2Sampling and Interpreting LichenDiversity Data for BiomonitoringPurposesPaolo Giordani and Giorgio Brunialti

Contents

2.1 Introduction ................................................ 20

2.2 Sampling Lichen Diversity........................ 222.2.1 Variability of Lichen Diversity Across

Scales ........................................................... 222.2.2 Standardizing Lichen Diversity

Sampling ...................................................... 272.2.3 Quality Assurance........................................ 31

2.3 Interpreting Lichen Diversity ................... 332.3.1 The Concepts of Biomonitoring.................. 332.3.2 Interpretative Tools...................................... 37

2.4 Open Questions .......................................... 40

References ................................................................ 41

Abstract

Although lichen diversity values are broadlyused as bioindicators, mainly for air pollution,lichen communities can be substantially influ-enced by other ecological factors, such as treespecies and forest structure, and microclimaticconditions. In particular, species compositionmay be a suitable indicator for climate and land-use effects as well. For effective utilization oflichen diversity data in biomonitoring studiesincluding air pollution, ecosystem functioning,and forestry studies, standardized sampling pro-cedure and avoiding sampling and non-samplingerrors are the important aspects to be considered.Further interpretation of lichen diversity datarequires careful data analysis for providingafrmative results related to ambient air quality.In any lichen biomonitoring program, expecteddeliverables are based on a hypothesis, whichmay be achieved by standardization of thesampling procedures based on the functionalrequirement of the dependent environmentalvariables. The chapter discusses the proceduresand methodology for sampling and interpretinglichen diversity data for biomonitoring purposes.

Keywords

Air pollution Ecosystem functioning Qualityassurance LDV

P. Giordani (&)Botanic Centre Hanbury, DISTAV, University ofGenova, Corso Dogali, 1M, 16136 Genova, Italye-mail: giordani@dipteris.unige.it

G. BrunialtiTerra Data Environmetrics, Spin Off of theUniversity of Siena, Via Bardelloni 19, 58025Monterotondo Marittimo, Italye-mail: brunialti@terradata.it

Springer India 2015D.K. Upreti et al. (eds.), Recent Advances in Lichenology,DOI 10.1007/978-81-322-2181-4_2

19

2.1 Introduction

Due to their physiology, lichens are sensitive to aseries of environmental parameters (namely light,air humidity, UV-B radiation, temperature, andairborne chemicals such as SO2 and NOx makingthem useful indicators for air pollution and climatechange (Poikolainen et al. 1998; Tarhanen et al.2000; Cornelissen et al. 2001; Kricke and Loppi2002; Castello and Skert 2005). Biomonitoringmethods based on the diversity of epiphyticlichens are among the most used worldwide.Apart from assessing the effects of gaseous pol-lutants, these approaches were recently extendedto a suite of other anthropogenic disturbances(Nimis et al. 2002). Several aspects of lichendiversity (e.g., species richness and abundance,species composition, indicator species, functionaltraits and groups) are usually considered, each ofthem for a particular reason (Nimis et al. 2002). Insimplied terms, it is possible to identify threemain purposes to perform lichen monitoring:1. Air quality monitoring. The sensitivity of

lichens to phytotoxic gases (mainly SO2and NOx) underlies their use to assess andmonitor the effects of atmospheric pollution(Hawksworth and Rose 1970; Richardson1993). They have been used as long-termbiomonitors of air pollution both for small-scale (Nimis et al. 1990; Loppi et al. 2004;Giordani 2007) and large-scale surveys (Nimiset al. 1991; Van Dobben and DeBakker 1996;Bennett andWetmore 1999; Frati and Brunialti2006; Giordani 2006). Lichen biomonitoringsurveys are often used to integrate instrumentaldata of atmospheric pollution (Nimis et al.1990; Giordani et al. 2002; Giordani 2007;Pinho et al. 2004) and for developing forecastsin connection with human health (Cislaghi andNimis 1997).

2. Sustainable forestry. Epiphyte diversity isrelated to forest structure and dynamics. For-est management, directly or indirectly, affectsseveral environmental factors relevant to dis-persal, establishment, and maintenance oflichen species. Studies on lichen diversityclearly demonstrate dramatic losses of species

caused by forest management in Europeantemperate and boreal forests (Hauck et al.2013). In general, higher lichen diversity isrelated to lower management intensity, eventhough in some cases non-intensively man-aged forests may provide better conditions forepiphytic lichens than recently abandonedforests. For instance, several studies suggestthat selective cutting is less detrimental toforest lichens than the shelterwood system orclear cutting (Nascimbene et al. 2013b).

3. Ecosystem functioning. Epiphytic lichensplay an important role in both the forest watercycle, increasing canopy interception of pre-cipitations and forestnutrient cycling, aboveall for N-xing cyanolichens (lichens with acyanobacterial symbiont; for a review seeEllis 2012). In addition to nutrient cycling,lichens are a focal point in forest food webs.They influence the ecological success of for-est-dwelling animals in a wide variety ofways, e.g., for macrofauna in nest building,and as forage (Ellis 2012). In some cases,indicator species have been proposed asproxies for lichen species richness moregenerally. As an example, forest stands withthe flagship species Lobaria pulmonariatended to have more red-listed species andmore continuity indicator species than standswithout this lichen (Nilsson et al. 1995), ten-tatively pointing to covariation among conti-nuity indicators. Similarly, comparing standswith different abundance and fertility attri-butes for L. pulmonaria, these attributes wereassociated with higher tree and stand-levelmacrolichen species richness and with sig-nicant differences in species compositionamong L. pulmonaria site-types, i.e., cyan-olichens and rare species skewed to the L.pulmonaria stands (Nascimbene et al. 2010).In this chapter, we intend to explore a series of

key issues on the topic in order to provide aframework of the subject with an overview ofrecently published researches.

In this regard, Fig. 2.1 provides a step-by-stepflow diagram of operations required to address alichen monitoring program. As a rule, these are

20 P. Giordani and G. Brunialti

the typical questions to be answered: does airquality affect lichen diversity in our study area?Or else, does forest management (or fragmenta-tion) influence lichen communities in our inves-tigated forests? Or even, do lichen diversity playa role in ecosystem function of our natural park?

The rst step in the organization of a lichenmonitoring program is to establish the initial nullversus alternative hypotheses. For example, wemay state the basic hypothesis that the expectedlichen diversity is equal for impact and controlsites (with respect to air pollution or to forestmanagement). However, this is where environ-mental variability comes into play. Thus, thesecond step is to qualify, quantify, and control themain ecological factors that can affect ourresponse variable, both including those of ourinterest and also the ones which may represent asource of noise (Brunialti et al. 2010a). Indeed, theeffect of large-scale (such as climate) or local scale(as tree substrate) natural factors may representrelevant sources of variability and noise in relationto the phenomenon we are studying (Giordaniet al. 2013; Ellis 2012). Also, some of these effectsmay be cumulative (Jovan and McCune 2004;Pinho et al. 2004; Giordani 2006; Caruso and Thor2007). For instance, we should study air pollution

effect on lichen diversity in forest sites with dif-ferent forest management: We will have to set thestudy in order to disentangle the effect of indi-vidual factors and determine which of the two actsas main driver affecting lichen diversity or thepresence/abundance of indicator species (Gior-dani 2007; Giordani et al. 2014a).

Our opinion is that this issue may have acentral role in biomonitoring studies. For thisreason, an entire Sect. 2.2.1 of the chapter focuseson these aspects, where we explore the mainsources of variability affecting lichen diversity atdifferent spatial scales. This variability beingconsidered to adequately respond to our initialquestion, and the third step of our scheme con-cerns the adoption of an appropriate samplingdesign, from the selection of objectives and targetpopulation to the development of an adequatesampling strategy. We address this topic inSect. 2.2.2, where we explore the main stan-dardized methods for lichen diversity assessment,and analyze their suitability for various purposes.

Section 2.2.3 focuses on quality assuranceprocedures that are closely linked to samplingdesign and should be taken into account to obtainreproducible, reliable, and defensible data (Fer-retti 2011). In this respect, we must remember

Fig. 2.1 Schematicdiagram of the operationsrequired to address a lichenmonitoring program

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 21

that there is a strong link between data quality,information quality, and decision-making pro-cesses (Crumbling 2002).

The fourth step is then the phase of dataprocessing. A broad range of statistical tools maybe adopted to test the initial hypothesis. A sub-sequent interpretation of the results is essential toverify that the statistical signicance of the testhas a meaning even from an ecological point ofview (Sect. 2.3). Starting from raw data, itis possible to adopt a series of interpretativedevices that allow analyzing various aspects ofthe phenomenon under investigation. Only in thisway, we may obtain useful information formanagement and decision making.

In this regard, the last step of our processinvolves the rejection or acceptance of the startingnull hypothesis. On the basis of this result, we getessential information to proceed with the processof planning and management (policies to reduceair pollution, forest and landscape management,etc.). Indeed, without reliable information onchanges to ecological systems and on the causesof those changes, decision making cannot dealeffectively with these issues (Ferretti 2011).

The main objectives reported above areaddressed in most cases with management anddecision-making intent. In this perspective, ifbiomonitoring should be taken as a serious basisfor decision making, it needs to produce robust,defensible data and documented quality (Brunialtiet al. 2004). With this in mind, it is essential toapproach the issue by means of a suitable sam-pling design and also appropriate statistical andinterpretative tools that should be carefullyselected to address properly each of these pur-poses (Ferretti and Erhardt 2002).

2.2 Sampling Lichen Diversity

There is no doubt that ecological monitoring has afundamental role in providing baseline data onthe status and trend of environmental resources.However, as underlined by Legg and Nagy(2006), in several cases, ecological conservationprograms suffer from the lack of details of goaland hypothesis formulation, survey design, data

quality and statistical power at the start. As aresult, sometimes, they are likely to fail to reachthe necessary standard of being capable ofrejecting a false null hypothesis with reasonablepower. This is mainly related to the adoption of anunsuitable sampling design, and this is basicallythe reason why sampling has frequently beenconsidered as the weakest point of monitoringprograms. This is unfortunate because failure toprovide convincing evidence of the overall qual-ity of environmental information can have seriousconsequences in environmental decision making.Indeed, we must consider that the defensibility ofthe data supporting the decision is fundamental(Ferretti 2011). To avoid this and to promote acorrect process, the selection of a suitable sam-pling design represents the rst step to reducedata variability due to sampling error. Whenselecting the proper design, the objectives (bio-monitoring purposes, Fig. 2.1) of the survey andenvironmental variability should be taken intoaccount carefully (Brunialti et al. 2004).

With respect to lichen biomonitoring, envi-ronmental variability acting at different spatialscales has been taken into account during thestandardization process of the sampling protocols(Giordani et al. 2001; Brunialti and Giordani2003; Nimis et al. 2002), thereby providinguseful information for the selection of suitablesampling designs.

This section provides useful insights on sev-eral issues relevant to environmental variability,sampling design, and quality assurance proce-dures, which are the basis of a proper environ-mental monitoring program.

2.2.1 Variability of Lichen DiversityAcross Scales

Variability is an inherent property of ecologicalsystems and every attempt to measure andinterpret the environment should consider it(Brunialti et al. 2004). Thus, in the assessment ofenvironmental quality by means of biomonitors,it is important to understand the environmentalprocesses driving the variability of biologicaldata, since this may affect the forecasting

22 P. Giordani and G. Brunialti

precision of these techniques (Laskowky andKutz 1998). Environmental factors such as geo-morphology, climatic variables, and substratecould have a great impact on the ecosystemproperty being studied in order to assess envi-ronmental quality, such as the rate of indicatorspecies, the biodiversity of a community, or thepresence of injuries on organisms. In simpleterms, environmental heterogeneity may controlthe variability of lichen diversity and the com-position of lichen communities at least at threespatial scales: (1) tree level; (2) plot level, and (3)landscape level. In this section, we highlight thepossible effect of these ecological factors onlichen diversity by referring to the main studiesconducted so far.

2.2.1.1 Spatial Scale of Variationin Lichen Monitoring Data

Quantifying spatial and temporal variation is akey element for understanding mechanisms andprocesses that structure the species assemblagesfor assessing the environmental impact and foridentifying an appropriate scale of sampling(Levin 1992; Underwood et al. 2000). Nonethe-less, biodiversity assessments are stronglydependent on observation scales (Gaston 2000;Purvis and Hector 2000), and natural spatialvariation or the spacing of sampling units maydecisively affect results (Dungan et al. 2002).

Although several studies have investigatedspatial variation in plants (see, e.g., Palmer 1990;Gunnar and Moen 1998), lichens are still poorlyinvestigated from this point of view (Ferretti et al.2004), even though descriptors of lichen diversityare widely used in applicative studies for assessingalterations of the environment (Nimis et al. 2002),covering areas ranging from some dozen to somethousands square kilometers (e.g., Giordani et al.2002). In particular, bioindication techniquesusually focus on between-site variability in epi-phytic lichen diversity using a high samplingdensity and do not sufciently consider possiblewithin-site variability, probably because thiswould imply sampling more trees, thus increasingthe cost of the survey (Ferretti et al. 2004;Brunialti et al. 2010a). Controlling the spatial

variation at macroscales is necessary but not suf-cient, if the variation at more detailed levels (e.g.,considering areas

Fig. 2.2 Estimatedcomponents of variance(%) in lichen diversityvalues, calculated for eachspatial scale (modied fromGiordani et al. 2013)

Fig. 2.3 Conceptualmodel illustrating the mainecological factorspotentially affecting lichencommunities under thecanopy of a broad-leavedtree in winter and summerperiods (see explanation inthe text)

24 P. Giordani and G. Brunialti

inclination of the bole may contribute indirectlyto modify these ecological factors (see, e.g.,Giordani et al. 2001).

Figure 2.4 summarizes the main ecologicalfactors affecting lichen communities at plot andlandscape levels. They are mainly related to thedistribution and density of trees (e.g., the gradi-ent from isolated trees to near-natural foreststands) and to the land-use and geomorphologicalvariability (e.g., altitudinal range, agricultural, orforested landscapes). Light and humidity avail-ability generally play the role of direct factors inthis context.

Basically, all these aspects are very intercon-nected and show that the effects largely covariate.Canopy structure decisively controls light condi-tions at the lower bole (Moe and Botnen 1997;Fritz et al. 2009), and this may result as a negativeeffect of canopy shading on species richness, bothat tree level (Loppi and Frati 2004) and at plotlevel (Brunialti et al. 2010b). While focusing onedge effects in fragmented boreal and Mediter-ranean forests, several authors pinpointed light asa main driving factor for lichen colonization,affecting lichen abundance, diversity, and

composition at tree level (Esseen and Renhorn1998; Belinchn et al. 2007; Boudreault et al.2008; Brunialti et al. 2012b). In particular, higherlichen species richness was found at the side ofthe trunk facing the edge (Brunialti et al. 2012b),or in correspondence with the rst light peak(Belinchn et al. 2007). Loppi and Frati (2004)found higher lichen diversity values on Tiliaplatyphyllos compared to Quercus ilex. Thesendings were explained by signicantly higherwinter light conditions and water-holding capac-ity of the bark for Tilia trees. Bark-pH was notfound as discriminant parameter among the twotree species. On the contrary, many studies havedemonstrated a strong epiphyte response to bark-pH, explaining epiphyte community variationcompared among tree species in both boreal andtemperate systems (Hyvrinen et al. 1992; Gau-slaa 1995; Kuusinen 1996; Jriado et al. 2009;Lewis and Ellis 2010; Leppik et al. 2011).Moreover, intraspecic studies, among boles ofthe same tree species, have indicated that bark-pH/nutrient status may be modied by an inter-action with the soil environment (Gauslaa 1995;Gustafsson and Eriksson 1995; Kermit and

Fig. 2.4 Conceptual model illustrating the main ecological factors potentially affecting lichen communities at plot andlandscape levels (see explanation in the text)

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 25

Gauslaa 2001) and with a subsequent effect onepiphyte community composition. Interestingly,differences in tree age/size (dbh) correlate withbark-pH among individual boles of the samespecies, though the direction of this relationshipcannot be generalized. Bark-pH may decreasewith increasing tree circumference (Bates 1992;Kuusinen 1994) and/or age (Ellis and Coppins2007), or pH may increase with circumference(Jriado et al. 2009) and/or tree age (Fritz et al.2009). Also, bark texture was found to be animportant tree species factor influencing lichendiversity (Bates 1992; Ccares et al. 2007; Raniuset al. 2008; Fritz et al. 2009). This is subject to aneffect of tree age/size (dbh) and an interaction withvertical height on the trunk (Johansson et al. 2007;Ranius et al. 2008; Fritz 2009), and it is relatedwith a species-specic relationship betweenbark roughness and tree age/size (Uliczka andAngelstam 1999).

The standardization process of air quality bio-monitoring method took into account all thesesources of variability acting on lichen diversity attree level, making it possible to isolate environ-mental gradients in a simplied system (Asta et al.2002; EN 16413 2014; see paragraph interpretinglichen diversity). For this reason, the standardfeatures of the sampling trees (slope, pH of thebark, etc.) have been selected carefully (Table 2.1).

With respect to plot-level variability, Giordani(2006) carried out a hierarchical evaluation of theeffects of substrate- and environmental-relatedvariables on lichen diversity. Elevation, meanannual temperature, and latitude were the mainfactors influencing epiphytic lichen distribution,

together with disturbances (such as atmosphericpollution, forest res and agricultural practices)and habitat heterogeneity that covary along thestrongest gradient of lichen community compo-sition. Lichen species richness was positivelyassociated with mean annual rainfall and longi-tude, negatively associated with harvesting, andpositively associated, though weakly, with theoccurrence of past forest res (these burnt siteswere characterized by recolonization processes).

In a study of epiphytic lichens of forest eco-systems of Tuscany, Loppi et al. (1999) concludedthat habitat characteristics aremore important thanphorophyte properties and there is evidence thatthe epiphytic lichen vegetation of deciduousQuercus trees follows a distribution which isrelated to elevation and climate, with great dif-ferences in community structure along the altitu-dinal gradient (Loppi et al. 1997).

As for the aspects affecting lichen distributionat landscape level, many authors pointed out therole of climatic factors (McCune et al. 1997;Goward and Spribille 2005; Hauck and Spribille2005; Giordani 2006, 2007). In this respect,Giordani and Incerti (2008) found that the dis-tribution of more than 30 % of epiphytic specieswas associated to macroclimatic variables. Asignicant subset of epiphytic lichens in the studyarea has been proved to be efcient bioclimaticindicator for montane, humid sub-Mediterranean,and Mediterranean units.

Further, also land-use intensity drives thelocal variation of lichen diversity, both in Med-iterranean (Giordani et al. 2010) and in borealecosystems (Holt et al. 2008). Land-use intensity

Table 2.1 List of features to of a standard tree (see EN 16413 2014; Asta et al. 2002)

Feature Description

Suitable treespecies

The sampling tree belongs to one of the groups with similar bark physicochemical properties(EN 16413 2014; Asta et al. 2002). Indicatively, species belonging to the same group can beused interchangeably

Trunkcircumference

The sampling tree has a trunk circumference (at 130 cm from the ground level) between 50 and250 cm

Trunkinclination

Each exposition (N, E, S, W) has an inclination (at the center of each grid)

was detected as a good proxy for describing thelichen distribution and abundance under anthro-pogenic pressures. In particular, land-use cate-gories, mainly based on vegetational features,differed in epiphytic and epilithic lichen com-munities, and the strongest differences wereobserved among the forested sites versus man-aged agroforestry lands. These latter were mainlycharacterized by common lichen vegetation(mainly xerophyticnitrophytic species).

Similarly, in Scotland, lichen communitiesdiffered between natural ancient pine forests andthose of trees within managed sites both in forestand agricultural areas (Wolseley et al. 2006). Thealteration was mainly due to increasing nitro-phytes, most evident for epiphytes, but also sig-nicant for saxicolous communities. Accordingto Bergamini et al. (2005), the trend of modi-cation of lichen communities under changingland use is nearly constant along large latitudinalgradient, ranging from northern Europe to theMediterranean. Also, at this large scale, strongdifferences were detected between the forestedand the more open land-use types, especially forepiphytic crustose lichens.

2.2.2 Standardizing Lichen DiversitySampling

Differences between methods, difference in theapplication of the same method, measurementerror, sampling and non-sampling error, anderrors related to model applications are all termsof the whole error budget that inevitably affectsenvironmental surveys (Gertner et al. 2002). Inthis perspective, the extent to which the objectiveof the survey is matched depends very much onthe ability to manage the various sources ofvariability (Khol et al. 2000; Wagner 1995) byadopting suitable standard operating procedures(Brunialti et al. 2004, 2012b).

A recent example of standardization processrelates to the sampling protocol to assess epy-phitic lichen diversity for air quality. In fact, theEuropean Committee for Standardization (CEN,Comit Europen de Normalization) has recentlypublished an European standard, reporting the

sampling protocol for lichen diversity assessment(EN 16413 2014). The process of standardizationstarted in 2007 and took into account the previousEuropean and national guidelines (Asta et al.2002; VDI 2005; AFNOR2008). In themeantime,some eld tests have been performed to obtaininformation on the type and size of errors and theuncertainty of the methodologies under standard-ization (Brunialti et al. 2012a, Cristofolini et al.2014). Toll this gap, the tests dealt with the entireprocess from survey design to eldmeasurements.In particular, the comparative tests, consisting inmultiple exercises, were run with the aim ofcomparing the results obtained by different, well-experienced operators faced with the same prob-lem, at the same time, under the same eldconditions and following the same standardoperating procedures (SOPs).

A similar standardization sampling procedurehas been adopted in forest monitoring. Anexample is represented by the sampling protocolof the EU project Forest BIOdiversity Testphase Assessments (ForestBIOTA), carried outby 12 European countries in the framework ofthe ICP Forests Expert Panel on Biodiversity andGround Vegetation Assessments (Fisher et al.2009). In this context, Stofer et al. (2003, 2012)prepared a standardized sampling protocol forlichen diversity assessment that takes intoaccount all the main steps of the sampling fromthe tree- to the stand level.

2.2.2.1 Sampling ObjectivesA clear denition of objective is important;otherwise, any statement about data qualityremains elusive (Ferretti 2011). As a conse-quence, although it is not a typical source of errorper se, ambiguous monitoring objectives can begreat promoters of errors: On the one hand, themonitoring design cannot be properly addressedif the objective is not clear, and incorrect designmay jeopardize the whole monitoring. For thisreason, objective must be explicit with respect totarget population and its attributes, spatial andtemporal domains, desired precision level andminimum detectable change, and type I and IIacceptable error rates.

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 27

With respect to the CEN standard for biomon-itoring ambient air, the sampling objective is toobtain an estimate of the parameter of the responsevariable (e.g., mean species richness or meanlichen diversity value, LDV) over the studydomain with a given precision (EN 16413 2014).The precision level should be expressed in terms ofcondence intervals for a dened probability level.It is required that the sampling objective is denedfor each study. For example, we may want toobtain an estimate of the mean LDV for the studydomain with a condence interval 10 % of themean value at a probability (P) level of 95 %.

Since the computation of estimates and con-dence intervals depends on the sampling designadopted, each study shall dene precision andprobability levels, taking into account therequirements of the study framework and consid-ering the available resources (EN 16413 2014).

2.2.2.2 Target PopulationIn statistical terminology, the target populationrefers to a different concept than that associatedwith the community, population, individual, andgenetic concept of biological systems. The targetpopulation is the collection of elements aboutwhich information is wanted (Cochran 1977).Both the target population for which informa-tion is wanted and the elements that make upthe target population must be rigorously dened.As clearly stated, target population should becarefully selected before starting each monitoringprogram. It must be representative of the wholepopulation in order to extend the results obtainedby the selection of a sample of it to the entirepopulation. In fact, we must consider that dataobtained outside a formal sampling design can-not be considered representative for the entirepopulation, and therefore, conclusions cannot beextended without making assumptions about it(Ferretti 2011). Similarly, models built uponnon-representative data can be seriously biased.A classical example is the biodiversity data col-lected by taxonomists who are inclined to con-centrate their efforts in the localities thatguarantee success in the collection of as many

species as possible, thus resulting in species-richness bias (Sastre and Lobo 2009).

For instance, a recent comparative test toidentify critical issues in lichen biomonitoringdemonstrated that different teams may selectdifferent target populations when planning thework (Brunialti et al. 2012a). Although all ofthem may be formally correct, these differencesare a source of inconsistency in the results andcould potentially lead to differing conclusions. Itis therefore extremely important that every stepof the design process is properly documented andreported in survey documents so that futurerepetitions may be made.

2.2.2.3 Sampling Design and SamplingSchemes

For broader applications (e.g., at the Europeanlevel), a standard protocol should be flexibleenough to preserve the representativeness of thedata under different ecological conditions. Ingeneral, it is difcult, if not impossible, to nd asingle sampling design that is suitable for eachsituation. Different options of sampling schemesmay be adopted, mainly in relation to the com-plexity of the survey area and with the distribu-tion of the trees to be sampled (Elzinga et al.2001; EN 16413 2014; Ferretti and Erhardt2002). With respect to air monitoring pollution,the recent CEN standard proposes the adoption ofdifferent schemes in relation to ecologicallyhomogeneous and heterogeneous areas. In par-ticular, in the rst case, we may have tree options:1. When standard trees are abundantly and

homogeneously distributed over the studydomain, a simple random or systematic designis recommended. Plot sampling is recom-mended, with sample plots allocated accord-ing to a regular grid, with the starting point ofthe grid chosen at random (e.g., Giordaniet al. 2002).

2. When standard trees are abundantly scatteredin clusters over the study domain, tree-basedcluster sampling or two stage sampling isrecommended. A criterion to identify clustersshould be initially set, then identify, and list

28 P. Giordani and G. Brunialti

all the clusters and obtain a random sample ofthem. It should be noted that a two-stagesampling requests a subsampling of the sam-pling unit, and it introduces a further source ofvariability which may affect the quality of thedata. It is important to take this into accountwhen performing statistical analysis of thedata (EN 16413 2014).

3. When standard trees are infrequently scatteredover the study domain, a simple tree-basedrandom sampling is recommended. It is pos-sible to obtain a list of the individual trees onthe basis of the aerial photo and to selectrandomly the sample trees.Similar options may be selected also in eco-

logically heterogeneous areas. In this case,however, a stratied random sampling design isrecommended both with standard trees homoge-neously distributed and scattered in clusters overthe study domain. A tree-based stratied randomdesign is suitable when we have standard treesinfrequently scattered over the study domain.

Notwithstanding this great variability ofsolutions, it is not so obvious that each of these isproperly taken into account when a monitoringprogram is designed, since the experts in the eldmay be more influenced by examples from thepast (seen as best practices) that by newly sug-gested options. The results of a comparative testcarried out of operators on the same area andwith the same standard procedures by ve inde-pendent groups partly conrm this statement. Inthis study, in fact, most teams adopted a stratiedrandom sampling (Brunialti et al. 2012a).Although they largely agreed on selection of thesampling scheme, considerable differencesoccurred in subsequent steps of the samplingdesign: For example, the number of selected landcover categories ranged from 2 to 8 and thesampling density 2343. The authors observedthat training courses on sampling design andbasic statistics may be successful in reducing thissource of variability in routine eld studies.

2.2.2.4 Sampling UnitsAs reported above, both plot-based and tree-based sampling schemes can be adopted.

According to the latter option, the sampling unitsare represented by the standard trees available inour survey area. Instead, when considering theplot-based sampling, the selection of the optimalshape and size of the plot remains an openquestion. In this case, the goal is to nd anacceptable trade-off between a good representa-tiveness of the study area and a cost-effectivesampling effort.

Concerning the shape of sampling units, cir-cular or square plots are used in most cases. Ingeneral, the former are suggested in air pollutionmonitoring programs (Giordani et al. 2002),while the latter are mainly adopted in forestsurveys (Giordani 2006). The basis for a properselection of the shape of sampling units reliesboth on probabilistic issues and on practicalquestions (Elzinga et al. 2001). For rectangularsampling units, the associated edge error is largerthan for circular plots. Moreover, a circular plotrequires only one measurement (radius) to beinstalled. In special cases, such as for theassessment of the effect of forest fragmentation,rectangular transects were taken into account, asthey could t better with the natural shape offorest fragments, which were further divided intosubplots (Brunialti et al. 2012b).

As for the plot dimensions, McCune andLesica (1992) found trade-offs between speciescapture and accuracy of cover estimates for threedifferent within-site sample designs for inventoryof macrolichen communities in forest plots. Onaverage, whole-plot surveys captured a higherproportion of species than did multiple micro-plots, while giving less accurate cover estimatesfor species. The reverse was true for microplots,with lower species captures and much bettercover estimates for common species. Belt tran-sects fell between the over two sampling designs.

Ravera and Brunialti (2013) showed that aprobabilistic sampling based on the selection ofonly three trees within small circular plots (14 mdiameter) can be effective for assessing theoccurrence of species of conservation concern inold growth forests in an Italian National Park.They found that most of the species were presentin a few sampling sites and only a small groupof species were common to more than 50 % of

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 29

the plots. This may suggest that the adoptedsampling design allowed not only to determinethe most local common species but also to detectthe presence of sporadic species (Ravera andBrunialti 2013). To conrm this, their lichen listreported several new and interesting species andrepresented 30 % of the indicators of forestcontinuity known from the region. This is aconsiderable result if we consider the nature ofthe adopted sampling and also the fact thatregional floras mainly report records from pref-erential surveys. We must keep in mind, how-ever, that floristic surveys, mainly carried out bymeans of a preferential sampling, are biased bythe fact that they are primarily carried out todiscover rare species in specic habitats or eco-logical niches. Humphrey et al. (2002) obtainedsimilar results in a study carried out to comparelichen and bryophyte communities betweenplanted and seminatural stands. A high percent-age of species was recorded only once, and veryfew species were common to more than half theplots. This local rarity phenomenon has beennoted in other studies (Vitt et al. 1995; Collinsand Glenn 1997; Qian et al. 1999; Humphreyet al. 2000) and is partly related to sampling area.The authors of that study observed that it ispossible that a 1-ha sampling plot used is toosmall to capture a representative sample of lowerplant diversity in forest stands. For example,Rose (1993) recommends a minimum samplingarea of 1 km2, but this depends on the objectiveof the survey. Hence, these ndings suggest toadopt one of the following approaches for futureinvestigations: (1) extend the surveys to plot witha broad area and, at the same time, detect ahigher number of trees and/or substrates (rocks,soil, etc.). Although this approach could be verytime consuming, it should ensure the nding of alarge number of species; (2) improve the numberof small plots in the study area. In this way, witha lower sampling effort at the individual plot, it ispossible to obtain information on a wider terri-tory. Moreover, this second approach should bemostly useful for preliminary lichen surveys inpoorly studied wild areas (Ravera and Brunialti2013).

2.2.2.5 Sampling DensityOptimal sampling density is an essentialrequirement of lichen biomonitoring surveys forobtaining precise and unbiased estimates ofpopulation parameters and maps of known reli-ability (Ferretti et al. 2004). This aspect is often asore point of biomonitoring projects and shouldbe carefully addressed to respond correctly to thesampling objective and to select a sample with aproper sample size to avoid jeopardizing theeffectiveness of the investigation. Apart fromcatching a sufcient amount of variability, adecision on a proper sample size forcely con-siders the sampling effort and its costs and itoften happens to reduce the number of samplingsites in order to limit the costs.

Both the ecological complexity of a givenarea and the desired level of precision (in termsof condence interval) drive the sampling design.In heterogeneous areas, a greater number ofsampling units and more detailed land coverstratication are needed to obtain accurate esti-mates. For broader application (e.g., at theEuropean level), a standard protocol should beflexible enough to preserve the representative-ness of the data under different ecological con-ditions (Brunialti et al. 2012a). In this respect,Ferretti et al. (2004), starting from the results oftwo large-scale surveys undertaken in Italy, car-ried out a study on the effects of different sam-pling densities (the number of sampling units inthe study area) and on the reliability (in terms ofcondence intervals and relative error in themean values) of the estimates of the lichendiversity values of a given area and of lichendiversity maps. An iterative approach was takeninto account to generate subsets with lowersampling density with respect to the originalsampling units (ordinary kriging interpolations).Obviously, a higher sampling density may leadto a low error rate but may be nanciallyunsustainable. A very low sampling density, onthe other hand, may provide uncertain data as tobe of no real use (Khl et al. 1994; Ferretti andErhardt 2002). Their ndings suggested that alarge-scale lichen diversity spatial pattern can bedetected with a much more relaxed grid density

30 P. Giordani and G. Brunialti

than those originally applied (Ferretti et al.2004). This is extremely important, since reduc-ing the sampling effort can result in considerablesavings in resources that can be diverted toadditional, more detailed investigation withdenser sampling in those areas identied asproblematic by the study based on the lowsampling density. These results were also con-rmed by a similar study (Frati and Brunialti2006), within a long-term monitoring program,that showed the possibility to reduce the sam-pling effort in future monitoring surveys, result-ing in a considerable reduction of the samplingeffort maintaining a good data quality.

2.2.2.6 Lichen SurveyAs discussed in the previous sections, a series ofmicro- and macroenvironmental variables canaffect the composition and diversity of epiphyticlichen communities. For this reason, a standard-ized sampling strategy to count lichen diversity isextremely important to ensure comparable andaccurate results. The recently developed Euro-pean standards for lichen biomonitoring (Stoferet al. 2012; EN 16413 2014) have been developedas the result of a standardization process carriedout in the last few years starting from the Index ofAtmospheric Purity (IAP) approach (Hawks-worth and Rose 1970; Nimis et al. 1990, 1991)and upgrading it with previous guidelines (Astaet al. 2002; VDI 3957 2005). Furthermore, alsoeld experience of several European researchersand the results of recent comparative tests(Brunialti et al. 2012a; Cristofolini et al. 2014)were useful to obtain the current sampling strat-egy for lichen survey. At the end of this process,to reduce the effect of several possible sources oferror (e.g., different size of explored area on thetrunk, subjectivity in the positioning of the sam-pling grid, etc.), several parameters have beenstandardized. Hence, the abundance of eachlichen species is currently sampled by means of asampling grid consisting of a 10 50 cm ladderdivided into 5 10 10 cm quadrants. This laddergrid is placed systematically on the N, E, S, andW side of the bole of each tree (4 per tree), withthe top edge 1.5 m above ground, following the

standards suggested by Asta et al. (2002). Sum-mary measurements of species richness andabundance are usually calculated for each plot:mean number of species per tree; total number ofspecies within the plot; and mean lichen diversityvalue (LDVAsta et al. 2002) calculated as thesum of the abundance of each species within thesampling grids on a tree, averaged for all treeswithin a plot.

2.2.3 Quality Assurance

Biomonitoring investigations are subjected to avariety of error sources that need to be acknowl-edged and documented in order to be managedproperly (Brunialti et al. 2004). The quality of thedata originating from biological measurementsdepends heavily on at least three factors (see, e.g.,Kovacs 1992; Brunialti et al. 2004): (1) vari-ability of the biomonitoring organisms (interac-tions between the organisms and environmentalfactors); (2) type of sampling (sampling design,density of sampling points); and (3) operatorsinvolved in assessing lichen diversity, which is amethod requiring a relatively high taxonomicknowledge.

The rst issue has already been discussedabove (Sect. 2.2.1), and we have seen that thissource of variability can be controlled with agood knowledge of the ecological characteristicsof biological indicators considered and adoptingproper sampling designs. Regarding the type ofsampling, the sampling errors are closely asso-ciated with the sampling design adopted and itsquality. As for the third factor, the effect ofoperators subjectivity and expertise has beenwidely addressed within lichen biomonitoring soas to obtain useful information to further improvestandardized protocols.

To respond to these issues and to take all thesteps of the monitoring survey into account, theadoption of quality assurance procedures isstrongly recommended. QA is an organizedgroup of activities dening the way in whichtasks are to be performed to ensure an expressedlevel of quality. The main benet of a QA planis the improved consistency, reliability, and

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 31

cost-effectiveness of a program through time(Ferretti 1998). A QA plan is essential since itforces program managers to identify and evaluatemost of the factors involved in the program. Inaddition, the assessment of data quality enablesmathematical management of uncertainty due tothe method used (Ferretti 1998; Cline andBurkman 1989).

It is therefore important that environmentalbiologists and eld ecologists consider QA as akey attribute of their work in order to providerobust and defensible data to decision makers(Brunialti et al. 2004).

2.2.3.1 Sampling ErrorsSampling errors are generated by the nature ofthe sampling itself and by the degree of vari-ability in the target population. As reportedabove, such kinds of errors always occur but canbe controlled by appropriate sampling design(Cochran 1977; Khl et al. 2000; Ferretti andErhardt 2002).

A good sampling design is essential either tocollect data amenable to statistical analyses andto control errors in relation to the costs (Brunialtiet al. 2004). An important issue, in this respect, isselectivity, which seems particularly important inecological measurement. A protocol is selectiveif the response provided as a measurementdepends only on the intended ecosystem property(Olsen et al. 1999). Regarding this aspect,Yoccoz et al. (2001) suggest that quantitativestate variables characterizing the system wellshould be privileged. For example, when den-ing management objectives in terms of changesof densities of indicator species, the programshould incorporate tests to ensure that selectedspecies are indeed indicators of the process andvariables of interest (Yoccoz et al. 2001). For thisreason, it is important to establish a priori thevariables of interest in a sampling protocol. Thecriteria for this selection should be based ondata quality, applicability, data collection,repeatability, data analysis and interpretation,and cost-effectiveness. In particular, the use ofquantitative state variables is recommended in

order to reduce the error in data collecting due tothe subjectivity of the operators.

2.2.3.2 Non-sampling ErrorsNon-sampling errors include measurement, clas-sication, and observer errors, which are rootedin how the standard operating procedures (SOPs)are prepared and applied and how well-trainedand skilled the eld crews are (Ferretti 2009). Ingeneral, non-sampling errors can occur when themethodology is poorly standardized, when teamshave insufcient skills or insufcient care istaken in applying the method, or when there areproblems with instrument calibration (Ferrettiand Erhardt 2002). Many papers have focused onthese topics and have shown that non-samplingerrors can be a signicant source of variability inmonitoring studies (Gertner and Khl 1992;McCune et al. 1997; Giordani et al. 2009;Francini et al. 2009; Gottardini et al. 2009;Marchetto et al. 2009; Sastre and Lobo 2009).

In the specic case of epiphytic lichen assess-ment, non-sampling error may basically occur attwo stages (Brunialti et al. 2012a): the identica-tion of standard trees and the counting of lichenspecies. As for the former case, an imprecisedenition of suitable trees in the SOPs might beone of the main reasons for differences in thenumber of suitable trees found by the differentteams. As for the counting of lichen species, flo-ristic knowledge is a crucial issue that needs to beaddressed in lichen biomonitoring where theprotocol is based on assessment of all lichen spe-cies, including groups of lichens which are hard toidentify in the eld, such as crustose lichens (Astaet al. 2002). Differences in the floristic skills ofthe teams can cause serious errors (Brunialti et al.2002; Giordani et al. 2009). Brunialti et al. (2012a)conrmed that considerable underestimation ofspecies richness may occur even when samplingwithin an a priori positioned grid. Floristic skill iseven more important when assessing crustose-dominated communities, where poorly developedthalli often occur (Giordani et al. 2009).Undoubtedly, variability among crews couldbe reduced with intercalibration courses and

32 P. Giordani and G. Brunialti

harmonization procedures (McCune et al. 1997;Brunialti et al. 2002, 2004). In fact, there is evi-dence that operators often improve in accuracyduring the same test and that their accuracyimproves with taxonomic training and, above all,continuous eldwork (Brunialti et al. 2002).

As far as precision is concerned, very highlevels are usually registered among operators,suggesting a high reproducibility of the lichendiversity counts (Brunialti et al. 2002). This isextremely important for correct evaluation oftime series in biomonitoring studies. However, itshould be borne in mind that changes in opera-tors in long-term monitoring of permanent plotscan give misleading results (McCune et al. 1997)and should be carefully addressed.

2.3 Interpreting Lichen Diversity

As described in the previous paragraphs, a con-siderable effort has been made in the recent yearsfor standardizing the sampling design and strat-egies of lichen biomonitoring. In the followingparagraphs, we will describe some of the mostused approaches for interpreting lichen diversitydata in terms of effects of various anthropogenicdisturbances. In general, the interpretation ofgeographic patterns and temporal trends of lichendiversity may be assisted by using ecologicalindicator values (Hawksworth and Rose 1970;Wirth 2010; Nimis and Martellos 2001, 2002),multivariate statistics, such as numerical analysisof matrices of species (Giordani et al. 2002;Giordani 2006), nonparametric models (McCuneand Mefford 2004; Giordani 2007), or otherstatistical tools.

2.3.1 The Concepts of Biomonitoring

We can dene monitoring as the process ofgathering information about some systemstatevariables at different points in time for the pur-pose of assessing the state of the system andmaking inferences about changes in state overtime (Yoccoz et al. 2001). If our focus is onthe monitoring of biological diversity, the

systems of interest to us are typically ecosystemsor components of such systems (communitiesand populations), and the variables of interestinclude quantities such as species richness, spe-cies diversity, biomass, and population size.

2.3.1.1 Lichen Diversity Value,-Diversity

According to the recently developed standardsfor lichen biomonitoring (Stofer et al. 2012; EN16413 2014), the basic results of lichen diversitysampling are aggregated matrices of the speciesfrequencies at nested spatial levels of sampling,i.e., a matrix of species at aspects of each tree; attrees of each sampling units; and, nally, atsampling unit level. Several diversity indices arecalculated a posteriori basing on these basicmatrices. This recommendation comes to the factthat several interpretative tools may be applied tobasic data, which are used in a various manner indifferent countries.

As a simple but effective approach, -diver-sity (number of species at plot level) is a robustparameter for interpreting patterns of epiphyticlichen communities along gradient of anthropo-genic effects, such as pollution or forestmanagement.

Among the possible descriptors obtained fromthe basic species sampling unit matrix, thelichen diversity value (LDV) by Asta et al.(2002) is by far one of the most used in appli-cative biomonitoring surveys. The value for agiven sampling unit is calculated as the sum ofthe frequencies of all lichen species found oneach tree within the unit and averaged by thenumber of sampled trees. Relevant differences inlichen growth may be expected on different car-dinal aspects of the trunks; therefore, it is sug-gested that frequencies are also summedseparately for each aspect, and possibly, addi-tional analysis might be carried out in this sense.LDV in its basic denition has been extensivelyused in applicative studies all over Europe (Paoliet al. 2006; Svoboda et al. 2010; Giordani et al.2014a), and its relationships with pollution andother environmental factors had been analyzed indetails. In the Mediterranean, together with

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 33

pollution, the mean annual rainfall of the surveyarea seems to be the best predictor for LDVvariability (Giordani 2006). However, the deci-sive variables affecting the lichen diversity areapparently different in urban versus forestedareas (Giordani 2007). In these latter, harvestingand forest res showed a predominant effect.Contrarily, in urban areas, air pollutants, mainlySO2, are the main limiting factors, even if thisrelationship is lowering under ameliorating con-ditions of atmospheric pollution. Similarly,Svoboda et al. (2010) found that lichen diversityin Central Europe responded differently to envi-ronmental predictors depending on differenthuman impact. These authors observed that inindustrial regions, air pollution was the strongestfactor affecting lichen diversity, whereas inagricultural to highly forested regions, LDV wasmainly influenced by forest age and forestfragmentation.

Starting from the basic species samplingunit matrix, further parameters can be derivedfrom the data set of the species frequencies,including the relative LDVs of morphofunctionalgroups of lichens associated to particular sourcesof atmospheric pollution (e.g., nitrophilus versusacidophilus speciessee Sect. 3.2.3).

2.3.1.2 Rapid Biodiversity Assessment(RBA) Based on Morphospecies

The application of biomonitoring methods basedon high levels of taxonomic knowledge, such asthe recently standardized lichen biomonitoringmethod (EN 16413 2014), requires an adequatenumber of specialists that are not always avail-able, especially in large-scale biodiversityassessments (Wilkie et al. 2003). The use ofguilds or morphological groups as indicators ofchanges in ecosystem function has been consid-ered by several authors as a good compromisebetween the need for specialized knowledge andrapid eld procedures employing non-specialisttechnicians, thus providing a possible shortcut inassessing total species richness (see, e.g., Pharoet al. 2000; Giordani et al. 2009). This issue hasbeen considered in several ecological monitoringelds to explore the possibility of using surrogate

species for estimate total biological diversity. Asan example, several studies have been performedto assess the congruence among vascular plants,vertebrate, invertebrate, bryophytes, and lichenswithin large-scale biomonitoring surveys usingsimplied assessment methods (see Oliver et al.1998; Pharo et al. 2000; Wilkie et al. 2003; Santiet al. 2010). However, some conflicting resultsobtained from the works listed above suggest thatthis is not always the ideal solution and severalcritical issues emerge: Among the others, thereare sources of variation coming from the fact thatthe communities do not always behave in a linearand unambiguous manner. Also, the data qualityin surveys involving non-specialist crews mayvary at such a level which could drasticallycompromise the reliability of the results. In thisrespect, Giordani et al. (2009) carried out a studyto compare data obtained by non-specialiststhrough simplied methods based on morpho-species (RBA), with those collected by special-ists using the lichen diversity value (LDV)method. They found that lichen diversity esti-mated by means of Rapid Biodiversity Assess-ments (i.e., based on morphospecies) showedgood correlations with the results of a classical,systematic identication of species only whenperformed by operators with high taxonomicknowledge. Furthermore, the use of samplinglists based on highly simplied morphospeciesdid not lead to signicant advantages in terms oftime needed for eldwork (Fig. 2.5).

In some other cases, Rapid BiodiversityAssessments of lichen biodiversity led to inter-esting results even at large scale. Recently, citi-zen science approaches have been applied tolichen biomonitoring of the effects of atmo-spheric pollution. These are voluntary schemesengaging members of the public in the collectionof scientic information. The OPAL Air Surveyin the UK (Davies et al. 2011) used presence andabundance data for 9 selected lichens, collectedby more than 4,000 volunteers in a public surveyof lichens on trees, to examine the response ofindividual species and groups of indicator lichensto air pollution and climate drivers on a nationalscale. The use of these macrolichen indicatorshas shown to have robust relationships with

34 P. Giordani and G. Brunialti

modeled nitrogenous pollutants at the nationalscale (Seed et al. 2013). In the USA, Casanovaset al. (2014) proposed a citizen scientist-basedsurvey methodology for macrolichen diversity inwhich parataxonomic units (PUs), as identied inlichen photographs, served as species surrogatesto estimate lichen diversity. Although in mostcases the authors showed that the observed andestimated cumulative richnesses from both tech-niques were not statistically signicantly differentfrom each other, the extensive use of theseapproaches in biomonitoring surveys should becarefully evaluated, as misidentications ofmorphologically similar species could led towrong interpretation of data, e.g., in terms ofrelative abundance of functional groups. At thisregards, it has been shown that the use of elec-tronic devices and identication tools may help toincrease the quality of RBA. The project Dryadesdeveloped interactive identication keys in theform of applications for mobile devices. Keyswere generated from databases of morphoana-tomical characters. The applications were testedEurope wide during the project KeyToNature andhave proved to be useful in education and inprojects of citizen science (Nimis et al. 2012).

2.3.1.3 -DiversityMost biomonitoring applicative surveys andresearches using epiphytic lichens have focusedon analyzing plot-level species richness (e.g.,alfa-diversity), abundance, and/or compositionpatterns along ecological gradients. Recently, ithas been suggested that also analyses of-diversity may provide insights into mecha-nisms and drivers influencing lichen communi-ties, thus contributing to a better interpretation ofthe results. Beta-diversity has been interpretedmostly as the extent of change in communitycomposition (Whittaker 1960). Basically, beta-diversity patterns are originated from two distinctprocesses, the replacement and the loss of species(Carvalho et al. 2012). With the aim of differ-entiating the relative influences of these compo-nents on beta-diversity, various measures havebeen proposed, which relied on an additive ratherthan multiplicative approach (Baselga 2010). Asa rst application to lichen communities,Nascimbene et al. (2013a) used the conceptualscheme by Podani and Schmera (2011) to eval-uate the relative importance of -diversity, nest-edness, and agreement in species richness inpresenceabsence data matrices via partitioning

Fig. 2.5 Time (minutes)required for each lichendiversity assessment (treelevel), using differentsampling lists. Lichendiversity value (samplingof lichens at species level)was compared with twodifferent simplicationmethods of rapidbiodiversity assessment(RBA, simplied methodsA and B). Data fromGiordani et al. (2009)

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 35

pairwise gamma diversity into additive compo-nents. Podani and Schmera (2011) consideredthree complementary indices that measure simi-larity (S), relative species replacement (R), andrelative richness difference (D) for given pairs ofobservations. In particular, according to theseauthors, -diversity is dened as the additiveresult of R and D, whereas other descriptors,namely nestedness and richness agreement, resultfrom the additive effects of similarity with theother two complementary components (S + Dand S + R). By analyzing the diversity inL. pulmonaria communities in Italian forests,Nascimbene et al. (2013a) showed that bothspecies replacement and similarity were gener-ally associated with forest structure predictors,such as the number of trees in the plot and thedistance between trees, while richness differencewas mainly associated with geographic predic-tors, with special reference to longitude andaltitude. Giordani et al. (2014b) coupled theanalysis of -diversity with the approach basedon the relative abundance of functional groups.These authors explored the shift in functionalgroups for nitrogen tolerance along a gradient ofincreasing cattle load in epilithic lichen com-munities of alpine pasturelands. An increasingcattle load caused a decreasing replacement ofoligotrophic species and consequently a decreasein -diversity. Conversely, when considering adata set with only N-tolerant species, there wasvery high pairwise similarity among samplingplots, irrespectively by the cattle load gradient towhich they were exposed.

2.3.1.4 Indicator SpeciesDetecting species that best characterize some setof sites is an important step in evaluating classi-cations in community ecology. With referenceto lichen biomonitoring, this approach has beenmostly applied for detecting the effects of forestmanagement or to assess the ecological continuityof ancient woodlands. In some cases, lists ofindicators species were compiled on the basis ofexpert assessments (e.g., Rose 1976). As for otherbiologically based approaches, also for biomoni-toring, the use of proper methods for measuring

the explanatory power of species is essential,whenever priority must be given to species thatbest reflect environmental quality (Dufrne andLegendre 1997; Podani and Csnyi 2010).

Whittet and Ellis (2013) tested 29 indicatorspecies of forest continuity, as proposed by thecurrent suite of British lichen indicators (Coppinsand Coppins 2002) for different biogeographicregions in the UK. In accordance with previousstudies (e.g., Stersdal et al. 2005; Giordani andIncerti 2008), these authors conrmed that indi-cator species are likely to have a restricted geo-graphic scope. Moreover, they suggested thatonly some of the studied taxa could be actuallyaccurate indicators of ecological continuitythrough a dependency on the long-term persis-tence, whereas in other cases, several speciesmay possibly be associated with specialistmicrohabitats under a sub-optimal climate,whereas they would not be signicantly associ-ated with ancient woodlands. Giordani (2012)compared the performance of four potentialindicators for monitoring the effects of forestmanagement on epiphytic lichens in broad-leaved Mediterranean forests. Indicators includedtotal lichen diversity (LDV) and the abundancesof species associated with intensive management,species associated with aged coppiced wood-lands and indicator species ratio (ISR). ISR wasdened as the ratio between the differencebetween the species associated to aged coppiceforests and those associated to intensively man-aged forests and the total abundance. At each of50 sampling sites, the four indicators were cal-culated using indicator value analysis (Dufrneand Legendre 1997) and compared through cor-respondence analysis. By balancing the partialinformation provided by both sensitive andresistant species, ISR was shown to be a moreeffective indicator, being independent of floristiccomposition and the occurrence of rare species.The main drawback of the indicator speciesapproach is that evaluation of the effects of agiven stress (e.g., forest management) is possiblybiased due to the fact that lichen species arestrongly threatened by several anthropogenicdisturbances occurring at the same time (e.g.,high levels of air pollutants or forest res).

36 P. Giordani and G. Brunialti

2.3.2 Interpretative Tools

2.3.2.1 Interpretative Scalesof Alteration

Following the denition of Nimis (1999), bio-monitoring techniques estimate the degree ofalteration from normal conditions resulting fromthe effects of pollution on the reactive compo-nents (e.g., lichens) of the ecosystems. However,dening normal conditions in ecology is extre-mely critical and can only rely on an operationalbasis. At this regard, some authors proposed aninterpretation of lichen diversity data in terms ofpercentile deviations from an observed maximumdiversity. Loppi et al. (2002a, b) sampled lichendiversity on about 3,000 trees in northern Tyr-rhenian, Italy, and considered as natural thosevalues the 98 percentile of their frequencydistribution. The average of these values wastaken as an operational denition of naturality,which represents a maximum potential lichendiversity in a given area. These authors consid-ered that a 25 % deviation from normal condi-tions could still be regarded as natural, owingto natural fluctuations of lichen diversity. Startingfrom this point, an interpretative scale of natu-rality/alteration was build basing the differentdegrees of naturality/alteration on progressive25 % deviations from normal conditions.

As lichen distribution strongly depends onmacro- and mesoclimatic factors (see Ellis 2012),such interpretative scales of lichen diversity interms of deviation from maximum potentialconditions information should be referred to aregional level. Thus, direct comparisons betweenbiomonitoring surveys carried out in differentbioclimatic regions are often poorly informativeand can also led to misleading interpretations. Asimilar approach of regionalization has beenfollowed and debated in the case of other bio-monitoring techniques (e.g., Moog et al. 2004).For example, the EU Water Framework Directivesuggested qualitative reference for evaluating theecological status of water bodies, assessing thehighest potential quality, based on the composi-tion of aquatic communities. It was possible topredict how the response of diatom communities

to anthropogenic pressure in each French hydro-ecoregion was predicted (Tison et al. 2007), inrelation to the topology of running waters andvalidated with benthic macroinvertebrates fromreference sites (Wasson et al. 2002).

The delimitation of eco-regions or bioclimaticregions bases upon differences in the composi-tion of their lichen flora. At this regards, Gior-dani and Incerti (2008) applied a nonparametricmultiplicative regression model to the lichenflora of a climatically heterogenous area anddetected 59 species which were signicantlyassociated to macroclimatic variables (i.e.,annual rainfall and temperature). A cluster anal-ysis grouped the taxa into four subsets that wererelated to different climatic niches (warmhumid,coldhumid, mesothermichumid, warmdry)corresponding to distinct bioclimatic regions.

As an example of regionalization of interpre-tative scales, Castello and Skert (2005) providedevaluation scales of environmental alterationbased on lichen diversity in the North Adriaticsub-Mediterranean bioclimatic region. Theseauthors sampled deciduous oaks in 11 referencesites in open stands within natural woods or nearvery small isolated villages in rural or naturalareas, far from large urban areas, industrialzones, and long-distance transport of air pollu-tants. As a result of relatively dry conditionoccurring in North Adriatic, the LDV thresholdfor naturality class was slightly lower than theone calculated for the more humid ThyrrenianItaly (Giordani 2004). Regionalized interpreta-tions are also strongly recommended by theGerman guidelines (VDI 3957 2005), whichstates that a comparison of surveys of differentregions is only possible if the surveyed areashave a similar climate and, therefore, a compa-rable lichen flora, as is the case with CentralEurope (excepting the Alps) for which the eval-uation scale was calibrated.

In some cases, ad hoc interpretative scaleshave been calculated for local situations. Due tothe lack of an interpretative scale for semiaridMediterranean bioclimatic region, Paoli et al.(2006) developed a calibrated scale for asmall area (36 km2) according to the protocolsuggested by Loppi et al. (2002a, b) for the

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 37

assessment of environmental deviation fromnatural conditions. Macroregional- and local-based interpretations of lichen diversity are notmutually exclusive and can provide integrateinformation on the actual environmentalconditions.

Thought that nitrogen is becoming one of themost relevant limiting factors for lichen com-munities, especially in Central Europe, the Ger-man standard for lichen biomonitoring (VDI3957 2005) proposed an interesting integrateapproach for interpretation, which is basicallybased on the relative diversity of functionalgroups for nitrogen requirements. According tothe German guidelines, nitrophytic species whichrespond positively and oligotrophic specieswhich respond negatively to eutrophication(referred to as reference species) are calculatedseparately. The two partial LDVs of a samplingunit are combined to form the air quality index.The LDV of the reference species is enteredalong the ordinate axis, whereas the LDV of theindicators of eutrophication is entered along theabscissa. The quality class assigned to the eld inwhich the crossing point of the lichen diversityvalues comes to lie gives the evaluation of the airquality of the sampling unit. According to theGerman approach, the thresholds of the airquality classes in the matrix are chosen so thatthe class width is approximately equal to threetimes of standard deviation of the LDV in thestudy area. In this way, the lichen cover ofsampling units belonging to different air qualityclasses is signicantly different if the respectiveclasses are separated by at least one other class.

The use of interpretative scales of lichendiversity based on deviation from maximumpotential diversity also presents some weakpoints. From a theoretical point of view, it is wellknown that natural situations are not necessarilyassociated with maximum values of biodiversity.According to influential ecological theories, suchas the intermediate disturbance hypothesis(Connell 1978), intermediate levels of distur-bance, in terms of frequency and/or intensity ofthe phenomena, will maximize species diversity.The question has been shown to be more com-plex and less generalizable than this, with recent

studies demonstrating that diversity could showmonotonic, unimodal, or even flat response todisturbance, depending on the studied organismsand on the disturbance aspects considered (Hallet al. 2012). In an applicative perspective, thelacking of undisturbed reference situations insome eco- or bioclimatic regions (e.g., the PoPlain in Italy or many anthropized areas ofCentral Europe and North America) makes dif-cult to dene properly the values associated tothe highest classes of the scale, thus affecting theentire interpretation process.

2.3.2.2 Mapping Lichen DiversityMapping lichen diversity is an attractiveapproach, which allows an immediate represen-tation of the results. Showing spatial distributionpatterns of the studied descriptors, maps of lichenbiodiversity, or abundance enable a quick andclear identication of areas with different levelsof disturbance (Pinho et al. 2004; Asta et al.2002). Spatial mapping of lichen diversity orassociated measurements had been extensivelyused both in research and applicative lichenbiomonitoring works (Giordani et al. 2002;Pinho et al. 2004, 2008a; Geiser et al. 2010). Alarge suite of GIS softwares provide tools forinterpolating (i.e., estimating) the values of theresponse variable (e.g., lichen diversity) in non-measured parts of the survey area, basing on thedata collected at sampling sites. In this chapter,we do not review the technical aspects of thisapproach. Basically, it follows the principle ofgeostatistical modeling theory originally devel-oped for applied geology and more recentlyapplied to ecology (see, e.g., Perry et al. 2002 fora detailed description). However, when applyingthese techniques, lichenologists should be awarethat incorrect settings of the geostatistical modelcould lead to misleading results and interpreta-tion and that the error associated to a poor spatialmodel might be also larger than those imputableto other sources of errors, such as samplingdesign or taxonomic misidentication.

Estimated values in lichen biodiversity mapsare related to sampling densities (the number ofsampling units in the study area), which in turn

38 P. Giordani and G. Brunialti

affect the reliability of the collected data in termsof condence intervals and relative error in themean values (Geiser and Neitlich 2007). Byanalyzing two large surveys of lichen mappingcarried out in Italy (Nimis et al. 1991; Giordaniet al. 2002), Ferretti et al. (2004) showed thateven a considerable reduction (up to 50 %) of theoriginal sampling effort led to a much smallerincrease in mapping errors (

ammonia concentrations into classes and focus-ing on the highest diversity values. Based on thesignicant correlations between ammonia andbiodiversity, the CLE of ammonia for Mediter-ranean evergreen woodlands was found to be0.69 g m3, below the currently accepted pan-European CLE of 1.0 g m3.

Concerning the detection of CLOs, theoccurrence of oligotrophic lichen species pro-vided information on the actual impact of reducednitrogen compounds (mainly ammonia) in theforest plots of the European network ICP Forests(Giordani et al. 2014a). The critical load causing asignicant change of the expected composition ofepiphytic lichen vegetation occurred at nitrogendeposition = 2.4 kg N ha1 year1. Lichen func-tional groups for eutrophication and/or, morespecically, nitrogen tolerance have been exten-sively used to assess the critical level and criticalload of nitrogen compounds in several forestecosystems all over the world (Fenn et al. 2008;Geiser et al. 2010; Pinho et al. 2008b, 2011).Interestingly, the results for European plots werein accordance with those of other areas of theworld. For example, in conifer forests of thePacic North West of USA, Geiser et al. (2010)found, for wet deposition, a critical load rangingfrom 0.7 to 4.4 kg ha1 year1, depending on theamount of precipitation. The concept of criticalload for lichen communities has been also appliedto other sources of disturbance. Giordani et al.(2014b) established the cattle critical load inalpine pasturelands, in terms of Adult Cattle Units(ACU) per hectare per year. These authorsshowed that the relative frequency of oligotrophicepilithic lichen species signicantly decreasedas ACU increased. The cattle critical load was setfor ACU = 0.12 ACU ha1 year1.

2.4 Open Questions

Despite the relevant number of researches in theeld of lichen biomonitoring carried out in the last25 years, there are still open points which have notyet been sufciently addressed. In some cases,these questions are far from being minor and

responses are urged in order to make the inter-pretation of lichen biomonitoring more robust.Among the others, the insufcient knowledge onthe timescale of the response of epiphytic lichencommunities to disturbances and the lacking ofan adequate integration about lichen biomonitor-ing and other standard biomonitoring techniquesparticularly call for more attention by researchersand stakeholders. According to its technical de-nition, monitoring is the collection and analysis ofrepeated observations or measurements to evalu-ate changes in condition and progress toward amanagement objective (Elzinga et al. 2001). Thus,also for lichen biomonitoring studies, an accurateknowledge on the temporal variation of theobserved phenomenon is crucial for getting reli-able results. However, despite the huge literatureon physiological effects of disturbances on lichenthalli, the time span between the disturbance (e.g.,pollution) and the alteration on epiphytic lichencommunities in terms of species loss, recover, orchanges in species composition has not been fullyexplored. Such effects result from complex inter-actions between temporal trends of limiting fac-tors (e.g., phytotoxic gases) and natural dynamicsof the communities, which include re-coloniza-tion processes driven by dispersal, substrateavailability, establishment of propagules, andintra- and interspecic competition (Werth et al.2006).

Data from long observation periods showedclear trends in lichen diversity (Lisowska 2011).The diversity of epiphytic species in London hascontinued to increase from the 1970s to 2004 as aresponse to decreasing NOx atmospheric concen-trations (Hawksworth and Rose 1970; Davies2007). In the urban area of Turin (N-Italy), con-trasting trends of the numbers of both lichen spe-cies presence and abundance were observed over aperiod of 200 years as a result of changing pollu-tion scenarios (Isocrono et al. 2007). A dramaticspecies loss was detected in seminatural broad-leaved forests in northwestern Germany from thenineteenth century to date (Hauck et al. 2013).Up to 70 % of the species became rarer during the100- to 150-year long observation period, and anextinction rate of 28 % was estimated.

40 P. Giordani and G. Brunialti

Against these robust evidences of long-termvariations, information on short-term trends oflichen diversity and composition are surprisinglyscarce and urge more detailed researches. Loppiet al. (2004) carried out 5 repeated surveys oflichen from 1993 to 2000 with time spans rang-ing from 1 to 3 years. They showed that despitetheir slow growth rate, lichens respond rapidly todecreasing concentrations of air pollutants,allowing annual changes to be detected. Totalspecies richness increased from 1993 to 1999 andthen decreased again in 2000, while the meannumber of species per station increased from1993 to 1999 and remained constant in 2000.The -diversity decreased linearly from 1993 to1999, indicating that sampling stations becamefloristically more similar in time.

From an applicative perspective, when the aimis the evaluation of change between subsequentmeasurements, there are several implicationsrelated to the statistical analysis for detectingchanges, which should carefully considered (EN16413 2014). Among them, one should decidewhether to make the sampling units temporary orpermanent. When sampling units are temporary,the random sampling procedure is carried outindependently at each sampling period (Elzingaet al. 2001). The principal advantage of usingpermanent instead of temporary sampling units isthat the statistical tests for detecting change fromone time period to the next in permanent samplingunits are much more powerful than the tests usedon temporary sampling units.

References

AFNOR (2008) Biosurveillance de lenvironnementDtermination dun indice biologique de lichenspiphytes (IBLE). NF X43-903

Asta J, Erhardt W, Ferretti M, Fornasier F, Kirschbaum U,Nimis PL, Purvis OW, Pirintsos S, van Scheidegger C,Haluwyn C, Wirth V (2002) Mapping lichen diversityas an indicator of environmental quality. In: Nimis PL,Scheidegger C, Wolseley PA (eds) Monitoring withlichensmonitoring lichens. Kluwer Academic Pub-lishers, Dordrecht, pp 273279

Baselga A (2010) Partitioning the turnover and nestednesscomponents of beta diversity. Glob Ecol Biogeogr19:134143

Bates JW (1992) Influence of chemical and site factors onQuercus and Fraxinus epiphytes at Loch Sunart,western Scotland: a multivariate analysis. J Ecol80:163179

Belinchn R, Martnez I, Escudero A, Aragn G,Valladares F (2007) Edge effects on epiphytic com-munities in a Mediterranean Quercus Pyrenaica forest.J Veg Sci 18:8190

Bennett JP, Wetmore CM (1999) Changes in elementcontents of selected lichens over 11 years innorthernMinnesota, USA. Environ Exp Bot 41:7582

Bergamini A, Scheidegger C, Stofer S, Carbalho P, DaveyS, Dietrich M, Dubs F, Farkas E, Groner U, KrkinenK, Keller C, Lks L, Lommi S, Mguas C, MitchellR, Pinho P, Rico VJ, Aragn G, Truscott A-M,Wolseley P, Watt A (2005) Performance of macroli-chens and lichen genera as indicators of lichen speciesrichness and composition. Conserv Biol 19:10511062

Boudreault C, Bergeron Y, Drapeau P, Mascara Lpez L(2008) Edge effects on epiphytic lichens in remnantstands of managed landscapes in the eastern borealforest of Canada. Forest Ecol Manage 255:14611471

Brunialti G, Giordani P (2003) Variability of lichendiversity in a climatically heterogeneous area (Liguria,NW Italy). The Lichenologist 35:5569

Brunialti G, Giordani P, Isocrono D, Loppi S (2002)Evaluation of data quality in lichen biomonitoringstudies: the Italian experience. Environ Monit Assess75:271280

Brunialti G, Giordani P, Ferretti M (2004) Discriminatingbetween the good and the bad: quality assurance iscentral in biomonitoring studies. Environ Monit20:443464

Brunialti G, Frati L, Incerti G, Rizzi G, Vinci M, GiordaniP (2010a) Lichen biomonitoring of air pollution:issues for applications in complex environments. In:Romano GC, Conti AG (eds) Air quality in the 21stcentury. Nova Science Publishers Inc., New York

Brunialti G, Frati L, Alef M, Marignani M, Rosati L,Burrascano S, Ravera S (2010b) Lichens and bryo-phytes as indicators of old-growth features in Medi-terranean forests. Plant Biosyst 114(1):221233

Brunialti G, Frati L, Cristofolini F, Chiarucci A, GiordaniP, Loppi S, Benesperi R, Cristofori A, Critelli P, DiCapua E, Genovesi V, Gottardini E, Innocenti G,Munzi S, Paoli L, Pisani T, Ravera S, Ferretti M(2012a) Can we compare lichen diversity data? A testwith skilled teams. Ecol Ind 23:509516

Brunialti G, Frati L, Loppi S (2012b) Fragmentation ofMediterranean oak forests affects the diversity ofepiphytic lichens. Nova Hedwig 96(12):265278

Ccares MES, Lcking R, Rambold G (2007) Photophytespecicity and environmental parameters versus sto-chasticity as determinants for species composition ofcorticolous crustose lichen communities in the Atlan-tic rain forest of north-eastern Brazil. Mycol Prog6:117136

Cape JN, van der Eerden LJ, Sheppard LJ, Leith ID,Sutton MA (2009) Evidence for changing the criticallevel for ammonia. Environ Pollut 157:10331037

2 Sampling and Interpreting Lichen Diversity Data for Biomonitoring Purposes 41

Caruso A, Thor G (2007) Importance of different treefractions for epiphytic lichen diversity on Picea abiesand Populus tremula in mature managed boreonemor-al Swedish forests. Scand J Forest Res 22:219230

Carvalho JC, Cardoso P, Gomes P (2012) Determiningthe relative roles of species replacement and speciesrichness differences in generating beta-diversity pat-terns. Glob Ecol Biogeogr 21:760771

Castello M, Skert N (2005) Evaluation of lichen diversityas an indicator of environmental quality in the NorthAdriatic submediterranean region. Sci Total Environ336:201214

Casanovas P, Lynch HJ, Fagan WF (2014) Using citizenscience to estimate lichen diversity.Biol Cons 171:18

Cislaghi C, Nimis PL (1997) Lichens, air pollution andlung cancer. Nature 387:463464

Cline SP, Burkman WG (1989) The role of qualityassurance in ecological programs. In: Bucher JB,Bucher-Wallin J (eds) Air pollution and forest decline.Birmensdorf, IUFRO, p 361

Cochran WG (1977) Sampling techniques, 3rd edn.Wiley, New York

Collins SL, Glenn SM (1997) Effects of organismal anddistance scaling on analysis of species distribution andabundance. Ecol Appl 7:543551

Connell JH (1978) Diversity in tropical rain forests andcoral reefs. Science 199:13021310

Coppins AM, Coppins BJ (2002) Indices of ecologicalcontinuity for woodland epiphytic lichen habitats inthe British Isles. British Lichen Society, London

Cornelissen JHC, Callaghan TV, Alatalo JM, MichelsenA, Graglia E, Hartley AE, Hik DS, Hobbie SE, PressMC, Robinson CH, Henry GHR, Shaver GR, PhoenixGK, Gwynn Jones D, Jonasson S, Chapin FS III,Molau U, Neill C, Lee JA, Melillo JM, Sveinbjrns-son B, Aerts R (2001) Global change and arcticecosystems: is lichen decline a function of increases invascular plant biomass? J Ecol 89:984994

Cristofolini F, Brunialti G, Giordani P, Nascimbene J,Cristofori A, Gottardini E, Frati L, Matos P, Bati F,Caporale S, Fornasier MF, Marmor L, Merinero S,Nuez Zapata J, Trra T, Wolseley P, Ferretti M(2014) Towards the adoption of an internationalstandard for biomonitoring with lichensconsistencyof assessment performed by experts from six Euro-pean countries. Ecol Ind 45(2014):6367

Crumbling D (2002) In search of representativeness:evolving the environmental data quality model. QualAssur 9:179190

Davies L, Bates JW, Bell JNB, James PW, Purvis WO(2007) Diversity and sensitivity of epiphytes to oxidesof nitrogen in London. Environ Pollut 146:299310

Davies L, Bell JNB, Bone J, Head M, Howard C, HobbsSJ, Jones DT, Power SA, Rose N, Ryder C, Seed L,Stevens G, Toumi R, Voulvoulis N, White PCL(2011) Open air laboratories (OPAL): a communitydriven research programme. Environ Pollut 159(9):22032210

de Hauck M, Bruyn U, Leuschner C (2013) Dramaticdiversity losses in epiphytic lichens in temperatebroad-leaved forests during the last 150 years. BiolConserv 157:136145

Dufrne M, Legendre P (1997) Species assemblages andindicator species: the need for a flexible asymmetricalapproach. Ecol Monogr 67:345366

Dungan JL, Perry JN, Dale MRT, Legendre P, Citron-Pousty S, Fortin MJ et al (2002) A balanced view ofscale in spatial statistical analysis. Ecography 25:626640

Ellis CJ (2012) Lichen epiphyte diversity: a species,community and trait-based review. Perspect PlantEcol, Evol Syst 14:131152

Ellis CJ, Coppins BJ (2007) Reproductive strategy andthe compositional dynamics of crustose lichen com-munities on aspen (Populus tremula L.) in Scotland.The Lichenologist 39:377391

Elzinga CL, Salzer DW, Willoughby JW, Gibbs PJ (2001)Monitoring plant and animal populations. BlackwellScience, Malden 360 pp

EN 16413 (2014) Ambient airbiomonitoring withlichensassessing epiphytic lichen diversity. ComitEuropen de Normalisation, 33 pp

Esseen P-A, Renhorn K-E (1998) Edge effects on anepiphytic lichen in fragmented forests. Conserv Biol12:13071317

Fenn ME, Jovan S, Yuan F, Geiser L, Meixner T, GimenoBS (2008) Empirical and simulated critical loads fornitrogen deposition in California mixed conifer for-ests. Environ Pollut 155:492511

Ferretti M (1998) Potential and limitations of visualindices of tree condition. Chemosphere 36:1031

Ferretti M (2009) Quality assurance in ecological mon-itoringtowards a unifying perspective. J EnvironMonit 11:726729

Ferretti M (2011) Quality assurance: a vital need inecological monitoring. CAB Rev 6(2011). http://www.cabi.org/cabreviews

Ferretti M, Erhardt W (2002) Key issues in designingbiomonitoring programmes. Monitoring scenarios,sampling strategies and quality assurance. In: NimisPL, Scheidegger C, Wolseley P (eds) Monitoring withlichens-monitoring lichens. Kluwer Academic Pub-lishers, Dordrecht, pp 111139

Ferretti M, Brambilla E, Brunialti G, Fornasier F, MazzaliC, Giordani P, Nimis PL (2004) Reliability of differentsampling densities for estimating and mapping lichendiversity as a bioindicator of environmental quality.Environ Pollut 127(2):249256

Fischer R, Granke O, Chirici G, Meyer P, Seidling W,Stofer S, Corona P, Marchetti M, Travaglini D (2009)Background, main results and conclusions from a testphase for biodiversity assessments on intensive forestmonitoring plots in Europe. iForest 6774

Francini A, Pellegrini E, Lorenzini G, Nali C (2009) Non-sampling error in ozone biomonitoring: the role ofoperator training. J Environ Monit 11:736744

42 P. Giordani and G. Brunialti

Frati L, Brunialti G (2006) Long-term biomonitoring withlichens: comparing data from different samplingprocedures. Environ Monit Assess 119:391404

Frati L, Santoni S, Nicolardi V, Gaggi C, Brunialti G,Guttova A, Gaudino S, Pati A, Pirintsos SA, Loppi S(2007) Lichen biomonitoring of ammonia emissionand nitrogen deposition around a pig stockfarm.Environ Pollut 146:311316

Fritz O, Brunet J, Caldiz M (2009) Interacting effects oftree characteristics on the occurrence of rare epiphytesin a Swedish boreal forest. Bryologist 112:488505

Gaston KJ (2000) Global patterns in biodiversity. Nature405:220227

Gauslaa Y (1995) The Lobarion, an epiphytic lichencommunity of ancient forests threatened by acid rain.The Lichenologist 27:5976

Geiser LH, Neitlich PN (2007) Air pollution and climategradients in western Oregon and Washington indi-cated by epiphytic macrolichens. Environ Pollut145:203218

Geiser LH, Jovan SE, Glavich DA, Porter MK (2010)Lichen-based critical loads for atmospheric nitrogendeposition in Western Oregon and Washington For-ests, USA. Environ Pollut 158:24122421

Gertner G, Khl M (1992) An assessment of some non-sampling errors in a national survey using an errorbudget. Forest Sci 41:758

Gertner G et al (2002) Mapping and uncertainty ofpredictions based on multiple primary variables fromjoint co-simulation with Landsat TM image andpolynomial regression. Remote Sens Environ 83:498

Giordani P (2004) Licheni epiti come biomonitorsdellalterazione ambientale. Influenza delle variabiliecologiche sulla diversit lichenica. Tesi di dottorato.Universit di Trieste

Giordani P (2006) Variables influencing the distributionof epiphytic lichens in heterogeneous areas: a casestudy for Liguria, NW Italy. J Veg Sci 17:195206

Giordani P (2007) Is the diversity of epiphytic lichens areliable indicator of air pollution? A case study fromItaly. Environ Pollut 146: