Introduction

Dry forests are ecosystems with a marked seasonality, with dry periods interrupted by rainy months. Formally, a tropical dry forest (TDF) is defined as one with a closed canopy, where the average annual precipitation is less than 1 800 mm, with periods of three to six months with precipitation of less than 100 mm per month, during which the vegetation is deciduous (^{Gentry, 1995}; ^{Murphy & Lugo, 1986}; ^{Sánchez-Azofeifa et al., 2005}). The topical dry forest also encompasses taxonomic, compositional, and structural variations that depend on climate, topography, and hydrogeography (^{Ramos & Silverstone, 2018}). As these ecosystems normally occur in agriculturally fertile areas, often large human populations have settled there, exacerbating the pressures on these forests, and drastically reducing their extension to only 10 % of the original. In Colombia, dry forest is the most threatened ecosystem, with only 5 % of their original extension left (^{Ramos & Silverstone, 2018}).

The Colombian dry forest is restricted to the inter-Andean valleys, the Caribbean region, part of the plains and rivers. The floristic composition and physiognomy of TDF in Colombia differs markedly between regions, with several endemic species (^{Ramos & Silverstone, 2018}). However, dry forests in Colombia are being destroyed at accelerated rates due to intensive agriculture such as the cultivation of sugarcane, rice and soybeans and transformation into pastures for cattle (^{Ramos & Silverstone, 2018}). This destruction is very noticeable in the region of the geographic valley of the Cauca River, where the sugarcane cultivation and cattle ranching reduced the dry forest to small patches, with an estimated total extension of less than 500 ha. These remnants are largely isolated due to the persistence of the agricultural lands (^{Ramos & Silverstone, 2018}).

One of the most threatened floristic components in the tropics are the epiphytes, plants that use other plants as substrate without parasitizing the host and can be classified into vascular and non-vascular epiphytes (algae, bacteria, bryophytes, and lichens) (^{Mendieta-Leiva, Porada et al., 2020}). Despite the taxonomic heterogeneity of the group, epiphytes show ecological and functional similarities. Epiphytes have important functions in ecosystems ranging from the cycling of nutrients and water to serving as food source for other organisms (^{Benzing, 1992}). The greatest diversity of these plants occurs in the neotropics, with cloud forests being one of the most diverse habitats (^{Gentry, 1995}; ^{Soto et al., 2019}). These organisms depend on tree structure, climatic factors, and other epiphytes for their establishment (e.g., non-vascular epiphytes form litter that allows vascular establishment) (^{Gentry, 1995}). Epiphytes also interact with each other (vascular and non-vascular), with variations according to the type of ecosystem, as in cloud forests, where vascular and bryophytes are dominant compared to lichens (^{Soto et al., 2019}; ^{Wolf, 1993}). However, there is no consensus regarding the relationship between the diversity of vascular and non-vascular epiphytes, showing patterns of positive, negative, or absent relationship (^{Affeld et al., 2008}; ^{Bacaro et al., 2019}). Currently, progress has been made in the study of epiphytes, which has allowed a better understanding of their ecology, conservation, distribution, and large-scale patterns, as well as standardizing sampling protocols (^{Wolf, 1993}; ^{Wolf et al., 2009}). However, due to both taxonomic and sampling complexity of these groups, there are few studies that include vascular and non-vascular epiphytes (^{Affeld et al., 2008}; ^{Bacaro et al., 2019}, ^{Ramos & Silverstone, 2018}; ^{Soto et al., 2015}). Therefore, there are still several gaps of knowledge about the interaction and ecological dynamics of the relationship between vascular and non-vascular epiphytes (e.g., relationship between the diversity of vascular vs. non-vascular epiphytes and their responses to environmental gradients).

Currently, vascular epiphytes in the Neotropics are of high interest due to their high diversity, but not well known in ecology, biogeography and macroecology (^{Mendieta-Leiva, Porada et al., 2020}). However, it is known that epiphytes are less diverse in dry ecosystems, but little is known about their ecology (^{Gentry, 1995}; ^{Mendieta-Leiva, Ramos et al., 2020}; ^{Werner & Gradstein, 2009}). On the other hand, non-vascular epiphytes show growing interest, but are perhaps less well known in the neotropics both taxonomically and ecologically (^{Benitez et al., 2019}; ^{Cornelissen & Steege, 1989}). There are also very few ecological studies that integrate vascular and non-vascular epiphytes, despite their obvious interaction (^{Ramos & Silverstone, 2018}; ^{Soto et al., 2015}). Interest in the conservation of these plants has increased, with greater representation in management plans and conservation priorities. Because of this, it is necessary to contribute to the knowledge of these groups of organisms, especially in poorly studied ecosystems, such as the dry forest.

Bryophytes and lichens are organisms that are an important part of the primary ecological succession, as well as of the water cycle through horizontal rainfall (^{Blum, 1973}; ^{Kappen, 1973}; ^{Proctor, 1982}). Furthermore, they can be used as bioindicator organisms of pollution and disturbance since they are poikilohydrous, that is, their water content depends on the environment (^{Conti & Cecchetti, 2001}). Non-vascular epiphytes strongly depend on deterministic factors such as microclimate and stochastic factors such as dispersal, already showing changes in their community composition with small environmental changes. Similarly, vascular epiphytes also depend on factors such as climate, plant structure, characteristics of the phorophyte and dispersal capacity. Due to these similarities, vascular epiphytes have been used as surrogates for establishing nature reserves, identifying conservation priorities, and for environmental planning (^{Ryti, 1992}). In the dry forests of Colombia, fauna and flora are mainly used as priorities for conservation and to establish guidelines for management plans, under the principle that these groups are substitutes for other groups of organisms. However, there are studies that show that the relationship between the diversity of vascular and non-vascular epiphytes is not consistent, with positive or negative relationships, for which there is not a high congruence between these taxa (^{Bacaro et al., 2019}). Thus, understanding the relationship between the diversity and composition of vascular epiphytes, bryophytes and lichens would help improve conservation strategies in the Colombian dry forest.

The objective of this study was to compare the diversity patterns and species composition of vascular and non-vascular epiphytes along tropical dry forest remnants. The hypothesis is that vascular and non-vascular epiphytes will show similar patterns of diversity with respect to an environmental and disturbance gradient of dry forest.

Materials and methods

Study zone: Sampling was carried out in the dry forest of Cauca Valley, a department in the Southwestern part of Colombia. This department ranges from the Pacific coast to the central and western mountain ranges, with elevations above 4 000 m. The dry forest of the geographic Cauca valley is found along the Cauca River, which is located between the two aforementioned mountain ranges. Ten dry forest remnants (sites) were chosen as sampling sites, mainly in the North-central part of Cauca Valley: Chimbilaco (Yotoco), PNR El Vínculo (Buga), Gitana (Riofrío), Higueroncito (Roldanillo), PNR Mateguadua (Tuluá), El Medio (Zarzal), Valenzuela (Tuluá), La Venta (Bugalagrande), Verdun (Andalucia) and El Volcán (Yotoco) (Table 1). The ten localities present dry forests in different states of conservation, with an elevation range between 993-1 150 m and precipitation between 1 057 and 1 593 mm.

Sampling: For vegetation sampling (forest structure) plots of 50 x 20 m² were established and all the trees with a diameter at breast height (DBH) > 2.5 cm were counted. For each tree DBH, height and number of branches were noted down. Phorophytes with a DBH > 20 cm were selected from the dominant species of each locality. The phorophytes were at a distance > 10 m from the forest edges. The species and number of vascular epiphyte individuals were counted in three vertical strata (stratum 1: from the base up to the first branching; stratum 2: between the first and second branching; stratum 3: after the second branching) (^{Gradstein et al., 2003}). In total, 50 trees were selected for each plot. One specimen of each species was deposited in the TULV Herbarium.

For the sampling of non-vascular epiphytes, three strata were established: base (between 0.5-1 m), trunk (between 1-2 m) and branches (branches were collected from the upper parts) (^{Gradstein et al., 2003}). In each stratum, a 30 x 20 cm² quadrant (divided into 1 cm² cells) was placed and the coverage of each non-vascular species was estimated (^{Déleg et al., 2021}). One specimen of each species was deposited in the TULV Herbarium. For more information on the species found, consult ^{Soto & Moncaleano (2021}). Additionally, Annual Mean Temperature (Bio 1), Annual Precipitation (Bio 12) and Temperature seasonality (Bio 4) were estimated for each forest remnant using the Worldclim databases (^{Fick & Hijmans, 2017}). Layers were used at a resolution of 2.5 minutes and values of the variables were extracted by site (ten forest remnants). Only these variables were used because they were the ones with the highest correlation between them (Pearson R > 0.7) and variation among the remnants (since the study sites are in the same geographical valley).

Statistical analysis: Initially, the abundance of bryophytes and lichens was measured as the cover (cm²) per sampled quadrant, while the abundance of vascular plants was measured as the number of individuals. Alpha diversity of epiphytes was evaluated by the number of species (Hill number q0) and Shannon diversity indices (Hill number q1) per sampling unit (phorophyte) (^{Chao et al., 2014}). Alpha diversity per site was calculated as the average of q0 and q1 of the phorophytes. Gamma diversity was calculated as q0 (richness) and (q1) (index Shannon’s diversity) for the entire remnant (^{McCune & Grace, 2002}) and beta diversity as the Whitaker index for each remnant, which considers the average alpha diversity and gamma diversity of each site (^{Hammer et al., 2001}). In this way, an alpha (average between phorophytes), gamma (Hill number) and beta diversity value was obtained for each remnant. These indices were calculated with the PAST 2.0 ® software (^{Hammer et al., 2001}).

General Linear Models (GLM) were performed to evaluate the relationship between abundance, alpha, beta and gamma diversities, and the environmental variables (temperature, precipitation, and Temperature seasonality, abundance, diversity, DBH and tree height). Previously, Spearman’s correlation analysis (between the predictor variables) was performed to select variables with high correlation (R > 0.7) and avoid collinearity. These analyzes were also made between the alpha, beta, and gamma diversities. The variables followed a normal distribution (Shapiro Wilk: P > 0.05). Statistical analyzes were conducted with the software STATISTICA® 7.0 (^{StatSoft, 2004}).

To evaluate the relationship between the remnants based on their epiphyte composition (similarity), a multidimensional non-metric scaling (NMDS) was performed. The Bray-Curtis similarity index was used to make the NMDS (^{McCune & Grace, 2002}). To evaluate the relationship between environmental variables and epiphyte composition, a Spearman correlation was performed between the NMDS axes and environmental variables. These analyses were conducted with the PAST 2.0 ® software (^{Hammer et al., 2001}).

The 10 remnants were also grouped according to whether they had a conservation status assigned (natural reserve or a similar status) or not (Remnants on farms or without protection), to establish differences in alpha, beta, and gamma diversities of the epiphytes. For this purpose, t-tests were used to compare the remnants with a conservation status vs. no conservation in terms of diversity and abundance of epiphytes. Alpha and gamma diversities were compared between remnants using t-tests (^{Hammer et al., 2001}).

Results

There were 50 morphospecies of vascular epiphytes, 77 of bryophytes and 290 of lichens. The most abundant lichen species were Cryptothecia striata (15 743 cm²), Coenogonium magdalenae (11 887 cm²), Porina imitatrix (10 229 cm²) and Zwackhia viridis (Ach.) Poetsch & Schied (8 451 cm²), while Emmanuellia tenuis, Bacidia sp.2, Bulbothrix isidiza, Byssoloma leucoblepharum, Herpothallon confluenticum, Hyperphyscia minor, Lopezaria sp., Physcia sp., Porina sp., Lichen sp. and Usnea sp. had 1 cm². The most frequent species in the ten remnants were Coenogonium linkii (9 remnants), Herpothallon rubrocinctum, Diorygma australasicum (Elix) Lücking et al., Herpothallon roseocinctum, Zwackhia viridis, Porina imitatrix, Coenogonium magdalenae and Cryptothecia striata (8 remnants), while 175 species were very rare (singletons: occurred in only 1 remnant).

For bryophytes, Lejeunea sp., Radula pallens, Helicodontium capillare, Schiffnerolejeunea polycarpa and Lejeunea subsessilis Spruce were most abundant (13 074, 4 755, 3 488, 3 313 and 3 230 cm², respectively), whereas Plagiochila deflexirama, Bryophyte sp. and Leskeadelphus angustatus, were the least abundant (5, 3 and 3 cm², respectively). The most frequent species were Sematophyllum subpinnatum (Brid.) E.Britton, Neckeropsis undulata, Frullania ericoides (Mart.) Mont., Radula tectiloba and Frulania sp.2., occurring between in 5-6 remnants, while 10 were rare (doubletons: occurred in 2 remnants) and 55 species (71 %) were very rare.

The most abundant species among the vascular epiphytes were Tillandsia recurvata (2 221 individuals), Pleopeltis astrolepis (160), Tillandsia usneoides (159) and Tillandsia juncea (115). The most frequent species were Monstera adansoni (8 remnants), Tillandsia recurvata (7 remnants), Tillandsia juncea (6 remnants), Philodendron barrosoanum and Tillandsia elongata (5 remnants), while 34 species were only present in one remnant.

Diversity 𝛼, β and 𝛾 in dry forest remnants: The remnants with the highest bryophyte gamma diversity (q0) were Mateguadua (24 species) and Verdun (23 species), while El Vínculo was the place with the lowest diversity (six species). In terms of q1, Verdun was the site with the highest diversity, while the Medio had the lowest bryophyte diversity. Alpha diversity was highest in Mateguadua and lowest in Vínculo (Table 2). Verdun was the remnant with the highest beta diversity, while Vínculo had the least beta diversity. The abundance of bryophytes was highest in the Gitana and lowest in the Volcan.

Regarding lichens, it was found that Higueroncito was the place with the greatest diversity 𝛾, while Chimbilaco was the one with the least diversity. Regarding the diversity 𝛼, Mateguadua was the remnant that showed the highest values, while the Medio presented the lowest q1 and Chimbilaco the lowest q0. Diversity β was highest in Verdun, while Vínculo showed the lowest values. The abundance of lichens was highest in Valenzuela (total abundance) and El Volcan (average abundance).

Vascular epiphytes presented a higher diversity 𝛾 in Verdun, while it was lower in Vínculo. Diversity 𝛼 was highest in Verdun (average richness) and Mateguadua (average q1), while it was lowest in Vínculo. Regarding β diversity, the Valenzuela remnant was the one that showed the greatest diversity, while the Vinculo had the lowest values (Table 2). The abundance of vascular epiphytes was highest in Higueroncito (total abundance) and El Vínculo (average abundance per tree), in contrast, La Gitana was the site with the lowest total and average abundance.

Diversities 𝛼, β, and 𝛾 and environmental variables: The gamma diversity of epiphytic bryophytes from each remnant was significantly explained by alpha and beta diversities (Table 3). The diversities 𝛼 and 𝛾 were related with temperature and precipitation, indicating that the sites with the highest precipitation and lowest temperature tended to present the highest diversity of corticolous bryophytes. The average DBH was significant to the 𝛼 diversity, suggesting that the sites with the largest trees tended to present the highest diversity of bryophytes per tree.

The diversity 𝛾 of corticolous lichens was explained by 𝛼 and β diversities. The abundance of lichens per tree was significantly related with the DBH and tree height of each remnant (Table 4). Alpha diversity was affected by temperature.

The alpha diversity of vascular epiphytes was explained by temperature with precipitation (Table 5). Gamma diversity was related to precipitation. Gamma diversity was explained only by alpha diversity. The total abundance of vascular epiphytes per tree showed a negative relationship with the DBH.

When comparing the diversity among the remnants with some figure of conservation vs. without conservation, it was observed that the richness and diversity of trees was greater for these last (Table 6). Regarding the epiphytes, no significant differences were found for the diversities 𝛼, β and 𝛾.

Composition of epiphytes: The NMDS for vascular epiphytes, lichens, and bryophytes, showed different clustering patterns related to the different environmental variables evaluated (Fig. 1). The bryophytes were grouped according to precipitation and temperature, with a group formed by Chimbilaco, El Medio and El Vínculo, and another by Volcán, Venta, Mateguadua and Valenzuela. Regarding lichens, the grouping patterns were related to the coordinates (latitude and longitude), and the parameters related to the vegetation. The NMDS of vascular epiphytes showed that temperature, temperature seasonality and precipitation were the variables associated with the grouping patterns of the remnants, with Chimbilaco, Venta, Higueroncito, El Vínculo forming a group and Valenzuela, Volcán, Gitana and Mateguadua the other.

Fig. 1 NMDS epiphytes from TDF Valle del Cauca. A. vascular, B. bryophytes and C. lichens. 

Discussion

This study is the first to explore the epiphyte diversity in the dry forest of the geographic valley of the Cauca River. The diversity of vascular epiphytes was low compared to other tropical ecosystems (^{Mendieta-Leiva, Porada et al., 2020}), which may be since the conditions of high temperature and low rainfall negatively impacted the abundance and richness of these organisms (^{Gentry, 1992}). Regarding bryophytes, a low diversity was found compared to the Andean life zones, but like other dry forests in Colombia (^{García-Martínez & Mercado-Gómez, 2017}; ^{Lombo-Sánchez & Suarez-Contento, 2018}; ^{Ramírez-Morán et al., 2016}). Regarding lichenized fungi, high diversity was observed compared to other dry forests in Colombia (^{Lücking et al., 2019}, ^{Soto et al., 2021}). However, this diversity is still low compared to the Andean zone and humid forests at low elevations, which may be because the forests have low precipitation and high temperatures (^{Sipman, 1996}; ^{Soto et al., 2019}).

The most frequent and abundant vascular epiphyte species found in this study are common in the dry forest of the geographic valley of the Cauca River, both in artificial and natural ecosystems (^{Benzing et al., 1992}). These species were dominant in remnants with a plant structure with greater dominance than few species, which suggests that these epiphytes are favored by the disturbance.

The most abundant lichen species were Cryptothecia striata, C. magdalenae, Porina imitatrix and Zwackhia viridis. The most frequent species in the ten remnants were Coenogonium linkii, Herpothallon rubrocinctum, Diorygma australasicum, H. roseocinctum, Z. viridis, Porina imitatrix, Coenogonium magdalenae y C. striata, while 175 species were very rare. Thus, the most important species of lichens belong to the Arthoniaceae, Coenogoniaceae, Porinaceae and Lecanographaceae families, typical families in open and disturbed ecosystems. It should be noted that species of the Graphidaceae, Ramalinaceae and Trypetheliaceae families were not dominant, families that are considered the most diverse in the Neotropics (^{Menezes et al., 2018}). On the other hand, there is a high presence of rare to very rare species, which is a common pattern in communities of lichenized fungi in the Neotropics, since many species of lichens have microclimate and microsite specialists (^{Cáceres et al., 2007}).

The abundance of bryophytes was highest in the Gitana and lowest in the Volcan. The greater abundance of bryophytes in La Gitana may be since this site is a dry forest-premontane transition, for which climatic conditions favor a greater abundance and richness of bryophytes, which are photophobic organisms (^{Glime, 2007}; ^{Proctor, 1999}; ^{Proctor & Tuba, 2002}). In contrast, Verdun was the site with the highest diversity, although it was not the one with the highest abundance and richness, which indicates that it is the site with the greatest diversity of bryophytes since the Shannon index corrects the effects of abundances. In this way, it is possible that Verdun is a forest with a higher degree of conservation even though it is immersed in a cattle farm, since bryophytes are indicators of habitat quality (^{Czerepko et al., 2021}).

In contrast, it was found that Higueroncito was the place with the greatest diversity 𝛾 of corticolous lichens, while Chimbilaco was the one with the least diversity. The β diversity was highest in Verdun, while the Vinculo showed the lowest values. The abundance of lichens was highest in Valenzuela (total abundance) and Volcán (average abundance). These results differ from those of bryophytes because the diversity of lichens is modulated by other factors. Higueroncito was one of the sites with the highest temperature of the ten remnants, in addition to a possible effect of a fog current, which favors a high abundance and diversity of lichens (^{Stanton, 2015}). The sites with the lowest rainfall tended to have the lowest lichen diversity.

Vascular epiphytes presented greater diversity 𝛼 and 𝛾 in Verdun, while it was lower in Vínculo. Regarding β diversity, the Valenzuela remnant was the one that showed the greatest diversity, while the Vinculo had the lowest values. The abundance of vascular epiphytes was higher in Higueroncito (total abundance) and Vínculo (average abundance per tree), in contrast, La Gitana was the site with the lowest total and average abundance. The greater diversity of vascular epiphytes in Verdun may be due to the fact that it is a site with premontane forest characteristics, which are characterized by high epiphytism (^{Gentry, 1992}). On the other hand, the Vínculo was a site with a low diversity and high abundance of epiphytes, which suggests that this site presents a high disturbance.

Regarding the diversity of epiphytic bryophytes, 𝛾 diversity of each remnant was significantly related with precipitation and temperature. The diversity 𝛼 was explained by temperature, indicating that the sites with the highest precipitation and lowest temperature tended to present the highest diversity of corticolous bryophytes. The foregoing is in line with what was previously said that the remnants with higher precipitation and lower temperature tended to present the greatest diversity and abundance of bryophytes (^{Grass et al., 2015}). Bryophytes are poikilohydrous organisms since they do not have mechanisms for regulating their water content. This condition makes them organisms with high water / humidity requirements to maintain adequate hydration and maintain biological processes (^{Glime, 2007}; ^{Proctor, 1999}; ^{Proctor & Tuba, 2002}). The average DBH was positively related to the diversity 𝛼 and 𝛾, suggesting that the sites with the largest trees tended to present the greatest diversity of bryophytes. This may indicate that the diversity of bryophytes is higher in potentially more conserved sites (with larger trees), since they are forests with favorable climatic conditions for these organisms (^{Alvarenga & Pôrto, 2007}; ^{Bates et al., 2005}; ^{Bisang & Hedenäs, 2005}; ^{Czerepko et al., 2021}). This response is because bryophytes show particular physiological and ecological aspects related to tolerance to light (photophobic), habitat specialization and reproductive strategies, therefore an alteration of their habitat negatively affects their diversity.

With respect to lichenized fungi, diversity 𝛾 was only explained by 𝛼 and β diversities. The abundance of lichens per tree was positively correlated with the abundance of trees in each remnant. In lichens it has been observed that the total diversity of a place is strongly influenced by the exchange of species among phorophytes (^{Cáceres et al., 2007}; ^{Lücking, 1999}). This pattern is since lichens have a strong microclimate and microsite specialization, which is why a high presence of rare and restricted species is observed.

In general terms, in bryophytes and lichens 𝛾 diversity was explained by 𝛼 and β diversities, but in vascular epiphytes it was only alpha diversity. This pattern is because the sites that have a high disturbance, present a lower diversity, is usually due to a homogenizing effect, that is, a low turnover of species between sampling units (^{Ardila-Ríos et al., 2015}). In this way, even though a high diversity 𝛼 is conserved, the diversity β falls and this is expressed in a low diversity 𝛾; However, in vascular epiphytes, alpha diversity was more important for the total diversity of each site, which can be explained by the fact that microenvironmental factors are less important in modulating the diversity of these epiphytes at a site (^{Soto-Medina et al., 2015}). On the other hand, precipitation and temperature affected the diversity of bryophytes and vascular epiphytes, while it did not show a relationship with the lichen’s diversity. In this way, the results show that there are contrasting patterns between the two groups of epiphytes, which is very important when making conservation decisions: e.g., using only vascular epiphytes to take conservation measures could negatively affect epiphytic lichens. In this sense, these results are an important basis for establishing conservation strategies in tropical dry forests.

Ethical statement: the authors declare that they all agree with this publication and made significant contributions; that there is no conflict of interest of any kind; and that we followed all pertinent ethical and legal procedures and requirements. All financial sources are fully and clearly stated in the acknowledgments section. A signed document has been filed in the journal archives.