Light drives nitrogen fixation in tropical montane cloud forests in Costa Rica

Bryophyte-associated N 2 fixation is a major source of new N in several northern environments, but their contributions to the N cycle in other ecosystems is still poorly understood. In this work, we evaluated N 2 fixation rates associated with epiphytic bryophytes growing along the stems of pumpwood trees ( Cecropia sp.) as well as in surrounding litter and soil from a primary and a secondary cloud forests in the Talamanca Mountain Range, Costa Rica. Nitrogen fixation was significantly higher in substrates from the secondary forest compared to those from the primary forest. Overall, N 2 fixation rates associated with epiphytic bryophytes were 57 times those of litter and 270 times what was measured in soil. Further, light intensity was the major factor influencing N 2 fixation rates in all substrates. Increased access to light in disturbed cloud forests may therefore favor bryophyte-associated N 2 fixation, potentially contributing to the recovery of these ecosystems.

epiphytic bryophytes than litter or soil.
• N 2 fixation was higher in natural regrowth than old growth cloud forests.• Light increased N 2 fixation rates associated with epiphytic bryophytes.• Light influenced the effect of soil nutrients on N 2 fixation in the forest floor.Tropical montane cloud forests are high altitude ecosystems characterized by very high ambient humidity, which favors organisms that depend on the environment for their water status, such as bryophytes and their nitrogenfixing symbionts.Bryophyte-associated N 2 fixation is a major source of new N in several northern environments, but their contributions to the N cycle in other ecosystems is still poorly understood.In this work, we evaluated N 2 fixation rates associated with epiphytic bryophytes growing along the stems of pumpwood trees (Cecropia sp.) as well as in surrounding litter and soil from a primary and a secondary cloud forests in the Talamanca Mountain Range, Costa Rica.Nitrogen fixation was significantly higher in substrates from the secondary forest compared to those from the primary forest.Overall, N 2 fixation rates associated with epiphytic bryophytes were 57 times those of litter and 270 times what was measured in soil.Further, light intensity was the major factor influencing N 2 fixation rates in all substrates.Increased access to light in disturbed cloud forests may therefore favor bryophyte-associated N 2 fixation, potentially contributing to the recovery of these ecosystems.

Introduction
Bryophyte-associated nitrogen fixation is widely recognized as a crucial source of new nitrogen (N) into ecosystems like boreal forests (Rousk et al., 2013), arctic tundra (Rousk and Michelsen, 2017) and peatlands (Vile et al., 2014).Most research in this field has focused on samples from high latitudes in the global north (Alvarenga and Rousk, 2022), leaving the potential contributions of bryophyte associations to the N balance of other regions mostly unexplored.There have been a few reports of significant contributions by bacteria associated with bryophytes to the N pool of alpine forests (Kubota et al., 2023); temperate forests (Zheng et al., 2019), grasslands (Calabria et al., 2020) and bogs (Rousk et al., 2018); as well as tropical rainforests (Cusack et al., 2009;Van Langenhove et al., 2021) and cloud forests (Markham and Otárola, 2021;Fan et al., 2022;Permin et al., 2022;Clasen et al., 2023).Nonetheless, the role of bryophytes and their symbionts in the N balance of these neglected ecosystems remains poorly understood.
As poikilohydric organisms, bryophytes and their N 2 -fixing symbionts are incapable of regulating their water content, thus completely relying on their environment for hydration.In addition, rates of N 2 fixation associated with bryophytes are highly dependent on their wetness levels, and increase with higher temperatures as long as the hosts remain moist (Rousk et al., 2017).Warmer environments capable of providing high humidity could therefore be optimal for bryophyte physiology and N 2 -fixing activities.Such features are found in tropical montane cloud forests, highly diverse ecosystems characterized by elevated humidity levels stemming from a persistent cloud cover or mist.They harbor not only a huge number of endemic species, but also possibly the greatest diversity of mosses and epiphytes in the world (Karger et al., 2021).However, these ecosystems are threatened by the rising temperatures and drier weather resulting from climate change (Permin et al., 2022).
Tropical montane cloud forests can have high N concentrations in soil, which inhibit the expression of nitrogenases and, thereby, N 2 fixation activity (Clasen et al., 2023).This is, nonetheless, highly dependent on the conservation state and age of the forest.The availability of N decreases dramatically after disturbance, being lower in the early stages of succession than mature forests in several climates (Tu et al., 2022), suggesting that older forests may have lower N 2 fixation rates.Additionally, N 2 fixation rates differ considerably between forest compartments like soil, litter, endosphere and episphere (Cusack et al., 2009;Zheng et al., 2017;Zheng et al., 2019;Van Langenhove et al., 2021).This variability results from the intricate interplay between the unique features of these distinct habitats and the environmental conditions they encounter in different ecosystems.In the case of bryophytes, light intensity is possibly one of the most important factors, considering that the N 2 fixation activities associated with these organisms is usually attributed to symbiotic cyanobacteria, photosynthetic microbes that fix large amounts of N in several ecosystems (Markham and Otárola, 2021;Alvarenga and Rousk, 2022).
Bryophytes growing on tree stems exhibit differences in species distribution and associated N 2 -fixing activities along a vertical epiphytic gradient and are shaped by several environmental factors, of which light appears to be the most crucial (Fan et al., 2020).The N 2 fixation rates associated with epiphytes also show fluctuations that are often reported to be controlled by environmental factors like temperature, nutrient availability, moisture and light (Cleveland et al., 2022).However, in comparison with the huge amount of research that has focused on vascular plants, there is currently a very low level of understanding about the different aspects of N 2 fixation associated with epiphytic bryophytes (Cleveland et al., 2022).There are indications that the N 2fixing cyanobacteria associated with epiphytic bryophytes may be a considerable source of new N into tropical montane cloud forests, and could even make up for the N lost in stream discharges (Markham and Otárola, 2021;Permin et al., 2022).Nevertheless, the factors controlling N 2 fixation associated with epiphytic bryophytes as well as their contribution to N budgets in different ecosystems and habitats remain unclear.
In this study, we set out to investigate N 2 fixation rates across a vertical gradient composed of soil, litter and epiphytes along the stems of pumpwood trees (Cecropia sp.) as well as their relationship with light intensity and nutrients in old growth and natural regrowth cloud forests from Costa Rica.We hypothesized that 1) higher N 2 fixation rates would be found in association with epiphytic bryophytes than litter and soil, as trunks are decoupled from the soil and, therefore, have no direct access to soil nutrients.We also predicted that 2) overall N 2 fixation rates would be lower in the old growth forest than the natural regrowth forest, as the latter is at earlier stages of succession.Further, since bryophyteassociated N 2 fixation is usually attributed to cyanobacteria, we expected that 3) their rates would increase with light intensity and, consequently, 4) N 2 fixation would increase with trunk height.On the other hand, as non-photosynthetic bacteria are likely responsible for N 2 fixation in the dark forest floor, we anticipated that 5) light would not influence the N 2 fixation rates in soil and litter samples.

Site description
Sampling sites were established in a primary forest (9  1).Cloudbridge is a private reserve that was established in 2002 for the purpose of regenerating and conserving the tropical montane cloud forest biome.The area within the reserve contains oldgrowth forests as well as formerly deforested cattle pastures and plantations that were either replanted or left to recover by natural regeneration.
The target primary forest is a natural old-growth oak forest with high diversity and dense vegetation, whereas the natural regrowth forest was a former plantation which was left to regenerate since 2008; therefore, its vegetation is much younger and less dense, with more light and fewer trees.The region has a rainy season from the middle of May until November, approximately, and a dry season from December to April.The average relative humidity in the area is higher than 85 % (Kappelle and Van Uffelen, 2006).Altitude is 1947 m above sea level in the old growth cloud forest and 1697 m in the natural regrowth forest.

Monitoring of environmental parameters
TMS-4 microclimate sensors (TOMST, Prague, Czechia) were installed in both cloud forests during the years of 2021 and 2022 to record temperatures.Annual deposition of ammonium (NH 4 ) and phosphate (PO 4 ) in the sites was measured with resin lysimeters (Ackermann et al., 2012;Goth et al., 2019).The lysimeters were prepared by placing an ion exchange resin capsule (Unibest, International, USA) between a 1 cm-thick glass wool in 50 mL centrifuge tubes that were open at the bottom to allow water flow.Two lysimeters were placed in each plot; one hanging from trunks at chest level to catch stemflow nutrients, and another at moss level height in the soil (n = 4).The lysimeters were placed on November 2021 and removed on January 2023.

Sampling and measurement of environmental factors
A conspicuous tree species that was found in both old growth and natural regrowth forests in the Cloudbridge Nature Reserve, the pumpwood (Cecropia sp.), common in humid areas of South and Central America (Berg et al., 2005), was selected as sampling target.Several pumpwood trees presented clear, abundant growth of bryophytes along D.O.Alvarenga et al. their trunks (Suppl.Fig. 1d, e), and were thus selected for sampling in this study.Samples of the epiphytes growing on the pumpwood tree trunks and their adjacent litter and soil were collected in January 2023, roughly in the middle of the dry season.
The pumpwood epiphytes were sampled by carefully removing them with a pocketknife from the tree bark in a 10 × 10 cm square.Samples were taken at five heights up the tree trunk: 0-10, 50-60, 100-110, 150-160 and 200-210 cm, measuring from the soil level.All epiphytic material within these squares was collected, and photosynthetic active radiation (PAR) was measured with an MQ-200 quantum meter (Apogee, Logan, USA) at each height.Four trees from each forest were sampled.The epiphytic communities were moss-dominated, but we did not distinguish between bryophyte groups (mosses or liverworts), as our objective was to gather samples that represented the dominant epiphytic community on the sites.
The litter in an area within 10 cm of the tree roots was collected by hand, and samples of the top 10 cm of the soil underneath were taken with a hand shovel.At each height and location near the trunk, measurements were taken for photosynthetic active radiation, relative humidity and temperatures in the air, soil and epiphyte surface.The samples were placed in individual plastic bags and stored in the dark at 4 • C until processed.

Incubation and acetylene reduction assays
The nitrogenase activity in the different samples was estimated with the acetylene reduction assay (Hardy and Knight, 1967) using four replicates.For this, 0.1 g of the epiphyte samples were weighed in 20 mL glass vials.Litter samples were cut into smaller pieces with scissors and mixed to ensure homogenization, after which 1 g was transferred to 20 mL glass vials.After sieving, 1 g of the soil samples was transferred to 20 mL vials.The vials were kept in an Aralab FitoClima 1200 PLH Plus growth chamber (ARAlab, Rio de Mouro, Portugal) set to 20 • C and 80 % humidity.To prevent the samples from drying out, 1 mL of distilled water was added to them as needed to make sure that moisture did not limit N 2 fixation.
The samples were initially incubated in the dark, at 0 μmol photo-ns⋅m − 2 ⋅s − 1 .After a week, the samples were incubated under a photoperiod of 12 h of light intensities that doubled every subsequent week, ranging from 50 to 100, 200, 400 and 800 μmol photons⋅m − 2 ⋅s − 1 , to verify whether light could be limiting N 2 fixation in the sampling sites.The light in the growth chamber was provided by Solray385 high power LEDs (Valoya, Helsinki, Finland), which have a color temperature of 4500 K and wide spectrum lights with wavelengths ranging from UVA (<400 nm) to far red (800 nm).
At the end of each week, the samples were assessed for nitrogenase activity with acetylene reduction assays, which started with sealing the vials with rubber septas and replacing 2 mL of air (10 % of the headspace) with 2 mL of acetylene gas using sterile syringes and needles.The samples were placed back in the growth chamber for 18-24 h, after which ethylene production in the vials was measured using the Agilent 8890 Gas Chromatography System (Agilent, Santa Clara, USA).These measurements were conducted along with a set of vials containing ethylene gas standards.Empty vials containing 10 % acetylene gas were also analyzed to verify if there was any residual ethylene gas in the acetylene gas, which was subtracted from the assay results.Once the gas chromatography was finished, the rubber septas were removed from the vials and the samples returned to the growth chamber to continue with their regular weekly incubation.

Analysis of C, N and plant-available nutrient content
After the end of the main experiment, the samples were observed under the Olympus BX61 epifluorescence microscope (Olympus, Tokyo, Japan) to visually inspect the presence of cyanobacteria.Next, the samples were transferred to smaller glass vials and dried at 70 • C for one day.The dry weight was recorded on the next day.The epiphytes were pulverized with sterile scissors, while the litter and soil were processed in a crushing machine.The pulverized samples were transferred to tinfoil cups and weighed (8-10 mg for mosses, 10-12 mg for litter and 12-15 mg for soil).Next, the samples were analyzed for total C and N content in the Euro EA elemental analyzer (EuroVector, Pavia, Italy) using orchard leaves (Leco, St. Joseph, USA) as standard.For plantavailable soil nutrient analyses, 10 g of each soil sample (n = 3) were shaken in 50 mL of double distilled water.After the extracts were filtered, the water-soluble phosphate (PO 4 ), ammonium (NH 4 ) and nitrate (NO 3 ) content was determined in the SEAL AA500 continuous flow analyzer (Seal Analytical, Mequon, USA) to evaluate the effects of plantavailable soil nutrients.

Statistical analyses
The data was analyzed with R version 4.2.2 (https://www.R-project.org/) using the RStudio IDE 2022.12.0 (http://www.rstudio.com/)and visualized with the packages ggplot2 3.4.3(Wickham, 2016) and patchwork 1.1.3(https://patchwork.data-imaginist.com).The normality of the data was verified with Q-Q plots and Shapiro-Wilk's test (Shapiro and Wilk, 1965).The relationship between acetylene reduction, trunk height, light intensity and nutrients was verified with linear regressions and non-linear least squares (Levenberg, 1944).The Kruskal-Wallis test (Kruskal and Wallis, 1952) was used for evaluating the differences between nitrogenase activity associated with different substrates, and Dunn's test (Dunn, 1964) was used as a post-hoc analysis to identify groups presenting significant differences.One-way analysis of variance (Fisher, 1918) tests were performed to evaluate differences between the nitrogenase activity in the two forest types.Differences between means in ANOVAs were identified with Tukey's Honestly Significant Different post-hoc test (Tukey, 1949).

Characterization of environmental factors
The natural regrowth and old growth cloud forests differed significantly in some environmental factors.At the time of sampling, the old growth forest had an average light intensity of 8.1 ± 1.7 (standard error) μmol photons⋅m − 2 ⋅s − 1 , which was significantly darker than that of the natural regrowth forest (16.4 ± 1.2 μmol photons⋅m − 2 ⋅s − 1 ) (one-way ANOVA, p < 0.001, F 1,38 = 15.45).The old growth forest also had a slightly lower ambient temperature (13.8 ± 0.1 • C on average) than the natural regrowth forest (15.9 ± 0.2 • C) at the time of sampling (one-way ANOVA, p < 0.001, F 1,6 = 92.03).These temperatures were within the annual average temperatures obtained with TOMST sensors, which indicated an average of 15-18 • C for the natural regrowth site and 14-16 • C for the old growth forest.Differences between the sampled cloud forests were also observed in the light intensities the epiphytic bryophytes were exposed to along the pumpwood trunk: while the height range of 0 to 200 cm was not sufficient to produce a gradient of differing light intensities in the old growth forest (Fig. 1a), the epiphytes in the natural regrowth forest were exposed to increasing light intensities in higher heights (Fig. 1b).
On the other hand, samples from the old growth and natural regrowth forests did not differ significantly in nutrient content.Atmospheric deposition of NH 4 averaged 1.5(±0.3)and 1.6(0.6)kg⋅ha − 1 ⋅year − 1 in the old growth and natural regrowth cloud forests, respectively, while these forests, respectively, received 2.8(±0.6)× 10 − 2 and 2.4(±0.3)× 10 − 2 kg⋅ha − 1 ⋅year − 1 of PO 4 on average by deposition (Table 1).The concentrations of available NH 4 , NO 3 and PO 4 were similar across soil samples collected in the different forests (oneway ANOVA, p = 0.68, p = 0.56 and p = 0.27, respectively) (Table 2).Likewise, no significant differences between the C and N content in the old growth and natural regrowth forests were observed in most of the D.O.Alvarenga et al. substrates, except for C content in litter samples, which had a larger average percentage in the old growth forest (54.1 ± 0.6 %) when compared against the natural regrowth forest (48.2 ± 1.9) (one-way ANOVA, p = 0.023, F 1,6 = 9.11) (Table 2).

N 2 fixation between substrates and forest types
Independent of successional stages and light intensities during incubations, N 2 fixation rates associated with pumpwood epiphytes were significantly higher than those in adjacent litter and soil, whose rates did not differ significantly from each other (Fig. 2a).The mean ethylene production between all measurements was 459(±80.1)nmol ethyl-ene⋅g − 1 ⋅h − 1 for epiphytes, which was approximately 57 times higher than the average of 7.95(±1.9)nmol ethylene⋅g − 1 ⋅h − 1 for litter, and 270 times higher than soil, with an average of 1.70(±0.9)nmol ethyl-ene⋅g − 1 ⋅h − 1 (one-way ANOVA, p < 0.001, F 2,381 = 31.98).
Regarding cloud forests with distinct conservation statuses, significantly higher N 2 fixation rates were found in the natural regrowth forest than the old growth forest across all light intensities (Fig. 2b).Considering all substrates together, the average N 2 fixation activity in the natural regrowth forest was 557(±45.6)nmol ethylene⋅g − 1 ⋅h − 1 , a number approximately 5.5 times higher than the average of 101(±10.8)nmol ethylene⋅g − 1 ⋅h − 1 in the samples from the old growth forest (oneway ANOVA, p < 0.001, F 1,334 = 83.13).
Differences in N 2 fixation rates between the cloud forests were reflected in the epiphytic samples, but not in litter and soil.Epiphytes from the natural regrowth forest produced an average of 778(±56.9)nmol ethylene⋅g − 1 ⋅h − 1 , significantly higher than the 140(±14.7)nmol ethyl-ene⋅g − 1 ⋅h − 1 produced by epiphytes from the old growth forest (one-way ANOVA, p < 0.001, F 1,238 = 117.77).For litter samples, no significant differences were found between the average of 10.3(±3.3)nmol ethyl-ene⋅g − 1 ⋅h − 1 for those from the regrowth forest and the 5.6(±1.6)nmol ethylene⋅g − 1 ⋅h − 1 found in the old growth forest samples (p = 0.21).Similarly, soil samples from the natural regrowth forest had an average production of 3.35(±1.8)nmol ethylene⋅g − 1 ⋅h − 1 , not differing significantly from those in the old growth forest, which produced 0.348(±0.1)nmol ethylene⋅g − 1 ⋅h − 1 (p = 0.09).

N 2 fixation and environmental factors
No significant relationship between trunk height and N 2 fixation in samples from either forest was found.Nevertheless, we did find a positive relationship between N 2 fixation and the light intensity each epiphytic sample was originally exposed to in the field (Fig. 3a, Suppl.Fig. 2).Further, there was a significant, positive relationship between N 2 fixation and increasing incubation light intensity for epiphytic samples (Fig. 3b, Suppl.Fig. 3) as well as soil from the natural regrowth forest (Fig. 4a) and litter from the primary forest (Fig. 4b).
Light intensities were also related to how soil nutrients interacted with N 2 fixation rates in some samples.Non-linear least square analyses suggested that there was a negative relationship between PO 4 concentrations and N 2 fixation in soil samples when acetylene reduction assays were incubated only under the high light intensities of 400 and 800 μmol photons⋅m − 2 ⋅s − 1 (Suppl.Fig. 4a).Soil PO 4 was also negatively related to N 2 fixation associated with epiphytic bryophytes growing at a trunk height of 0-10 cm irrespective of light intensity (Suppl.Fig. 4b).Such relationships were not found in litter samples.

Table 1
Average annual atmospheric deposition of ammonium and phosphate on habitats related to pumpwood trees from Costa Rican cloud forests.The numbers in brackets represent standard errors.n = 4.

N 2 fixation in different substrates and forests
In accordance with our first hypothesis, we observed much higher N 2 fixation rates associated with epiphytic bryophytes than adjacent litter or soil in both successional stages (Fig. 2a).Cusack et al. (2009) also observed that N 2 fixation in two Puerto Rican humid tropical forests was highest in epiphytic and ground-dwelling mosses, followed by litter, and lowest in epiphylls, lichens and soil.Similarly, Zheng et al. (2017) found higher N 2 fixation rates in epiphytic bryophytes than in litter and soil of a subtropical forest.On the other hand, while Van Langenhove et al. (2021) also found that soil of a lowland tropical forest from French Guiana presented the lowest N 2 fixation rates when compared with litter and epiphytes, they observed higher rates in litter than trunk bryophytes.This discrepancy may be related to the abundance and ecological role of bryophytes in these different ecosystems, as montane forests (such as cloud forests) usually host more epiphytes than lowland forests (Van Langenhove et al., 2021).
We have found higher N 2 fixation rates in the natural regrowth forest than the old growth forest (Fig. 2b), thus confirming our second hypothesis.This suggests that N 2 fixation associated with epiphytic bryophytes may have a relatively larger contribution to total N input in cloud forests during earlier stages of succession.This pattern has been reported before for root nodule symbionts in some tropical forests (Gehring et al., 2005;Batterman et al., 2013;Taylor et al., 2019), but is unlike what has been observed in boreal forests for moss-associated N 2 fixation.Disturbance in boreal forests causes a dramatic reduction in N 2 fixation, which commonly increases with succession due to lower nutrient levels and larger moss biomass in older sites (Zackrisson et al.,   2004).As boreal forest bryophytes are usually more abundant on forest floors than tree trunks (Turetsky et al., 2010;Renaudin et al., 2022), differences in ecosystem characteristics, environmental conditions, nutrient content and habitat traits might explain these contrasting contributions.

Influence of light on N 2 fixation
Though different light intensities were verified along the natural regrowth forest tree trunks, the height limit of 200 cm was likely not enough to establish a vertical light gradient in the old growth forest (Fig. 1).Light gradients likely stretch vertically as forests grow older and trees become taller, thus requiring wider ranges to properly represent differences in luminosity along primary forest trunks.Nevertheless, when all data was considered, our hypothesis that N 2 fixation rates associated with epiphytic bryophytes would increase with light intensity was indeed confirmed both when light conditions originally in the field were considered as with those in laboratory experiments (Fig. 3).Further, samples that were adapted to higher light intensities in the field (i.e., those collected in the natural regrowth forest) were able to remain active later in the acetylene reduction assay even under dark conditions (Suppl.Fig. 2a), suggesting a lingering effect of light exposure.
Van Langenhove et al. ( 2021) have found that the amount of N 2 fixed in association with canopy bryophytes was indeed almost four times the amount on tree trunks, possibly also resulting from higher light intensities in the canopy.Light is fundamental for the growth and development of photoautotrophic organisms like cyanobacteria, as it drives the production of chemical energy and biomass (Gris et al., 2017;Tiwari et al., 2022).As cyanobacteria are the only known photosynthetic N 2 fixers, our results suggest that symbiotic cyanobacteria might be the main N 2 fixers also in this habitat.Phyllosphere cyanobacteria were indeed shown to introduce significant amounts of new N into Costa Rican rainforests and to increase their activity with higher light intensities (Freiberg, 1998), and they could possibly persist on fallen leaves in the litter and occasionally fix more N 2 when exposed to light.Cyanobacteria are also found as efficient N 2 fixers in soil and rhizospheres, increasing soil fertility (Mishra et al., 2021).
Our hypothesis that light would not affect N 2 fixation in soil and litter was disproved (Fig. 4), even though this influence was dependent on the conservation status of the investigated cloud forests.Further, high light intensity negatively impacted the relationship between N 2 fixation in parts of the forest floor with plant-available PO 4 (Suppl.Fig. 4).P was shown to stimulate the activity of N 2 -fixing cyanobacteria, for which P can be a limiting element in several environments (Chen et al., 2012;Zhang et al., 2022;Clasen et al., 2023).Thus, it is surprising that higher PO 4 concentrations would be related to a decrease in N 2 fixation in the epiphytic bryophyte samples growing close to soil level.It is likely that this is not a direct effect of PO 4 on bryophyte-associated cyanobacteria, however, but rather an indirect consequence caused by a response from the microbial communities around them.Further research is necessary to properly evaluate the complex interplay between P availability and environmental factors influencing cyanobacterial activity and N 2 fixation in cloud forest floors.

Potential contribution to N budgets in cloud forests
Epiphyte biomass varies according to forest type, conservation level and tree species, for example.The biomass of epiphytic bryophytes in different cloud forests has been estimated as varying from 0.5 to 1253 kg⋅ha − 1 (Gotsch et al., 2016).Considering these estimated values and the widely used theoretical factor of 3:1 to convert from nmols of acetylene reduced to nmols of N fixed (Hardy et al., 1968), our results suggest that microbes associated with epiphytic bryophytes could contribute to primary cloud forests with anywhere between 0.002 and 7 kg N⋅ha − 1 ⋅year − 1 .Secondary forests, on the other hand, could receive the much more impressive amounts of 40 to 279 kg N⋅ha − 1 ⋅year − 1 in total from these organisms.These estimations, however, assume constant and homogeneous activity across complex habitats under ideal conditions.The actual contributions of these organisms to N 2 fixation in cloud forests are likely to significantly vary according to environmental conditions, seasonal variation, disturbance levels and differences in host composition and colonization patterns.
These figures could potentially be larger, as it is possible that more N 2 is fixed by epiphytes growing on higher trunk heights.Further, the extended surface area in taller trees from primary forests could host larger epiphytic biomass, thus potentially enabling higher contributions to N 2 fixation.However, the distribution of bryophytes in the episphere does not follow light so closely, as vertical migration presents considerable challenges for these organisms.The vertical distribution of bryophytes depends on adaptations to different episphere microhabitats, which are reflected in their morphological, physiological and photosynthetic traits (Fan et al., 2020), and these do not necessarily translate into abundance.Indeed, Markham and Otárola (2021) surveyed a high elevation montane cloud forest in Costa Rica and found that 67 % of epiphytic bryophytes occurred on lower branches, 24 % occurred on the tree trunk and 7 % occurred on the upper branches.Nardy et al. (2024) have also found higher richness of vascular epiphytes in cloud forest canopies than trunks, resulting from the influence of biotic and abiotic factors.Hence, though our results do suggest that microbes associated with epiphytic bryophytes are an important source of new N into the early successional stages of tropical montane cloud forests, complementary investigations into the bryophyte biomass and associated N 2 fixation rates of these ecosystems are highly recommended.

Conclusions
Light has a strong, positive influence on N 2 fixation associated with epiphytic bryophytes in tropical montane cloud forests, but the extent of its impact may be dependent on conservation status.Potentially, N 2 fixation can be favored by increased light intensity in early succession cloud forests and contribute to their faster recovery, thus resulting in higher N inputs when compared with old growth forests.This heightened N input in early succession forests may play a crucial role in ecosystem resilience and recovery following disturbances.The complex interplay between light availability, canopy structure and N 2 fixation dynamics in cloud forest ecosystems remains ambiguous, however.A comprehensive survey of the abundance, biomass and area covered by epiphytic bryophytes in the cloud forests is therefore necessary, although difficult, for an accurate upscaling of the N 2 fixation rates estimated in this work to the ecosystem level.Understanding the spatial distribution and ecological role of these organisms can therefore provide valuable knowledge into the functioning of cloud forest ecosystems and inform conservation and management strategies.

Fig. 1 .
Fig. 1.Linear regressions evaluating the relationship between light intensity and sampling height on pumpwood tree trunks from an old growth (a) and a natural regrowth (b) cloud forests from the Talamanca Range, in Costa Rica.n = 20.

Fig. 2 .
Fig. 2. Nitrogenase activity estimated by the acetylene production assay in samples from Costa Rican cloud forests.Averages are represented by bars and standard errors by lines.(a) N 2 fixation in different substrates associated with pumpwood trees.Significant differences between substrates according to Dunn's test (p < 0.05) are represented by letters.n = 48, 48, 240.(b) N 2 fixation in old growth and natural regrowth cloud forests.Significant differences are indicated by letters (Tukey's HSD test, p < 0.05).n = 168.

Fig. 3 .
Fig. 3. Linear regressions indicating relationship between (a) light intensity measured at the moment of sampling or (b) light intensity used for incubations in growth chambers with nitrogenase activity in epiphytic bryophytes from Costa Rican cloud forests.n = 240.

Fig. 4 .
Fig. 4. Relationship between N 2 fixation in soil (a) and litter (b) from cloud forests and photosynthetic active radiation in incubations.n = 24.

Table 2
Averages and standard errors for carbon and nitrogen content in soil, litter and epiphytic bryophytes from cloud forests in Costa Rica.Letters represent statistically significant differences (p < 0.05) in ANOVA tests.n = 4.