assessing ozone and nitrogen impact on net primary productivity with a generalised non-linear model

14
Invited paper Assessing ozone and nitrogen impact on net primary productivity with a Generalised non-Linear Model Alessandra De Marco a , Augusto Screpanti a , Fabio Attorre b , Chiara Proietti b , Marcello Vitale b, * a Italian National Agency for New Technologies, Energy and the Environment (ENEA), C.R. Casaccia, Via Anguillarese 301, 00123 S. Maria di Galeria, Rome, Italy b Department of Environmental Biology, SapienzaUniversity of Rome, Piazzale Aldo Moro, 5, 00185 Rome, Italy article info Article history: Received 4 April 2012 Received in revised form 17 August 2012 Accepted 28 August 2012 Keywords: Generalized Linear/non-Linear Model Mediterranean climate Ozone Net primary productivity Nitrogen deposition abstract Some studies suggest that in Europe the majority of forest growth increment can be accounted for N deposition and very little by elevated CO 2 . High ozone (O 3 ) concentrations cause reductions in carbon xation in native plants by offsetting the effects of elevated CO 2 or N deposition. The cause-effect relationships between primary productivity (NPP) of Quercus cerris, Q. ilex and Fagus sylvatica plant species and climate and pollutants (O 3 and N deposition) in Italy have been investigated by application of Generalised Linear/non-Linear regression model (GLZ model). The GLZ model highlighted: i) cumulative O 3 concentration-based indicator (AOT40F) did not signicantly affect NPP; ii) a differential action of oxidised and reduced nitrogen depositions to NPP was linked to the geographical location; iii) the species-specic variation of NPP caused by combination of pollutants and climatic variables could be a potentially important drive-factor for the plant speciesshift as response to the future climate change. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction European forests play a major role in carbon sequestration (Nabuurs et al., 1997) and the increases in net primary productivity (NPP) in past decades are associated with changes in forest management (Lawler et al., 2010), inuencing the standing growing stock (Nabuurs et al., 2001). Moreover, the increase in forest productivity seems to be also due toother drivers such as rises in air-free CO 2 concentration (Matthews, 2007), temperature (Myneni et al., 1997; Luo, 2007), and nitrogen deposition (Thomas et al., 2010). However, there is no consensus on what the main environ- mental drivers are behind forest carbon sequestration, despite hundreds of papers on this topic (Long et al., 2004; Ainsworth and Long, 2005; Hyvönen et al., 2007; Norby et al., 2010). Some studies have suggested that in Europe the majority of forest growth increment can be accounted for N deposition (Solberg et al., 2004; Ciais et al., 2008) and very little by elevated CO 2 , but this does not seem to apply in all regions. However, other studies emphasize the key role played by high concentration of CO 2 on forest growth (Norby et al., 2005), although free air CO 2 experiments (FACE) reveal that tree speciesresponses differ. Lavorel et al. (1998) reported that Mediterranean regions are transitional zones where climatic changes may have the greatest effects and where intense feedbacks from the land to the atmo- sphere are expected (Scarascia-Mugnozza et al., 2000). In Mediterranean-type ecosystems and in other water-limited areas, global change factors act concurrently with additive effects rather than interactive ones (Heimann and Reichstein, 2008; Matesanz et al., 2008; Niu et al., 2009; Wu et al., 2011). In order to clarify the causes of the increased forest growth the EU-project RECOGNITION recently concluded that increased avail- ability of nitrogen was the main driver for increased growth, although it could not determine whether this was due to increased nitrogen deposition or increased availability in the soil (Kahle et al., 2008). Rehfuess et al. (1999) reported that the combination of CO 2 rise and elevated N deposition accounted for a 15e20% increase in forest net primary productivity. However, the N deposition effect on growth is expected to saturate or even decline in ecosystems with high N inputs (Brumme and Khanna, 2008). Several meth- odologies for estimation of C sequestration occurring under N additions in both above-ground biomass and soil organic matter for forests have recently been summarized by De Vries et al. (2009), who demonstrated that total carbon sequestration was in the range of 5e75 kgC kgN e1 deposition for forest and heathlands, with a most common range of 20e40 kgC kgN 1 . Modelling studies * Corresponding author. E-mail addresses: [email protected] (A. De Marco), [email protected] (A. Screpanti), [email protected] (F. Attorre), [email protected] (C. Proietti), [email protected] (M. Vitale). Contents lists available at SciVerse ScienceDirect Environmental Pollution journal homepage: www.elsevier.com/locate/envpol 0269-7491/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.envpol.2012.08.015 Environmental Pollution 172 (2013) 250e263

Upload: alessandra-de-marco

Post on 25-Nov-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

at SciVerse ScienceDirect

Environmental Pollution 172 (2013) 250e263

Contents lists available

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

Invited paper

Assessing ozone and nitrogen impact on net primary productivity witha Generalised non-Linear Model

Alessandra De Marco a, Augusto Screpanti a, Fabio Attorre b, Chiara Proietti b, Marcello Vitale b,*

a Italian National Agency for New Technologies, Energy and the Environment (ENEA), C.R. Casaccia, Via Anguillarese 301, 00123 S. Maria di Galeria, Rome, ItalybDepartment of Environmental Biology, “Sapienza” University of Rome, Piazzale Aldo Moro, 5, 00185 Rome, Italy

a r t i c l e i n f o

Article history:Received 4 April 2012Received in revised form17 August 2012Accepted 28 August 2012

Keywords:Generalized Linear/non-Linear ModelMediterranean climateOzoneNet primary productivityNitrogen deposition

* Corresponding author.E-mail addresses: alessandra.demarco@casacci

[email protected] (A. Screpanti)(F. Attorre), [email protected] (C. Proietti)(M. Vitale).

0269-7491/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.envpol.2012.08.015

a b s t r a c t

Some studies suggest that in Europe the majority of forest growth increment can be accounted for Ndeposition and very little by elevated CO2. High ozone (O3) concentrations cause reductions in carbonfixation in native plants by offsetting the effects of elevated CO2 or N deposition. The cause-effectrelationships between primary productivity (NPP) of Quercus cerris, Q. ilex and Fagus sylvatica plantspecies and climate and pollutants (O3 and N deposition) in Italy have been investigated by application ofGeneralised Linear/non-Linear regression model (GLZ model). The GLZ model highlighted: i) cumulativeO3 concentration-based indicator (AOT40F) did not significantly affect NPP; ii) a differential action ofoxidised and reduced nitrogen depositions to NPP was linked to the geographical location; iii) thespecies-specific variation of NPP caused by combination of pollutants and climatic variables could bea potentially important drive-factor for the plant species’ shift as response to the future climate change.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

European forests play a major role in carbon sequestration(Nabuurs et al., 1997) and the increases in net primary productivity(NPP) in past decades are associated with changes in forestmanagement (Lawler et al., 2010), influencing the standing growingstock (Nabuurs et al., 2001). Moreover, the increase in forestproductivity seems to be also due toother drivers such as rises inair-free CO2 concentration (Matthews, 2007), temperature (Myneniet al., 1997; Luo, 2007), and nitrogen deposition (Thomas et al.,2010). However, there is no consensus on what the main environ-mental drivers are behind forest carbon sequestration, despitehundreds of papers on this topic (Long et al., 2004; Ainsworth andLong, 2005; Hyvönen et al., 2007; Norby et al., 2010). Some studieshave suggested that in Europe the majority of forest growthincrement can be accounted for N deposition (Solberg et al., 2004;Ciais et al., 2008) and very little by elevated CO2, but this does notseem to apply in all regions. However, other studies emphasize thekey role played by high concentration of CO2 on forest growth

a.enea.it (A. De Marco),, [email protected], [email protected]

All rights reserved.

(Norby et al., 2005), although free air CO2 experiments (FACE)reveal that tree species’ responses differ.

Lavorel et al. (1998) reported that Mediterranean regions aretransitional zones where climatic changes may have the greatesteffects and where intense feedbacks from the land to the atmo-sphere are expected (Scarascia-Mugnozza et al., 2000). InMediterranean-type ecosystems and in other water-limited areas,global change factors act concurrently with additive effects ratherthan interactive ones (Heimann and Reichstein, 2008; Matesanzet al., 2008; Niu et al., 2009; Wu et al., 2011).

In order to clarify the causes of the increased forest growth theEU-project RECOGNITION recently concluded that increased avail-ability of nitrogen was the main driver for increased growth,although it could not determine whether this was due to increasednitrogen deposition or increased availability in the soil (Kahle et al.,2008). Rehfuess et al. (1999) reported that the combination of CO2

rise and elevated N deposition accounted for a 15e20% increase inforest net primary productivity. However, the N deposition effecton growth is expected to saturate or even decline in ecosystemswith high N inputs (Brumme and Khanna, 2008). Several meth-odologies for estimation of C sequestration occurring under Nadditions in both above-ground biomass and soil organic matter forforests have recently been summarized by De Vries et al. (2009),who demonstrated that total carbon sequestrationwas in the rangeof 5e75 kgC kgNe1 deposition for forest and heathlands, witha most common range of 20e40 kgC kgN�1. Modelling studies

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263 251

presented similar results where, the total C accumulation in 22European forests was within the range 41e54 kgC kgN�1 (Milneand Van Oijen, 2005; Sutton et al., 2008).

Wamelink et al. (2009) simulated NPP for the European foreststesting different scenarios of CO2, climate change and N deposi-tions, singly and in combination. Simulations of NPP performed inthe frame of an increase of CO2 to 537 ppm for the 2070, pointedout that NPP increased at all latitudes, but above a latitude of 58 Csequestration is always below 500 kg ha�1 yr�1, whereas it rangedfrom approximately 100e2000 kg ha�1 yr�1 at lower latitudes(Wamelink et al., 2009). This variation was mainly due to thedifference in tree species with latitude.

The simulated effect of climate change, under the IPCC A2climate change scenario for 2070, on NPP as compared to thereference run (1960e1990 scenario) was always positive abovelatitude 52, but below this latitude, in the Mediterranean area, theeffect could also be negative, although yearly changes in bothtemperature and precipitation lead to year-to-year changes incarbon sequestration (Wamelink et al., 2009).

With respect to N deposition in 2070, the impact was lowest inNorthern Europe where N deposition changes were smallest takingthe N deposition of 1990 as the reference.

Finally, simulation carried out by combination of CO2 increase,climate change and a decreasing N deposition scenario generallylead to an increase in C sequestration except for parts of SouthernEurope. However, C sequestration was relatively insensitive to thedecrease in nitrogen deposition compared to the increase of Csequestration at increasing of N deposition (Wamelink et al., 2009).

Another important factor affecting forest productivity is tropo-spheric ozone (O3), both singly and in combinationwith the others.Furthermore, O3 is an important air pollutant and greenhouse gasthat has been found to affect forest trees through visible injury(Schaub et al., 2010), changes in plant physiology and carbon allo-cation (Novak et al., 2007), acceleration of leaf senescence (Bussottiet al., 2011) and decreasing growth and productivity of forests(Karnosky. et al., 2007; Matyssek et al., 2010a, b), with possibleconsequences for altered carbon sequestration potentials of forestecosystems (Sitch et al., 2007; Bytnerowicz et al., 2007). Current O3levels across Europe are considered high enough to constitute a riskfor forests across the region (Ashmore, 2005; Matyssek et al., 2008).Due to its central location in the Mediterranean area, Italy may beconsidered a hot-spot for O3 and representative of O3 effects onMediterranean ecosystems (Paoletti, 2006). Environmental factorssuch as temperature, soil moisture, leaf-to-air vapour pressuredeficit and irradiance modulate O3 uptake by plants, influencingconcentrationeresponse relationships. Research on plantresponses to the uptake of O3 by plant canopies and individualplants (flux) should aid in resolving some of the uncertaintiesassociated with interacting environmental and biological factors(Matyssek et al., 2008; Gerosa et al., 2009). However, under-standing how plants cope with O3 toxicity is essential for accuratepredictions of ambient O3 impacts on vegetation. There continuesto be a critical need to obtain quantitative data on the relationshipbetween O3 exposure and response of a variety of plant speciesunder ambient and changing climatic conditions.

The impact assessment of air pollutants (such as ozone andnitrogen deposition), meteorological parameters (temperature,rainfall, frost and hot day number, solar radiation), and geograph-ical coordinates (latitude and longitude) on above-ground netprimary productivities (NPPs) of Fagus sylvatica, Quercus cerris andQuercus ilex represent the main objective of this paper. NPP datahave been estimated by the semi-empirical MOCA model (Vitaleet al., 2007, 2003), and spatially represented within Italian terri-tory. Pollutant data have been estimated by RAINS-Italy model, asdescribed in Material and Methods section. At the aim to

reconstruct the cause-effect path between NPP, environmentalclimate parameters and air pollutants, the Generalised Linear/non-Linear regression model (GLZ model) has been applied, and therobustness’ coefficient of the GLZ model has been also evaluated.

2. Material and methods

2.1. Integrated assessment modelling system

The Integrated Assessment Modelling (IAM) is used in Europe to assess thenitrogen and O3 impact on forest ecosystems and to design effective policies in termsof both abatement costs and environmental improvements (Kangas and Syri, 2002).IAMs are trans-disciplinary tools in combining the scientific research results indifferent field like atmospheric chemistry, meteorology, economy, abatementtechnologies, policies and influences on human health (Schöpp et al., 1999; Amannet al., 2011). The Regional Air Pollution Information and Simulation e RAINS-Italymodel (Vialetto et al., 2005; D’Elia et al., 2009), the national version of the RAINS-Europe model (Amann et al., 1993) and developed within a joint research projectENEA-IIASA, is an Integrated Assessment Model, that provides maps of AOT40F(defined as accumulated exposure over a threshold of 40 ppb over the growingseason (April to September), expressed in ppb h, for forest protection; LRTAP e ICP,2004) and Nitrogen deposition (oxidized and reduced state), at 20 km � 20 km ofspatial resolution. Reference year used in this analysis is 2005. The RAINS-Italymodel is part of the MINNI project, which is a national IAM based on AtmosphericModelling System (AMS), a model chain composed by a meteorological model(RAMS) (Pielke et al., 1992), an emission pre-processor (Emission Manager) anda multiphase chemical transformation model (Flexible Air Quality Regional Model)(ARIANET, 2004).

2.2. The MOCA model

The MOCA (Modelling for Carbon Assessment) model has been mainly devel-oped to estimate the above-ground total net primary production of forest stands(NPP, tC ha�1 yr�1) by using climatic and physiological parameters as driver variables(Vitale et al., 2003, 2005, 2007). MOCA is based on the big-leaf approach (Kulland Jarvis, 1995; Dai et al., 2003), which considers the canopy structure as a bigleaf functionally characterised by time-dependent Leaf Area Index (LAI,m2

leaf m�2ground). MOCA can be adapted to different species and vegetation types

and requires as input: annual mean temperature and geographical coordinates tocalculate daily irradiance values above canopy (Qo(i), mmole photons/m2/s) and airtemperature (Tair(i)). Additional parameters must be obtained by a set of parametersacquired by laboratory and field data in order to adapt the model to different species(Appendix 1). Growth period has been selected to be from 100th to 281st Julian Day(JD) for deciduous species, and the overall JDs for the evergreen species. Finally,MOCA implemented a sub-routine generating pseudo-random numbers (ranging0e1) to simulate the daily clouds coverage, whichwas on average 42% as reported bythe Italian cloudiness annals regarding the 1961e1990 time period (SCIA, 2011). Therandom assessment of daily cloudiness implied that one run of the model wasslightly different to another. For this reason, the NPP values showed here wereaverages of 10 runs of the model. The NPP values estimated by MOCA model havebeen validated using the above-ground net primary production data measuredthrough both eddy covariance flux towers of the CarboEurope Programme suchas Roccarespampani (42�230N,11�510 E, Q. cerris forest) and Collelongo(41�520N,13�380E; , F. sylvatica forest), and Castelporziano estate (41�450N,12�220E,Q. ilex forest), and NPP values derived by remote sensing studies carried out on thesame plant species formations and locations.

2.3. GLZ: Generalised Linear/non-Linear Model

When analysing data coming from physiological responses to environmentones, it does not rare to observe that the distribution of the dependent or responsevariable is (explicitly) non-normal, and does not have to be continuous, or the effectof the predictors on the dependent variable may not be linear in nature. Under theseconditions, the dependent variable values are predicted from a linear combination ofpredictor variables, which are “connected” to the dependent variable via a linkfunction. In the Generalised Linear/non-Linear Model (GLZ) a response variable Y islinearly associated with values on the X variables by Eqn. (1)

Y ¼ gðb0 þ b1X1 þ b2X2 þ :::þ bkXkÞ þ e (1)

where e is the error, and g($) is a function. Formally, the inverse function of g($), f($),is called the link function; so that:

f�my

� ¼ b0 þ b1X1þ b2X2 þ :::þ bkXk (2)

where my stands for the expected value of y. Many link functions can be chosen(McCullagh and Nelder, 1989), depending on the assumed distribution of the yvariable values. In this case we chose a log-type link function between predictorsand Y variable being much more sensitive to the variations of predictors and Y. The

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263252

values of the parameters are obtained by maximum likelihood (ML) estimation,which required an iterative computational procedure such as the NewtoneRaphsonmethod (Dobson, 1990). Tests for the significance of the effects in the model havebeen performed via the Wald statistic (Dobson, 1990). Statistical analysis has beencarried out on both the overall data set available for each species and datasetbelonging to the three geographical areas defined as North, Central and South Italy.Predictor variables were: Longitude (Long), Latitude (Lat), Number of frost and hotdays, Rainfall (mm), Solar Radiation (W me2), Minimum and Maximum air Tempera-ture (�C), AOT40F (ppb h), N-oxidised (ox-N, kg hae1 yre1) and N-reduced (red-N, kghae1 yre1). Dependent variable was annual net primary production. Predictors asAOT40F and N deposition were the RAINS-Italy outputs, whereas climatic onesderived from National Meteorological Network UCEA (http://www.cra-cma.it/). NPPvalues resulted from application of the MOCA model to geographical pointsbelonging to the potential Italian distribution areas of the plant species corrected bysuitability map (Attorre et al., 2011) for each pant species. Statistical models derivedfrom GLZ have been validated by cross-validation technique performed on theoverall data set. The cross-validation involved partitioning a dataset into a trainingset (80% of the dataset randomly sampled) and a validation set (20% of dataset) thatis used to validate the GLZ model. Multiple rounds of cross-validation (five forF. sylvatica and Q. cerris, and six for Q. ilex) have been performed to test therobustness of the predictive model. The average value (�standard deviation) of eachsignificant coefficient has been compared with the relative significant coefficientvalue belonging to themodel built with the entire dataset. If the latter fell within thestandard deviation found then the averaged coefficient was belonging to the overallpopulation and the robustness of the model was verified. All analyses were made byStatistica 8.0 (Stat Soft Inc., USA).

3. Results and discussion

3.1. Ozone and nitrogen deposition

The AOT40F values from April to September were well above theestablished threshold of 5000 ppb h throughout the Italian territory(Fig. 1A). The nitrogen depositions ranged between 140 and6000 mg m�2 yr�1 (Fig. 1B) and were generally higher in theNorthern Italy, where the urea consumption reached the 70% of thetotal used in Italy (ISTAT, 2005). The highest consumption offertilizers and livestock concentration explained the greater N-reduced component percentage (Fig. 2A) than to N-oxidized one(Fig. 2B) in that area with respect to other parts of Italy. Because N

Fig. 1. Italian distribution of AOT40F (ppb h) (A) and tot

deposition and O3 have a common precursor (NOx), the relation-ship between nitrogen deposition (red-N and ox-N) and AOT40F hasbeen investigated for broadleaves, conifers, sclerophyll and mixedforests. The sclerophyll ecosystem showed a significant relationshipwith total nitrogen deposition (R2 ¼ 0.61) and ox-N deposition(R2 ¼ 0.64), whereas mixed ecosystem showed the weaker ones(R2 ¼ 0.43 for total N deposition and R2 ¼ 0.49 for ox-N deposition).

3.2. Modelled annual net primary production

Themodelled NPP distribution follows the potential distributionof plant species assessed by statistical approach based on randomforest modelling (Attorre et al., 2011) and corrected by suitabilitymap of the species derived by Corine Land Cover 4th level (CLC,2000). Fig. 3 (AeC) highlights the NPP distributions in a grid at20 � 20 km of resolution for the three plant species analysed here.The NPP of F. sylvatica ranges from 18 to 4750 tC 20 km�2 yr�1

whereas Q. cerris and Q. ilex show values ranging between 27 and3240 tC 20 km�2 yr�1, and 18 and 3204 tC 20 km�2 yr�1, respec-tively. High NPP values are observed in areas fully suitable for thespecies: Apennine for F. sylvatica, large part of central Italy forQ. cerris and Italian coasts and south-facing limestone slopes of thehinterland forQ. ilex. Table 1 representsmodelledNPP rangeswhichare compared with NPP calculated by remote sensing approach orhybrid methodologies between modelling and remote sensing.Modelled NPP values are in agreement to the calculated ones, eitherfor averages calculated for the overall Italian territory or for specificlocations where eddy covariance flux towers were placed.

3.3. Cause-effect assessment performed by Generalised non-LinearModels

The coefficient estimations for predictors performed by GLZ arereported in Appendices 2e4. The Wald statistic defined the statis-tical estimation of significant coefficients.

al nitrogen deposition (mg m�2 yr�1) (B) for 2005.

Fig. 2. Italian distribution of reduced (A) and oxidized (B) nitrogen deposition defined as percent of the total for 2005.

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263 253

3.3.1. Fagus sylvaticaThe estimation of coefficients derived from the GLZ application

to the overall dataset highlighted the linkage between predictorsand NPP. NPP of Fagus sylvatica is negatively and significantlyrelated to: maximum temperature (Tmax), non-linear increases ofFrost days number and Hot Days number. These relationships pointout a high sensitivity of F. sylvatica to hot environmental condi-tions, in agreement with Gutiérrez (1988) who reported a strongnegative impact of high temperatures, and consequently of limitedwater availability in the soil, on the growth of F. sylvatica. However,when predictors were individually analysed for Northern, Centraland Southern Italy they had different relationships with the NPP.As an example, among climatic predictors solar radiation and Tmaxplayed an important role to reduce NPP values for beech pop-ulation growing in the Central and Southern Italy, but they did notshow any significant relationship in the Northern Italy (Appendix2). AOT40F and its combination with N-reduced (red-N) weaklyaffected the NPP of Italian beech population. When the GLZ wereperformed for geographical partitions, air pollutants had a relevantrole for affecting the NPP. The AOT40F had only a limited reducingeffect on NPP in Central and Southern Italy. Ox-N and red-Nchanged in values from north to south Italy in a complementarymanner, acting as positive or negative drivers in affecting NPP(Appendix 2). In Southern Italy red-N was the most negativelyinfluencing predictor for productivity. red-N, although in smallerquantities than the rest of Italy, exerted its damaging effects morestrongly than Central and Northern Italy, perhaps because itsaction had much more impact in a not optimal climatic environ-ment for beech. In fact, the susceptibility of F. sylvatica to theextreme temperature increases with decreasing water availabilityin the soil (Lebourgeois et al., 2005), that occurs in Southern Italy.Ox-N had a negative impact on productivity in southern Italy, but ithad not impact when the overall data set were considered; itappeared that its fertilisation effect was greatly reduced by

Mediterranean climate (Wamelink et al., 2009). Validation of GLZmodel performed by cross-validation technique is shown inFig. 4A; it highlighted a correlation coefficient of 0.745 betweenobserved NPP dataset (testing set) and predicted one, which wasa good result, taking in account the non-linear nature of the NPPpredicting model.

3.3.2. Quercus cerrisGLZ modelling performed on the overall dataset of Quercus

cerris showed a negative role on the final NPP values of thecombination between number of frost days and Tmin, and thenumber of hot days and Tmax, other than the number of frost days 2

and Tmin2, all altering the phenological period of Q. cerris(Appendix 3). It was confirmed the important role played byclimatic predictors in combination with pollutants (Tmin � N-red;number of hot days � N-ox; number of frost day � N-ox) allreducing the NPP values. It is well known that a rapid increase ofminimum temperature can induce stress in Q. cerris, due to itsinability to respond quickly to environmental changes (Maneset al., 2006), and the presence of red-N in air reduces resistanceto water stress (Krupa, 2003), leading thus to a decrease inproductivity. The AOT40F exerted a significant and negative effecton NPP also, although when it acted in combination with theother predictors, it showed a weak effect on NPP. Interestingly, thenegative action of red-N in northern Italy was reversed when itwas in combination with ox-N in central Italy. Moreover, red-Npollution had a local action, because when it was combined to thelatitude (Lat) a negative effect on NPP was observed in Northernand Southern Italy, but not in the central Italy data set where thiscombination of predictors was positive on NPP. High values ofmodelled NPP in central Italy could be due to non-limiting envi-ronmental conditions (such as water availability), explaining thusthe positive effect of N-red observed in this area on the NPP ofQ. cerris. AOT40F was a weak significant predictor for NPP,

Fig. 3. Italian distribution of net primary production (NPP, tC 20 kme2 yre1) for Fagus sylvatica (A), Quercus cerris (B) and Quercus ilex (C).

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263254

although it occurred in many combinations with other predictors(Appendix 3). However, red-N combined with Tmax had a positiveeffect on NPP, probably because Tmax strongly affected NPP ina positive way. When the dataset was divided in Northern, Central

Table 1Modelled and measured NPP values. Asterisk denotes the NPP average value calcu-lated for the 2000e2005 time period from data measured by eddy covariancemethodology. MOCA run for five times for each year (2000e2005) and the NPPaverage has been calculated.

Species Location NPPmodelled

(gC m�2 yr�1)NPP(gC m�2 yr�1)

Source

F. sylvatica Collelongo 442e510;426

Maselli et al.,2009; Chiriciet al., 2007Q. cerris Roccarespampani;

San Rossore372e496;428

Q. ilex Castelporziano 396; 496F. sylvatica Collelongo 1047* 917* G. Matteucci,

personalcommunication

Q. cerris Roccarespampani 731* 868*Q. ilex Castelporziano 515* 430*F. sylvatica Italy (averages) 200e520 442 Chirici et al.,

2007Q. cerris 370e620 427Q. ilex 100e400 497

and Southern parts, the important role played by geographicalvariables in combination with pollutants appeared evident(Appendix 3). Moreover, it should be emphasized the differentactions carried out by ox-N and red-N in the Northern Italy can bereversed in the Southern Italy. This suggests that nitrogen actslocally. Erisman and De Vries (2000) reported that an increase ofN deposition caused a significant intensification in productivity,but above a certain threshold was expected a decrease of NPP. It isinteresting to note that ox-N and rainfall in combination madea positive effect on NPP (in the overall data set), whereas con-trasting effects on the NPP in Central Italy (negative effect) or inNorthern Italy (no-effect) were evident. Probably, multiple non-linear and indirect interactions with other climatic predictorsdetermined these results, showing thus an important aspect inthe cause-effect mechanisms, which were at the basis of theItalian NPP distribution of Q. cerris. The action of red-N noticeablypromoted NPP in Southern Italy. This could be because the red-N,being present in limited quantities (1.43 � 0.71 kg ha�1 yr�1),could not exert its harmful action. Ox-N instead was present inhigher amounts (5.04 � 2.12 kg ha�1 yr�1), exceeding thethreshold for this ecosystem and reducing NPP (Erisman and De

Fig. 4. Cross e Validation carried out on the validation data set of each plant species;Fagus sylvatica (A), Quercus cerris (B) and Quercus ilex (C).

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263 255

Vries, 2000). Globally, rainfall played an important and affectingrole in the NPP of Q. cerris. The latter is a water spender speciesand, as a consequence, it uses all water availability in the soil forsupporting high gas exchange rates but decreasing them once thewater supply is gradually reduced. Validation of GLZ model

performed by cross-validation technique is shown in Fig. 4B;a correlation coefficient of 0.781 between observed NPP dataset(testing set) and predicted one is found.

3.3.3. Quercus ilexTmax was not limiting for NPP of Q. ilex in Italy, although the

Tmax2 negatively affected it. Tmin differently affected NPP valueswith respect to different geographical locations (negative effect inNorth Italy, but positive ones in Central and Southern Italy).Interestingly, Tmin2 changed the above reported relationships(Appendix 4). Globally, the relationships among environmentalpredictors and NPP magnified the Mediterranean ecological char-acteristics of Q. ilex, showing thus interesting physiologicalresponses linked with air pollutants and climate. The minimumtemperature seems to have an important role in the C sequestra-tion performance of Q. ilex, representing a linkage with localenvironmental conditions for modelling gas exchanges. red-N hada positive effect on NPP in Italy, whereas ox-N showed a negativeone. Different behaviours of these two predictors were observedfor the three geographical locations, where red-N2 had significantand positive effect in Central and Southern Italy but not in theNorth (Appendix 4). ox-N showed a negative effect in North andCentral Italy, but ox-N2 reverted in a positive action on the NPP ofQ. ilex population of Central Italy, due to the “fertilising” effect ofox-N and to non-limiting environmental conditions there. AOT40Fhad a positive effect on NPP of Q. ilex, but looking at thegeographical distribution significant different effects on produc-tivity were found in the Northern and Southern Italy (negativeeffect) respect to Central Italy (positive effects). The cross-validation technique highlighted a very good correspondencebetween the observed and predicted NPP values (correlationcoefficient of 0.833), (Fig. 4C).

3.4. Evaluation of the GLZ model robustness

Multiple runs of the GLZ model have been carried out for theseplant species to demonstrate the coefficients’ robustness of theGLZ models found. The coefficient values of the predictorsbelonging to GLZ model built with the entire data set fell insidethe statistical variations (standard deviations) of the averagedones, except for Tmax � red-N of Q. cerris and some predictors ofF. sylvatica (see marked predictors with asterisk in the Appen-dices). However, by considering the non-linear nature of the GLZmodels, the performance of cross-validations, and the low reso-lution of spatial data, the model’s robustness for Italy could bepositively accepted.

3.5. General considerations and conclusion

Cause-effect relationships between pollutants and primaryproduction can be difficult to find in the field because a multitudeof interactions concurrently could act on the response variable. Forthis reason, non-linear statistical techniques are becoming ofincreasing interest (Ball et al., 2000; Manes et al., 2005). In thiscontext, the data analysis performed by Generalized Linear/non-Linear Models well described the non-linear relationshipsbetween NPP, meteorological, geographical and pollutantspredictors. The GLZ analyses did not clearly establish a negativeeffect of AOT40F on the NPP, although it is more significantlyrelated to the combination of AOT40F with the other predictors.Higher significant relationships were instead found with nitrogendepositions and climatic parameters singly or in combination. Thishighlights the synergic or antagonistic roles of some limitingfactors (such as high temperature, high ozone concentration) inaffecting physiological processes (gas exchange, stomatal

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263256

conductance). Further, local variations of ox-N and red-N affectedNPP in a complementary manner, acting as positive or negativedrivers. Moreover, combination of pollutant and climatic predic-tors established much more information for predicting NPP indifferent parts of Italy than single predictor. Paoletti and Manning(2007) suggest an effect-based approach for evaluating the impactof pollutant predictors on the NPP. However, although it is relevantto estimate the effective dose of pollutant up-taken by plants(Karlsson et al., 2007; Matyssek et al., 2007), higher ozoneconcentrations occur during the hottest months of the year, whenMediterranean tree species (especially Quercus ilex) close stomatafor avoiding water losses (Ferretti et al., 2007). These physiologicalmechanisms enduring summer conditions represent key-factorsfor the cause-effect modelling of pollutants and primary produc-tivity. N deposition is predicted to alter N and C storage in livingand dead biomass of Mediterranean ecosystems (Vourlitis et al.,2009) and, because water is another limiting factor for plantsgrowing in Mediterranean ecosystems, the growth vs. storagestrategy of evergreen and deciduous shrub species suffering ofchronic N deposition is predicted to be modulated by wateravailability (Sanz-Pérez et al., 2007). Finally, it should be taken inconsideration that red-N is often more toxic than ox-N for plants,which is associated with changes in competition for resources(Dias et al., 2011).

Many considerations made above confirm the foreseen shiftsof plant species as resulting by dynamic space-time integration ofeffects due to climate change (Di Traglia et al., 2011), nutrientresources (Sanz-Pérez et al., 2007), inter-specific competition(Emmett, 2007) and air pollutants (Langley and Megonigal, 2010)in the Mediterranean area. How to preserve old-growth forests inthis dynamic situation remains an outstanding challenge (Bauhuset al., 2009). However, climate change may provide an opportu-nity to increase the relevance of protected areas for conservation,keeping track of biodiversity hotspots and assistingalso the migration of conservation activities (Hole et al., 2009;Hannah et al., 2007; Pressey et al., 2007). However, a long-termperspective is needed also for recommendations for landscapeplanning.

Acknowledgements

This research has been supported by the ATENEO Sapienza 2009grants and by Italian Ministry for the Environment, the Land andthe Sea (MATTM).

Name Symbol Value/species Units

Dark respiration Rd 1.03 F. sylvatica1.03 Q. cerris0.30 Q. ilex

mmolCO2$m�2s�1

Leaf area index(initial value)

LAI 0.1 F. sylvatica0.1 Q. cerris2.2 Q. ilex

m2leaf$m�2

ground

Maximum photosynthesisrate at light saturation

Amax 20 F. sylvatica21 Q. cerris15 Q. ilex

mmolCO2$ m�2s�1

Quantum yield efficiencyof photosynthesis

QY 0.055 F. sylvatica0.036 Q. cerris0.044 Q. ilex

mmolCO2$mmolphotons�1

Appendix 1

Mathematical equations used in the MOCA model.Daily net photosynthetic rate (A(i), mmolCO2 m�2s�1) is calcu-

lated as a response to the mean light radiation inside the canopy atthe ith day of year (Qi(i), mmole photons m�2s�1), according to DeWit et al. (1978):

AðiÞ ¼ ðAmax � RdÞ�1� e

�QYQiðiÞAmax

��þ Rd (A1)

where Amax is the maximum rate of net photosynthesis at lightsaturation (mmolCO2$ m�2s�1), Rd is the dark respiration rate(mmolCO2$ m�2s�1) integrated for the darkness period, and QY isthe quantum yield efficiency of photosynthesis (mmolCO2$mmolephotons�1).

Net primary productivity (gC$ m�2 yr�1) is calculated by anintegration with time (from 100th to 281st Julian Day (JD) for

deciduous species, and the overall JDs for the evergreen species) ofA(i) as:

NPP ¼Xi

½AðiÞ � KcðiÞ��12$10�6

�(A2)

where Kc(i) is the maintenance respiration calculated as a functionof leaf temperature by the Q10 temperature coefficient (that isameasure of the rate of change of a biological or chemical system asa consequence of increasing the temperature by 10 �C):

KcðiÞ ¼ R10$Q10

��Tleaf ðiÞe10

10

�(A3)

where R10 is the respiration rate at 10 �C (1.5 mmolCO2$m�2s�1) andQ10 is 1.7, which was in the range found by Granier et al. (2000) forthe beech (1.6 � Q10 � 1.8), and applied also for Quercus cerris. Q10was set to 2.2 for Q. ilex.

Light attenuation through the canopy Qi(i) (mmolphotons$ m�2s�1) is assumed to be depending on the LAI(i)according to the BeereLambert law:

QiðiÞ ¼ Q0ðiÞ$e�ðK$LAIðiÞÞ (A4)

where K (¼0.72) is the coefficient of light extinction through thecanopy. The daily LAI (m2

leaf$ m�2ground) is modelled by the

following equation:

LAIðiÞ ¼ SLAðiÞ$b$Aði� 1Þ (A5)

where A(i-1) is the leaf net productivity (gDry Weight$ m�2) ofprevious (ith-1) day of the year, b is the leaf partition coefficient ofthe accumulated biomass (¼0.25), and SLA(i) is the specific leaf areaat the ith day of the year (m2$ g�1

Dry Weight) defined by a saturationgrowth equation fitting empirical data derived from measurementcampaigns in beech and mixed forests of the Apennine mountains:

SLAðiÞ ¼� ðc$timeÞðdþ timeÞ

(A6)

where c and d are fitted parameters (c¼ 0.0116 and d¼�52.563 forQ. cerris; c¼ 0.0059 and d¼�95.250 for F. sylvatica). The SLA is keptconstant during the year (0.005m2/gDry Weight) only for Quercus ilex.

Values reported above are derived from field and laboratorymeasurements performed in F. sylvatica and Q. cerris forestsgrowing in the Italian Central Apennine e Abruzzi and MoliseRegions, whereas values concerning Q. ilex derived from Maneset al. (1997), 2007; Faria et al., 1998.

Appendix 2. Fagus sylvatica. Parameter estimates for predictors. Distribution: Normal; Link function: Log; Dependent parameter: Net Primary Production (NPP). Runsnumber N [ 5

Predictors Italy Averagedparameter coeff.

� S.D. North Italy Central Italy South Italy

Parametercoeff.

� S.E. p Predictorcoeff.

� S.E. p Predictorcoeff.

� S.E. p Predictorcoeff.

� S.E. p

AOT40 �4.86E-03 2.40E-03 0.043 �4.18E-03 2.45E-03 �4.33E-02 1.28E-02 0.0007 �4.24E-02 6.83E-03 0.000AOT40*ox-N 4.07E-05 1.23E-05 0.001AOT40*red-N �2.37E-05 1.12E-05 0.034 �2.69E-05 1.26E-05 �3.32E-05 1.12E-05 0.003 �9.60E-04 3.02E-04 0.0015AOT40

ˇ

2 �1.64E-08 5.99E-09 0.006Hot day num �1.61Eþ02 6.78Eþ01 0.018Hot day num*AOT40 7.71E-05 2.29E-05 0.001 6.81E-05 3.70E-05 3.94E-04 1.20E-04 0.001 3.06E-04 8.09E-05 0.000Hot day num*frost day num 2.88E-01 1.46E-01 0.048 2.83Eþ00 5.89E-01 0.000Hot day num*ox-N 4.24E-01 1.72E-01 0.014Hot day num*rainfall 1.73E-02 6.81E-03 0.011 5.84E-02 9.21E-03 0.000Hot day num*solar radiation �4.07E-03 1.47E-03 0.006 5.37E-03 2.05E-03 0.0089Hot day num*red-N �6.61E-01 1.77E-01 0.000 3.80Eþ00 1.82Eþ00 0.0363Hot day num*Tmax 2.74Eþ00 9.46E-01 0.004Hot day num*Tmin 1.01Eþ01 2.78Eþ00 0.000Hot day num

ˇ

2 �1.70E-01 8.35E-02 0.041 �2.04E-01 1.30E-01 �3.02Eþ00 6.90E-01 0.000 �3.98Eþ00 8.11E-01 0.0000 3.07Eþ00 4.69E-01 0.000Frost day num �1.77Eþ02 2.13Eþ01 0.000Frost day num*AOT40 1.87E-05 5.44E-06 0.001 1.98E-05 5.84E-06 5.90E-05 1.76E-05 0.001 2.02E-04 9.49E-05 0.0335 �1.39E-04 5.83E-05 0.017Frost day num*rainfall 3.64E-02 5.57E-03 0.000Frost day num*solar radiation 1.17E-03 5.52E-04 0.034Frost day num*red-N 4.23E-02 1.08E-02 0.000 4.15E-02 4.86E-03Frost day num*Tmin* �2.06E-01 1.04E-01 0.047 �1.07E-01 1.36E-01Frost day num

ˇ2 �1.39E-02 5.26E-03 0.008 �1.16E-02 5.65E-03 �7.59E-02 2.55E-02 0.0029 �2.78E-01 6.78E-02 0.000

ox-N 1.29Eþ02 3.57Eþ01 0.0003 �8.44Eþ01 3.35Eþ01 0.012ox-N

ˇ

2 �3.87E-02 7.81E-03 0.000 �1.65Eþ00 3.05E-01 0.000Rainfall �1.19Eþ00 3.95E-01 0.003Rainfall*AOT40 3.36E-06 9.06E-07 0.000 8.20E-06 2.56E-06 0.001Rainfall*ox-N 3.33E-03 1.11E-03 0.003 �3.81E-02 1.02E-02 0.0002 7.29E-02 1.03E-02 0.000Rainfall*TmaxRainfall*Tmin* �1.29E-02 4.06E-03 0.002 �1.07E-02 3.13E-03Rainfall*red-N �2.83E-03 9.16E-04 0.002 6.20E-02 1.39E-02 0.0000 �8.75E-02 2.08E-02 0.000Rainfall

ˇ

2 1.78E-04 4.77E-05 0.000 �3.27E-04 1.35E-04 0.0158 �4.21E-04 8.21E-05 0.000Solar radiation �2.88E-02 8.04E-03 0.0003 �5.78E-01 1.09E-01 0.000Solar Radiation*AOT40Solar radiation*red-N 6.69E-04 3.31E-04 0.043 2.37E-02 4.56E-03 0.000Solar radiation*TminSolar radiation

ˇ

2 1.70E-05 3.62E-06 0.000red-N �1.57Eþ01 6.61Eþ00 0.017 �1.68Eþ02 7.07Eþ01 0.0175 �5.75Eþ02 1.73Eþ02 0.001red-N*ox-N 4.56E-02 1.33E-02 0.001 3.24Eþ00 8.87E-01 0.000Tmax* �7.06Eþ01 1.87Eþ01 0.000 �6.04Eþ01 7.97Eþ00 �4.66Eþ00 2.08Eþ00 0.0247 �2.86Eþ01 7.86Eþ00 0.000Tmax*red-NTmax

ˇ

2 1.08Eþ00 3.23E-01 0.001 9.03E-01 3.12E-01Tmin �3.30Eþ02 4.24Eþ01 0.000Tmin*AOT40 1.95E-04 4.39E-05 0.000 1.92E-04 3.75E-05 1.04E-03 3.67E-04 0.005 2.23E-03 1.10E-03 0.0430Tmin*ox-N* �2.36E-01 1.15E-01 0.040 �1.33E-01 9.32E-02 1.07Eþ01 2.36Eþ00 0.000Tmin*red-N 3.74E-01 1.41E-01 0.008 2.22E-01 4.15E-02Tmin

ˇ

2 7.92Eþ00 2.72Eþ00 0.0036 7.06Eþ00 1.22Eþ00 0.000long �6.79Eþ01 1.64Eþ01 0.0000 8.17Eþ01 1.64Eþ01 0.000long*AOT40 �4.58E-05 1.87E-05 0.014 4.19E-04 1.22E-04 0.0006 9.37E-05 3.63E-05 0.010long*hot day num �7.17E-01 2.67E-01 0.007

(continued on next page)

A.D

eMarco

etal./

EnvironmentalPollution

172(2013)

250e263

257

(continued )Appendix 3. (continued )

Predictors Italy Averagedparameter coeff.

� S.D. North Italy Central Italy South Italy

Parametercoeff.

� S.E. p Predictorcoeff.

� S.E. p Predictorcoeff.

� S.E. p Predictorcoeff.

� S.E. p

long*solar radiationlong*ox-N �1.05E-01 3.63E-02 0.004 �1.32Eþ00 3.32E-01 0.0001 �1.20Eþ00 1.83E-01 0.000long*Rainfall �5.71E-03 2.49E-03 0.022long*red-N 1.10E-01 4.38E-02 0.012 1.85Eþ00 4.62E-01 0.0001 3.21Eþ00 5.33E-01 0.000long*Lat 9.86E-02 4.96E-02 0.047 6.58E-01 2.91E-01 0.0238long*Tminlong

ˇ

2 �1.98E-02 9.53E-03 0.038 �2.78E-02 1.07E-02 5.14E-01 1.15E-01 0.0000 �7.99E-01 1.56E-01 0.000Lat 1.49Eþ02 2.09Eþ01 0.000Lat*AOT40 2.28E-04 1.13E-04 0.0433 1.89E-04 8.70E-05 0.030Lat*Hot daLat num �1.96Eþ00 4.55E-01 0.000 1.88Eþ00 6.71E-01 0.0052Lat*Frost daLat num 3.36Eþ00 3.60E-01 0.000Lat*ox-N �1.15Eþ00 4.34E-01 0.0081 2.47Eþ00 4.40E-01 0.000Lat*red-N 3.33Eþ00 1.68Eþ00 0.047Lat*TmaxLat

ˇ

2 �9.63E-02 4.42E-02 0.029 �3.83Eþ00 4.43E-01 0.000Intercept 4.62Eþ02 4.26Eþ02 0.028 2.89Eþ03 9.28Eþ02 0.0019 6.35Eþ03 1.09Eþ03 0.000

Appendix 3. Quercus cerris. Parameter estimates for predictors. Distribution: Normal; Link function: Log; Dependent parameter: Net Primary Production (NPP). Runsnumber N [ 5

Predictors Italy Averaged parametercoeff.

� S.D. North Italy Central Italy South Italy

Parameter coeff. � S.E. p Parameter coeff. � S.E. p Parameter coeff. � S.E. p Parameter coeff. � S.E. p

AOT40 �9.67E-03 2.18E-03 0.000 �1.14E-02 2.75E-03 2.09E-02 4.02E-03 0.000AOT40*ox-N �1.63E-05 7.80E-06 0.036 �1.41E-05 1.21E-05 1.49E-04 1.95E-05 0.000 �3.25E-05 1.62E-05 0.045AOT40*red-NAOT40

ˇ

2 �1.46E-08 7.11E-09 0.040 5.38E-09 2.51E-09 0.032Hot day num 1.55Eþ01 7.94Eþ00 0.050 9.72Eþ00 1.34Eþ01Hot day num*AOT40 �2.27E-05 1.10E-05 0.039 �2.91E-05 1.60E-05 �3.52E-04 1.25E-04 0.005Hot day num*frost day num �5.00E-01 1.60E-01 0.002 6.42E-01 9.06E-02 0.000Hot day num*ox-N �1.48E-01 3.14E-02 0.000 �1.20E-01 6.10E-02 1.20Eþ00 2.14E-01 0.000 �2.98E-01 1.46E-01 0.042 2.21E-01 8.24E-02 0.007Hot day num*rainfall 4.32E-03 8.03E-04 0.000 2.91E-03 1.78E-03 4.03E-03 1.69E-03 0.017Hot day num*solar radiation 4.61E-03 1.22E-03 0.000 2.36E-03 1.10E-03 0.032Hot day num*red-N 1.58E-01 5.70E-02 0.006 1.40E-01 2.88E-02 1.29E-02 3.35E-01 0.000Hot day num*Tmax �8.00E-01 2.74E-01 0.004 �7.96E-01 5.30E-01 2.37E-01Hot day num*Tmin 5.30Eþ00 8.78E-01 0.000Hot day num

ˇ

2 1.25E-01 5.32E-02 0.019 8.84E-02 5.21E-02 1.89Eþ00 9.50E-01 0.047 2.88E-01 1.30E-01 0.027Frost day num 5.30Eþ00 2.61Eþ00 0.042 5.65Eþ00 6.25Eþ00 2.01Eþ01 1.95Eþ00 0.000 �2.96Eþ01 8.90Eþ00 0.001Frost day num*AOT40 �4.61E-05 2.26E-05 0.041Frost day num*ox-N �2.35E-02 1.13E-02 0.038 �3.72E-03 2.74E-02 �3.12E-01 5.75E-02 0.000Frost day num*rainfall �8.42E-04 2.84E-04 0.003 �5.36E-04 5.86E-04 �5.56E-03 1.62E-03 0.001 1.24E-02 1.47E-03 0.000Frost day num*red-N 2.43E-01 7.99E-02 0.002 �6.30E-01 1.84E-01 0.001Frost day num*Tmin �3.27E-01 7.76E-02 0.000 �2.81E-01 1.61E-01

A.D

eMarco

etal./

Environmental

Pollution172

(2013)250

e263

258

Frost day num

ˇ

2 �1.16E-02 4.89E-03 0.017 �1.28E-02 4.18E-03 1.30E-01 1.70E-02 0.000 �4.33E-02 1.28E-02 0.001 �8.90E-02 4.03E-02 0.027ox-N 3.59Eþ01 6.23Eþ00 0.000 �8.49Eþ01 1.19Eþ01 0.000ox-N

ˇ

2 �7.62E-02 1.86E-02 0.000 �5.20E-02 3.35E-02 �3.16E-02 1.27E-02 0.013Rainfall 5.73E-01 1.30E-01 0.000 4.83E-01 4.15E-01Rainfall*AOT40 �2.36E-06 1.16E-06 0.042 �1.35E-06 2.87E-07 0.000Rainfall*ox-N 1.82E-03 6.15E-04 0.003 6.54E-04 1.76E-03 �4.66E-03 2.26E-03 0.040 4.11E-03 1.02E-03 0.000Rainfall*red-N 1.59E-02 4.66E-03 0.001 �1.36E-02 2.75E-03 0.000Rainfall*solar radiation �2.37E-05 4.87E-06 0.000 �2.02E-05 1.19E-05 5.30E-06Rainfall*Tmax �1.10E-02 2.84E-03 0.000 �9.48E-03 6.78E-03 3.03E-03Rainfall*Tmin 2.99E-03 7.42E-02 1.08E-02 0.000Rainfall

ˇ

2 3.12E-05 8.35E-06 0.000 2.56E-05 1.34E-05 5.98E-06 5.80E-05 0.000 5.85E-05 1.58E-05 0.000Solar radiation �2.65E-02 5.20E-03 0.000 1.28E-01 5.53E-02 0.021Solar radiation*AOT40 3.33E-07 6.08E-08 0.000 3.09E-07 6.88E-08 3.35E-07 1.39E-07 0.016Solar radiation*Tmin �2.63E-03 9.47E-04 0.006 �1.65E-03 2.67E-03Solar radiation*ox-N 1.91E-03 4.24E-04 0.000Solar radiation

ˇ

2 �3.16E-06 1.50E-06 0.035red-N �6.99Eþ01 1.66Eþ01 0.000 2.32Eþ02 3.11Eþ01 0.000red-N*ox-N 8.52E-01 3.23E-01 0.008red-N

ˇ

2 �9.01E-02 4.48E-02 0.045 �8.62E-01 3.82E-01 0.024 9.11E-01 3.17E-01 0.004Tmax 1.17Eþ02 3.06Eþ01 0.000 9.76Eþ01 4.67Eþ01 7.88Eþ01 2.91Eþ00 0.000 2.07Eþ01 2.61Eþ00 0.000Tmax*red-N* 8.90E-01 2.83E-01 0.002 1.22Eþ00 2.39E-01 2.73Eþ00 2.70E-01 0.000Tmax

ˇ

2 �2.37Eþ00 5.62E-01 0.000 �2.08Eþ00 9.69E-01Tmin 4.51Eþ02 4.95Eþ01 0.000Tmin*AOT40 1.02E-04 3.53E-05 0.004 1.09E-04 6.16E-05 �1.76E-03 4.28E-04 0.000Tmin*ox-N 4.29E-01 1.21E-01 0.000 5.28E-01 2.62E-01 1.17E-01 8.75E-01 0.000 2.84Eþ00 5.37E-01 0.000Tmin*red-N �1.03Eþ00 2.45E-01 0.000 �1.11Eþ00 2.27E-01 1.02E-01 �7.13Eþ00 2.09Eþ00 0.001 �4.25Eþ00 1.52Eþ00 0.005Tmin

ˇ

2 �1.72Eþ00 4.76E-01 0.000 �1.36Eþ00 1.59Eþ00 7.10E-01 3.31Eþ00 0.000 �8.42Eþ00 1.55Eþ00 0.000Long �1.12Eþ01 3.75Eþ00 0.003 �1.42Eþ01 1.10Eþ01 4.93Eþ00 1.41Eþ01 6.26Eþ00 0.024 7.95Eþ00 4.07Eþ00 0.050Long*AOT40 4.18E-05 5.90E-06 0.000 4.09E-05 6.28E-06 2.81E-06 2.98E-05 1.49E-05 0.045Long*frost day num 1.70E-01 4.20E-02 0.000Long*ox-N 9.97E-02 1.74E-02 0.000 7.95E-02 4.13E-02 �4.55E-01 7.77E-02 0.000 2.76E-01 4.84E-02 0.000Long*rainfall �7.63E-04 3.23E-04 0.018 �4.49E-04 1.13E-03 �1.61E-02 3.63E-03 0.000 �2.47E-03 1.12E-03 0.027Long*solar radiation �6.31E-04 2.77E-04 0.023Long*red-N �2.52E-01 3.61E-02 0.000 �1.89E-01 6.82E-02 5.61E-01 1.29E-01 0.000Long*Lat �1.31E-01 5.74E-02 0.023 �4.08E-01 5.87E-02 0.000Long*Tmax 5.27E-01 9.09E-02 0.000 5.46E-01 2.12E-01 9.48E-02Long

ˇ

2 �2.57E-02 8.05E-03 0.001 �2.63E-02 1.44E-02 6.46E-03 6.65E-02 0.038 �3.75E-02 1.90E-02 0.048Lat 5.57Eþ01 7.10Eþ00 0.000 4.92Eþ01 6.95Eþ00 0.000Lat*AOT40 5.47E-05 1.03E-05 0.000 6.33E-05 1.14E-05 �1.22E-04 4.91E-05 0.013Lat*frost day num �5.29E-01 4.54E-02 0.000 1.90E-01 6.36E-02 0.003 3.35E-01 7.38E-02 0.000Lat*ox-N 1.60E-01 3.07E-02 0.000 9.48E-02 1.07E-01 8.87E-01 1.06E-01 0.000 8.35E-01 1.07E-01 0.000Lat*solar radiation �3.04E-04 1.12E-04 0.006 2.33E-05 6.51E-04 �9.59E-04 3.26E-04 0.003Lat*red-N �3.22E-01 5.48E-02 0.000 �1.82E-01 1.51E-01 �5.75E-01 1.51E-01 0.000 8.24E-01 2.82E-01 0.003 �1.73Eþ00 3.14E-01 0.000Lat*Tmin 1.82Eþ00 5.97E-01 0.002Lat

ˇ

2 �1.75E-02 8.20E-03 0.033 �1.09E-02 3.38E-02 �1.78E-01 5.55E-02 0.001 �4.18E-01 6.15E-02 0.000Intercept �1.75Eþ03 4.62Eþ02 0.000 �8.70Eþ02 1.78Eþ03 �4.50Eþ03 3.39Eþ02 0.000 �1.47Eþ03 6.89Eþ02 0.033

A.D

eMarco

etal./

EnvironmentalPollution

172(2013)

250e263

259

Appendix 4. Quercus ilex. Parameter estimates for predictors. Distribution: Normal; Link function: Log; Dependent parameter: Net Primary Production (NPP). Runsnumber N [ 6

Predictors Italy Averagedparameter coeff.

� S.D. North Italy Central Italy South Italy

Parameter coeff. S.E. p Parameter coeff. S.E. p Parameter coeff. S.E. p Parameter coeff. S.E. p

AOT40 5.95E-03 1.59E-03 0.000 6.36E-03 2.87E-03 �1.28E-02 3.32E-03 0.000 2.44E-02 1.60E-03 0.000 �1.12E-02 4.22E-03 0.008AOT40*ox-N �6.70E-05 1.31E-05 0.000AOT40*red-N 6.01E-05 1.64E-05 0.000AOT40

ˇ

2 4.60E-08 5.53E-09 0.000 �9.86E-09 1.74E-09 0.000Hot day num �2.43Eþ02 6.29Eþ01 0.000 �2.01Eþ02 2.82Eþ01 0.000 �4.52Eþ01 2.05Eþ01 0.027Hot day num*AOT40Hot day num*frost day num 3.47E-01 1.09E-01 0.002 �5.50E-01 1.78E-01 0.002 5.61E-01 1.21E-01 0.000Hot day num*ox-N �1.56Eþ00 2.24E-01 0.000 �3.88E-01 1.62E-01 0.017Hot day num*rainfall �8.52E-03 1.63E-03 0.000 �6.24E-03 4.77E-03 2.94E-02 7.99E-03 0.000Hot day num*solar radiation 9.75E-03 3.55E-03 0.006 1.81E-03 5.04E-04 0.000Hot day num*red-N 3.83E-01 1.58E-01 0.015 4.88E-01 2.16E-01 1.05Eþ00 3.45E-01 0.002 1.39Eþ00 5.65E-01 0.014Hot day num*Tmin 6.18Eþ00 1.40Eþ00 0.000 �4.60Eþ00 1.70Eþ00 0.007Hot day num

ˇ

2 �2.82E-01 9.63E-02 0.003 �2.72E-01 2.23E-01 3.28Eþ00 5.55E-01 0.000 �2.03Eþ00 3.84E-01 0.000Frost day num 9.16Eþ00 3.35Eþ00 0.006 7.50Eþ01 9.84Eþ00 0.000Frost day num*AOT40 �7.76E-05 1.90E-05 0.000 �1.97E-04 3.35E-05 0.000 �7.08E-05 1.71E-05 0.000Frost day num*ox-N �1.35E-01 4.41E-02 0.002 �2.49E-01 7.51E-02 0.001Frost day num*rainfall �5.84E-03 9.95E-04 0.000 1.18E-02 2.25E-03 0.000Frost day num*solar radiation 1.59E-03 4.27E-04 0.000Frost day num*red-N �1.73E-01 4.10E-02 0.000 �5.64E-02 2.11E-01 1.51E-01 3.78E-02 0.000 7.17E-01 2.90E-01 0.013Frost day num*Tmax �1.97E-01 7.90E-02 0.013 �4.84Eþ00 7.40E-01 0.000Frost day num

ˇ

2 �6.90E-02 7.41E-03 0.000 �7.50E-02 2.48E-02 0.002 �3.59E-01 5.31E-02 0.000ox-N �3.78Eþ01 7.60Eþ00 0.000 �2.85Eþ01 2.61Eþ01 �2.54Eþ01 7.38Eþ00 0.001 �7.64Eþ01 7.73Eþ00 0.000ox-N

ˇ

2 �9.41E-02 2.26E-02 0.000 �1.06E-01 6.67E-02 4.24E-01 9.23E-02 0.000Rainfall 7.19E-01 1.50E-01 0.000 3.72E-01 3.46E-01 �4.04E-01 1.41E-01 0.004 5.92E-01 2.02E-01 0.003 �1.13Eþ00 2.59E-01 0.000Rainfall*AOT40 �1.17E-06 1.54E-07 0.000 �1.18E-06 5.69E-07 �2.90E-06 1.22E-06 0.018 4.00E-06 8.60E-07 0.000 2.76E-06 4.38E-07 0.000Rainfall*ox-N �1.65E-03 8.03E-04 0.040 �4.16E-04 1.95E-03 �5.23E-03 2.53E-03 0.039Rainfall*red-N 1.61E-02 2.31E-03 0.000 1.07E-02 7.82E-03 4.06E-03 1.79E-03 0.023 3.15E-02 5.30E-03 0.000Rainfall*solar radiation �3.46E-05 4.24E-06 0.000 �2.46E-05 1.61E-05Rainfall*Tmin 6.80E-02 1.87E-02 0.000Rainfall

ˇ

2 �2.01E-04 8.74E-05 0.022 �1.39E-04 5.62E-05 0.014 �1.75E-04 4.61E-05 0.000Solar radiation �8.10E-02 2.69E-02 0.003 �2.43E-02 6.57E-02 8.92E-01 2.48E-01 0.000 2.26E-01 4.79E-02 0.000Solar radiation*AOT40 �1.32E-07 4.64E-08 0.004 �1.36E-07 8.17E-08 2.54E-07 9.29E-08 0.006Solar radiation*ox-N 1.13E-03 2.07E-04 0.000 6.09E-04 5.97E-04Solar radiation*red-N �4.72E-03 6.10E-04 0.000 �2.93E-03 2.35E-03 �4.61E-03 4.54E-04 0.000 �7.02E-03 1.85E-03 0.000Solar radiation

ˇ

2 1.88E-06 7.57E-07 0.013 5.17E-07 2.06E-06 �3.09E-05 8.74E-06 0.000 �5.88E-06 1.31E-06 0.000red-N 4.51Eþ01 1.55Eþ01 0.004 1.80Eþ01 7.64Eþ01 6.06Eþ01 8.39Eþ00 0.000red-N*ox-N 4.50E-01 9.86E-02 0.000 4.19E-01 2.39E-01 �3.25Eþ00 5.11E-01 0.000red-N

ˇ

2 �3.09E-01 1.05E-01 0.003 �1.93E-01 3.61E-01 3.40Eþ00 6.90E-01 0.000 1.87Eþ00 5.03E-01 0.000Tmax 2.11Eþ02 3.44Eþ01 0.000 1.89Eþ02 1.73Eþ02 �3.43Eþ02 6.00Eþ00 0.000 �2.94Eþ01 6.53Eþ00 0.000 �2.09Eþ01 3.62Eþ00 0.000Tmax*AOT40 3.83E-04 1.47E-04 0.009Tmax*ox-N 6.21E-01 2.56E-01 0.015Tmax*red-N 1.54Eþ00 4.32E-01 0.000 1.23Eþ00 1.93Eþ00 6.72Eþ00 1.52Eþ00 0.000Tmax

ˇ

2 �3.22Eþ00 6.44E-01 0.000 �2.89Eþ00 2.71Eþ00Tmin �1.87Eþ02 2.51Eþ01 0.000 1.49Eþ02 2.46Eþ01 0.000 3.61Eþ02 5.84Eþ01 0.000Tmin*AOT40 �1.29E-04 3.28E-05 0.000 �1.72E-04 1.04E-04 �7.09E-04 2.85E-04 0.013 �2.43E-03 3.35E-04 0.000 �5.18E-04 1.14E-04 0.000Tmin*ox-N 4.76E-01 1.56E-01 0.002 1.32E-01 3.46E-01 �3.67Eþ00 7.03E-01 0.000Tmin*red-N �2.02Eþ00 3.77E-01 0.000 �1.18Eþ00 1.00Eþ00 2.80Eþ00 6.29E-01 0.000 1.08Eþ01 2.77Eþ00 0.000 �4.87Eþ00 1.70Eþ00 0.004Tmin

ˇ

2 1.19Eþ01 1.83Eþ00 0.000 6.48Eþ00 2.08Eþ00 0.002 �1.51Eþ01 2.32Eþ00 0.000Long 4.04Eþ01 1.45Eþ01 0.005 3.20Eþ01 4.42Eþ00 0.000Long*AOT40 �5.99E-05 1.89E-05 0.001Long*hot day num 3.27E-01 4.72E-02 0.000 3.59E-01 1.89E-01 �8.24E-01 2.54E-01 0.001 1.17Eþ00 1.57E-01 0.000 2.50E-01 7.72E-02 0.001Long*frost day num 2.80E-02 1.29E-02 0.030 2.22E-02 1.87E-02 2.50E-01 4.30E-02 0.000 �1.38E-01 6.58E-02 0.036 �2.82E-01 5.00E-02 0.000

A.D

eMarco

etal./

Environmental

Pollution172

(2013)250

e263

260

Long*rainfall

�2.34E

-03

3.96

E-04

0.00

0�1

.88E

-03

7.22

E-04

5.17

E-03

1.45

E-03

0.00

0Lo

ng*solarradiation

6.11

E-04

9.86

E-05

0.00

04.36

E-04

3.19

E-04

�4.86E

-03

8.31

E-04

0.00

0Lo

ng*ox

-N8.54

E-02

1.87

E-02

0.00

05.70

E-02

4.11

E-02

Long*red-N

�2.80E

-01

4.86

E-02

0.00

0�2

.08E

-01

1.22

E-01

�2.72E

-01

7.62

E-02

0.00

0�5

.47E

-01

1.44

E-01

0.00

0Lo

ng*Tm

ax�5

.98E

-01

9.34

E-02

0.00

0�7

.15E

-01

2.55

E-01

1.16

Eþ00

3.22

E-01

0.00

0Lo

ng*Tm

in3.63

Eþ00

6.63

E-01

0.00

0�1

.71E

þ00

6.68

E-01

0.01

0�1

.93E

þ00

3.28

E-01

0.00

0Lo

ng*Lat

2.68

E-02

1.13

E-02

0.01

7�4

.58E

-03

4.44

E-02

�7.44E

-01

7.30

E-02

0.00

0�3

.13E

-01

6.14

E-02

0.00

0Lo

ng

ˇ 2�1

.37E

-02

4.81

E-03

0.00

4�2

.81E

-02

1.50

E-02

�1.48E

-01

2.18

E-02

0.00

0�6

.00E

-02

1.88

E-02

0.00

1Lat

9.53

Eþ00

2.01

Eþ00

0.00

01.78

Eþ01

8.49

Eþ00

�2.81E

þ01

4.41

Eþ00

0.00

0�2

.04E

þ01

8.03

Eþ00

0.01

11.06

Eþ02

1.78

Eþ01

0.00

0Lat*AOT4

0�2

.13E

-05

7.62

E-06

0.00

5�2

.90E

-05

2.54

E-05

2.63

E-04

4.10

E-05

0.00

07.55

E-05

3.81

E-05

0.04

8Lat*Hot

day

num

1.86

Eþ00

2.46

E-01

0.00

02.61

Eþ00

2.84

E-01

0.00

0Lat*Frostday

num

5.52

E-02

2.50

E-02

0.02

72.65

E-02

6.66

E-02

2.40

E-01

2.03

E-02

0.00

0�4

.68E

-01

1.07

E-01

0.00

0Lat*ox

-N1.78

E-01

3.96

E-02

0.00

01.30

E-01

8.51

E-02

8.91

E-01

2.12

E-01

0.00

0Lat*rainfall

�4.61E

-03

8.44

E-04

0.00

0�3

.88E

-03

2.77

E-03

1.85

E-02

2.57

E-03

0.00

05.35

E-03

2.63

E-03

0.04

2Lat*solarradiation

7.44

E-04

1.17

E-04

0.00

05.01

E-04

8.30

E-04

�0.003

9329

0.00

0729

0Lat*red-N

�2.52E

-01

8.66

E-02

0.00

4�1

.84E

-01

3.03

E-01

�1.99E

þ00

5.13

E-01

0.00

0�1

.51E

þ00

3.09

E-01

0.00

0Lat*Tm

in3.16

E-01

1.45

E-01

0.03

0�1

.36E

-02

6.10

E-01

�2.15E

þ00

6.01

E-01

0.00

0Lat

ˇ 2�2

.10E

-01

8.78

E-02

0.01

7Intercep

t�2

.57E

þ03

5.35

Eþ02

0.00

0�1

.20E

-02

5.15

E-02

�4.26E

þ03

6.81

Eþ02

0.00

0

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263 261

References

Ainsworth, E.A., Long, S.P., 2005. What have we learned from 15 years of free-airCO2 enrichment (FACE)? A meta-analytic review of the responses of photo-synthesis canopy. New Phytologist 165, 351e371.

Amann, M., Bertok, I., Borken-Kleefeld, J., Cofala, J., Heyes, C., Höglund-Isaksson, L.,Klimont, Z., Nguyen, B., Posch, M., Rafaj, P., Sandler, R., Schöpp, W., Wagner, F.,Winiwarter, W., 2011. Cost-effective control of air quality and greenhouse gasesin Europe: modeling and policy applications. Environmental Modelling &Software 26 (12), 1489e1501.

Amann, M., Bertok, I., Cofala, J., Klimont, Z., Schöpp, W., (1993). Structure of theRAINS 7.0 energy- and emissions database. IIASA Working Paper WP-93e67.

Ashmore, M.R., 2005. Assessing the future global impacts of ozone on vegetation.Plant. Cell and Environment 28 (8), 949e964.

ARIANET, 2004. Farm (Flexible Air Quality Regional Model) e Model Formulationand User Manual e Version 2.2. Milano. Arianet report R2004.04.

Attorre, F., Francesconi, F., Valenti, R., Vitale, M., Alfò, M., Bruno, F., 2011. Evaluatingthe effects of climate change on Italian forests trough abundance measures andspecies composition indexes. Applied Vegetation Science 14 (2), 242e255.

Ball, G.R., Palmer-Brown, D., Fuhrer, J., Skärby, L., Gimeno, B.S., Mills, G., 2000.Identification of non-linear influences on the seasonal ozone dose response ofsensitive and resistant clover clones using artificial neural networks. EcologicalModelling 129, 153e168.

Bauhus, J., Puettmann, K., Messier, C., 2009. Silviculture for old-growth attributes.Forest Ecology & Management 258, 525e537.

Brumme, R., Khanna, P.K., 2008. Ecological and site historical aspects of Ndynamics and current N status in temperate forests. Global Change andBiology 14, 125e141.

Bytnerowicz, A., Omasa, K., Paoletti, E., 2007. Integrated effects of air pollution andclimate change on forests: a northern hemisphere perspective. EnvironmentalPollution 147 (3), 438e445.

Bussotti, F., Desotgiu, R., Cascio, C., Pollastrini, M., Gravano, E., Gerosa, G.,Marzuoli, R., Nali, C., Lorenzini, G., Salvatori, E., Manes, F., Schaub, M.,Strasser, R.J., 2011. Ozone stress in woody plants assessed with chlorophylla fluorescence. A critical reassessment of existing data. Environmental andExperimental Botany 73, 19e30.

CLC, (2000). Corine land cover update 2000. Technical guidelines. Prepared by:George Büttner, Jan Feranec, Gabriel Jaffrain. Technical report 89. pp. 56. http://www.eea.europa.eu/publications/technical_report_2002_89.

Chirici, G., Barbati, A., Maselli, F., 2007. Modelling of Italian forest net primaryproductivity by the integration of remotely sensed and GIS data. ForestEcological Management 246, 285e295.

Ciais, P., Schelhaas, M.J., Zaehle, S., Piao, S.L., Cescatti, A., Liski, J., Luyssaert, S., Le-Maire, G., Schulze, E.-D., Bouriaud, O., Freibauer, A., Valentini, R., Nabuurs, G.J.,2008. Carbon accumulation in European forests. Nature Geoscience 1, 425e429.

D’Elia, I., Bencardino, M., Ciancarella, L., Contaldi, M., Vialetto, G., 2009. Technicaland non-technical measures for air pollution emission reduction: the integratedassessment of the regional air quality management plans through the Italiannational model. Atmospheric Environment 43, 6182e6189.

Dai, Y., Zeng, X., Dickinson, R.E., Baker, I., Bonan, G.B., Bosilovich, M.G., Denning, A.S.,Dirmeyer, P.A., Houser, P.R., Niu, G., Oleson, K.W., Schlosser, C.A., Yang, Z.L.,2003. The common land model. Bulletin of the American MeteorologicalSociety 84, 1013e1023.

De Vries, W., Solberg, S., Dobbertin, M., Sterba, H., Laubhann, D., Van Oijen, M.,Evans, C., Gundersen, P., Kros, J., Wamelink, G.W.W., Reinds, G.J., Sutton, M.A.,2009. The impact of nitrogen deposition on carbon sequestration by Europeanforests and heath-lands. Forest Ecology and Management 258, 1814e1823.

De Wit, C.T., Goudriaan, J., van Laar, H.H., Penning de Vries, F.W.T., Rabbinge, R., vanKeulen, H., Louwerse, W., Sibma, L., de Jonge, C., 1978. Simulation of Assimila-tion, Respiration and Transpiration of Crops. Pudoc, Wageningen, TheNetherlands, ISBN 90-220-0601-8, 141 pp.

Di Traglia, M., Attorre, F., Francesconi, F., Valenti, R., Vitale, M., 2011. Is cellularautomata algorithm able to predict the future dynamical shifts of tree species inItaly under climate change scenarios? A methodological approach. EcologicalModelling 222 (4), 925e934.

Dias, T., Malveiro, S., Branquinho, C., Tenreiro, R., Chaves, S., Martins-Loução, M.A.,Sheppard, L., Cruz, C., 2011. Effect of increased nitrogen availability inMediterranean-type ecosystems: a case study in a Natura 2000 site in Portugal.In: Hicks, W.K., Whitfield, C.P., Bealey, W.J., Sutton, M.A. (Eds.), NitrogenDeposition and Natura 2000: Science and Practice in Determining Environ-mental Impacts. Springer, ISBN 978-91-86125-23-3, pp. 173e180.

Dobson, A.J., 1990. An Introduction to Generalized Linear Models. Chapman & Hall,New York.

Emmett, B.A., 2007. Nitrogen saturation of terrestrial ecosystems: some recentfindings and their implications for our conceptual framework. Water, Air andSoil Pollution 7, 99e109.

Erisman, J.W., De Vries, W., 2000. Nitrogen deposition and effects on Europeanforests. Environmental Reviews 8, 65e93.

Faria, T., Silvério, D., Breia, E., Cabral, R., Abadía, A., Abadía, J., Pereira, J.S.,Chaves, M.M., 1998. Differences in the response of carbon assimilation tosummer stress (water deficits, high light and temperature) in four Mediterra-nean tree species. Physiologia Plantarum 102, 419e428.

Ferretti, M., Fagnano, M., Amoriello, T., Badiani, M., Ballarin-Denti, A., Buffoni, A.,Bussotti, F., Castagna, A., et al., 2007. Measuring, modelling and testing ozone

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263262

exposure, flux and effects on vegetation in southern European conditions ewhatdoes not work? A review from Italy. Environmental Pollution 146, 648e658.

Gerosa, G., Finco, A., Mereu, S., Vitale, M., Manes, F., Ballarin Denti, A., 2009.Comparison of seasonal variation of ozone exposure and fluxes in a Mediter-ranean Holm oak forest between the exceptionally dry 2003 and the followingyear. Environmental Pollution 157, 1737e1744.

Granier, A., Ceschia, E., Damesin, C., Dufrêne, E., Epron, D., Gross, P., Lebaube, S., LeDantec, V., Le Goff, N., Lemoine, D., Lucot, E., Ottorini, J.M., Pontailler, J.Y.,Saugier, B., 2000. The carbon balance of a young beech forest. FunctionalEcology 14, 312e325.

Gutiérrez, E., 1988. Dendroecological study of Fagus sylvatica L. in the MontsenyMountains (Spain). Acta Oecologica 9, 301e309.

Hannah, L., Midgley, G., Andelman, S., Araujo, M., Hughes, G., Martinez-Meyer, E.,Pearson, R., Williams, P., 2007. Protected area needs in a changing climate.Frontiers in Ecology and Environment 5, 131e138.

Heimann, M., Reichstein, M., 2008. Terrestrial ecosystem carbon dynamics andclimate feedbacks. Nature 451, 289e292.

Hole, D.G., Willis, S.G., Pain, D.J., Fishpool, L.D., Butchart, H.M., Collingham, Y.C.,Rahbek, C., Huntley, B., 2009. Projected impacts of climate change on a conti-nent-wide protected area network. Ecology Letters 12, 420e431.

Hyvönen, R., Ågren, G.I., Linder, S., Persson, T., Cotrufo, M.F., Ekblad, A., Freeman, M.,Grelle, A., Janssens, I.A., Jarvis, P.G., Kellomäki, S., Lindroth, A., Loustau, D.,Lundmark, T., Norby, R.J., Oren, R., Pilegaard, K., Ryan, M.G., Sigurdsson, B.D.,Strömgren, M., Van Oijen, M., Wallin, G., 2007. The likely impact of elevated[CO2], nitrogen deposition, increased temperature and management on carbonsequestration in temperate and boreal forest ecosystems: a literature review.New Phytologist 173, 463e480.

ISTAT, 2005. Statistiche Dell’Agricoltura. Anno 2000. Annuario n� 48. IstitutoNazionale di Statistica (ISTAT), Rome, Italy.

Kahle, H.P., Karjalainen, T., Schuck, A., Agren, G.I., Kellomäki, S., Mellert, K.,Prietzel, J., Rehfuess, K.E., Spiecker, H., 2008. In: Causes and Consequences ofForest Growth Trends in Europe. European Forest Institute Research Project 21.Results of the RECOGNITION Project, vol. 21. Boston, Brill, Leiden, p. 261.

Kangas, L., Syri, S., 2002. Regional nitrogen deposition model for integratedassessment of acidification and eutrophycation. Atmospheric Environment 36,1111e1122.

Karlsson, P., Braun, S., Broadmeadow, M., Elvira, S., Emberson, L., Gimenod, B.S., LeThiec, D., Novak, K., Oksanen, E., Schaub, M., Uddling, J., Wilkinson, M., 2007.Risk assessments for forest trees: the performance of the ozone flux versus theAOT concepts. Environmental Pollution 146, 608e616.

Karnosky, D.F., Werner, H., Holopainen, T., Percy, K., Oksanen, T., Oksanen, E.,Heerdt, C., Fabian, P., Nagy, J., Heilman, W., Cox, R., Nelson, N., Matyssek, R.,2007. Free-air exposure systems to scale up ozone research to mature trees.Plant Biology 9, 181e190.

Krupa, S.V., 2003. Effects of atmospheric ammonia (NH3) on terrestrial vegetation:a review. Environmental Pollution 124, 179e221.

Kull, O., Jarvis, P.G., 1995. The role of nitrogen in a simple scheme to scale upphotosynthesis from leaf to the canopy. Plant Cell and Environment 18,1174e1182.

Langley, J.A., Megonigal, J.P., 2010. Ecosystem response to elevated CO2 levelslimited by nitrogen-induced plant species shift. Nature e Letters 466, 96e99.

Lavorel, S., Canadell, J., Rambal, S., Terradas, J., 1998. Mediterranean terrestrialecosystems: research priorities on global change effects. Global Ecology andBiogeography 7, 157e166.

Lawler, J.J., Tear, T.H., Pyke, C., Shaw, M.R., Gonzalez, P., Kareiva, P., Hansen, L.,Hannah, L., Klausmeyer, K., Aldous, A., Bienz, C., Pearsall, S., 2010. Resourcemanagement in a changing and uncertain climate. Frontiers in Ecology and theEnvironment 8, 35e43.

Lebourgeois, F., Bréda, N., Ulrich, E., Granier, A., 2005. Climate-tree-growth rela-tionships of European beech (Fagus sylvatica L.) in the French Permanent PlotNetwork (RENECOFOR). Trees 19, 385e401.

Long, S.P., Ainsworth, E.A., Rogers, A., Ort, D.R., 2004. Rising atmospheric carbondioxide: plants face the future. Annual Review of Plant Biology 55, 591e628.

LRTAP e ICP, 2004. In: Spranger, T., Lorenz, U., Gregor, H.D. (Eds.), Manual onMethodologies and Criteria for Modelling and Mapping Critical Loads andLevels and Air Pollution Effects, Risks and Trends. Federal EnvironmentalAgency, Berlin. http://www.umweltdaten.de/publikationen/fpdf-l/2837.pdf(accessed 08.01.12.).

Luo, Y., 2007. Terrestrial carbon-cycle feedback to climate warming. Annual Reviewof Ecology, Evolution, and Systematics 38, 683e712.

Manes, F., Vitale, M., Fabi, M.A., De Santis, F., Zona, D., 2007. Estimates of potentialozone stomatal uptake in mature trees of Quercus ilex in a Mediterraneanclimate. Environmental and Experimental Botany 59, 235e241.

Manes, F., Vitale, M., Donato, E., Giannini, M., Puppi, G., 2006. Different ability ofthree Mediterranean oak species to tolerate progressive water stress. Photo-synthetica 44 (3), 387e393.

Manes, F., Vitale, M., Di Traglia, M., 2005. Tropospheric ozone impact on plants andmonitoring in natural and urban areas characterized by Mediterranean climate.Plant Biosystems 139 (3), 265e278.

Manes, F., Seufert, G., Vitale, M., 1997. Eco-physiological studies of Mediterraneanplant species at the Castelporziano Estate. Atmospheric Environment 31 (SI),51e60.

Maselli, F., Chiesi, M., Moriondo, M., Fibbi, L., Bindi, M., Running, S.W., 2009.Modelling the forest carbon budget of a Mediterranean region through theintegration of ground and satellite data. Ecological Modelling 220, 330e342.

Matesanz, S., Escudero, A., Valladares, F., 2008. Additive effects of a potentiallyinvasive grass and water stress on the performance of seedlings of gypsumspecialists. Applied Vegetation Science 11, 287e296.

Matthews, H.D., 2007. Implications of CO2 fertilization for future climate change ina coupled climate-carbon model. Global Change & Biology 13, 1068e1078.

Matyssek, R., Wieser, G., Ceulemans, R., Rennenberg, H., Pretzsch, H., Haberer, K.,Löw, M., Nunn, J.J., Werner, H., Wipfler, P., Oßwald, W., Nikolova, P., Hanke, D.,et al., 2010a. Enhanced ozone strongly reduces carbon sink strength of adultbeech (Fagus sylvatica) e Resume from the free-air fumigation study atKranzberg Forest. Environmental Pollution 158, 2527e2532.

Matyssek, R., Karnosky, D., Kubiske, M., Oksanen, E., Wieser, G., 2010b. Advances inunderstanding ozone impact on forest trees: messages from novel phytotronand free-air fumigation studies. Environmental Pollution 158 (6), 1990e2006.

Matyssek, R., Sandermann, H., Wieser, G., Booker, F., Cieslik, S., Musselman, R.,Ernst, D., 2008. The challenge of making ozone risk assessment for forest treesmore mechanistic. Environmental Pollution 156, 567e582.

Matyssek, R., Bytnerowicz, A., Karlsson, P.E., Paoletti, E., Sanz, M., Schaub, M.,Wieser, G., 2007. Promoting the O3 flux concept for forest trees. EnvironmentalPollution 146, 587e607.

McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models, second ed. Chapman &Hall, New York.

Milne, R., Van Oijen, M., 2005. A comparison of two modelling studies of envi-ronmental effects on forest carbon stocks across Europe. Annals of ForestScience 62, 911e923.

Myneni, R.B., Keeling, C.D., Tucker, C.J., Asrar, G., Nemani, R.R., 1997. Increased plantgrowth in the northern high latitudes from 1981 to 1991. Nature 386, 698e702.

Nabuurs, G.J., Päivinen, R., Schanz, H., 2001. Sustainable management regimes forEurope’s forests e a projection with EFISCEN until 2050. Forest Policy andEconomics 3, 55e173.

Nabuurs, G.J., Päivinen, R., Sikkema, R., Mohren, G.M.J., 1997. The role of Europeanforests in the global carbon cycle: a review. Biomass and Bioenergy 13, 345e358.

Niu, S., Yang, H., Zhang, Z., Wu, M., Lu, Q., Li, L., Han, X., Wan, S., 2009. Non-additiveeffects of water and nitrogen addition on ecosystem carbon exchange ina temperate steppe. Ecosystems 12, 915e926.

Norby, R.J., Warren, J.M., Iversen, C.M., Medlyn, B.E., McMurtrie, R.E., 2010. CO2enhancement of forest productivity constrained by limited nitrogen availability.Proceedings of the National Academy of Sciences USA 107, 19368e19373.

Norby, R., De Lucia, E., Gielen, B., Calfapietra, C., Giardina, C., King, J., Ledford, J.,McCarthy, H., Moore, D., et al., 2005. Forest response to elevated CO2 isconserved across a broad range of productivity. Proceedings of the NationalAcademy of Sciences USA 102, 18052e18056.

Novak, K., Cherubini, P., Saurer, M., Fuhrer, J., Skelly, J.M., Kräuchi, N., Schaub, M.,2007. Ozone air pollution effects on tree ring growth, delta13C, visible foliarinjury and leaf gas exchange in three ozone-sensitive woody plant species. TreePhysiology 27, 941e949.

Paoletti, E., 2006. Impact of ozone on Mediterranean forests: a review. Environ-mental Pollution 144, 463e474.

Paoletti, E., Manning,W.J., 2007. Toward a biologically significant andusable standardfor ozone that will also protect plants. Environmental Pollution 150, 85e95.

Pielke, R.A., Cotton, W.R., Walko, R.L., Tremback, C.J., Lyons, W.A., Grasso, L.D.,Nicholls, M.E., Moran, M.D., Wesley, D.A., Lee, T.J., Copeland, J.H., 1992.A comprehensive meteorological modeling system e RAMS. Meteorology andAtmospheric Physics 49, 69e91.

Pressey, R.L., Cabeza, M., Watts, M.E., Cowling, R.M., Wilson, K.A., 2007. Conserva-tion planning in a changing world. Trends in Ecology and Evolution 22 (11),583e592.

Rehfuess, K.E., Ågren, G.I., Andersson, F., Cannell, M.G.R., Friend, A., Hunter, I., Kahle,H-.P., Prietzel, J., Spiecker, H., (1999). Relationships between recent changes ofgrowth and nutrition of Norway spruce, Scots pine and European beech forestsin Europe-RECOGNITION. Working Paper 19, European Forest Institute, Joensuu,Finland, 94 p.

Sanz-Pérez, V., Castro Díez, P., Valladares, F., 2007. Growth versus storage: responsesof Mediterranean oak seedlings to changes in nutrient and water availabilities.Annals of Forest Science 64, 201e210.

Scarascia-Mugnozza, G., Oswald, H., Piussi, P., Radoglou, K., 2000. Forests of theMediterranean region: gaps in knowledge and research needs. Forest Ecologyand Management 132, 97e109.

Schaub, M., Calatayud, V., Ferretti, M., Brunialti, G., Lövblad, G., Krause, G., Sanz, M.J.,2010. Monitoring of ozone injury. manual part X. In: Manual on Methods andCriteria for Harmonized Sampling, Assessment, Monitoring and Analysis of theEffects of Air Pollution on Forests. UNECE ICP Forests Programme Co-ordinatingCentre, Hamburg, p. 22. http://www.icp-forests.org/Manual.htm.

Schöpp, W., Amann, M., Cofala, J., Heyes, C., Klimont, Z., 1999. Integrated assessmentof European air pollution emission control strategies. Environmental Modelingand Software 14 (1), 1e9.

SCIA, 2011. National System for Collecting, Processing and Dissemination ofClimatological Data of Environmental Interest.

Sitch, S., Cox, P.M., Collins, W.J., Huntingford, C., 2007. Indirect radiative forcing ofclimate change through ozone effect on land carbon sink. Nature 448, 791e794.

Solberg, S., Andreassen, K., Clarke, N., Torseth, K., Tveito, O.E., Strand, G.H.,Tomter, S., 2004. The possible influence of nitrogen and acid deposition onforest growth in Norway. Forest Ecology and Management 192, 241e249.

Sutton, M.A., Simpson, D., Levy, P.E., Smith, R.I., Reis, S., van Oijen, M., de Vries, W.,2008. Uncertainties in the relationship between atmospheric nitrogen deposi-tion and forest carbon sequestration. Global Change Biology 14, 2057e2063.

A. De Marco et al. / Environmental Pollution 172 (2013) 250e263 263

Thomas, R.Q., Canham, C.D., Weathers, K.C., Goodale, C.L., 2010. Increased treecarbon storage in response to nitrogen deposition in the US. Nature Geoscience3, 13e17.

Vialetto, G., Contaldi, M., De Lauretis, R., Lelli, M., Mazzotta, V., Pignatelli, T., 2005.Emission scenarios of air pollutants in Italy using integrated assessmentmodels. Pollution Atmosphérique 185, 71.

Vitale, M., Capogna, F., Manes, F., 2007. Resilience assessment on Phillyrea angus-tifolia L. maquis undergone to experimental fire through a big-leaf modelingapproach. Ecological Modelling 203, 387e394.

Vitale, M., Gerosa, G., Ballarin-Denti, A., Manes, F., 2005. Ozone uptake by anevergreen Mediterranean forest (Quercus ilex L.) in Italy. Part II: flux modelling.Up scaling leaf to canopy ozone uptake by a process-based model. AtmosphericEnvironment 39, 3267e3278.

Vitale, M., Scimone, M., Feoli, E., Manes, F., 2003. Modelling leaf gas exchanges topredict functional trends in Mediterranean Quercus ilex forest under climaticchanges in temperature. Ecological Modelling 166, 123e134.

Vourlitis, G.L., Pasquini, S.C., Mustard, R., 2009. Effects of dry-season N input on theproductivity and N storage of Mediterranean-type shrublands. Ecosystems 12,473e488.

Wamelink,G.W.W.,Wieggers, H.J.J., Reinds,G.J., Kros, J.,Mol-Dijkstra, J.P., VanOijen,M.,DeVries,W., 2009.Modelling impacts of changes in carbondioxide concentration,climate and nitrogen deposition on carbon sequestration by European forests andforest soils. Forest Ecology and Management 258, 1794e1805.

Wu, Z., Dijkstra, P., Koch, G.W., Peñuelas, J., Hungate, B.A., 2011. Responses ofterrestrial ecosystems to temperature and precipitation change: a meta-analysis of experimental manipulation. Global Change and Biology 17, 927e942.