. See Plates 1–5 for images of indicator taxa.

Observed Soil fertility conditions Indicators species P IV Low fertility specialists Lindsaea arcuata 1.00 0.006 (rhyolite sites) Cyathea divergens 0.97 0.013 Chorro specialists Alsophila salvinii 1.00 0.039 Cyathea rojasiana 1.00 0.039 Honda specialists Austroblechnum stoloniferum 1.00 0.032 Blechnum binervatum 1.00 0.032 Cyathea pinnula 0.99 0.031 Intermediate fertility specialists 1.00 0.003 (mafic volcanic sites) Danaea moritziana Stigmatopteris heterophlebia 1.00 0.003 Mickelia oligarchica 0.97 0.022 Cyathea eggersii 0.87 0.048 High fertility specialists (basal, dacite and undifferentiated volcanic Diplazium plantaginifolium 0.816 0.049 sites) Generalists Didymochlaena truncatula Diplazium urticifolium Megalastrum atrogriseum Olfersia cervina

FIGURES

Figure 3.1. Location of the 12 one-hectare plots surveyed at Fortuna and Palo Seco forests.

Figure 3.2. Species richness of Fortuna ferns species by family. Data source: McPherson et al.

(2010) and Viana field collections.

Figure 3.3. Distribution of fern species habits across fern families in Fortuna.

Epiphyte/terrestrial/rupicolous species can adopt epiphytic, rupicolous, or terrestrial habits; epiphyte/terrestrial species can be epiphytic or terrestrial.

Figure 3.4. (Top) Species richness and Fisher’s alpha diversity. (Bottom) Tree fern, herbaceous

fern, and total fern abundance across 12 one-hectare plots at Fortuna. Sites were ordered from low- fertility sites on the left to high-fertility sites on the right (Prada et al., 2017). See Plates 1–13 for images of the floristic composition of fern species from Fortuna.

Figure 3.5. The first two dimensions of a nonmetric multidimensional scaling analysis of the

floristic composition of 77 fern species collected from 12 one-hectare plots. Sites are color coded by geological substrate. Herbaceous and tree fern species scores are color and shape coded. Vectors indicate environmental variables significantly correlated with the fern community (p ≤ 0.05).

Confidence ellipses (95% confidence limit) are shown for geological substrates with more than

three plots. Total N:P (soil N:P), total nitrogen to phosphorus ratio; Total P, total phosphorus; BD,

bulk density; Mg, magnesium; ECEC, effective cation exchange capacity; TEB, total

exchangeable bases; Ca, calcium; Mn, manganese; R:FR, red to far-red light ratios; Fe, iron.

Figure 3.6. The first two dimensions of nonmetric multidimensional scaling analysis using raw abundance data. Confidence ellipses (95% confidence limit) were calculated for soil fertility categorical variables with three levels: low (rhyolite geological substrates), intermediate (mafic- volcanic), and high (basalt, dacite and undifferentiated volcanics). The Samudio plot, which includes both rhyolite and mafic microsites, was excluded. Confidence ellipses are shown for both tree ferns (T) and herbaceous ferns (H). Site scores are based on the floristic composition of the herbaceous ferns. Site codes are colored according to the geological substrate indicated in the legend.

Figure 3.7. Selaginella correae, a lycophyte species endemic of Fortuna. Morphological features of S. correae. 1: (a) plant habit, (b) median leaf, (c) lateral leaf hairs, (d) fertile portions (strobili) of the stem, (e and f) lower and upper surface of stem. Modified from Valdespino (1993). 2:

Exsiccate of the specimen collection from the New York Botanical Garden (I.A. Valdespino 1335 with E. Ríos-Levy, V. Young & D. Chacón, NY).

Figure 3.8. Silhouettes of four Diplazium taxa. (a) Diplazium plantagineum (G.J.Gastrony et al.

651, MO). (b) Diplazium werckleanum (I.deMartinez 58, MO). (c) D. x verapax (W.L.Testo 1045,

NY). (d) D. riedelianum (E.Schmalz s.n.,VT). (e) Reproductive leaf of D. x verapax from a specimen collected at Fortuna.

REFERENCES

Alfonso-Moreno, R. A., Cadena-Vargas, C. E., Morales, G., Peña, N., and Pérez, B. (2011).

Conservación Integral de Dicksonia sellowiana Hook., en Bogotá D.C. y su Área de

influencia. Rev. Acad. Colomb. Cienc., 134(35), 79–96.

Andersen, K. M., Turner, B. L., and Dalling, J. W. (2010). Soil-based habitat partitioning in

understorey palms in lower montane tropical forests. Journal of Biogeography, 37(2), 278–

292. https://doi.org/10.1111/j.1365-2699.2009.02192.x

Brock, J. M. R., Perry, G. L. W., Lee, W. G., and Burns, B. R. (2016). Tree fern ecology in New

Zealand: A model for southern temperate rainforests. Forest Ecology and Management,

375, 112–126. https://doi.org/10.1016/j.foreco.2016.05.030

Coomes, D. A., Bentley, W. A., Tanentzap, A. J., and Burrows, L. E. (2013). Soil drainage and

phosphorus depletion contribute to retrogressive succession along a New Zealand

chronosequence. Plant and Soil, 367(1–2), 77–91. https://doi.org/10.1007/s11104-013-

1649-5

Corrales, A., Mangan, S. A., Turner, B. L., and Dalling, J. W. (2016). An ectomycorrhizal

nitrogen economy facilitates monodominance in a neotropical forest. In Ecology Letters

(Vol. 19, Issue 4, pp. 383–392). https://doi.org/10.1111/ele.12570

Dalling, J. W., Prada, C. M., Heineman, K. D., Andersen, K. M., Morris, A., Pardo, J.,

Hollenbeck, E., Caballero, P., Espinosa, C., García, E., and Turner, B. L. (n.d.-a).

Introduction to the Fortuna Forest Reserve: soil and climate effects on the tree communities

of a premontane tropical forest. In Smithsonian Contributions to Botany.

Dalling, J. W., Prada, C. M., Heineman, K. D., Andersen, K. M., Morris, A., Pardo, J.,

Hollenbeck, E., Caballero, P., Espinosa, C., García, E., and Turner, B. L. (in press).

Introduction to the Fortuna Forest Reserve: Soil and Climate Effects on the Tree

Communities of a Premontane Tropical Forest. In J. W. Dalling and B. L. Turner (Eds.),

Fortuna Forest Reserve, Panama: Interacting Effects of Climate and Soils on the Biota of a

Wet Premontane Tropical Forest (Smithsonia). Smithsonian Institution Scholarly Press.

De Caceres, M. (2013). How to use the indicspecies package (ver. 1.7.1). http://cran.r-

project.org/web/packages/indicspecies/vignettes/indicspeciesTutorial.pdf

De Cáceres, M., Legendre, P., and Moretti, M. (2010). Improving indicator species analysis by

combining groups of sites. Oikos, 119(10), 1674–1684. https://doi.org/10.1111/j.1600-

0706.2010.18334.x

Dufrêne, M., and Legendre, P. (1997). Species Assemblages and Indicator Species: The Need for

a Flexible Asymmetrical Approach. Ecological Monographs, 67(3), 345–366.

https://doi.org/10.1890/0012-9615(1997)067[0345:SAAIST]2.0.CO;2

Farias, R. de P., Costa, L. E. N., Silva, I. A. A., and Barros, I. C. L. (2015). Phenological studies

of selected leaf and plant traits of Didymochlaena truncatula (Dryopteridaceae) in a

Brazilian submontane tropical rainforest. Nordic Journal of Botany, 33(2), 249–255.

https://doi.org/10.1111/njb.00656

Garcia, P. A., and Salino, A. (2008). Dryopteridaceae (Polypodiopsida) no estado de Minas

Gerais, Brasil. Lundiana, 9(1), 3–27.

Greer, G. K., and McCarthy, B. C. (1999). Gametophytic Plasticity among Four Species of Ferns

with Contrasting Ecological Distributions. International Journal of Plant Sciences, 160(5),

879–886. https://doi.org/10.1086/314188

Holdridge, L. (1947). Determination of world plant formations from simple climatic data.

Science, 105, 367–368. http://science.sciencemag.org/content/105/2727/367.short

Holmlund, H. I., Davis, S. D., Ewers, F. W., Aguirre, N. M., Sapes, G., Sala, A., and Pittermann,

J. (2020). Positive root pressure is critical for whole-plant desiccation recovery in two

species of terrestrial resurrection ferns. Journal of Experimental Biology, 71(3), 1139–1150.

https://doi.org/10.1111/nph.16067

Jones, M. M., Cicuzza, D., van Straaten, O., Veldkamp, E., and Kessler, M. (2014).

Determinants of fern and angiosperm herb community structure in lower montane rainforest

in Indonesia. Journal of Vegetation Science, 25, 1216–1224.

https://doi.org/10.1111/jvs.12181

Jones, M. M., Ferrier, S., Condit, R., Manion, G., Aguilar, S., and Erez, R. P. (2013). Strong

congruence in tree and fern community turnover in response to soils and climate in central

Panama. Journal of Ecology, 101, 506–516. https://doi.org/10.1111/1365-2745.12053

Jones, M. M., Ruokolainen, K., Martinez, N. C. L., and Tuomisto, H. (2016). Differences in

topographic and soil habitat specialization between trees and two understorey plant groups

in a Costa Rican lowland rain forest. Journal of Tropical Ecology, 32(6), 482–497.

Jones, M. M., Szyska, B., and Kessler, M. (2011). Microhabitat partitioning promotes plant

diversity in a tropical montane forest. Global Ecology and Biogeography, 20, 558–569.

https://doi.org/10.1111/j.1466-8238.2010.00627.x

Jones, M. M., Tuomisto, H., Clark, D. B., and Olivas, P. (2006). Effects of mesoscale

environmental heterogeneity and dispersal limitation on floristic variation in rain forest

ferns. Journal of Ecology, 94, 181–195. https://doi.org/10.1111/j.1365-2745.2005.01071.x

Kessler, M. (2010). Biogeography of ferns. In K. MEHLTRETER, L. R. WALKER, and J. M.

SHARPE (Eds.), Fern Ecology. Cambridge University Press. https://doi.org/10.5167/uzh-

42711

Klekowski, E. J. (1979). The genetics and reproductive biology of ferns. In A. F. Dyer (Ed.), The

experimental biology of ferns (pp. 133–170). Academic Press.

Kluge, J., and Kessler, M. (2006). Fern endemism and its correlates: Contribution from an

elevational transect in Costa Rica. Diversity and Distributions, 12(5), 535–545.

https://doi.org/10.1111/j.1366-9516.2006.00231.x

Korall, P., and Pryer, K. M. (2014). Global biogeography of scaly tree ferns (Cyatheaceae):

Evidence for Gondwanan vicariance and limited transoceanic dispersal. Journal of

Biogeography, 41(2), 402–413. https://doi.org/10.1111/jbi.12222

Kramer, K. U., and Green, P. S. (1990). Pteridophytes and Gymnosperms. In K. Kubitzki (Ed.),

The Families and genera of Vascular Plants. Springer-Verlag. https://doi.org/10.1007/978-

3-662-02604-5

Kreft, H., Jetz, W., Mutke, J., and Barthlott, W. (2010). Contrasting environmental and regional

effects on global pteridophyte and seed plant diversity. Ecography, 33(2), 408–419.

https://doi.org/10.1111/j.1600-0587.2010.06434.x

Large, M. F., and Braggins, J. E. (2004). Tree Ferns. Timber Press: Portland.

McPherson, Gordon Merello, M., and Fica, M. (n.d.). A Checklist of the Fortuna Dam Region,

Chiriqui/Bocas del Toro, Panama. Retrieved January 3, 2019, from

http://www.mobot.org/MOBOT/fortuna/fortunaDropdown.aspx

McPherson, G., Merello, M., and Fica, M. (2010). A checklist of the Fortuna dam region,

Chiriqui/Bocas del Toro, Panama. www.mobot.org/MOBOT/Research/fortuna.shtml

Moran, R. C. (2008). Diversity, Biogeography, and floristics. In T. A. Ranker and C. H. Haufler

(Eds.), Biology and Evolution of Ferns and Lycophytes (pp. 367–394). Cambridge

University Press. https://doi.org/10.1017/CBO9780511541827.015

Moran, R. C., and Riba, R. (1995). Psilotaceae a Salviniaceae. In M. Davidse, S. Sousa, and S.

Knapp (Eds.), Flora Mesoamericana. Volumen 1. (Volumen 1). Univ. Nacional Autónoma

de México, Ciudad Universitaria. http://www.mobot.org/mobot/fm/ ]

Nitta, J. H., Taputuarai, R., Meyer, J., and Davis, C. C. (2017). Life cycle matters: DNA

barcoding reveals contrasting community structure between fern sporophytes and

gametophytes. Ecological Monographs, 87(2), 278–296. https://doi.org/10.1111/ijlh.12426

Page, C. (2002). Ecological Strategies in Fern Evolution; a neoteridological overview. Reveiw of

Palaeobotany & Palynology, 119, 1–33.

Phillips, T. L. (1974). Evolution of Vegetative Morphology in Coenopterid Ferns. Annals of the

Missouri Botanical Garden, 61(2), 427–461. https://doi.org/10.2307/2395067

Prada, C. M., Morris, A., Andersen, K. M., Turner, B. L., Caballero, P., and Dalling, J. W.

(2017). Soils and rainfall drive landscape-scale changes in the diversity and functional

composition of tree communities in premontane tropical forest. Journal of Vegetation

Science, 28(4), 859–870. https://doi.org/10.1111/jvs.12540

Schneider, H., Schuettpelz, E., Pryer, K. M., Cranfill, R., Magallón, S., and Lupia, R. (2004).

Ferns diversified in the shadow of angiosperms. Nature, 428(6982), 553–557.

https://doi.org/10.1038/nature02361

Schuettpelz, E., and Pryer, K. M. (2007). Fern phylogeny inferred from 400 leptosporangiate

species and three plastid genes. Taxon, 56(4), 1037–1050. https://doi.org/10.2307/25065903

Schuettpelz, E., and Pryer, K. M. (2009). Evidence for a Cenozoic radiation of ferns in an

angiosperm-dominated canopy. Proceedings of the National Academy of Sciences, 106(27),

11200–11205. https://doi.org/10.1073/pnas.0811136106

Sessa, E. B., Testo, W. L., and Watkins, J. E. (2016). On the widespread capacity for, and

functional significance of, extreme inbreeding in ferns. New Phytologist, 211(3), 1108–

1119. https://doi.org/10.1111/nph.13985

Silva, C., Cárdenas, D., Londoño, L., Turner, B. L., and Jaramillo, C. (in press). Geological Map

of the Fortuna Forest Reserve. In J. W. Dalling and B. T. Turner (Eds.), Fortuna Forest

Reserve, Panama: Interacting Effects of Climate and Soils on the Biota of a Wet

Premontane Tropical Forest (Smithsonia). Smithsonian Institution Scholarly Press.

Testo, W. L., Sundue, M., and Vasco, A. (2017). Diplazium hybrids involving D.

plantaginifolium and D. ternatum from Mexico and Central America. Brittonia, 69(3), 295–

306. https://doi.org/10.1007/s12228-017-9476-2

Testo, W., and Sundue, M. (2016). A 4000-species dataset provides new insight into the

evolution of ferns. Molecular Phylogenetics and Evolution, 105, 200–211.

https://doi.org/10.1016/j.ympev.2016.09.003

Tuomisto, H. (2006). Edaphic niche differentiation among Polybotrya ferns in western

Amazonia: implications for coexistence and speciation. Ecography, 29(3), 273–284.

https://doi.org/10.1111/j.2006.0906-7590.04390.x

Tuomisto, H., and Poulsen, A. D. (1996). Influence of edaphic specialization on pteridophyte

distribution in neotropical rain forests. Journal of Biogeography, 23(3), 283–293.

http://www.jstor.org/stable/2845846

Tuomisto, H., and Ruokolainen, K. (1993). Distribution of Pteridophyta and Melastomataceae

along an edaphic gradient in an Amazonian rain forest. Journal of Vegetation Science, 4(1),

25–34. https://doi.org/10.2307/3235634

Turner, B. L., and Dalling, J. W. (in press). The Soils of Fortuna. In J. W. Dalling and B. L.

Turner (Eds.), Fortuna Forest Reserve, Panama: Interacting Effects of Climate and Soils on

the Biota of a Wet Premontane Tropical Forest (Smithsonia). Smithsonian Institution

Scholarly Press.

Turner, B. L., Wells, A., Andersen, K. M., and Condron, L. M. (2018). Consequences of the

physical nature of the parent material for pedogenesis, nutrient availability, and succession

in temperate rainforests. Plant and Soil, 423(1–2), 533–548. https://doi.org/10.1007/s11104-

017-3514-4

Valdespino, I. A. (1993). Notes on Neotropical Selaginella (Selaginellaceae), Including New

Species from Panama. Brittonia, 45(4), 315–327. https://doi.org/10.2307/2807605

Vasco, A., Moran, R. C., and Ambrose, B. a. (2013). The evolution, morphology, and

development of fern leaves. Frontiers in Plant Science, 4, 1–16.

https://doi.org/10.3389/fpls.2013.00345

Viana, J. L., Turner, B. L., and Dalling, J. W. (2020). Compositional variation in understory fern

and palm communities along a soil fertility and rainfall gradient in a lower montane tropical

forest. Journal of Vegetation Science, jvs.12947. https://doi.org/10.1111/jvs.12947

Vitousek, P. M. (2004). Nutrient Cycling and Limitation: Hawai’i as a Model System. Princeton

University Press.

Watkins, J. E., Cardelús, C., Colwell, R. K., and Moran, R. C. (2006). Species richness and

distribution of ferns along an elevational gradient in Costa Rica. American Journal of

Botany, 93(1), 73–83. https://doi.org/10.3732/ajb.93.1.73

Watkins, J. E., Churchill, A. C., and Holbrook, N. M. (2016). A site for sori: Ecophysiology of

fertile–sterile leaf dimorphy in ferns. American Journal of Botany, 103(5), 845–855.

https://doi.org/10.3732/ajb.1500505

Wolf, P. G., Robison, T. A., Johnson, M. G., Sundue, M. A., Testo, W. L., and Rothfels, C. J.

(2018). Target sequence capture of nuclear-encoded genes for phylogenetic analysis in

ferns. Applications in Plant Sciences, 6(5), 1148. https://doi.org/10.1002/aps3.1148

Zuquim, G., Tuomisto, H., Jones, M. M., Prado, J., Figueiredo, F. O. G., Moulatlet, G. M.,

Costa, F. R. C., Quesada, C. A., and Emilio, T. (2014). Predicting environmental gradients

with fern species composition in Brazilian Amazonia. Journal of Vegetation Science, 25(5),

1195–1207. https://doi.org/10.1111/jvs.12174

CHAPTER 4: TRAIT DISPERSION AND PHYLOGENETIC RELATEDNESS IN A GUILD OF TROPICAL TERRESTRIAL FERNS

ABSTRACT

An essential goal of functional ecology is to identify the degree of trait-dependency in species occurrence patterns in relation to environmental conditions, which in combination with phylogenetic data improves the understanding of major assembling mechanisms. Fern community composition responds to nutrient limitation gradients, but it is unknown whether functional traits are influenced similarly, and which individual traits are most strongly associated with these gradients. We hypothesized that nutrient limitation acts as a strong environmental filter, structuring trait and phylogenetic diversity in fern communities. We asked (a) whether functional trait composition values and functional dispersion values and (b) trait and phylogenetic diversity decrease towards sites with lower nutrient availability. We measured 15 functional traits from 33

fern species occurring in 12 one-ha plots across nutrient limitation and rainfall gradients in a lower

montane tropical rainforest in Panama. We applied RLQ and fourth-corner analyses to assess trait-

environment relationships, used beta regression to evaluate functional dispersion responses to

environmental factors, and investigated trait diversity distributions with respect to their phylogeny

to link ecological processes and lineage-dependent evolutionary factors across the community.

Functional composition was predicted by soil variables and dry season rainfall. Leaf traits

contributed to most of the variation among all trait variables, with leaf phosphorus (leaf P) and

leaf carbon to nitrogen ratio (leaf C:N) decreasing with increasing soil nitrogen to phosphorus ratio

(soil N:P). Functional dispersion decreased with soil N:P and Mn, and increased with dry season

soil moisture, suggesting that low soil fertility and high dry season moisture stress can affect

functional diversity in similar ways. Trait and phylogenetic diversity showed clustered patterns, as

a direct effect of phylogenetic constrains. Our results indicated that environmental filtering

influences trait assembly of terrestrial fern communities, with functional composition and

functional dispersion responding similarly to environmental gradients, decreasing towards the

extreme ends of the soil fertility gradient.

INTRODUCTION

Ecologists have long sought to understand the mechanisms by which species assemble into

communities and the rules that may explain species coexistence (Cody and Diamond, 1975;

MacArthur and Wilson, 1967; Volterra, 1926). Trait-based approaches hold particular promise for explaining the controls on coexistence since trait distribution patterns reflect the interplay between evolutionary and ecological assembly processes, and respond to the effects of both biotic and abiotic interactions (Tilman, 1982; Valladares et al., 2007).

Contrasting trait dispersion patterns within communities, such as clustering (non-random

accumulation of similar trait values) and overdispersion (non-random accumulation of different

trait values), have been employed to interpret the relative importance of the ecological processes

influencing species abundance and phylogenetic structure (Weiher and Keddy 1995; Kraft and

Ackerly 2010). The theoretical predictions of trait dispersion have indicated that environmental filtering is expected when strong environmental conditions select for species with similar traits based on shared environmental requirements. On the other hand, limiting similarity is expected when intense interspecific competition constrains the ability of species with similar environmental and biotic requirements to co-occur (Kraft and Ackerly, 2010; Weiher and Keddy, 1995).

However, the idea that competition only results in overdispersed traits has been challenged

(Mayfield and Levine, 2010), possibly leading to trait clustering when the results of competition

removes species that have more distinct trait values and are less phylogenetic related other the

other species.

Similarly, the combination of phylogenetic and trait data can improve our understanding

of observed trait-environment relationships (Cadotte et al., 2009; Cavender-Bares et al., 2004;

Leibold et al., 2010) and facilitate our understanding of the relative contributions of ecological processes under which communities assemble (Webb et al., 2002). Furthermore, contrasting predictions may rise when evolutionary and ecological processes interact to determine the structure of communities (Pausas and Verdú, 2010). For instance, co-occurring species exhibiting a combination of phylogenetic overdispersion and high functional trait diversity might result from

competition limiting the coexistence of closely related species, whereas co-occurring species

exhibiting a combination of the phylogenetic clustering and low functional trait diversity indicates

the influence of environmental filtering (Grime, 2006; Weiher et al., 1998). Thus, testing for

phylogenetic signal allows for the interpretation of evolutionary influences on communities and

provides further information on how functional trait diversity is distributed across phylogeny (de

Bello et al., 2015).

For fern communities in lower montane forest in Panama, previous work has shown that

compositional turnover is linked to changes in nitrogen and phosphorus availability with

herbaceous ferns and tree ferns exhibiting distinct abundance patterns. While herbaceous ferns

were abundant in sites with lower soil nitrogen to phosphorus ratio (soil N:P) and high total soil

phosphorus, tree ferns were abundant in the opposite soil conditions (Viana et al., 2020). Here, we

combine trait and phylogenetic approaches to determine the importance of niche-based processes

in determining the trait assembly of these fern communities. Compared with woody taxa,

phylogenetic and functional traits of fern communities along environmental gradients have

received much less attention (Carvajal-Hernández et al., 2018; Dai et al., 2020; Kluge and Kessler,

2011; Tanaka and Sato, 2015). These studies are mainly focused on the effects of elevation (Kluge

& Kessler, 2011; Tanaka & Sato, 2015), disturbance (Dai et al., 2020), or a combination of both

(Carvajal-Hernández et al., 2018). We are unaware of other studies looking at fern functional traits along soil fertility gradients. However, as a dominant component of a warm temperate rainforest from New Zealand, tree ferns have been included among the study of 13 functional traits among

30 tree species across soil fertility gradients (Jager et al., 2015).

More specifically, we focus on determining the role of environmental filtering in

determining trait-environment relationships, and the trait and phylogenetic structure of terrestrial

fern assemblages, as environmental filtering has been found to influence fern, understory palm,

and canopy tree species turnover in our study area (Andersen et al., 2010; Prada et al., 2017).

Although the composition of fern species assemblages is related to abiotic factors in our study site,

the degree of trait-dependency in species occurrence patterns remains an open question.

Furthermore, high plasticity in functional traits in ferns in response to abrupt changes

environmental conditions (Zhu et al., 2016), combined with a high dispersal ability may result in

distinct patterns from those of angiosperm trees. We measure community weighted mean of traits

values (sensu Lavorel et al. 2008), which determines the mean trait value of a community while

accounting for species relative abundances, functional dispersion (sensu Laliberté & Legendre

2010), and Rao’s quadratic entropy based on Botta‐Dukát (2005), to evaluate the phenotypic and phylogenetic structure of fern communities, thereby determining the responses of functional traits to environmental conditions. We hypothesize that nutrient limitation at our study acts as a strong environmental filter, structuring trait and phylogenetic diversity in fern communities. We asked

(a) whether functional trait composition change across the environmental gradient (b) whether

functional dispersion values and (c) trait and phylogenetic diversity decrease towards sites with

lower nutrient availability.

MATERIAL AND METHODS

Study site

Our study is based on a species distribution analysis of 77 terrestrial fern species recorded

in 12 permanent one-hectare plots established across two humid to super-humid rainforest

(Holdridge, 1947) reserves: Fortuna Forest Reserve and the contiguous Palo Seco Protected Forest in western Panama (Figure 4.1). The plots occupy a natural gradient in rainfall, and in nutrient availability determined by the underlying parent material (Figure 4.1; Table C.1). We categorized sites according to three levels of soil fertility (high, intermediate and low) based on effective cation exchange capacity (ECEC) which co-varies with resin extractable P (resin P) and available nitrogen in the form of ammonium and nitrate (Figure C.1). High fertility sites combine high values of ECEC and high values of nitrogen and phosphorus and but low values of total soil N:P. Mean annual rainfall ranges between 4,600 and 6,300 mm. Elevation ranges from 700 and 1,500 m asl and includes both Pacific and Caribbean slopes as well as a central valley surrounding Lake

Fortuna. Fog and cloud cover are frequent, varying with elevation and rainfall (Cavelier et al.,

1996; Dalling et al., in press). For brevity the two sites are referred to collectively as Fortuna.

Environmental variables

We obtained 21 soil and rainfall variables from Prada et al. (2017) (Tables C.2). Soil and rainfall variables were associated with compositional turnover of both the overstory tree community and understory palm community in previous studies (Andersen et al., 2010; Prada et al., 2017). In Fortuna, soils vary in soil pH, nitrogen (N), phosphorus (P) and ECEC among plots and parent materials. Plots also vary in the amount and seasonality in rainfall and cloud cover,

ranging from aseasonal conditions in Caribbean slope plots (Palo Seco, Verrugosa A and B),

occasional dry periods (monthly rainfall <100 mm) in sites surrounding Lake Fortuna, and a longer

dry season on the plot on the Pacific slope (Alto Frio; Table 4.1). Soil variables were analyzed from 13 soil samples collected at 0-10 cm soil depth in each of the 12 plots. Soil samples were analyzed for pH, bulk density, total soil carbon, N and P, extractable inorganic N (NH4 and NO3),

resin extractable P (resin P), extractable cations (Al, Ca, Fe, K, Mg, Zn), ECEC ( Al, Ca, Fe, K,

Mg, Mn, Na), and total exchangeable bases (TEB - Ca, Mg, K, Na). In addition, we measured

gravimetric soil moisture content during the mid-dry season (Feb-Mar) in 15 soil samples in each

plot.

Rainfall variables were collected once every two weeks for seven years (2007 to 2009 and

2011 to 2014) in a large gap adjacent to a subset of plots. For Bonita, Chorro B, and Verrugosa A

& B, we used rainfall data from the nearest plots (<3 km away and with comparable slope position

following Prada et al., (2017). For two plots (Pinola and Alto Frio) rainfall data were only available

from 2013 and 2014. We measured light conditions through canopy openness using hemispherical

photographs. A total of 15 photos were taken for each plot at a constant height of 1 m using Opteka

fisheye lens attached to a Nikon D3300 camera. All photos were taken under uniformly cloudy

conditions. Canopy openness was analyzed with Gap Light Analyzer v. 2.0 (Frazer et al., 2000).

Soil variables values for plots were obtained by taking the average from the values obtained from

the 13 soil samples (and 15 samples for soil moisture) collected in each plot.

Trait data

Prior to trait sampling, fern species abundances were determined in fifteen 25 m2 subplots

in each of the 12 one-ha plots. Subplots were spaced 20 m apart along three 5 m wide transects

each 30 m apart. We selected the five to six most abundant species representing 16.81% to 92.47%

of fern individuals from each plot (which varied in richness and abundance) to ensure a reasonable

representation of trait values. We measured three vegetative, one regenerative and eleven leaf functional traits (Table C.2) on six to ten individuals occurring within the plots and in similar light

conditions following Cornelissen et al. (2003). Growth form is associated with plant growth strategies and may reflect habitat association (Amatangelo and Vitousek, 2008; Brock et al., 2016).

Clonality and dimorphism can affect plant community assembly and leaf carbon cost, respectively

(Britton and Watkins, 2016; Herben et al., 2014; Klimeš et al., 1997). Leaf production and plant

maximum height are whole plant traits related with both competitive ability of individuals and

with trade-offs between plant height and environmental stress (Chabot and Hicks, 1982; Grime,

1991). Leaf nutrient content is associated with mass-based maximum photosynthetic rate (leaf

carbon – leaf C), and resource-use efficiency (Wright et al., 2001; 2004). Trait values for the

species were averaged across subplots.

Functional dispersion

For each site, we calculated a multidimensional index of functional dispersion (FDis) to

represent functional diversity. Functional dispersion measures how species are dispersed in trait

space, taking into account the relative abundances of species by computing the weighted centroid

of the species × trait matrix and is not affected by the number of species (Laliberté and Legendre,

2010). FDis can be used to model changes in functional diversity over environmental gradients in

beta regression models. Gower distance was used to calculate the functional distance between the

species (Legendre and Legendre, 1998; Podani, 1999). To compute FDis we used the function

dbFD in the FD package (Laliberté et al., 2014; 2010).

Phylogenetic tree

To construct a phylogenetic tree for our taxa, we retrieved rbcl sequences (Table C.3) from

GenBank (Benson et al., 2009). To deal with missing phylogenetic information for 9 of the 33

species in our data, we used phyndr (swap algorithm) and taxonlookup (builds a taxonomic table

for vascular plants) packages (Pennell et al., 2016). The objective is to maximize the overlap of

phylogenetic data by swapping taxa lacking phylogenetic information (without rbcl sequences) to

phylogenetically related species using topological and taxonomic information. This approach

minimizes the increased or decreased phylogenetic signal as a result of the arbitrary selection of

congeneric species used as replacements. We applied phyndr functions to a published time- calibrated phylogeny of fern species (Testo and Sundue, 2016). Sequence alignment was done using the MUSCLE algorithm (Edgar, 2004b) and maximum likelihood methods were used to estimate the phylogenetic tree. We selected the following parameters for tree estimation: 1) the

Jukes-Cantor substitutions model, 2) Gamma distributed with Invariant Sites (G+I) as the option for substitution rates among sites, and 3) partial deletion as a method for dealing with gaps and

missing data. We estimated the reliability of our phylogenetic tree by bootstrapping with 999

replications. We used MEGA X: Molecular Evolutionary Genetics Analysis across computing

platforms (Kumar et al., 2018) for sequence alignment and tree estimation. The final phylogenetic

tree (Figure C.2) was mostly constrained at the family level by the Pteridophyte Phylogeny Group

I classification (PPG I, 2016).

100 a

Statistical analysis

Functional composition and environmental predictors

We combined RLQ (Dolédec et al., 1996) and fourth-corner analyses (Dray et al., 2014;

Dray and Legendre, 2008) to examine the association between functional traits and environmental predictors. The RLQ analysis is an exploratory ordination method that provides a detailed overview of trait-environment associations. Hypotheses for the relationship between functional composition (expressed as the CWM - community-weighted mean of traits) and environmental variables are assessed through permutation. For the RLQ, we performed separate ordinations for each data table before running RLQ analysis: a principal components analysis (PCA) on the standardized environmental variables matrix (R table), an unconstrained correspondence analysis

(CA) on the species-abundance matrix (L table), and a Hill and Smith ordination (Hill and Smith,

1976) on the species traits matrix (Q table) for mixed quantitative and binary variables (Dray et al., 2003, 2014; Dray and Legendre, 2008). The PCA (R table) and Hill and Smith (Q table) ordinations were respectively weighted by the sites scores and by the species scores weights derived from the CA. We compared the scores from the separate ordinations to the RLQ generated scores to determine the strength of environment-trait relationships and the amount of variation considered by the RLQ analysis. We ran 4999 permutations to test for the global significance of the trait-environment relationship based on the total inertia of the RLQ analysis.

We also performed a fourth-corner analysis to test the direct links between individual traits and environmental predictors, and traits/environmental predictors and RLQ axes. The fourth- corner approach applied here was proposed by Dray and Legendre (2008), in which the concept is to predict species abundance as a function of environmental variables, species traits and their interaction. The significance of these relationships was tested using a permutation procedure that

101

included two permutation models: Model 2 and Model 4, as suggested in Dray & Legendre (2008).

The former tests whether the environment influences the distribution of species with fixed traits

by permuting the sites; the latter tests whether the species composition of sites with fixed

environmental conditions is influenced by the species traits. The models were combined to obtain

the correct level of Type I error. We used 4,999 permutations and the false discovery rate method

(FDR) to adjust p-values for multiple testing issue. All data were analyzed using the ade4 library

(Dray and Dufour, 2007) in the open source R environment version 3.6.1 (R Core Team, 2019).

Functional diversity and environmental predictors

Functional trait dispersion (FDis) indicates the degree of separation of individuals in the space shaped by functional traits, and is analogous to functional diversity (Laliberté and Legendre,

2010). In our study, a multivariate FDis value was calculated for each of our sites. FDis was then modeled with beta distributions using model averaging approaches to test the association between

FDis and environmental predictors. Beta regression is appropriate to model continuous response variables that assume values in the standard unit interval (Ferrari and Cribari-Neto, 2004), which is the case of our response variable – FDis. To avoid multicollinearity, we performed pairwise

Pearson correlation on quantitative traits and environmental variables before model construction and removed highly correlated variables (|r| > 0.7), except for dry season soil moisture (DSS moisture) and soil potassium (K). Dry season rainfall, canopy openness, DSS moisture, bulk density (BD), extractable resin phosphorus (resin P), soil N:P, K and manganese (Mn) were kept for model construction. Then, environmental variables were standardized to mean of zero and variance of one. We used model-averaging to assess the relative importance of each of the variables, an approach that takes into account model uncertainty and increases the robustness of the parameter estimates (Grueber et al., 2011). The betareg package (Cribari-Neto and Zeileis,

102

2010) was used to fit a global model that included all the selected environmental variables cited

above. The dredge function of the MuMIn package (Barton, 2020) was applied to generate a model

selection table of models with combinations of fixed effect terms in a global model. Then, we compared models based on AICc (Akaike's Information Criterion with correction for small sample size – (Hurvich and Tsai, 1989) to select the “best” models where the Akaike difference, delta, was less than 2 (ΔAICc < 2). Before proceeding with model averaging, we examined the predictor variables for collinearity by calculating variance inflation factors. Next, we applied the model.avg function to generate averaged parameter coefficients from the selected models and to quantify the relative importance of predictors. The sum of Akaike weights across selected models was used to quantify the relative importance of the variables. Zero weight was given to models lacking a given parameter (Burnham and Anderson, 2002).

Trait and phylogenetic diversity patterns

We evaluated whether there was phylogenetic and trait structuring at Fortuna. First, we measured trait diversity from the global trait distances among the 15 traits recorded in 33 species, accounting for their relative abundance, from all sites. We used the Rao’s quadratic entropy to calculate trait diversity (Rao, 1982), which then is decomposed across the species phylogeny nodes using the topological information from the phylogenetic tree, a methodology proposed by Pavoine et al. (2010). The main goal is to characterize key among-clade variations in the trait states that contribute significantly to the overall trait diversity of the species considered. In addition, trait and phylogenetic clustering or overdispersion can be detected. Each node of the phylogenetic tree has a trait diversity value (node contribution to trait diversity), with the sum of all internal nodes equal to the total trait diversity, and trait diversity of a clade the sum of its internal nodes. Branch length information was then integrated to test for phylogenetic signal. By partitioning trait diversity

103

among nodes, we could determine whether trait diversity values were accumulated (a) on a phylogenetic root (Ro test = root/tips skewness test); (b) on a single node (SN test = single‐node skewness test ), (c) or on a few nodes (FN test = few‐nodes skewness test) (Pavoine et al. 2010).

We used 999 permutations to test for significance. We applied the same approach on individual

traits. These tests were adapted by Pavoine et al. (2010) from the Pst statistic, representing the proportion of the overall phylogenetic diversity expressed among sites (Hardy and Senterre, 2007).

The PQE β test (for phylogeny) and TQE β test (for traits), as referred by Pavoine et al. (2010), evaluates the differences among local communities in relation to phylogenetic composition using quadratic entropy (QE) on phylogenetic distances among species. Likewise, the TQE β test uses

QE on trait distances (see details on Pavoine et al. 2010). These tests measure whether phylogenetic and trait composition within local sites is lower than expected from the pool of species and determines the degree of environmental filtering (versus limiting similarity). The null models used in PQE β and TQE β tests allow a determination of whether there are significant

differences among local communities with respect to phylogenetic and trait composition by the

permutation of the species across the tips of the phylogenetic tree. Both PQE and TQE β tests were

performed using all traits from all sites. In addition, we performed PQE and TQE β tests in sets of

sites of three levels of soil fertility: low-fertility sites (Chorro A and B, Honda A and B),

intermediate-fertility sites (Samudio, Verrugosa A and B, Palo Seco, and Bonita), and high-fertility

sites (Alto Frio, Hornito, Pinola). We used the statistical codes provided by Pavoine et al. (2010)

to perform theses analyses.

104

RESULTS

Functional composition in response to environmental predictors (RLQ and fourth-corner)

We used RLQ analysis to detect the primary covariance between traits and environmental variation, while we used the fourth corner as a complementary analysis to assess the significance of individual trait-environment associations (e.g., leaf SLA × total soil N). The total variance - the

measure of the global link between species traits and environmental factors – captured by the RLQ

analysis was 3.42. The co-inertia explained by the first two RLQ axes was 90.68%, with the first

accounting for 76.95% and the second for 13.73% (Table C.4). The environmental variables that

most contributed to the RLQ total variance were soil N:P (26.37%), BD (18.01%), DSS moisture

(17.66%), dry season rainfall (12.45%) and Mn (11.38%). For traits, the most important variables

were leaf C:N (20.89%) and leaf P (14.57%). All other environmental and trait variables

contributed < 10% to the RLQ total inertia (Table C.4). We found a high correlation between the

environment (R table) and species traits (Q table) for both RLQ Axis 1 (0.57) and RLQ Axis 2

(0.42). The global relationship between species traits and environmental variables was significant

(p < 0.01).

We detected grouping of sites in our analysis. Sites with high soil N:P and high dry season soil moisture were associated with the positive scores on RLQ Axis 1, which corresponded to sites with deep layers of rhyolitic tuff with a shallow organic layer (Chorro A and B). Sites with high

BD, Mn and resin P were associated with the negative side of the RLQ Axis 1, which corresponds

to sites with basalt, dacite or undifferentiated volcanics with more organic-rich topsoil (Pinola,

Hornito and Alto Frio). Sites with high dry season rainfall were associated with the negative side

of the RLQ Axis 2, which correspond to the majority of the andesite sites featuring clay-rich soils

and shallow or no organic horizons located in the Caribbean slope (Palo Seco, Verrugosa A and

105

B). Sites surrounding Lake Fortuna featuring rhyolite-derived soils with deep organic horizons

(Honda A and B), rhyolite to andesite transition soil (Samudio), and an adjacent andesite site

(Bonita) were grouped in the center of the RQL Axis 1 and Axis 2.

For the quantitative traits, leaf C:N and leaf thickness were positively correlated, and leaf

P negatively correlated with the RLQ Axis 1; leaf dry matter content (LDMC) and leaf manganese

(leaf Mn) were positively correlated, and leaf aluminum (leaf Al) and leaf calcium (leaf Ca) were negatively correlated with the RLQ Axis 2 (Figure 4.2a). For the binary traits, growth form was associated with RLQ Axis 1, while dimorphism and clonality were negatively correlated with the

RLQ Axis 2. We observed that species grouped according to their associations with parent material based on traits (Figure 4.2b). Species with high SLA, notably Diplazium striatastrum, were more commonly found at sites with high soil fertility.

We found significant correlations of individual traits with environmental variables under the fourth-corner analysis. The traits leaf P and leaf C:N were correlated with soil N:P (r = -0.45, p < 0.05; r =0 .50, p < 0.05, respectively). The combination of fourth-corner and RLQ statistics also revealed a significant correlation between RLQ axes and environmental variables and between

RLQ axes and species traits. There were significant positive correlations between the first RLQ traits axis (RLQ AxcQ1) and soil N:P (r = 0.53, p < 0.01) and DSS moisture (r = 0.46, p < 0.01), and negative correlations with BD (r = – 0.53, p < 0.01) and soil Mn (r = – 0.34, p < 0.05). The second RLQ trait axis (RLQ AxcQ2) was negatively correlated with dry season rainfall (r = – 0.36, p < 0.01). There was also a significant positive correlation between the first RLQ environmental axis (RLQ AxcR1) and leaf C:N (r = 0.48, p < 0.01) and a negative correlation with leaf P (r = –

0.40, p < 0.05) (Figure 4.2c).

106

Functional dispersion responses to environmental predictors

Functional dispersion ranged among sites from 0.09 to 0.20 (mean = 0.14, SD = 0.04). A mafic volcanic site (Bonita) obtained the highest FDis, whereas both a low-fertility rhyolite

(Chorro B) and a high-fertility dacite (Hornito) site obtained the lowest FDis. Two out of 256 models evaluated by the model selection approach were retained with ΔAICc < 2. The first-best model included soil N:P and Mn, while the second-best model included soil N:P, DSS moisture and K (Table C.5). The rank order of the importance of the predictors based on the sum of model weights over conditioned models was soil N:P, soil Mn, DSS moisture and soil K (Table 4.2).

When we considered only the models retained with ΔAICc < 2, the rank order was soil N:P, soil

Mn, soil K and DSS moisture. Soil N:P appeared in 144 evaluated models and in both retained models. Although not selected by our criterion, the predictor dry season rainfall appeared in all models with ΔAICc > 2 and < 3, and may also be considered an important predictor. Model averaging from our two retained models indicated that estimated coefficients for soil Mn, soil N:P and soil K were negatively associated with FDis, while DSS moisture was positively associated with FDis (Table 4.2, Figure 4.3). The best model yielded a pseudo R-squared of 0.66, and the second-best model yielded a pseudo R-squared of 0.60.

Trait and phylogenetic diversity patterns

When all functional traits were combined, trait diversity values were not randomly distributed across the tips of our phylogenetic tree, indicating phylogenetic signal. Node contribution (NC) to trait diversity was significantly skewed toward the root nodes of the phylogenetic tree (Ro test, p = 0.001), such that trait diversity was clustered on nodes closer to the root instead of being evenly dispersed across the tips. The NC pattern indicates that the non-

Polypodiales clades (tree fern clade – Cyatheaceae - plus the Danaea clade - Marattiaceae)

107

contribute with a higher trait diversity value than all other Polypodiales clades (including most of

the herbaceous ferns) together (Figure 4.4). The single node test (SN test, p = 0.001) and the few nodes test (FN test, p = 0.001), were also significant on trait diversity when all traits were combined, suggesting that both a single node and few nodes contribute more to global trait diversity, than the random expectation, where diversity values are distributed randomly across the tips of the phylogenetic tree. When we considered each trait separately, 4 of the 15 traits were significantly correlated with phylogeny: two binary traits, growth form (1:tree ferns and 0: terrestrial herbs) and dimorphism (1:holodimorphic and 0:non-holodimorphic); and two quantitative traits (leaf Al, leaf C:N) (Table C.6). Growth form and leaf Al showed high NC accumulated in the roots of the tree. For trait clonality, values were skewed toward the tips, and trait values from dimorphism and leaf C:N traits were skewed towards a few nodes and a single node (trait values clustered in a single node and in few nodes – Figure C.3). For all traits combined, we found that both trait diversity (TQE β test = 0.319, p = 0.001) and phylogenetic diversity (PQE

β test = 0.278, p = 0.01) exhibited a clustered pattern. Trait and phylogenetic diversity also exhibited distinct patterns when analyzed for subsets of sites defined by soil fertility. In communities with low-fertility soils (rhyolite sites) both traits and phylogenetic diversity were clustered (TQE = 0.363, p = 0.002; PQE = 0.322, p = 0.018). In communities with intermediate- fertility soils (mafic volcanic sites plus the rhyolite to mafic volcanic transition site) trait diversity distribution did not differ from random distributions, but phylogenetic diversity was clustered

(TQE = 0.102, p = 0.872; PQE = 0.155, p = 0.048). In communities with high-fertility soil

(undifferentiated volcanics, dacite and basalt sites), trait diversity distribution was clustered, but phylogenetic diversity did not differ from random distributions (TQE = 0.215, p = 0.003; PQE =

0.085, p = 0.0846).

108 a

DISCUSSION

Using multivariate tests, we evaluated the significance of trait-environment relationships of a fern community in lower montane forests in western Panama. Consistent with our hypothesis, the functional composition of fern communities was associated with soil and rainfall variables, as indicated by significant correlations between RLQ axes and environmental variables. The first

RLQ trait axis (RLQ AxcQ1) was represented by species with high leaf C:N and leaf thickness values and species with low leaf P and SLA values. Therefore, this axis represents variation along the leaf economics spectrum (Wright et al., 2004). Leaf trait responses to soil nutrient gradients are linked to resource acquisition strategies and characterize general trade-offs, known as slow versus fast-growing plant strategies (Ordoñez et al., 2009; Wright et al., 2001). Soil N:P and DSS moisture were also positively correlated with RLQ AxcQ1, while BD and soil Mn were negatively correlated with RLQ AxcQ1. As a result, communities composed of species with high leaf C:N and leaf thickness, and low leaf P and SLA values were found at the low end of the soil fertility gradient. Shifts in community-wide functional composition along environmental gradients are recognized as evidence for environmental filtering affecting trait assembly (Cornwell and Ackerly,

2009). Our results are consistent with previous studies showing shifts in functional traits with soil fertility gradients, which is considered a strong predictor of functional traits on plant communities at both global and local scales (Liu et al., 2012; Ordoñez et al., 2009), and among distinct ecosystems (Jager et al., 2015; Libalah et al., 2017).

Both community weighted mean traits and individual functional traits are expected to be influenced by environmental predictors. Based on fourth-corner analyses, we provide evidence to support this expectation for two leaf traits. Significant associations were found between both leaf

C:N and leaf P and soil N:P, with leaf P decreasing and leaf C:N increasing with increasing soil

109

N:P. Therefore, leaf traits represented most of the functional composition variation and responded

independently to changes along the soil fertility gradient. The corresponding responses of

multidimensional functional traits and single traits to environmental sorting can strengthen our

trait assembly inferences. Andersen et al. (2012) found similar results studying trait-based

community assembly of understory palms in our same study area. In that study community means of leaf N, P, and SLA increased, and leaf C:N decreased with increasing soil nutrient availability.

In addition, analysis of the traits of a single habitat-generalist palm species, Geonoma cunneata, showed similar trends to the community mean trait values.

Functional trait dispersion (FDis) varied among sites. Interestingly, the lowest values of

FDis were obtained by sites that represented both ends of the soil fertility gradient in our study site

(Choro B, a low fertility, high soil moisture site, and Hornito, a high-fertility more seasonal site).

Conversely, the highest values were obtained for sites with intermediate soil fertility, high dry seasonal rainfall and bulk density. Functional dispersion decreased significantly with soil N:P, indicating that communities that occupy low-fertility sites have species with a narrow range of ecological traits. Both functional composition and dispersion showed similar responses to the soil nutrient gradient (decreasing with soil N:P), suggesting that these two trait dimensions are coupled and capture the same ecological mechanism (environmental filtering). Both trait composition and dispersion have been shown to respond to environmental variables, but previous work has also indicated that factors other than soil nutrients are most important. For example, Schellenberger

Costa et al. (2017) found that community weighted mean and FDis of distinct vegetation types were strongly related to precipitation and disturbance gradients on Mount Kilimanjaro. On the other hand, trait composition and diversity can diverge in their responses, suggesting that they

110

might be affected by different ecological mechanisms, such as the findings in an Italian coastal dune plant community (Carboni et al., 2013).

Our conditioned model averaging also indicated that FDis was negatively correlated with soil Mn and soil K, which suggests that high-fertility sites select communities with species exhibiting low values of FDis. This finding may appear to disagree with the productivity hypothesis, which suggests that local functional diversity should increase in the direction of the benign portion of a gradient, given the increased number of available niches (e.g., Laliberté et al.,

2013). Nevertheless, this pattern might reflect species responses to complex interactions between soil Mn availability and genotypic variation in Mn tolerance. Manganese toxicity is associated with soils of pH 5.5 or lower (El-Jaoual and Cox, 1998). The Hornito site obtained the highest

value of soil Mn among all sites, which, combined with soil pH 5, might have some influence on

this observed pattern and play a secondary role of filtering species with high Mn tolerance. While

multiple environmental factors may interact and mediate Mn redox status and Mn toxicity, the

poorly known range of Mn tolerance among species could hinder our understanding of how Mn

toxicity might play as an essential stress factor in natural systems (Fernando and Lynch, 2015).

With respect to soil K influence on FDis, we believe that it might be a consequence of

multicollinearity predictors on the regression analysis. When predictors are correlated, estimated

coefficients of a given variable may depend on other predictors included in the model (Graham,

2003). In our environmental data, soil N:P and soil moisture, and soil K and soil Mn are highly

correlated. When we re-run model selection and model averaging without soil moisture and soil

K, soil N:P and soil Mn were still negatively correlated with FDis, and dry season rainfall variable

emerged as a predictor of positively correlated FDis (Table C.7). Nonetheless, the multicollinearity

111

among predictor variables did not prevent correct predictions of the FDis response with respect to important predictors like soil N:P and soil Mn within the scope of the model.

Comparison of the global trait and phylogenetic diversity distribution to null models indicated that species trait values are more similar than expected by chance, indicative of phenotypic clustering, and that those traits were generally conserved across the fern phylogeny, indicating phylogenetic clustering. Early papers linking phylogenetic methods to community ecology (Jeannine Cavender-Bares et al., 2009; Kraft et al., 2007; Webb et al., 2002), have suggested that phylogenetic clustering is expected when environmental filtering limits the spread of ecological strategies at a site. In addition, trait decomposition across the phylogeny showed that the highest node contribution to trait diversity values comes from the roots of the tree, with early evolutionary differentiation among the non-Polypodiales clades encompassing Danaea

(Marattiales) and tree ferns (Cyatheales) having an important contribution to trait diversity. Thus, fern trait structure is largely determined by the basal phylogenetic position of genera specializing on low-fertility soils. This becomes more evident when sets of sites with distinct parent material were analyzed separately. While intermediate and high-fertility sites exhibited mixed trait and phylogenetic patterns, sites with rhyolite low-fertility soils exhibited trait and phylogenetic clustering. Over-representation of older plant lineages in low-fertility soils has been previously observed for ferns and canopy trees in western lowland Amazonian and in Peruvian Amazonian forests respectively (Fine and Kembel, 2011; Lehtonen et al., 2015). Older lineages can introduce bias into phylogenetic analyses given their long phylogenetic branches, which can obscure other phylogenetic patterns present in the community (Kembel and Hubbell, 2006). Therefore, inferences from phylogenetic patterns should be made with caution, particularly when older

112

lineage taxa are locally rare in the community (N. J. B. Kraft and Ackerly, 2010). In our sites, tree

ferns were particularly abundant and diverse in the low-fertility sites (Viana et al., 2020).

Sites with intermediate and high soil fertility had mixed trait and phylogenetic patterns.

Random trait diversity distribution and clustered phylogeny in sites with intermediate-fertility soils

might suggest that critical conserved traits affected by environmental filtering have been omitted

(Pavoine et al., 2010). In contrast, clustered trait diversity distribution and no phylogenetic signal

in sites with high-fertility soil might be indicative of environmental filtering acting on labile traits

and low trait diversity randomly distributed across the nodes of the tree (Pavoine et al., 2010). The

few studies focusing on trait and phylogenetic diversity in ferns are mostly limited to epiphytic

taxa. Kluge and Kessler (2011) found that trait diversity of epiphytic fern assemblages have low

trait spread at the ends of an elevation gradient and high trait spread in the center. Similarly, low

trait diversity was found at the extreme ends of the environmental gradient in our study, which

according to Austin (1999) is a pattern reflecting the strong deterministic effects of extreme environmental conditions on species composition.

In summary, our study of trait assembly of terrestrial fern communities demonstrated evidence for trait-environment relationships in terms of functional composition and functional

dispersion in Fortuna forests. Functional composition was best predicted by soil variables and to

some extent by dry season rainfall indicated by RLQ multivariate analysis. Leaf traits such as leaf

P and leaf C:N contributed to most of the variation captured by ordination analysis of all trait

variables. Moreover, for these two leaf traits, significant associations with soil N:P were identified.

Functional composition and functional dispersion responded similarly to environmental gradients,

but differently from functional composition, functional dispersion appeared to have a non-linear

relationship to the environmental variables tested, decreasing towards the extreme ends of the soil

113 a fertility gradient. In our model assessment, soil N:P was identified as the most important variable associated with functional dispersion variation. Trait and phylogenetic diversity from all species in the community was clustered, suggesting that the main patterns of trait variation were directly affected by phylogenetic constrains. However, when trait variation was analyzed separately in three sets of sites with similar parent material and soil fertility, trait and phylogeny patterns departed from being only clustered. Trait diversity distribution across the phylogenetic tree was biased toward the roots, with non-Polypodiales clades (Cyatheaceae and Marattiaceae) contributing with the most variation in trait values.

CONCLUSIONS

Our results suggest that the combination of soil and rainfall variables, reflecting the strong nutrient limitation and seasonality across sites, act as an environmental filter affecting functional composition of terrestrial fern communities in Fortuna. In addition, soil variables also influenced functional dispersion responses across sites. This suggests there might be a partial connection between functional composition and dispersion given that both trait measures were affected by similar and distinct environmental variables. Altogether, traits and phylogeny were clustered, with the highest contribution on trait diversity being derived from the root of the phylogenetic tree, indicating that the rate of evolutionary change was higher in the nodes containing the earliest clades in our phylogeny. Although we found trait and phylogeny clustering on the overall community, distinct combinations of trait and phylogenetic distribution patterns were observed across sets of sites with different parent material and associated soil fertility. This may imply that the relative importance of specific evolutionary and ecological processes may change across the soil fertility gradient and spatial scale.

114

TABLES

Table 4.1. Plot environmental variables and functional dispersion of terrestrial ferns at each of the 12 one-hectare plots. Rainfall is the annual mean

from 2007-9 and 2011-14. Dry rainfall data are the monthly means for the dry season (1 January - 30 April). Soil Moisture is gravimetric soil

moisture content; all other soil variables are expressed on a volume basis. ECEC: Effective cation exchange capacity; TEB: Total exchangeable

bases (see Prada et al., 2017 for more details on the means and standards errors and significant differences among plots).

Plot CHA CHB HOA HOB SAM BON PAS VEA VEB HOR ALT PIN FDis 0.127 0.087 0.144 0.171 0.117 0.203 0.162 0.170 0.167 0.088 0.123 0.109 Elevation (m) 1100 1240 1155 1240 1215 1300 880 970 850 1330 1100 1135

Rainfall (mm) 5507 6159 6255 6159 4833 5507 6257 6257 6257 5164 4641* 4964*

Dry rainfall (mm) 351 332 381 332 215 351 445 445 445 203 94* 159* 115 Canopy openness (%) 9.74 11.79 9.06 9.68 11.09 12.34 7.42 10.05 11.44 13.65 11.89 10.12

Soil moisture (g g-1) 2.98 2.35 1.07 0.84 1.04 2.31 1.02 0.92 1.10 0.67 0.31 0.64

pH (H2O) 3.7 4.7 3.6 3.8 4.2 4.5 4.4 4.6 5.1 5 5.6 5.4 Bulk Density (g cm-3) 0.13 0.31 0.29 0.17 0.4 0.34 0.41 0.42 0.4 0.26 0.66 0.5

NH4-N‡ 0.8 2.4 2.25 1.78 0.6 1.9 0.8 1.6 0.85 1.8 3.76 6.98

NO3-N‡ 0.2 1.17 1.2 0.43 1.17 1.77 0.8 0.98 0.67 1.2 2.5 2.3 Total N† 1.8 2.7 2.9 2.3 3.6 3.8 2.2 2.7 2.2 2.8 4.7 4.5 Resin P‡ 0.1 0.01 0.2 1.9 0.4 1.1 0.4 0.5 0.3 2.2 1.4 1.8 Total P‡ 57.2 86.3 180.6 127.7 270 351 369.2 241 317 280 503 617.5 Soil N:P 26.95 31.53 17.46 16.1 13.76 10.75 7.67 11.16 8 10.38 9.6 7.52 Total C 29.6 50.6 43.9 40.9 51.3 47.6 32.5 37.9 28.7 35 51.1 54.2

Al† 0.8 1.3 1.1 1.2 1.5 2.5 1.1 1.7 1.17 0.48 0.03 0.3

Table 4.1. (cont.) Plot CHA CHB HOA HOB SAM BON PAS VEA VEB HOR ALT PIN Ca§ 0.5 1.2 0.04 0.1 1.1 0.3 1 0.4 0.31 4.9 8.47 6.6

Fe§ 0.1 0.04 0.02 0.03 0.06 0.05 0.01 0.05 0.005 0.005 0.005 0.005

K§ 0.1 0.07 0.02 0.02 0.07 0.11 0.06 0.06 0.04 0.18 0.12 0.1

Mg§ 0.28 0.07 0.04 0.06 0.32 0.35 0.33 0.17 0.15 1.01 2.29 1.09

Mn§ 0.01 0.04 0.01 0.01 0.14 0.03 0.19 0.04 0.07 0.71 0.42 0.15

Na§ 0.01 0.03 0.01 0.01 0.05 0.03 0.04 0.03 0.04 0.03 0.02 0.02

ECEC§ 1.8 3.1 1.3 1.5 3.4 3.4 2.8 2.5 1.8 7.4 11.4 8.5

TEB§ 0.9 1.7 0.1 0.2 1.6 0.8 1.5 0.6 0.6 6.2 10.9 8

-3 -3 -1 *Data available only for 2013-2014; ‡units µg cm ; †units mg cm ; §units CM L ; ꭞ undifferentiated volcanics.

116

Table 4.2. Model-averaged standardized coefficients based on conditioned models and standard errors (ΔAICc < 2), 95% confidence intervals, importance of each environmental predictors of terrestrial fern functional dispersion, based on per-variable sum of the Akaike weights (0 to 1 variation) over all conditioned averaged models, and probability. Standardized coefficients estimated from beta distribution errors.

Standardized Std. Relative Predictor 2.5% 97.5% P estimate Error importance intercept -1.846 0.050 -1.943 -1.748 – < 0.0001 soil Mn -0.253 0.066 -0.382 -0.124 1 0.0001 soil N:P -0.237 0.083 -0.399 -0.0744 0.73 < 0.01 soil K -0.205 0.049 -0.301 -0.1084 0.27 < 0.0001 soil moisture 0.261 0.064 0.1360 0.387 0.27 < 0.0001 phi (Φ) 299.223 134.264 36.070 562.375 – < 0.05

117 a

FIGURES

Figure 4.1. Map of the 12 one-hectare plots in Fortuna forests and their underlying parent material.

118

Figure 4.2. Results of the RQL and fourth-corner analyses. (a) the first two RLQ axes with environmental variables and species traits components. Vectors from environmental variables are represented by filled circles, trait variables by vectors. (b) fern species row scores from trait data

(Q table), colors refer to the parent material where species were most abundant. (c) representation of the results of the fourth-corner tests between the first two RLQ axes for environmental variables

(AxR1/AxR2) and species traits (AxQ1 and AxQ2). Black squares broe. Total inertia = 3.423.

Values of d are the grid size.

119 a

Figure 4.3. Standardized mean effect sizes of four soil variables significantly correlated to functional dispersion. Error bars represent 95% confidence intervals (CIs).

120 a

Figure 4.4. Decomposition of terrestrial fern trait diversity (all traits) across the nodes of the hypothesized phylogenetic tree. Each circle represents the amount of trait variation accounted for by each node. The variation coming from the root explained most of the variation between fern species.

121 a

REFERENCES

Amatangelo, K. L., and Vitousek, P. M. (2008). Stoichiometry of ferns in Hawaii: Implications

for nutrient cycling. Oecologia, 157(4), 619–627. https://doi.org/10.1007/s00442-008-1108-9

Andersen, K. M., Endara, M. J., Turner, B. L., and Dalling, J. W. (2012). Trait-based community

assembly of understory palms along a soil nutrient gradient in a lower montane tropical

forest. Oecologia, 168(2), 519–531. https://doi.org/10.1007/s00442-011-2112-z

Andersen, K. M., Turner, B. L., and Dalling, J. W. (2010). Soil-based habitat partitioning in

understorey palms in lower montane tropical forests. Journal of Biogeography, 37(2), 278–

292. https://doi.org/10.1111/j.1365-2699.2009.02192.x

Austin, M. P. (1999). The potential contribution of vegetation ecology to biodiversity research.

Ecography, 22(5), 465–484. https://doi.org/10.1111/j.1600-0587.1999.tb01276.x

Barton, K. (2020). MuMIn: mulit-model inference. R package version 1.43.17. https://cran.r-

project.org/package=MuMIn

Benson, D. A., Karsch-Mizrachi, I., Lipman, D. J., Ostell, J., and Sayers, E. W. (2009).

GenBank. Nucleic Acids Research, 37(SUPPL. 1), D26–D31.

https://doi.org/10.1093/nar/gkn723

Botta‐Dukát, Z. (2005). Rao’s quadratic entropy as a measure of functional diversity based on

multiple traits. Journal of Vegetation Science, 16(5), 533–540.

https://doi.org/10.1111/j.1654-1103.2005.tb02393.x

Britton, M. R., and Watkins, J. E. (2016). The economy of reproduction in dimorphic ferns.

Annals of Botany, 118(6), 1139–1149. https://doi.org/10.1093/aob/mcw177

Brock, J. M. R., Perry, G. L. W., Lee, W. G., and Burns, B. R. (2016). Tree fern ecology in New

Zealand: A model for southern temperate rainforests. Forest Ecology and Management,

122 a

375, 112–126. https://doi.org/10.1016/j.foreco.2016.05.030

Burnham, K. P., and Anderson, D. R. (2002). Model Selection and Inference: A Practical

Information-Theoretic Approach. (2nd ed.). Springer‐Verlag.

https://doi.org/10.1002/sim.769

Cadotte, M. W., Cavender-Bares, J., Tilman, D., and Oakley, T. H. (2009). Using phylogenetic,

functional and trait diversity to understand patterns of plant community productivity. PLoS

ONE, 4(5), 1–9. https://doi.org/10.1371/journal.pone.0005695

Carboni, M., Acosta, A. T. R., and Ricotta, C. (2013). Are differences in functional diversity

among plant communities on Mediterranean coastal dunes driven by their phylogenetic

history? Journal of Vegetation Science, 24(5), 932–941. https://doi.org/10.1111/jvs.12095

Carvajal-Hernández, C. I., Gómez-Díaz, J. A., Kessler, M., and Krömer, T. (2018). Influence of

elevation and habitat disturbance on the functional diversity of ferns and lycophytes. Plant

Ecology and Diversity, 11(3), 335–347. https://doi.org/10.1080/17550874.2018.1484526

Cavelier, J., Solis, D., and Jaramillo, M. A. (1996). Fog interception in montane forests across

the Central Cordillera of Panamá. Journal of Tropical Ecology, 12, 357–369.

https://doi.org/10.1017/S026646740000955X

Cavender-Bares, J., Ackerly, D. D., Baum, D. A., and Bazzaz, F. A. (2004). Phylogenetic

overdispersion in Floridian oak communities. American Naturalist, 163(6), 823–843.

https://doi.org/10.1086/386375

Cavender-Bares, Jeannine, Kozak, K. H., Fine, P. V. A., and Kembel, S. W. (2009). The merging

of community ecology and phylogenetic biology. Ecology Letters, 12, 693–715.

https://doi.org/10.1111/j.1461-0248.2009.01314.x

Chabot, B. F., and Hicks, D. J. (1982). The ecology of leaf life spans. Annual Review of Ecology

123

and Systematics. Volume 13, 229–259.

https://doi.org/10.1146/annurev.es.13.110182.001305

Cody, M. L., and Diamond, J. M. (1975). Ecology and evolution of communities. Belknap Press,

Harvard University Press.

Cornelissen, J. H. C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N., Gurvich, D. E., Reich, P.

B., Ter Steege, H., Morgan, H. D., Van Der Heijden, M. G. A., Pausas, J. G., and Poorter,

H. (2003). A handbook of protocols for standardised and easy measurement of plant

functional traits worldwide. Australian Journal of Botany, 51, 335–380.

https://doi.org/10.1071/BT02124

Cornwell, W. K., and Ackerly, D. D. (2009). Community Assembly and Shifts in Plant Trait

Distributions across an Environmental Gradient in Coastal California. Ecological

Monographs, 79(1), 109–126. https://doi.org/10.1890/07-1134.1

Cribari-Neto, F., and Zeileis, A. (2010). Beta Regression in R. Journal of Statistical Software,

34(2), 1–24. https://doi.org/10.18637/jss.v069.i12

Dai, X., Chen, C., Li, Z., and Wang, X. (2020). Taxonomic, Phylogenetic, and Functional

Diversity of Ferns at Three Differently Disturbed Sites in Longnan County, China.

Diversity, 12(4), 135. https://doi.org/10.3390/d12040135

Dalling, J. W., Prada, C. M., Heineman, K. D., Andersen, K. M., Morris, A., Pardo, J.,

Hollenbeck, E., Caballero, P., Espinosa, C., García, E., and Turner, B. L. (n.d.).

Introduction to the Fortuna Forest Reserve: Soil and Climate Effects on the Tree

Communities of a Premontane Tropical Forest. In J. W. Dalling and B. L. Turner (Eds.),

Fortuna Forest Reserve, Panama: Interacting Effects of Climate and Soils on the Biota of a

Wet Premontane Tropical Forest (Smithsonia). Smithsonian Institution Scholarly Press.

124

de Bello, F., Berg, M. P., Dias, A. T. C., Diniz-Filho, J. A. F., Götzenberger, L., Hortal, J., Ladle,

R. J., and Lepš, J. (2015). On the need for phylogenetic ‘corrections’ in functional trait-

based approaches. Folia Geobotanica, 50(4), 349–357. https://doi.org/10.1007/s12224-015-

9228-6

Dolédec, S., Chessel, D., ter Braak, C. J. F., and Champely, S. (1996). Matching species traits to

environmental variables: a new three-table ordination method. Environmental and

Ecological Statistics, 3(2), 143–166. https://doi.org/10.1002/ccd.27419

Dray, S., Chessel, D., and Thioulouse, J. (2003). Co-inertia analysis and the linking of ecological

data tables. Ecology, 84(11), 3078–3089. https://doi.org/10.1890/03-0178

Dray, S., Choler, P., Dolédec, S., Peres-Neto, P. R., Thuiller, W., Pavoine, S., and Ter Braak, C.

J. F. (2014). Combining the fourth-corner and the RLQ methods for assessing trait

responses to environmental variation. Ecology, 95(1), 14–21. https://doi.org/10.1890/13-

0196.1

Dray, S., and Dufour, A. B. (2007). The ade4 package: Implementing the duality diagram for

ecologists. Journal of Statistical Software, 22(4), 1–20.

https://doi.org/10.18637/jss.v022.i04

Dray, S., and Legendre, P. (2008). Testing the species traits environment relationships: The

fourth-corner problem revisited. Ecology, 89(12), 3400–3412. https://doi.org/10.1890/08-

0349.1

Edgar, R. C. (2004). MUSCLE: A multiple sequence alignment method with reduced time and

space complexity. BMC Bioinformatics, 5(1), 1–19. https://doi.org/10.1186/1471-2105-5-

113

El-Jaoual, T., and Cox, D. A. (1998). Manganese toxicity in plants. Journal of Plant Nutrition,

125 a

21(2), 353–386. https://doi.org/10.1080/01904169809365409

Fernando, D. R., and Lynch, J. P. (2015). Manganese phytotoxicity: New light on an old

problem. Annals of Botany, 116(3), 313–319. https://doi.org/10.1093/aob/mcv111

Ferrari, S. L. P., and Cribari-Neto, F. (2004). Beta regression for modelling rates and

proportions. Journal of Applied Statistics, 31(7), 799–815.

https://doi.org/10.1080/0266476042000214501

Fine, P. V. A., and Kembel, S. W. (2011). Phylogenetic community structure and phylogenetic

turnover across space and edaphic gradients in western Amazonian tree communities.

Ecography, 34(4), 552–565. https://doi.org/10.1111/j.1600-0587.2010.06548.x

Frazer, G. W., Canham, C. D., and Lertzman, K. P. (2000). Gap light analyzer (GLA) Version

2.0: Image-processing software to analyze true-color hemispherical canopy photographs.

Bulletin of the Ecological Society of America, 81(3), 191–197.

Graham, M. H. (2003). Confronting multicollinearity in ecological multiple regression. Ecology,

84(11), 2809–2815. https://doi.org/10.1890/02-3114

Grime, J. P. (1991). Nutrition, environment and plant ecology: an overview. In J. Porter and D.

Lawlor (Eds.), Plant growth: interactions with nutrition and environment (pp. 249–267).

Cambridge University Press.

Grime, J. P. (2006). Trait convergence and trait divergence in herbaceous plant communities:

Mechanisms and consequences. Journal of Vegetation Science, 17(2), 255–260.

https://doi.org/10.1111/j.1654-1103.2006.tb02444.x

Grueber, C. E., Nakagawa, S., Laws, R. J., and Jamieson, I. G. (2011). Multimodel inference in

ecology and evolution: Challenges and solutions. In Journal of Evolutionary Biology (Vol.

24, Issue 4, pp. 699–711). John Wiley & Sons, Ltd. https://doi.org/10.1111/j.1420-

126 a

9101.2010.02210.x

Hardy, O. J., and Senterre, B. (2007). Characterizing the phylogenetic structure of communities

by an additive partitioning of phylogenetic diversity. Journal of Ecology, 95(3), 493–506.

https://doi.org/10.1111/j.1365-2745.2007.01222.x

Herben, T., Nováková, Z., and Klimešova, J. (2014). Clonal growth and plant species abundance.

Annals of Botany, 114(2), 377–388. https://doi.org/https://doi.org/10.1093/aob/mct308

Hill, M. O., and Smith, A. J. E. (1976). Principal Component Analysis of Taxonomic Data with

Multi-State Discrete Characters. Taxon, 25(2), 249–255.

https://www.jstor.org/stable/1219449

Holdridge, L. (1947). Determination of world plant formations from simple climatic data.

Science, 105, 367–368. http://science.sciencemag.org/content/105/2727/367.short

Hurvich, C. M., and Tsai, C.-L. (1989). Regression and Time Series Model Selection in Small

Samples. Biometrika, 76(2), 297. https://doi.org/10.2307/2336663

Jager, M. M., Richardson, S. J., Bellingham, P. J., Clearwater, M. J., and Laughlin, D. C. (2015).

Soil fertility induces coordinated responses of multiple independent functional traits.

Journal of Ecology, 103(2), 374–385. https://doi.org/10.1111/1365-2745.12366

Kembel, S. W., and Hubbell, S. P. (2006). The phylogenetic structure of a neotropical forest tree

community. Ecology, 87(7 SUPPL.), 86–99. https://doi.org/10.1890/0012-

9658(2006)87[86:tpsoan]2.0.co;2

Klimeš, L., Klimešova, J., Hendriks, R., and Van Groenendael, J. (1997). Clonal plant

architecture: a comparative analysis of form and function. In H. De Kroon and J. Van

Groenendael (Eds.), The ecology and evolution of clonal plants (pp. 1–29). Backhuys

Publishers: Leiden.

127 a

Kluge, J., and Kessler, M. (2011). Phylogenetic diversity, trait diversity and niches: Species

assembly of ferns along a tropical elevational gradient. Journal of Biogeography, 38(2),

394–405. https://doi.org/10.1111/j.1365-2699.2010.02433.x

Kraft, N. J. B., and Ackerly, D. D. (2010). Functional trait and phylogenetic tests of community

assembly across spatial scales in an Functional tests of trait and community assembly

phylogenetic across scales in an Amazonian forest spatial. Ecological Monographs, 80(3),

401–422. https://doi.org/10.1890/09-1672.1

Kraft, N. J. B. B., Cornwell, W. K., Webb, C. O., and Ackerly, D. D. (2007). Trait evolution,

community assembly, and the phylogenetic structure of ecological communities. American

Naturalist, 170(2), 271–283. https://doi.org/10.1086/519400

Kumar, S., Stecher, G., Li, M., Knyaz, C., and Tamura, K. (2018). MEGA X: Molecular

Evolutionary Genetics Analysis across computing plataforms. Molecular Biology and

Evolution, 35, 1547–1549. https://doi.org/https://doi.org/10.1093/molbev/msy096

Laliberté, E., and Legendre, P. (2010). A distance-based framework for measuring functional

diversity from multiple traits. Ecology, 91(1), 299–305. https://doi.org/10.1890/08-2244.1

Laliberté, E., Legendre, P., and Shipley, B. (2014). FD: measuring funtional diversity from

multiple traits, and other tools for functional ecology. R package version 1.0-12.

http://www.elaliberte.info/publications

Laliberté, E., Norton, D. A., and Scott, D. (2013). Contrasting effects of productivity and

disturbance on plant functional diversity at local and metacommunity scales. Journal of

Vegetation Science, 24(5), 834–842. https://doi.org/10.1111/jvs.12044

Lavorel, S., Grigulis, K., McIntyre, S., Williams, N. S. G., Garden, D., Dorrough, J., Berman, S.,

Quétier, F., Thébault, A., and Bonis, A. (2008). Assessing functional diversity in the field -

128

Methodology matters! Functional Ecology, 22(1), 134–147. https://doi.org/10.1111/j.1365-

2435.2007.01339.x

Legendre, P., and Legendre, L. (1998). Numerical ecology (2nd Englis). Elsevier.

Lehtonen, S., Jones, M. M., Zuquim, G., Prado, J., and Tuomisto, H. (2015). Phylogenetic

relatedness within neotropical fern communities increases with soil fertility. Global Ecology

and Biogeography, 24, 695–705. https://doi.org/10.1111/geb.12294

Leibold, M. A., Economo, E. P., and Peres-Neto, P. (2010). Metacommunity phylogenetics:

Separating the roles of environmental filters and historical biogeography. Ecology Letters,

13(10), 1290–1299. https://doi.org/10.1111/j.1461-0248.2010.01523.x

Libalah, M. B., Droissart, V., Sonké, B., Hardy, O. J., Drouet, T., Pescador, D. S., Kenfack, D.,

Thomas, D. W., Chuyong, G. B., and Couteron, P. (2017). Shift in functional traits along

soil fertility gradient reflects non-random community assembly in a tropical African

rainforest. Plant Ecology and Evolution, 150(3), 265–278.

https://doi.org/10.5091/plecevo.2017.1318

Liu, X., Swenson, N. G., Wright, S. J., Zhang, L., Song, K., Du, Y., Zhang, J., Mi, X., Ren, H.,

and Ma, K. (2012). Covariation in plant functional traits and soil fertility within two

species-rich forests. PLoS ONE, 7(4), e34767. https://doi.org/10.1371/journal.pone.0034767

MacArthur, R. H., and Wilson, E. O. (1967). The theory of island biogeography. In Monographs

in Population Biology (p. 203). Princeton University Press.

Mayfield, M. M., and Levine, J. M. (2010). Opposing effects of competitive exclusion on the

phylogenetic structure of communities. Ecology Letters, 13(9), 1085–1093.

https://doi.org/10.1111/j.1461-0248.2010.01509.x

Ordoñez, J. C., Van Bodegom, P. M., Witte, J. P. M., Wright, I. J., Reich, P. B., and Aerts, R.

129

(2009). A global study of relationships between leaf traits, climate and soil measures of

nutrient fertility. Global Ecology and Biogeography, 18(2), 137–149.

https://doi.org/10.1111/j.1466-8238.2008.00441.x

Pausas, J. G., and Verdú, M. (2010). The jungle of methods for evaluating phenotypic and

phylogenetic structure of communities. BioScience, 60(8), 614–625.

https://doi.org/10.1525/bio.2010.60.8.7

Pavoine, S., Baguette, M., and Bonsall, M. B. (2010). Decomposition of trait diversity among the

nodes of a phylogenetic tree. Ecological Monographs, 80(3), 485–507.

https://doi.org/10.1890/09-1290.1

Pennell, M. W., FitzJohn, R. G., and Cornwell, W. K. (2016). A simple approach for maximizing

the overlap of phylogenetic and comparative data. Methods in Ecology and Evolution, 7(6),

751–758. https://doi.org/10.1111/2041-210X.12517

Podani, J. (1999). Extending Gower’s general coefficient of similarity to ordinal characters.

Taxon, 48(2), 331–340. https://doi.org/10.2307/1224438

PPG I. (2016). A community-derived classification for extant lycophytes and ferns. Journal of

Systematics and Evolution, 54(6), 563–603. https://doi.org/10.1111/jse.12229

Prada, C. M., Morris, A., Andersen, K. M., Turner, B. L., Caballero, P., and Dalling, J. W.

(2017). Soils and rainfall drive landscape-scale changes in the diversity and functional

composition of tree communities in premontane tropical forest. Journal of Vegetation

Science, 28(4), 859–870. https://doi.org/10.1111/jvs.12540

R Core Team. (2019). R: A language and environment for statistical computing. R Foundation

for Statistical Computing. https://www.r-project.org/.

Rao, C. R. (1982). Diversity and dissimilarity coefficients: A unified approach. Theoretical

130

Population Biology, 21(1), 24–43. https://doi.org/10.1016/0040-5809(82)90004-1

Schellenberger Costa, D., Gerschlauer, F., Pabst, H., Kühnel, A., Huwe, B., Kiese, R.,

Kuzyakov, Y., and Kleyer, M. (2017). Community-weighted means and functional

dispersion of plant functional traits along environmental gradients on Mount Kilimanjaro.

Journal of Vegetation Science, 28(4), 684–695. https://doi.org/10.1111/jvs.12542

Tanaka, T., and Sato, T. (2015). Taxonomic, phylogenetic and functional diversities of ferns and

lycophytes along an elevational gradient depend on taxonomic scales. Plant Ecology,

216(12), 1597–1609. https://doi.org/10.1007/s11258-015-0543-z

Testo, W., and Sundue, M. (2016). A 4000-species dataset provides new insight into the

evolution of ferns. Molecular Phylogenetics and Evolution, 105, 200–211.

https://doi.org/10.1016/j.ympev.2016.09.003

Tilman, D. (1982). Resource Competition and Community Structure. Princeton University Press.

Valladares, F., Gianoli, E., and Gómez, J. M. (2007). Ecological limits to plant phenotypic

plasticity. In New Phytologist (Vol. 176, Issue 4, pp. 749–763). Wiley/Blackwell (10.1111).

https://doi.org/10.1111/j.1469-8137.2007.02275.x

Viana, J. L., Turner, B. L., and Dalling, J. W. (2020). Compositional variation in understorey

fern and palm communities along a soil fertility and rainfall gradient in a lower montane

tropical forest. Journal of Vegetation Science, 1, 1–14. https://doi.org/10.1111/jvs.12947

Volterra, V. (1926). Fluctuations in the Abundance of a Species considered Mathematically.

Nature, 118(2972), 558–560. https://doi.org/10.1038/118558a0

Webb, C. O., Ackerly, D. D., McPeek, M. A., and Donoghue, M. J. (2002). Phylogenies and

community ecology. Annual Review of Ecology and Systematics, 33, 475–505.

https://doi.org/10.1146/annurev.ecolsys.33.010802.150448

131

Weiher, E., Clarke, G. D. P., and Keddy, P. A. (1998). Community Assembly Rules,

Morphological Dispersion, and the Coexistence of Plant Species. Oikos, 81(2), 309.

https://doi.org/10.2307/3547051

Weiher, E., and Keddy, P. A. (1995). Assembly Rules, Null Models, and Trait Dispersion: New

Questions from Old Patterns. Oikos, 74(1), 159–164. https://doi.org/10.2307/3545686

Wright, I J, Reich, P. B., and Westoby, M. (2001). Strategy shifts in leaf physiology, structure

and nutrient content between species of high-and low-rainfall and high-and low-nutrient

habitats. Functional Ecology, 15(15), 423–434.

Wright, Ian J, Reich, P., Westoby, M., Ackerly, D. D., Baruch, Z., Bongers, F., Cavender-Bares,

J., Chapin, T., Cornelissen, J. H. ., Diemer, M., Flexas, J., Garnier, E., Groom, P. K., Gulias,

J., Hikosaka, K., Lamont, B. B., Lee, T., Lee, W., Lusk, C., … Villar, R. (2004). The

worldwide leaf economics spectrum. Nature, 428, 821–827.

https://doi.org/10.1038/nature02403

Zhu, S. D., Li, R. H., Song, J., He, P. C., Liu, H., Berninger, F., and Ye, Q. (2016). Different leaf

cost–benefit strategies of ferns distributed in contrasting light habitats of sub-tropical

forests. Annals of Botany, 117(3), 497–506. https://doi.org/10.1093/aob/mcv179

132

CHAPTER 5: CONCLUSIONS

Summary

This dissertation research contributes to the ecology of ferns by adding to our knowledge of the ecological factors and mechanisms that define how terrestrial ferns interact with their surrounding abiotic and biotic environments in lower montane rainforests. In chapter 2, it is demonstrated that despite responding to different soil and light variables, ferns and palms exhibit congruent floristic compositional associations, consistent with studies performed in lowland rainforests (Costa et al., 2005; Tuomisto et al., 2002; Zuquim et al., 2014). Despite large differences in dispersal abilities between ferns and palms distance-decay in compositional similarity did not differ significantly between the groups. However, tree ferns reached peak abundance in plots with high soil N:P, showing a contrasting response to palms and herbaceous ferns. Therefore, resource partitioning may potentially explain part of the coexistence of these major understory groups.

In chapter 3 the fern and lycophyte flora of Fortuna is reviewed, and major floristic components are highlighted. Overall, species that belong to the Dryopteridaceae family provide the highest contribution to the fern flora. Among vascular epiphytes in tropical rain forests, ferns are commonly the most diverse group (Cardelús et al., 2006; Gentry and Dodson, 1987). Indeed, the genus Elaphoglossum – generally epiphytes – was the richest in species among all life forms in Fortuna. Among terrestrial herbs, Diplazium and Thelypteris provided the largest number of species. More species names were added to the lycophyte and fern flora after in situ collections, which resulted in a 19% increase in species number.

133

Findings in chapter 4 indicate that the soil and rainfall variables influence trait assembly of terrestrial fern communities, with functional composition and functional dispersion responding similarly to environmental gradients, decreasing towards the extreme ends of the soil fertility gradient, promoting trait and phylogenetic clustering of the overall community. Moreover, functional composition and dispersion appear to be affected by similar environmental factors.

Ferns from non-polypod clades contributed with most of the trait value variation among all ferns, highlighting the relative importance of evolutionary and ecological processes across the soil fertility gradients.

Future directions

I expect to expand on my thesis to develop a more comprehensive view of the effects of nutrient limitation on the ecology of ferns. Insights on fern ecology can be gained from experimental manipulations of soil nutrients. Studies of the effects of nutrient limitation on fern- soil associations have been restricted to the sporophyte stage. These studies have shown that after controlling for the effects of climate variation, floristic composition is related to edaphic conditions

(Jones et al., 2006; Karst et al., 2005). This dissertation shows the role edaphic specialization and

edaphic heterogeneity in explaining most of the spatial patterning observed in fern distributions.

Nonetheless, the mechanisms underlying this specialization is often ignored, and the potential

effects of the gametophyte phase in regulating fern distributions has not been directly assessed.

Studying the impact of soil nutrient limitation on gametophytes is therefore the next step in

improving our understanding of this puzzling association between fern species and soils.

Beyond possible environmental impacts of soil variables on gametophyte recruitment,

there are many other aspects of fern ecology that deserve exploration. Genome size and ploidy

level are intrinsic traits of plant species with the potential to impact the performance of individuals

134

(Segraves and Anneberg, 2016). For instance, if DNA content increases without a countervailing

change in cell number, then polyploid species will have increased requirements for potentially

limiting nutrients such as soil N and soil P (Guignard et al., 2016). This might affect community composition via a reduction in the competitive ability of polyploid species, constraining their occurrence to environments with higher nutrient availability (Segraves, 2017).

In terms of the number of fern species and geographic range concerned, mycorrhizal interactions are by far the most common. Roots of 66 of 89 terrestrial and epiphytic ferns from

Hawaiian forests were found in to be infected by vesicular-arbuscular mycorrhizae (Gemma et al.,

1992). Environmental conditions may regulate mycorrhizal associations in ferns. For example,

Equisetum sporophytes are commonly mycorrhizal in dry environments, but in wetlands and moist soil environments mycorrhizal associations were absent (Brundrett, 2002; Marsh et al., 2000; Read et al., 2000). The heterogeneity in soil and rainfall conditions in Fortuna provide an opportunity to test mycorrhizal association in ferns.

135

REFERENCES

Brundrett, M. C. (2002). Coevolution of roots and mycorrhizas of land plants. New Phytologist,

154(2), 275–304. https://doi.org/https://doi.org/10.1046/j.1469-8137.2002.00397.x

Cardelús, C. L., Colwell, R. K., and Watkins Jr., J. E. (2006). Vascular Epiphyte Distribution

Patterns : Explaining the Mid-Elevation Richness Peak. Journal of Ecology, 94(1), 144–

156. https://www.jstor.org/stable/3599617

Costa, F. R. C., Magnusson, W. E., and Luizao, R. C. (2005). Mesoscale distribution patterns of

Amazonian understorey herbs in relation to topography, soil and watersheds. Journal of

Ecology, 93, 863–878. https://doi.org/10.1111/j.1365-2745.2005.01020.x

Gemma, J. N., Koske, R. E., and Flymn, T. (1992). Mycorrhizae in Hawaiian pteridophytes:

occurrence and evolutionary significance. American Journal of Botany, 79(8), 843–852.

http://www.jstor.org/stable/2444993

Gentry, A. H., and Dodson, C. H. (1987). Diversity and biogeography of neotropical vascular

epiphytes. Annals of the Missouri Botanical Garden, 74(2), 205.

https://doi.org/10.2307/2399395

Guignard, S., Nichols, R. A., Knell, R. J., Macdonald, A., Romila, C., Trimmer, M., Leitch, I. J.,

and Leitch, A. R. (2016). Genome size and ploidy influence angiosperm species ’ biomass

under nitrogen and phosphorus limitation. New Phytologist, 210, 1195–1206.

Jones, M. M., Tuomisto, H., Clark, D. B., and Olivas, P. (2006). Effects of mesoscale

environmental heterogeneity and dispersal limitation on floristic variation in rain forest

ferns. Journal of Ecology, 94, 181–195. https://doi.org/10.1111/j.1365-2745.2005.01071.x

Karst, J., Gilbert, B., and Lechowicz, M. J. (2005). Fern community assembly: The roles of

chance and the environment at local and intermediate scales. Ecology, 86(9), 2473–2486.

136 https://doi.org/10.1890/04-1420

Marsh, A. S., Aronomr II, J. A., Bormannt, T., and Gordon, J. C. (2000). The Role of Equisetum

in Nutrient Cycling in an Alaskan Shrub Wetland. Journal of Ecology, 88(6), 999–1011.

https://www.jstor.org/stable/2648408

Read, A. D. J., Duckett, J. G., Francis, R., Ligrone, R., and Russell, A. (2000). Symbiotic Fungal

Associations in “Lower” Land Plants. Philosophical Transactions: Biological Sciences,

355(1398), 815–831. https://www.jstor.org/stable/3066807

Segraves, K. A. (2017). The effects of genome duplications in a community context. New

Phytologist, 215(1), 57–69. https://doi.org/10.1111/nph.14564

Segraves, K. A., and Anneberg, T. J. (2016). Species interactions and plant polyploidy. American

Journal of Botany, 103(7), 1326–1335. https://doi.org/10.3732/ajb.1500529

Tuomisto, H., Ruokolainen, K., Poulsen, A. D., Moran, R. C., Quintana, C., Cañas, G., and Celi,

J. (2002). Distribution and diversity of pteridophytes and Melastomataceae along edaphic

gradients in Yasuni National Park, Ecuadorian Amazonia. Biotropica, 34(4), 516–533.

Zuquim, G., Tuomisto, H., Jones, M. M., Prado, J., Figueiredo, F. O. G., Moulatlet, G. M.,

Costa, F. R. C., Quesada, C. A., and Emilio, T. (2014). Predicting environmental gradients

with fern species composition in Brazilian Amazonia. Journal of Vegetation Science, 25(5),

1195–1207. https://doi.org/10.1111/jvs.12174

137

APPENDIX A: SUPPLEMENTARY MATERIAL FOR CHAPTER 2

Figure A.1. Individual-based species accumulation curves for each of the twelve one-hectare plots color coded by underlying geology (parent material) for ferns (a) and palms (b). Site labels: Chorro

A (CHA), Chorro B (CHB), Honda A (HOA), Honda B (HOB), Samudio (SAM), Bonita (BON),

Palo Seco (PAS), Verrugosa A (VEA), Verrugosa B (VEB), Hornito (HOR), Alto Frio (ALT),

Pinola (PIN).

138

Figure A.2. Principal Component Analysis (PCA) of the environmental variables measured in twelve one-hectare plots. Site labels: Chorro A (CHA), Chorro B (CHB),Honda A (HOA), Honda

B (HOB), Samudio (SAM), Bonita (BON), Palo Seco (PAS), Verrugosa A (VEA), Verrugosa B

(VEB), Hornito (HOR), Alto Frio (ALT), Pinola (PIN).

139

Table A.1. Principal Component Analysis (PCA) loadings of environmental factors on PC1 and

PC 2 axes.

Environmental variables PC1 PC2

Elevation 0.054 0.412 Rainfall -0.222 -0.202

Dry rainfall -0.246 -0.201

Canopy openness 0.124 0.229

Red: far red -0.100 0.390

Soil moisture -0.176 0.315

pH 0.233 -0.106

Bulk density 0.210 -0.189

Ammonium (NH4-N) 0.205 0.052

Nitrate nitrogen (NO3-N) 0.248 0.066

Total nitrogen 0.240 0.094

Resin phosphorus 0.189 0.023

Total phosphorus 0.236 -0.188

Soil N:P -0.149 0.351

Total carbon 0.149 0.211

Aluminum -0.178 0.045

Calcium 0.278 0.036

Iron -0.167 0.301

Potassium 0.177 0.219

Magnesium 0.266 0.032

Manganese 0.199 -0.004

Sodium 0.008 -0.175

Effective cation exchange capacity 0.282 0.058

Total exchangeable bases 0.278 0.041

140

Table A.2. Environmental variables and floristic composition coefficient of determination derived from multiple regression between environmental factor and Principal Coordinate

Analysis (PCoA) axes for ferns and palms. Coefficients of determination are displayed in descending order. Background shading indicates coefficients of determination that were significant at p ≤ 0.05 after Holm-Bonferroni analysis for multiple comparisons (Adj. P:

Adjusted p-values).

Ferns Palms

Environmental Environmental r2 P Adj. P r2 P Adj. P variables variables

Soil N:Ps 0.88 < 0.001 < 0.05 Bulk density 0.87 < 0.001 < 0.05 Red: far red 0.81 < 0.001 < 0.05 Soil N:P 0.72 < 0.05 0.546 Soil moisture 0.73 < 0.01 0.066 pH 0.66 < 0.005 0.115 Dry rainfall 0.52 < 0.05 0.776 Total phosphorus 0.63 < 0.05 0.308 Resin phosphorus 0.47 0.065 1 Magnesium 0.59 < 0.05 0.735 1 Total exchangeable Iron 0.47 0.063 0.57 < 0.05 0.721 bases 1 Effective cation Aluminum 0.42 0.089 0.57 < 0.05 0.721 exchange capacity Total phosphorus 0.41 0.105 1 Calcium 0.55 < 0.05 0.721 Elevation 0.38 0.108 1 Nitrate nitrogen 0.51 < 0.05 0.659 Manganese 0.33 0.163 1 Iron 0.51 < 0.05 0.735 Sodium 0.30 0.202 1 Soil moisture 0.50 0.055 0.769 Calcium 0.29 0.186 1 Total nitrogen 0.36 0.117 1 Total exchangeable 1 1 0.27 0.216 Red: far red 0.32 0.172 bases Rainfall 0.27 0.271 1 Sodium 0.29 0.221 1 Effective cation 1 1 0.23 0.289 Aluminum 0.29 0.236 exchange capacity Bulk density 0.22 0.350 1 Dry rainfall 0.26 0.255 1 Magnesium 0.18 0.428 1 Manganese 0.26 0.206 1 Ammonium 0.14 0.535 1 Ammonium 0.18 0.333 1 pH 0.14 0.499 1 Elevation 0.17 0.437 1 Total nitrogen 0.14 0.512 1 Total carbon 0.15 0.472 1 Canopy openness 0.12 0.570 1 Rainfall 0.12 0.574 1 Nitrate nitrogen 0.09 0.647 1 Canopy openness 0.08 0.634 1 Total carbon 0.08 0.666 1 Potassium 0.08 0.695 1 Potassium 0.07 0.726 1 Resin phosphorus 0.02 0.891 1

141

Table A.3. Mantel correlation coefficients between distance matrices of environmental variables and fern and palm floristic distances and their probabilities.

Ferns Palms Geographic distance Variables Coefficient Significance Coefficient Significance Coefficient Significance

All environmental 0.41 < 0.01 0.40 < 0.05 0.50 < 0.05 variables Geographic distance 0.36 < 0.05 0.34 < 0.05 . .

Soil PCA 0.42 < 0.01 0.45 < 0.05 . .

Precipitation 0.28 < 0.05 0.15 0.163 . .

Light 0.15 0.126 -0.02 0.560 . .

Ferns . . 0.53 < 0.001 0.36 < 0.01

Palms . . . . 0.34 < 0.05

Pure environmental 0.29 < 0.05 0.27 0.052 . .

Pure spatial 0.20 0.066 0.18 0.135

Controlling for environment Controlling for space

Coefficient Significance Coefficient Significance

Ferns × Palms 0.44 < 0.001 0.46 < 0.01

142

Table A.4. Parameter estimates of fixed effects derived from three individual generalized linear mixed models with Poisson errors. Each model tested the relationship between understory palm and terrestrial fern abundance (Groups) and soil N:P (NP), red: far-red (RFR) and bulk density

(BD). Models were fitted by maximum likelihood, and subplot was included as nested random term. Overall model: Abundance ~ Environmental factor + Groups + Environmental factor:

Groups + (1 | Plot) + (1 | Plot: Subplot). The null model included only random effects.

Environmental Fixed Effects Estimate Std. Error z-value P-value AIC Factors

Intercept 0.953 0.371 2.567 0.010

Model 1 NP 0.044 0.023 1.927 0.054 3253.3 Total N:P GroupsPalms 2.224 0.085 26.190 < 0.001

NP:GroupsPalms -0.115 0.006 -20.330 < 0.001

Intercept 0.878 0.500 1.756 0.079

Model 2 RFR 1.493 0.974 1.533 0.125 3684 Red: far-red GroupsPalms 1.412 0.102 13.821 < 0.001

RFR:GroupsPalms -1.511 0.190 -7.935 < 0.001

Intercept 3.500 0.625 5.604 < 0.001 Model 3 BD -5.709 1.642 -3.476 < 0.001 3250.3 Bulk density GroupsPalms -2.416 0.155 -15.634 < 0.001

BD:GroupsPalms 8.692 0.431 20.152 < 0.001

Model 4 Intercept 2.0065 0.1814 11.06 0.001 4122.9 Null model

143

Table A.5. Parameter estimates of fixed effects derived from three individual generalized linear mixed models with Poisson errors. Each model tested the relationship between tree fern and herbaceous fern abundance (Habit) and soil N:P (NP), red: far-red (RFR) and bulk density (BD).

Models were fitted by maximum likelihood, and subplot was included as nested random term.

Overall model: Abundance ~ Environmental factor + Habit + Environmental factor: Habit + (1 |

Plot) + (1 | Plot: Subplot).

Environmental Fixed effects Estimate Std. Error z-value P-value AIC Factors

Intercept 3.305 0.500 6.607 < 0.001

Model 1 NP -0.054 0.031 -1.732 0.083 2585.8 Total N:P HabitTree -3.982 0.106 -37.515 < 0.001

NP:HabitTree 0.163 0.006 27.011 < 0.001

Intercept 3.587 0.594 6.038 0.001

Model 2 RFR -2.147 1.161 -1.849 0.065 2518.9 Red: far-red HabitTree -4.765 0.134 -35.555 < 0.001

RFR:HabitTree 6.413 0.226 28.321 < 0.001

Intercept 3.318 0.573 5.788 < 0.001

Model 3 BD -1.900 1.501 -1.265 0.206 3272.4 Bulk density HabitTree 0.053 0.125 0.420 0.674

BD:HabitTree -4.908 0.389 -12.608 < 0.001

Model 4 Intercept 2.1428 0.2233 9.598 0.001 5449.1 Null Model

144

Figure A.3. Relationship between total abundance and bulk density from generalized linear mixed models. Palm and fern abundances (a), and tree fern and herbaceous fern abundances (b), in response to bulk density. C.I. = Confidence Intervals

145

APPENDIX B: SUPPLEMENTARY MATERIAL FOR CHAPTER 3

Table B.1. Terrestrial fern species present in the Fortuna Forest Reserve and adjacent upland areas of the Palo Seco reserve. This table

combines species from plot and floristic surveys, and from the Fortuna species list. Shading indicates parent material (R/M = rhyolite-

mafic volcanic transition, Undif. = undifferentiated volcanics). n = total number of individuals across all plots. * indicates species not

recorded in the Fortuna species list. x indicates the species occurs along trails or surrounding the plot

Rhyolite R/M Mafic volcanic Basalt Dacite Undif. Families & Species Cho.A Cho.B Hon.A Hon.B Samu. Bonita Palo Ver.A Ver.B Pinola Horn. Alto. n Anemiaceae Anemia hirsuta ------Anemia phyllitidis ------x - Aspleniaceae ------

146 Asplenium auriculatum ------Asplenium cirrhatum ------6 - 28 - 34

Asplenium cristatum ------17 - 17 Asplenium delitescens* ------x - - Asplenium dissectum ------Asplenium gomezianum ------Asplenium harpeodes ------Asplenium holophlebium ------Asplenium laetum* ------3 - - - - - 3 Asplenium miradorense ------Asplenium obtusifolium ------4 - 4 Asplenium pteropus ------x - - Asplenium pululahuae ------Asplenium radicans ------17 - 17 Asplenium repandulum ------Asplenium riparium ------Asplenium rutaceum ------Asplenium sulcatum ------Asplenium sp. - - - - - 5 ------5 Athyriaceae Diplazium atirrense - - - - - 5 3 - 1 - - - 9 Diplazium cristatum ------1 - - 1 Diplazium gomezianum ------

Table B.1. (cont.) Rhyolite R/M Mafic volcanic Basalt Dacite Undif. Families & Species Cho.A Cho.B Hon.A Hon.B Samu. Bonita Palo Ver.A Ver.B Pinola Horn. Alto. n Athyriaceae Diplazium hammelianum ------Diplazium lindbergii ------Diplazium lonchophyllum ------Diplazium macrophyllum - - - - 328 ------328 Diplazium obscurum* - - 4 - - 11 - 1 - - - - 16 Diplazium palmense - - - 1 - 17 ------18 Diplazium paucipinnum ------Diplazium plantaginifolium* ------1 59 - 60 Diplazium prominulum ------2 - 24 21 283 - 330 Diplazium seemannii ------Diplazium solutum* - - - - - 21 ------21 Diplazium striatastrum* ------22 - - 22 Diplazium subsilvaticum - - - 59 51 ------110 Diplazium urticifolium - - 7 3 166 23 22 - 84 21 1 - 327 Diplazium werckleanum ------Diplazium x verapax* ------x - - Blechnaceae 147 Austroblechnum divergens 14 3 2 2 102 9 ------132 Austroblechnum lherminieri - - - - 3 - - - - - 29 - 32

Austroblechnum stoloniferum - - 2 9 ------11 Blechnum binervatum - - 2 2 ------4 Blechnum falciforme ------Blechnum gracile* ------x - - Blechnum occidentale* ------22 22 Blechnum wardiae ------Blechnum werckleanum ------Parablechnum schiedeanum* x x x ------Salpichlaena volubilis - - x x ------x - - Cyatheaceae Alsophila erinacea - - 17 18 18 15 3 2 1 1 2 - 77 Alsophila firma* ------x - - - - - Alsophila polystichoides - - - - - 10 ------10 Alsophila salvinii 3 40 ------43 Cyathea bicrenata ------Cyathea chiricana 12 1 - - - 14 - 2 - - - - 29 Cyathea cocleana ------Cyathea delgagii ------1 - - - - - 1 Cyathea divergens 18 24 27 9 - 6 - 5 - - - - 89 Cyathea eggersii - - - - 3 36 4 12 - - - - 55 Cyathea fulva ------

Table B.1. (cont.) Rhyolite R/M Mafic volcanic Basalt Dacite Undif. Families & Species Cho.A Cho.B Hon.A Hon.B Samu. Bonita Palo Ver.A Ver.B Pinola Horn. Alto. n Cyatheaceae Cyathea horrida - - - 1 8 - 3 - 6 - - - 18 Cyathea multiflora - - - - 13 50 26 15 25 16 16 - 161 Cyathea mutica ------Cyathea nigripes - - 3 ------3 Cyathea pinnula - - 22 13 1 - - - - - 1 - 37 Cyathea rojasiana* 120 140 ------260 Cyathea schiedeana* ------Cyathea sp. - - 5 ------5 Cyathea williamsii ------Sphaeropteris brunei ------Dennstaedtiaceae Blotiella lindeniana ------Dennstaedtia dissecta - - - - - 1 - 1 - - 1 - 3 Dennstaedtia wercklei ------Hypolepis nigrescens ------Hypolepis stuebelii ------Hypolepis viscosa ------148 Desmophlebiaceae Desmophlebium lechleri 26 - - 11 - 2 ------39

Dicksoniaceae Dicksonia sellowiana - - - 1 23 ------24 Didymochlaenaceae Didymochlaena truncatula - - 52 13 62 - 24 14 75 46 77 - 363 Dryopteridaceae Arachniodes denticulate* ------x - - Arachniodes ochropteroides ------Bolbitis hastata* ------7 - 7 Ctenitis hemsleyana ------Ctenitis submarginalis* ------6 6 Elaphoglossum lingua* 1 - - 104 ------105 Elaphoglossum phoras* - - - - 1 ------1 Lastreopsis killipii* ------69 - - 69 Megalastrum atrogriseum - - - 4 12 2 2 4 5 78 89 - 196 Megalastrum biseriale* - - 56 ------56 Megalastrum pulverulentum - - 27 ------27 Mickelia hemiotis ------33 78 - - - 111 Mickelia nicotianifolia ------Mickelia oligarchica - - - 2 1 26 1 1 3 - - - 34 Olfersia cervina 2 - 26 38 - 91 - 1 1 - - - 159 Phanerophlebia juglandifolia ------x - -

Table B.1. (cont.) Rhyolite R/M Mafic volcanic Basalt Dacite Undif. Families & Species Cho.A Cho.B Hon.A Hon.B Samu. Bonita Palo Ver.A Ver.B Pinola Horn. Alto. n Dryopteridaceae Polystichum platyphyllum* ------Polybotrya polybotryoides - - - - 9 ------9 Polybotrya sp. - - - 1 ------1 Stigmatopteris heterophlebia - - - - - 4 28 12 4 - - - 48 Stigmatopteris lechleri ------Stigmatopteris longicaudata* ------Gleicheniaceae Sticherus bifidus ------Sticherus compactus ------Sticherus hastulatus ------Sticherus hypoleucus ------Sticherus intermedius* - - - - 3 - - x x - - - 3 Hymenophyllaceae Abrodictyum rigidum 5 - - - - 19 ------24 Trichomanes crinitum ------Trichomanes diversifrons ------Trichomanes elegans* - - - - - 7 - 2 5 - - - 14 149 Trichomanes ludovicinum ------

Lindsaeaceae Lindsaea arcuata 7 2 17 43 ------69 Lindsaea imrayana - 7 ------7 Lindsaea klotzschiana ------x - - Lindsaea pratensis ------Lindsaea quadrangularis - - - x ------Odontosoria gymnogrammoides ------Lomariopsidaceae Dracoglossum plantagineum ------8 123 - - - 131 Marattiaceae Danaea crispa ------x - - - - Danaea cuspidata ------Danaea elliptica ------Danaea moritziana* - - - - - 28 74 22 4 - - - 128 Danaea nodosa* ------58 - 27 - 3 - 88 Danaea sp. - 2 ------2 Marattia excavata ------Marattia interposita 1 - 2 ------3 Nephrolepidaceae Nephrolepis biserrata ------Nephrolepis multiflora ------

Table B.1. (cont.) Rhyolite R/M Mafic volcanic Basalt Dacite Undif. Families & Species Cho.A Cho.B Hon.A Hon.B Samu. Bonita Palo Ver.A Ver.B Pinola Horn. Alto. n Nephrolepidaceae Nephrolepis pectinata ------2 - - - - 2 Nephrolepis rivularis* - - - - x ------Pteridaceae Adiantum tetraphyllum* ------1 - 2 3 Jamesonia flexuosa ------Jamesonia glaberrima ------Pityrogramma calomelanos ------Pteris altissima - - - 5 1 - 6 3 1 7 - 15 38 Pteris praestantissima* ------x x - - - - Pteris quadriaurita* ------3 3 Pteris speciosa ------Pteris tripartita ------Pteris vittata* ------x - - - Saccolomataceae Saccoloma inaequale - - 24 - - 5 - 4 - - - - 33 Tectariaceae Tectaria chimborazensis ------150 Tectaria murilloana ------

Tectaria nicaraguensis ------Tectaria athyrioides* ------x x - - - - Tectaria pilosa ------3 - - - 3 Tectaria subebenea ------2 2 Tectaria trifoliata ------3 - 3 Thelypteridaceae Macrothelypteris torresiana* ------x - - - Steiropteris valdepilosa ------x - - - - Thelypteris andreana ------x - - - - Thelypteris atrovirens - x ------Thelypteris balbisii ------Thelypteris christensenii ------Thelypteris decussata* - - 10 - 5 10 - 2 - - - - 27 Thelypteris eggersii* ------79 - - 79 Thelypteris frigida* 29 39 - - - 34 - 1 - - 1 - 104 Thelypteris gigantea ------3 - - - - 3 Thelypteris glandulosa ------Thelypteris hatchii* ------189 - 189 Thelypteris lepidula ------Thelypteris leprieurii ------Thelypteris nicaraguensis* ------1 - - - 1 Thelypteris resinifera ------

Table B.1. (cont.) Rhyolite R/M Mafic volcanic Basalt Dacite Undif. Families & Species Cho.A Cho.B Hon.A Hon.B Samu. Bonita Palo Ver.A Ver.B Pinola Horn. Alto. n Thelypteridaceae Thelypteris tetragona* x x ------Thelypteris villana ------151

Table B.2. Coefficients of determination from environmental variables fitted onto a Non-metric

Multidimensional Scaling ordination of herbaceous and tree fern floristic composition.

Coefficients of determination are shown only for statistically significant variables at p ≤ 0.05.

Coefficient of determination (R2) Variables Tree fern Herbaceous fern

Soil moisture 0.90 0.88 Soil N:P 0.70 0.78 Iron 0.67 – Red: far red 0.64 0.50 Total phosphorus 0.62 0.73 Bulk density 0.59 0.69 Sodium 0.50 – Magnesium 0.49 0.67 Dry rainfall – 0.73 Calcium – 0.73 Total extractable bases – 0.72 Effective cation exchange capacity – 0.70 Manganese – 0.62 Aluminum – 0.58 Rainfall – 0.52 pH – 0.49

152

APPENDIX C: SUPPLEMENTARY MATERIAL FOR CHAPTER 4

Table C.1. Parent material assignment of the 12 one-hectare plots at Fortuna forests and associated soil fertility.

Site Code Soil fertility Parent material Description

Chorro A CHA low rhyolite deep layers of rhyolitic tuff with a shallow organic Chorro B CHB low rhyolite layer

Honda A HOA low rhyolite rhyolite with a deep organic Honda B HOB low rhyolite horizon rhyolite/mafic rhyolite to mafic volcanics Samudio SAM intermediate volcanics transition Bonita BON intermediate mafic volcanics Palo Seco PAS intermediate mafic volcanics mafic volcanics with clay- Verrugosa VEA intermediate mafic volcanics rich soils and shallow or no A organic horizons Verrugosa VEB intermediate mafic volcanics B Pinola PIN high basalt no organic horizon Hornito HOR high dacite undifferentiated organic-rich topsoil Alto Frio ALT high volcanics

153

Figure C.1. Principal Component Analysis (PCA) of the soil variables measured in twelve one- hectare plots. Site labels: Chorro A (CHA), Chorro B (CHB),Honda A (HOA), Honda B (HOB),

Samudio (SAM), Bonita (BON), Palo Seco (PAS), Verrugosa A (VEA), Verrugosa B (VEB),

Hornito (HOR), Alto Frio (ALT), Pinola (PIN).

154

limeš 982

2014; K 08

d Hicks, 1 and Watkins, 2016 ., 2014; al., et elissen usek, 20 References Chabot an Wright et al., 2001 Herben et al., Grime, 1991 Corn 997 1 al., et Britton i.e. Amatangelo and Vito use - ogical ogical

based - influence fertile leaf leaf fertile influence ,

offs between plant height and and height plant between offs -

y y

competitive ability of individuals and and individuals of ability competitive offs - de facilitate spore dispersal spore facilitate Functional significance affect plant productivity, plant community community plant productivity, plant affect and cycling nutrient storage, carbon assembly, econom water reflect trade significant stress environmental demographic and physiological and cost, carbon tra physiol with correlates content nutrient leaf mass as such individuals of aspects maximum photosynthetic rate, and resource and rate, photosynthetic maximum efficiency Associated with plant growth strategies

)

1 -

) 1

– (%) us fern - , 0: hemi mg

tration (%) 2 s ons leaves, 0:absence of herbaceo 0: dimorphism -

holo

manganese concentration (%)

Presence of bud Tree fern or sterile sterile Functional trait definition trait Functional 1: buds ratio concentration carbon Leaf Count leaves produced and died over a time interval The distance between leaves and theground (cm) level and abaxial the between dimension The (mm) leaf a of sides adaxial Leafdry matter content (mg g (mm area leaf Specific (%) concentration aluminum Leaf (%) concentration phosphorus Leaf concen potassium Leaf (%) concentration calcium Leaf Leaf concentration phosphorus and carbon Leaf 1: 1: dimorphism and monomorphism and dimorphism al fern species and their definition,unit measurement of are and significance functional

type

binary binary binary uantitative quantitative quantitative quantitative quantitative quantitative quantitative q quantitative quantitative quantitative quantitative quantitative Variable

tion

height

SLA thickness leaf P leaf leaf C leaf leaf K leaf leaf Al leaf leaf Ca leaf LDMC leaf Mn leaf imum leaf C:N leaf clonality x dimorphism eaf eaf growth form growth l leaf produc leaf Functional trait Functional ma

Functional traits from terrestri from traits 2. Functional

C. trait

traits eaf traits eaf le egenerative L Vegetative R Tab listed

155

Table C.3. List of fern species from where trait data were obtained, and accession numbers used to retrieve nucleotide sequence for building a phylogenetic tree. Taxon substitution was based on the published phylogeny from Testo and Sundue (2016)

Species rbcL Substitution taxon rcbL

Alsophila erinacea Alsophila cuspidata AM177323

Alsophila salvinii AM410184 Austroblechnum divergens KU898605

Austroblechnum lherminieri KU898606

Blechnum occidentale AB040565

Cyathea divergens AM177337 Cyathea multiflora AM410197

Cyathea pinnula Cyathea bicrenata KY684766

Cyathea rojasiana Cyathea fulva KY684773

Cyathea schiedeana AM410218

Danaea moritziana EU221756 Danaea nodosa EU221777

Desmophlebium lechleri KT329390

Didymochlaena truncatula AF425105

Diplazium macrophyllum KC254400 Diplazium plantaginifolium EF463314

Diplazium prominulum Diplazium lonchophyllum U05920

Diplazium striatastrum Diplazium striatum KC254395

Diplazium subsilvaticum Diplazium cristatum EF463310

Diplazium urticifolium Diplazium diplazioides KP318925 Dracoglossum plantagineum KC914564

Elaphoglossum lingua Elaphoglossum latifolium AY534860

Lastreopsis killipii KJ464448

Lindsaea arcuata Lindsaea divaricata KJ628724 Megalastrum atrogriseum AF537263

Mickelia hemiotis Mickelia scandens KJ464488

Mickelia oligarchica KJ464489

Olfersia cervina EF463213

Pteris altissima KF289659 Stigmatopteris heterophlebia KU521878

Thelypteris eggersii Goniopteris abrupta EF463280

Thelypteris frigida Thelypteris palustris AB575036

Thelypteris hatchii Goniopteris gemmulifera EF463285

156

Table C.4. Summary of the RLQ analysis and decomposition of inertia from the contributions of environmental factors (rows) and trait variables (columns) in multivariate RLQ analysis. The sum of both environmental factors and trait variables contributions equals a hundred percent.

Total inertia: 3.423 Projected inertia (%): RLQ Axis 1 RLQ Axis 2 76.95 13.73 Eigenvalues decomposition: eig covariance correlation eig1 2.63 1.72 0.57 eig2 0.47 1.35 0.42 Inertia & co-inertia R inertia max ratio eig1 2.70 3.48 0.85 eig1 + 2 4.78 5.50 0.87 Inertia & co-inertia Q inertia max ratio eig1 2.70 3.218 0.838 eig1 + 2 4.14 6.31 0.66 Correlation L: correlation max ratio eig1 0.57 0.83 0.69 eig2 0.42 0.80 0.53 Contribution to the RLQ total inertia [%] Environmental variables soil N: P 26.37 bulk density 18.01 soil moisture 17.67 dry seasonal rain 12.45 manganese 11.38 resin extractable phosphorus 6.68 potassium 4.82 canopy openness 2.63 Trait variables leaf C:N 20.89 leaf P 14.57 leaf thickness 9.42 leaf C 9.22 growth form 7.19 SLA 6.42 clonality 6.09 leaf Al 4.31 leaf Ca 3.96 LDMC 3.73 leaf Mn 3.49 maximum height 3.43 dimorphism 2.74 leaf Mg 2.33 leaf production 2.20

157

Table C.5. Top five ranked models explaining variation in fern community FDis in Fortuna

forests. AICc = Akaike Information Criteria corrected for small sample sizes. Model terms coded as: 1 = soil N: P; 2 = soil manganese; 3 = dry season rainfall; 4 = soil potassium; 5 = dry season

soil moisture. df = Degrees of freedom.

Model Rank Model # df Log-likelihood ratio AICc ∆AICc AICc weight terms

1 145 1, 2 4 29.47 -45.2 0 0.24

2 169 1, 4, 5 5 31.63 -43.3 1.96 0.09

3 5 3 3 26.06 -43.1 2.1 0.08

4 133 1, 3 4 28.35 -43 2.24 0.08

5 149 1, 2, 3 5 31.23 -42.5 2.77 0.06

158

Table C.6. Summary of the statistical tests measuring phylogenetic signals of individual traits. The table describe the variation of each trait along the phylogeny. The observed value is left-tailed tested against the theoretical values that are derived from null models, using the alpha level of 0.05. According Pavoine et al. (2010), the deviation from theoretical values represents the distance between the observed value of the statistic and the theoretical values expected randomly (observed value – mean (theoretical value))/standard deviation (theoretical value). A positive deviation means the statistic is higher than expected, and negative deviation means lower than expected. Data on 33 species were used in permutations.

Observed Deviation from theoretical Traits Test P-value value values SN test 1.00 16.40 0.001 growth form FN test 0.97 7.13 0.001 Ro test 0.03 -5.87 0.001 SN test 0.12 -0.14 0.416 Clonality FN test 0.49 0.33 0.342 Ro test 0.62 2.25 0.03 SN test 0.26 2.46 0.024 Dimorphism FN test 0.70 3.41 0.002 Ro test 0.41 -1.47 0.133 SN test 0.38 0.79 0.235 leaf production FN test 0.54 -0.04 0.521 Ro test 0.57 0.63 0.611 SN test 0.15 -0.79 0.748 max height FN test 0.45 -0.46 0.682 Ro test 0.61 1.10 0.277 SN test 0.18 -0.66 0.703 leaf thickness FN test 0.45 -1.12 0.877 Ro test 0.46 -0.44 0.679 SN test 0.22 -0.53 0.674 LDMC FN test 0.52 -0.76 0.785 Ro test 0.45 -0.58 0.592 SN test 0.16 -0.90 0.814 SLA FN test 0.43 -0.98 0.839 Ro test 0.46 -0.55 0.614 SN test 0.70 5.57 0.004 leaf Al FN test 0.82 2.98 0.003 Ro test 0.10 -3.18 0.002 SN test 0.25 -0.78 0.736 leaf P FN test 0.53 -0.34 0.634 Ro test 0.49 0.00 1 SN test 0.45 0.33 0.346 leaf Mn FN test 0.62 -0.70 0.739 Ro test 0.65 0.94 0.429 SN test 0.24 2.51 0.02 leaf C: N FN test 0.55 2.04 0.032 Ro test 0.46 -0.71 0.513 SN test 0.18 -0.02 0.466 leaf Mg FN test 0.43 -0.94 0.813 Ro test 0.52 0.25 0.816 SN test 0.37 0.62 0.245 leaf Ca FN test 0.58 -0.60 0.722 Ro test 0.48 -0.12 0.922 SN test 0.14 -0.68 0.735 leaf C FN test 0.47 0.10 0.453 Ro test 0.46 -0.57 0.596

159

160

Figure C.2. Phylogenetic tree of 33 species of terrestrial ferns with rbcL nucleotide sequences. Numbers in black are branch length and

in orange are bootstrapping

161

Figure C.3. Graphical representation of the phylogeny and traits of 33 species of terrestrial ferns. Standardized trait values are represented

by the white and black circles. The first three trait variables (from left to right) are represented by state 1 of the binary traits: growth form –

tree fern; clonality – presence of buds; dimorphism – holodimorphic.

Table C.7. Model-averaged standardized coefficients based on conditioned models and standard errors (ΔAICc < 2), 95% confidence intervals, importance of each environmental predictors of terrestrial fern functional dispersion, based on per-variable sum of the Akaike weights (0 to 1 variation) over all conditioned averaged models, and probability. Standardized coefficients estimated from beta distribution errors. Soil K and dry season soil moisture were excluded from model selection and averaging approaches.

Standardized Relative Predictor Std. Error 2.5% 97.5% P estimate importance soil Mn -0.246 0.066 -0.376 -0.117 0.64 < 0.001 soil N:P -0.192 0.069 -0.327 -0.057 1 < 0.01 dry season rainfall 0.204 0.057 0.092 0.316 0.36 < 0.001 phi (Φ) 251.340 103.335 48.808 453.8 – < 0.05

162

REFERENCES

Amatangelo, K. L., and Vitousek, P. M. (2008). Stoichiometry of ferns in Hawaii: Implications

for nutrient cycling. Oecologia, 157(4), 619–627. https://doi.org/10.1007/s00442-008-1108-9

Britton, M. R., and Watkins, J. E. (2016). The economy of reproduction in dimorphic ferns.

Annals of Botany, 118(6), 1139–1149. https://doi.org/10.1093/aob/mcw177

Chabot, B. F., and Hicks, D. J. (1982). The ecology of leaf life spans. Annual Review of Ecology

and Systematics. Volume 13, 229–259.

https://doi.org/10.1146/annurev.es.13.110182.001305

Cornelissen, J. H. C., Song, Y. Bin, Yu, F. H., and Dong, M. (2014). Plant traits and ecosystem

effects of clonality: A new research agenda. In Annals of Botany (Vol. 114, Issue 2, pp.

369–376). https://doi.org/10.1093/aob/mcu113

Grime, J. P. (1991). Nutrition, environment and plant ecology: an overview. In J. Porter and D.

Lawlor (Eds.), Plant growth: interactions with nutrition and environment (pp. 249–267).

Cambridge University Press.

Herben, T., Nováková, Z., and Klimešova, J. (2014). Clonal growth and plant species abundance.

Annals of Botany, 114(2), 377–388. https://doi.org/https://doi.org/10.1093/aob/mct308

Klimeš, L., Klimešova, J., Hendriks, R., and Van Groenendael, J. (1997). Clonal plant

architecture: a comparative analysis of form and function. In H. De Kroon and J. Van

Groenendael (Eds.), The ecology and evolution of clonal plants (pp. 1–29). Backhuys

Publishers: Leiden.

Wright, I. J., Reich, P. B., and Westoby, M. (2001). Strategy shifts in leaf physiology, structure

and nutrient content between species of high-and low-rainfall and high-and low-nutrient

habitats. Functional Ecology, 15(15), 423–434.

163

PLATES

164

165

166

167

168

169

170

171

172

173

174

175

176