Lichen Responses to Disturbance: Clues for Biomonitoring Land-use Effects on Riparian Andean Ecosystems

Abstract

The transformation of natural ecosystems due to anthropogenic land use is considered one of the main causes of biodiversity loss. Lichens, due to their poikilohydric nature, are very sensitive to natural and anthropogenic disturbances. Therefore, lichen communities have been widely used as bioindicators of climatic and environmental changes. In this study, we evaluated how the species richness and community composition of epiphytic lichens respond to land-use intensity in riparian ecosystems of the Andes in southern Ecuador. Additionally, we evaluate how the richness of six functional traits (photobiont type, growth form, and reproductive strategy) changed across the different land-use intensity. We selected 10 trees in twelve sites for a total de 120 trees, equally divided into four riparian land-use intensities (forest, forest-pasture, pasture and urban). We recorded a total of 140 lichen species. Species richness was highest in the forest sites and decreased towards more anthropogenic land uses. Lichen community composition responded to land-use intensity, and was explained by microclimate variables (e.g., precipitation, percentage forested area) and distance to the forest. Richness of functional traits of lichens also differed significantly among the four land-use intensity and decreased from forests to urban land-use. Taxonomic diversity and functional traits can be effectively applied as bioindicators to assess and monitor the effects of land-use changes in the riparian ecosystems of tropical montane regions.

1. Introduction

Ecosystem transformation due to anthropogenic disturbances such as land-use change is considered one of the main drivers of biodiversity loss [1,2,3,4]. Several studies state that current land-use practices will generate major impacts on the structure and functioning of ecosystems at different geographic scales [5,6,7], modifying biotic and abiotic conditions, increasing replacement rates and affecting local extinction [4,5,6,7,8]. Most of the tropical primary forests (e.g., montane Andean forests) were transformed to secondary forests, pastures, cropland, and urbanized areas to satisfy human needs [2,6,9]. In consequence, the diversity of plants, lichens, birds, arthropods, fish, reptiles and mammals has been diminished because of land use [4,10,11,12].

Epiphytic lichens are an essential component of tropical humid forests because of their important role in water and nutrient cycles [13]. Therefore, they are key organisms facilitating crucial ecosystem processes [10,14,15]. Epiphytic lichens are poikilohydric, thus they lack an active regulation of the loss and absorption of water [16], which in turn increases their sensitivity to environmental disturbances [17,18]. For these reasons, several biological traits of lichens as such as photobiont type, growth, reproduction, and development can be affected by environmental changes [19,20,21,22]. In this manner, these functional traits can be used as a complementary approach to better understand ecosystems because they allow us to assess the biodiversity and their relationship with ecosystem functioning [23,24].

As a general pattern, microclimatic variables (e.g., light, humidity, and temperature) that change in response to ecosystem transformation have been considered as constraining factors over the taxonomic and functional diversity of epiphytic lichens [21,25]. Therefore, several studies have used both taxonomic and functional diversity (e.g., richness of each functional trait) as indicators of land-use change [20,22,26,27], forest disturbance [14,21,28], global warming [29] and air quality [19,30,31,32].

In temperate ecosystems, most of the research has been focused on assessing the effects of different land uses on taxonomic and functional diversity of lichens [5,20,26,27]. Whereas in the tropics, studies have been restricted to determine the effects of disturbance and atmospheric pollution on lichen diversity [21,25,33,34]. This research is the first in Ecuador to quantify the response of taxonomic diversity and functional traits of epiphytic lichens to different land-use intensities along the riparian ecosystems of southern Andean streams. This area is affected by rapid forest transformation [9,35] and its riparian zones present a high degree of alteration [12,36,37]. Furthermore, we demonstrate, as in other regions (e.g., [20]), the complementary application of lichen’s taxonomic and functional traits to detect and monitor changes in the structure and functioning of tropical Andean ecosystems. We predicted that intensive land-use can decrease the taxonomical diversity of species (richness and composition) and functional traits, and that these changes are related to alterations in the microclimate. Specifically, we addressed the following questions: 1) how does the species richness and community composition respond to land-use intensity? And 2) how does the richness of each functional trait respond to land-use intensity?

2. Materials and Methods

2.1. Study Area

This study was conducted around the city of Loja (180,000 inhabitants) in the southern Ecuadorian Andes, at altitudes between 2200 and 2400 masl (Figure 1). The climate is humid tropical with a mean annual temperature between 7–20 °C, and annual rainfall between 800–2500 mm [38].

Since the 1960s, these native forests have been transformed into agricultural and urbanized land [31,36,37]. In this study, we distinguished four different land uses, i.e., forest, a combination of forest and pasture, pasture, and urban land use, where we established 12 sampling sites located in riparian margins influenced by those land uses mentioned before. Forest sites (Fo) are characterized by a dense canopy layer of evergreen montane tropical vegetation (ca. 72%–78% coverage). The upper canopy is composed of native trees of the genera Croton, Hedyosmum, Clusia, Morella, and Juglans. Forest-Pasture sites (FP) are characterized by a semidense canopy layer (ca. 52%–60% coverage). The disturbed forests are mixed with pastures dominated by planted trees of Alnus acuminata and Eucalyptus globulus. Pasture sites (Pa) are affected by deforestation. The canopy layer is ca. 31%–34% in coverage, mainly composed of planted trees of the genera Inga and Eucalyptus. Urban sites (Ur) are characterized by a very uniform structure and dominate sections of grassland with planted trees of Salix spp. The open canopy layer is ca. 31%–34% in coverage.

2.2. Data Collection

In each site (n = 12), we selected 10 mature trees with similar bark structure and diameter at breast height (DBH) over 20 cm within each site (120 trees total). They were selected about five meters from the river bank. We determined the occurrence of epiphytic lichens on 120 trees in total (10 trees per site). We used 20 × 50 cm quadrat on the bark of each selected tree, at the cardinal point with the most lichen abundance to 1.5 m above the ground. The sampling quadrat (20 × 50 cm) was divided into ten grids of a 10 cm × 10 cm, and the cover of each species in each grid was estimated as the proportion of the ten grids occupied by it. Lichen species cover was used as a surrogate of species abundance. For species identification, we used taxonomic and floristic papers [39,40,41,42]. In addition, we tested for specific secondary compounds using spot tests based on thallus fluorescence under ultraviolet light, with K (10% water solution of potassium hydroxide), Cl (bleach) and para-phenylenediamine (Pd). For the nomenclature of the species we followed mainly MycoBank. Finally, the specimens are stored in the Herbarium of Universidad Técnica Particular de Loja (HUTPL).

We evaluated six traits to perform the analysis of functional traits: (1) photobiont type; (2) growth form; (3) size; (4) reproduction type; (5) type of reproductive structure; and (6) thallus color (

Table 1. List of the six functional traits and their categories of the epiphytic lichen community in four riparian land uses of Andean streams in southern Ecuador.

Functional Trait	Categories
Photobiont type	C = Chlorococcoid; CY = Cyanobacteria: T = Trentepohlia
Growth form	CR = Crustose; F = Filamentose; FB = Foliose with broad lobes; FN = Foliose with narrow lobes; FR = Fruticose; G = Gelatinose
Thallus size	MA = Macrolichens; MI = Microlichens
Thallus colour	L = Light; D = Dark
Reproduction type	A = Asexual; S = Sexual; AS = Asexual and sexual
Reproductive structure	AP = Apothecia; I = Isidia; L = Lirellae; P = Perithecia; SO = Soredia

). The functional traits were selected based on previous studies, due to their relation to ecosystem functioning and land uses [5,19,21,22,25]. For instance, photobiont type and growth form (thallus morphology) are related with light, temperature and water requirements for the processes of photosynthesis and respiration [23,43,44,45], and with water uptake and loss [21,44]. Finally, reproductive strategy and reproductive structure are related to dispersion ability and establishment [22,25].

2.3. Environmental Variables

We quantified aerial forests cover (%) for each site from a 2016 land uses map obtained in other study from high-resolution (0.30–2 m) imagery (pers. comm.). Land use was classified into forest, pasture, crops, urban, bare surface. In a GIS, we extracted the land use information for all sites, and the area (m² and %) covered by forest was calculated. Also, we calculated the distance to forests for each site. Forest cover and distance to forests were obtained using the open source GIS software Quantum GIS 1.7.4 (QGIS). Light conditions were recorded by measuring canopy openness (%) using four digital hemispherical photographs per site. Digital photographs were always taken on overcast days and at breast height (1.3 meter), with a horizontally leveled digital camera and using a fish-eye lens. Photographs were analyzed using the software Gap Light Analyzer ver. 2.0 [46]. In addition, the following variables were measured: mean precipitation (mm) from the interpolated data of the meteorological stations located around the study area and tree diameter (DBH).

2.4. Data Analysis

Species richness of epiphytic lichens was defined as the total number of species found in each tree. We calculated sampling completeness with the rarefaction curves (95 % confidence intervals) and the Chao 2 species richness estimator. For the calculation of the rarefaction curves and species richness estimators, we used the R package ‘vegan’. Similarly, the species richness of each functional trait category was calculated as the total number of species with each trait category found in each tree. Thus, the effects of land use, % forests, distance to forest, precipitation, DBH and tree high on the species richness and the richness of each functional trait at the tree level were modeled by generalized linear mixed models (GLMMs) using Poisson distribution [47]. This modeling approach was chosen because our data present a hierarchical structure with sites nested within land use and trees nested within sites. Predictors were included as explanatory variables (fixed factors), and sites were included as random sources of variation. We used a logistic transformation for canopy openness and % forests variables [48]. Following this, we performed a stepwise best-model selection using a stepwise regression backwards, with predictors variables scaled and centered (mean = 0, SD = 1). Canopy openness was omitted in the model selection due to collinearity with % forests and precipitation. For GLMMs, the minimal adequate model was selected based on Akaike’s Information Criterion (AIC) provided for the model selection procedure. We used the package ‘nlme’ with lme function [49] for the mixed-effect model analyses in R environment [50].

Shifts in lichen species composition were evaluated through a canonical correspondence analysis (CCA) based on chi-square distances. CCA analysis was applied to an abundance matrix square-root transformed. Land-use intensity was established as the covariate and remaining variables were used as explicative variables. Prior to the analysis, we applied a logistic transformation to explanatory variables % forest and distance to the forest; and a square-root transformation to precipitation.

Variation in lichen species composition at the tree level in relation with measured environmental variables was explored by constrained ordinations [51]. As the first step, our data set (120 trees × 140 species) was subjected to a detrended correspondence analysis (DCA) to determine the most appropriate constrained ordination. Due to the length of the first DCA axis was 7.84 standard deviation units, we used a canonical correspondence analysis (CCA) to test the null hypothesis that species composition is independent of environmental variables [52]. CCA analyses were conducted between environmental variables and the epiphytic lichens abundances (square-root transformed). Land-use intensity was established as a covariate and remaining variables were used as explicative variables. Prior to the analysis, we applied a logistic transformation to explanatory variables % forest and distance to a forest; and a square-root transformation to precipitation. The model selection for the ordination was determined with a stepwise procedure based on a permutation test (using the ordistep function and 999 permutations). Then, the significance of explanatory variables on ordination was evaluated with ANOVA-like permutation test for Canonical Correspondence Analysis (using anova.cca function and 999 permutations). Additionally, the Pearson correlation coefficient of the explanatory continuous variables with the first two axes of the ordination (CCA1 and CCA2) was calculated. Also, in order to identify characteristic species in each ordination axis and relate them to predictor variables, we extracted the scores of the lichen species that were found at the end of the axes. For ordination analysis we used the ‘vegan3d’ package [53].

The IndVal function in the labdsv package [54] were used for indicator species analysis (ISA) [55] to determine individual species that are mainly associated with one land-use intensity. The indicator value ranges from 0 (one species was absent from one land-use intensity) to 1 (one species occurred in all trees of one land-use intensity and was absent from other trees). The significance was tested using a Monte Carlo permutation test with 1000 replicates.

3. Results

3.1. Species Richness and Community Composition

We recorded a total of 140 species of epiphytic lichens on 120 trees, distributed in 36 genera and 22 families. A total of 94 species were found in forest sites, 66 species in forest-pasture, 57 species in pasture, and 61 in urban sites. The Chao-2 richness estimator, confirming a high number of species estimated in forest, followed by forest-pasture, urban and pasture (Figure 2). Species richness of epiphytic lichen decreased along the land-use intensity, i.e., from forest to urban (

Table 2. Summary of the GLMM applied on species richness in four riparian land uses of Andean streams in southern Ecuador. Significant differences at p-value < 0.05. DBH: diameter at breast height.

Source of Variation	Coefficient	Standard Error	Z-value	p-value
Forest	2.651	0.297	8.908	<0.001
Forest-Pasture	−0.663	0.286	−2.313	0.020
Pasture	−0.976	0.336	−2.899	0.003
Urban	−1.661	0.537	−3.093	0.001
Distance to forest	0.445	0.160	2.767	0.005
DBH	−0.004	0.056	0.078	0.937

; Figure 3a).

Results GLMMs showed that lichen richness was lower in land uses with more intensity, i.e., forest-pasture, pasture and urban sites had a negative correlation with lichen richness. Conversely, distance to forests and forest land use showed a positive correlation (Table 2). The mean tree diameter (DBH) was not significant in any case (Table 2).

CCA indicated that precipitation, distance to forest and % forest were important factors to distinguish the taxonomic composition of the epiphytic lichen community among land uses (Figure 3b;

Table 3. Shift in lichen species composition in relation to predictor variables, measured with ANOVAlike permutation test in four riparian land uses of Andean streams in southern Ecuador. Significant differences at p-value (<0.05) are shown in bold. DBH: diameter at breast height.

Source of Variation	Degrees of Freedom	Chi Square	F-statistic	p-value
Land use	3	1.196	2.316	0.001
Precipitation	1	0.242	1.410	0.003
Distance to forest	1	0.283	1.6477	0.001
% Forest	1	0. 245	1.423	0.014

). Together, these variables explained a total variation of 10%. The % forest and precipitation were negatively correlated to axis 1, while distance to forest was positively correlated (Figure 3b, Table S1). Only distance to forest and precipitation were positively correlated to CCA axis 2. Across CCA axis 1, the lichen species Heterodermia leucomela, Dirinaria picta, Parmotrema arnoldi and Physcia aipolia showed more preference for disturbed sites (pasture and urban site) with less canopy cover, low humidity and more light availability. Conversely, lichen species such as Leptogium millegranum, Leptogium diffractum, Puntelia rudecta, Sticta fuliginosa and Sticta tomentosa preferred the conditions of more humidity and closed canopy provided by the forest sites. Similarly, across CCA axis 2 the lichen species Cococarpia palmicola., Leptogium coralloideum, Leptogium laceroides, Lobaria subexornata and Sticta fuliginosa were related to sites with closed canopy and high humidity; whereas, Heterodermia diademata, Heterodermia leucomela, Physcia aipolia, Physcia crispa, Telochistes exilis and Ramalina celastri were related to open canopy sites and less humidity (Figure 3b).

Eighteen species were the best indicators of forests land-use intensity, followed by forest-pasture with seven species and six and four species for urban and pasture respectively (Appendix A).

3.2. Functional Traits

Species richness of lichens with cyanobacteria and trentepohlia decreased from forest sites to urban sites (Figure 4a). A similar pattern was observed in the growth form, thus, crustose, foliose with broad lobes and gelatinose lichen species decreased along the land-use intensity (Figure 4b). Macrolichens with dark thallus decreased in pasture or urban sites (Figure 4b–d). Conversely, lichen with fruticulose growth form were more abundant in disturbed sites (Figure 4b). Lichens with sexual and asexual reproduction; and with apothecia, isidia and soredia as reproductive structure decreased in land use gradient (Figure 4e–f).

The GLMMs models showed that the most relevant predictor for richness of functional traits of lichens was land-use intensity (

Table 4. Summary of the GLMM applied on the functional traits of the epiphytic lichen community in four riparian land uses of Andean streams in southern Ecuador. Coefficient of variation with differences at p-value (<0.05) are indicated in brackets. DBH: diameter at breast height. Fo = Forest; FP = Forest-Pasture; Pa = Pasture; y Ur = Urban. DBH = diameter at breast height, Dist-F = distance to forest; %Fo = percentage of forest.

Functional Traits	Land-use Intensity				DBH	Dist-F	%Fo
	Fo	FP	Pa	Ur
Photobiont type
Chlorococcoid	1.883 (0.004)
Cyanobacteria	0.927 (0.001)	−0.871 (0.01)	−0.953 (0.01)	−1.195 (0.002)
Trentepohlia				−1.969 (0.03)
Growth form
Crustose	1.033 (0.005)	−0.704 (0.04)	−0.977 (0.02)	−1.847 (0.01)
Filamentose				−6.699 (0.04)		2.347 (0.02)
Foliose with broad lobes	1.277 (0.002)			−2.064 (0.01)		0.789 (0.04)
Foliose with narrow lobes	0.911 (0.003)
Fruticose	−0.844 (0.02)	1.007 (0.03)		1.013 (0.04)
Gelatinose		−1.028 (0.02)	−2.211 (0.01)	−2.13 (0.02)
Thallus size
Macrolichens	2.112 (0.009)
Microlichens						1.553 (0.01)	1.368 (0.006)
Thallus colour
Light	1.519 (0.001)
Dark	1.818 (0.001)	−1.091 (0.02)	−1.17 (0.03)
Reproduction type
Asexual	1.539 (0.001)	−1.263 (0.004)	−1.929 (0.002)	−2.226 (0.001)
Sexual	1.298 (0.001)		−0.818 (0.01)	−0.622 (0.04)
Asexual and sexual	0.775 (0.001)
Reproductive structure
Apothecia	1.769 (0.001)
Isidia	0.502 (0.006)	−0.492 (0.03)	−0.98 (0.005)	−1.281 (0.008)	0.271 (0.02)
Lirellae
Perithecia
Soredia	1.099 (0.03)

). The land-use intensity forest, % forests, and distance to forest (correlated with canopy cover and precipitation) showed a positive correlation on lichen species with cyanobateria as photobiont, and gelatinose and foliose with broad lobes growth forms and macrolichens (Table 4). Conversely, forest-pasture, pasture and urban land use showed positive correlation with lichens species with frutiulose thallus. Lichens with dark colour, apothecia, soredia and isidia reproductive structure were correlated positively with forest (Table 4). The tree diameter (DBH) showed a positive effect on isidia as reproductive structure (Table 4).

4. Discussion

4.1. Species Richness, Functional Traits and Community Composition

Our findings reveal a negative effect of land-use intensity on several metrics of taxonomic diversity and functional traits of the epiphytic lichen communities across riparian buffers in the Andes of southern Ecuador. There was a decrease in species richness and changes in community composition from forest to urban sites, as in other studies [5,20,56,57,58,59]. Similarly, the richness of functional traits such as photobiont type and growth form of epiphytic lichens diminished along the land-use intensity. Other studies have also demonstrated that land-use change has affected the functional traits of lichens [5,15,19,21,22,25,27,60]. Following this pattern for richness of functional traits Pinho et al. [20], Benitez et al. [21] and Koch et al. [22], showed that forest disturbance and urbanization have a strong correlation on the richness of functional traits of the epiphytic lichen communities. This phenomenon associated with species richness and functional traits can be explained by the ecological and physiological requirements related to water availability. Thus, a greater richness of sensitive species (e.g., Leptogium), species with cyanobacteria and gelatinose growth were present in forest sites with a closed canopy [5,14,18,19,61,62,63,64] than in pasture and urban sites. This is because they are intolerant to light and strongly depend on atmospheric humidity, in some cases they even need liquid water to do photosynthesis [65].

On the other hand, the highest occurrence of heliophytic lichen species with green algae and cortical pigments, microlichens, lichens with narrow lobes, light and fruticose thallus were related to sites with a more anthropogenic land use (e.g., forest-pasture, pasture and urban zones). These species show a higher tolerance to lower humidity and more light intensity promoted positively by an open canopy [5,14,18,19,20,23,63,64]. In our case, heliophytic lichen species (e.g., Parmotrema, Teloschistes, and Usnea), species with narrow lobes such as Physcia, Heterodermia and Hypotrachyna, and fruticose species (e.g., Usnea and Teloschistes) were present in more anthropogenic areas because these sites present high levels of solar radiation and water stress [22,63]. These lichen species can hydrate very fast, as well as rapidly lose water, because they occupy more surface and a have broad fixation structure to substrate [25,56]. In addition, most of the recorded species present secondary metabolites (i.e., atranorine or usnic acid) that provide protection against solar radiation typical in altered zones [21,22]. In the disturbed sites we also observed an increase of lichen species with chlorococcoid green algae. An explanation to this finding is that green algae species are better adapted to open forests because they avoid the photoinhibition by using a minimum amount of water in their thallus during photosynthesis [5,20,21,22,56,63].

Epiphytic lichen community composition was also influenced by the differences in land-use intensity. A forest cover reduction leads to less humidity and more intensity of light, thus lichen species restricted to undisturbed forests (shade epiphytes) are more affected [10,19,63]. In our study, the forest sites were dominated by species of the genera Leptogium, Sticta and Lobaria, which are species with cyanobacteria as a photobiont with high needs of water and forest cover [26,62,66]. These species are strictly associated with forest and can be considered good indicators of land-use intensity. On the other hand, another group of species were favored by the disturbance gradient, because they are capable of tolerating more light intensity and low humidity [10,14]. Thus, forest-pasture, pasture, and urban sites were dominated by xerophytic species (sun epiphytes) of the genera Heterodermia, Parmotrema, Physcia, Teloschistes and Usnea [15,22,59]. The adaptive advantage of sun epiphytes against shade epiphytes is their secondary metabolites that make them tolerant to high levels of radiation [21]. Both indicators and some non-indicator species have important ecological implications in the riparian ecosystems of the southern Ecuadorian Andes. For instance, most species of the genus Leptogium (e.g., L. burgesii, L. cochleatum, L. coralloideum, L. corticola, L. laceroides and L. marginellum), Lobaria (L. subexornata) and Sticta (S. ferax, S. fuliginosa and S. tomentosa); are associated with forest land use intensity and can be considered as good indicators of riparian forests with closed canopy and higher humidity [10]. Heterodermia hypoleuca, H. leucomela, Physcia aipolia and Ramalina celastri are the best indicators for urbanization, whereas that Dirinaria picta and Flavopunctelia flaventior for pasture. In accordance with other studies, the presence of these species indicates exposed conditions and can be considered a good indicator of the effects of land-use changes [10,20,21,22,28].

4.2. Application in Biomonitoring

Along our disturbance gradient, our measures of epiphytic lichen species richness, community composition, and functional traits richness, performed well when detecting the negative effect of land use intensity. However, functional traits have an advantage over community structure and species richness, of providing evidence of potential alterations in the biodiversity-ecosystem functioning relationships [21,22,24]. In this context, photobiont type and growth form of epiphytic lichens are easily measured and could be suitable indicators for detecting land-use intensity along riparian margins in Andean ecosystems. Likewise, growth form is a more sensitive indicator to changes in canopy cover facilitating a quantification of the effects of riparian forest disturbance, and potentially to assess the success of forest management in riparian buffers that promote and enhance the ecosystem health of Andean streams [36]. This first study in Ecuador, which assesses the suitability of a set of taxonomic and functional metrics of the epiphytic lichen community could be used as a complementary approach to understand the structure and functioning of riparian ecosystems in the tropical Andes. Therefore, these metrics could be effective indicators of land-use transformation in tropical Andes. Furthermore, the information provided here can contribute to better management practices in riparian margins of montane ecosystems in general.