The radiation of plants into epiphytic niches has largely been studied by comparing closely related taxa that contain both epiphytic and terrestrial species. While these studies have led to important insights, they leave unanswered an important question - what traits allow for the initial colonization of the epiphytic niche? Our study addressed this knowledge gap in an old-growth coast redwood (Sequoia sempervirens) forest which has few obligate epiphytes but an abundance of accidental epiphytes. We measured structural and water relations traits in four woody species that are largely terrestrial but also commonly exist as epiphytes in the canopy of coast redwood forests: Vaccinium ovatum, V. parvifolium, Gaultheria shallon and Tsuga heterophylla. We also measured these traits on one obligate epiphytic fern Polypodium scouleri, and one obligate terrestrial fern, Polystichum munitum. Traits were measured across a height gradient from ground level to the upper canopy (68.9-92.4 m). We found that accidental epiphytes exhibited many trait shifts as they occupied higher niches. Height above ground explained significant variation in stomatal density, specific leaf area (SLA), leaf thickness and stable isotope composition, which likely are responses to increases in light and decreases in water availability. Despite consistent variation of some traits with height, our study also highlights unique trait combinations for different taxa living as epiphytes. It is extremely rare to study within-species trait shifts that enable normally terrestrial species to occur as epiphytes, and our study uniquely leveraged the lengthy ground-to-canopy height gradient offered by this coast redwood forest to resolve how and why some species are able to make the transition to epiphytism. We found consistent variation in the expression of SLA, stomatal density, turgor loss point and minimum leaf conductance between terrestrial and epiphytic individuals, despite our focal taxa including a wide range of growth forms (i.e. ferns, shrubs and trees). These results highlight the potential importance of trait variation in supporting an evolutionary niche shift towards epiphytism. In contrast, some traits we measured exhibited inconsistent variation between epiphytic and terrestrial individuals, indicating that despite some coordinated trait shifts, unique trait combinations facilitate the colonization of the epiphytic niche.

Site Description

Our study took place in Jedediah Smith Redwoods State Park in Northern California. This property contains ~10,000 acres of old-growth forest at the northern edge of the range of Sequoia sempervirens. The region experiences cool, wet winters followed by warm, dry summers. The average high temperature, in August, is 18.9º C and the average low temperature, in January, is 4 C (weather.gov).  Annual precipitation averages 1803 mm per year (weather.gov). In addition, forest vegetation intercepts coastal fog that frequently moves through this habitat; these inputs comprise approximately 30+% of the overall water budget (Dawson 1998). Since epiphyte colonization is often associated with the time that supporting architectural features have been in place, these groves support the greatest density of vascular epiphytes (Sillett 1999; Williams & Sillett 2007).

Nine old trees were chosen for this study. Diameter, which was measured a few meters above the ground at a height above the basal trunk flare (~2-3 m), ranged from 3.5-5.1 m. These individuals ranged in height from 68.9 to 92.4 m. Prior to climbing, each tree underwent a risk assessment by a team of experienced research climbers and certified arborists. Lines were set using a crossbow with modified arrows connected to a thin monofilament line. Once the line was in place, first a 2mm cord, and then a static climbing rope, was connected to the monofilament line, which was then pulled through the crown (Williams and Sillett 2007). Thin nylon cords were kept in the trees during the study to secure the climbing path, but at the end of the study, this cordage and all research supplies were removed. The approximate coordinates of the study location are 41.76, -124.11. Details of the study trees are not provided to protect the identity of the trees.

Microclimate

In three of the study trees, 5-6 iButtons (DS1923-F5#) that measure temperature and relative humidity (RH) were placed along a vertical transect from the ground (~1m) to the upper crown of each tree (77-87m). Sensors were placed on a shrub near the ground, at the height of the lowest part of the tree’s crown,n and then at additional points along the vertical transect where epiphytes were abundant. Data were logged automatically every 20 minutes for the month of October in 2019. Though the trees were within 1 km of one another, they were grouped in three microsites. One tree in each microsite contained the iButtons. The vapor pressure deficit, which is an important metric of evaporative demand on plant tissues, was calculated using the following equation:

VPD = (1-(RH/100))*SVP                                                     Eq. 1

Where SVP is the saturated vapor pressure, which is calculated using the following equation (Monteith and Unsworth, 2008):

         SVP = 0.61078 e ^ (17.27T / T + 237.3).                                   Eq. 2

Study Species

Three of the five focal species were ericaceous shrubs and lived primarily as terrestrial species: Vaccinium ovatum, V. parvifolium, and Gaultheria shallon. Another focal species,s Tsuga heterophylla, western hemlock, grows primarily as a terrestrial tree that can achieve heights of 50-70 m.  Polypodium scoulei is the only one of our focal species that is an obligate epiphyte. Since it rarely occurs on the ground, we also included a common terrestrial fern, Polystichum munitum, for comparison. We recognize that these two fern species represent more genetic variation than would be expected within a species. While an examination of traits between two ferns and within the other taxa does not represent identical processes, we include this complementary pair of closely related species because it provides an opportunity to compare intra and interspecific trait variation across epiphytic and terrestrial populations.

Sample Collection

Within the crown of each study tree, we selected an approximately even spatial distribution of up to 15 individual epiphytes of each species for detailed study. Since the climbing and collection were labor-intensive, samples were collected from 1-2 trees per day. Samples were collected in the morning, and were then brought back to the field lab, where stems were cut under water and samples were placed in the dark to rehydrate overnight. Terrestrial individuals of the study species were also marked close (<50m) to each of the study trees. We attempted to match the size of the terrestrial and epiphyte plants included in the study. For woody species, collected samples consisted of branches with several terminal branchlets and many leaves,s so that a single sample could be used for all the trait measurements. For ferns, many leaves in a clump were collected and bagged together.

Size Measurements

The length, wid,th and depth of the crown, defined as the part of the plant with leaves, were measured for each of the study individuals for the four woody species. The multiplication of these variables resulted in the crown volume,e which was used to evaluate differences in size between terrestrial and epiphytic individuals. Sample sizes and crown dimensions can be found in Table S1.

Trait Measurements

One-time trait measurements

We made all leaf-based measurements on young mature leaves that were free of noticeable insect or pathogen damage. Leaf dry matter content (LDMC, g g-1), leaf water content (LWC, %) and saturated water content (SWC %) were calculated using leaf weights, while specific leaf area (SLA, cm2 g-1) was calculated by dividing the fresh leaf area (measured with a leaf area meter-CID Biosciences CI-202) by dry mass, which was obtained by oven drying samples at 60C for 3 days. We calculated LDMC as the dry mass divided by the fresh mass of the sample, LWC as the percentage of the dry mass of a fresh leaf sample that is water,er and SWC as the leaf water content calculated after the samples were rehydrated overnight. We measured stomatal density (SD, # mm-2) by coating the underside of the leaf with nail varnish. Once the coating was dry and firm, we peeled it from the leaf surface, placed it under a microscope, and counted the stomata using a 10x objective in a compound light microscope (Parco model 4008, Westland, MI). We calculated stomatal density as the number of stomata divided by the area of the field of view. We measured leaf thickness (LT, µm) from a fresh leaf section that was viewed under the microscope, and wood density (WD, g cm^-3) was measured with the displacement method on sections of the main stem of each sample. Dried and ground samples were analyzed for C and N content (% dry weight), and stable isotope ratios (δ¹³C and δ¹⁵N respectively) were extracted with an elemental analyzer/continuous flow isotope ratio mass spectrometry using a CHNOS Elemental Analyzer (vario ISOTOPE cube, Elementar, Hanau, Germany) coupled with an IsoPrime 100 mass spectrometer (Isoprime Ltd, Cheadle, UK). The long-term analytical precision is 0.08 ‰ for C and 0.14 ‰ for N. For C-isotope analyses, the standard used was V-PDB, and for N-isotope analyses, we used N₂-gas. All isotope analyses were conducted at the Center for Stable Isotope Biogeochemistry at the University of California, Berkeley, CA, USA.

Traits obtained with dehydration trials (minimum leaf conductance and pressure-volume curve metrics)

Using the methodology described in Williams et al. (2017), we generated pressure-volume (PV) curves on a subset of individuals from each species and study tree to understand the dynamics of water retention and drought tolerance. In analyzing the relationship between leaf water potential and the relative water content of the dehydrating sample, we estimated several important water relations traits including the turgor loss point (TLP, MPa), bulk elastic modulus (Ɛ, MPa) and the capacitance before the turgor loss point (CT, MPa^-1) (Tyree and Hammel 1972; Koide et al.  2000). In total, we completed 55 PV curves on a minimum of 10 individuals per species (or functional group in the case of the ferns). The samples were evenly spread across the study trees. Since the sample size was smaller for these metrics, we restricted our analyses to compare terrestrial versus epiphytic individuals.

We also measured minimum leaf conductance (g_min, mmol m^-2 s^-1) because this metric relates to the ability of a species to regulate water loss once the stomata are closed as much as possible (Duursma et al. 2019). To measure minimum leaf conductance, we removed leaves from their stems and sealed the petiole with parafilm to minimize water loss from the cut surface. We weighed the leaves every 30 minutes for 4-6 hours while recording the air temperature and relative humidity. We estimated g_min as the slope of the mass over time divided by the average mole fraction of the vapor pressure deficit taken over the measurement period (Scoffoni et al. 2018). To account for changes in leaf area during dehydration, we calculated g_min using the average leaf area from the beginning to the end of the experiment (Scoffoni et al. 2018).

Data Analyses

For functional traits and stable isotopes, we collected samples from individuals that occupied a range in vertical heights. Since some trait data were not normally distributed, we calculated the Spearman correlation coefficient to evaluate relationships between vertical height in the host tree where a plant was found and functional traits, including leaf thickness, stomatal density, leaf water content,t, and specific leaf area. (Table S2). Variation in stable carbon and nitrogen isotopes with height was also evaluated using Spearman correlation coefficients (Table S3). For water relations measures, which were more labor-intensive and yielded smaller sample sizes (i.,e. PV-curve parameters and g_min), variation between terrestrial and epiphytic niches was analyzed, as opposed to evaluating variation by height. We also performed additional analyses for functional traits, based on the niche occupied. Functional traits exhibited a non-normal distribution; therefore, the effect of niche was evaluated for each trait using a Wilcoxon test.  For g_min, which was normally distributed, we analyzed the effect of taxon and niche (terrestrial vs. epiphyte) using a two-way ANOVA followed by post-hoc one-way ANOVAs on each taxon. For the PV-curve parameters, for each species, we either performed an ANOVA or a Wilcoxon test depending on the distribution ofthe datat, where the effect of niche was analyzed (Table S4). For each group of analyses (i.,e. functional traits, PV-curve parameters, and stable isotopes), we applied a Benjamini-Hochberg procedure for multiple comparisons for each factor to determine the threshold for significance (Q=0.05). Lastly, the percent change of functional traits between epiphytic and terrestrial individuals of the focal taxawase was calculated (Table S5). All data analyses were performed in JMP (Version Pro 17, SAS Institute, Cary, NC, USA).