Effects of bryophyte and lichen cover on permafrost soil ... · simulates bryophyte and lichen cover on upland sites. Wet-lands are not included. We take into account the dynamic

The Cryosphere 10 2291ndash2315 2016wwwthe-cryospherenet1022912016doi105194tc-10-2291-2016copy Author(s) 2016 CC Attribution 30 License

Effects of bryophyte and lichen cover onpermafrost soil temperature at large scalePhilipp Porada12 Altug Ekici134 and Christian Beer12

1Department of Environmental Science and Analytical Chemistry Stockholm University Svante Arrhenius vaumlg 811418 Stockholm Sweden2Bolin Centre for Climate Research Stockholm University 10691 Stockholm Sweden3Earth System Sciences Laver Building University of Exeter Exeter UK4Uni Research Climate Bjerknes Centre for Climate Research Bergen Norway

Correspondence to Philipp Porada (philippporadaacessuse)

Received 4 December 2015 ndash Published in The Cryosphere Discuss 1 February 2016Revised 12 May 2016 ndash Accepted 6 September 2016 ndash Published 30 September 2016

Abstract Bryophyte and lichen cover on the forest floorat high latitudes exerts an insulating effect on the groundIn this way the cover decreases mean annual soil tempera-ture and can protect permafrost soil Climate change how-ever may change bryophyte and lichen cover with effects onthe permafrost state and related carbon balance It is there-fore crucial to predict how the bryophyte and lichen coverwill react to environmental change at the global scale Todate current global land surface models contain only empir-ical representations of the bryophyte and lichen cover whichmakes it impractical to predict the future state and functionof bryophytes and lichens For this reason we integrate aprocess-based model of bryophyte and lichen growth intothe global land surface model JSBACH (Jena Scheme forBiospherendashAtmosphere Coupling in Hamburg) The modelsimulates bryophyte and lichen cover on upland sites Wet-lands are not included We take into account the dynamicnature of the thermal properties of the bryophyte and lichencover and their relation to environmental factors Subse-quently we compare simulations with and without bryophyteand lichen cover to quantify the insulating effect of the organ-isms on the soil

We find an average cooling effect of the bryophyte andlichen cover of 27 K on temperature in the topsoil for theregion north of 50 N under the current climate Locally acooling of up to 57 K may be reached Moreover we showthat using a simple empirical representation of the bryophyteand lichen cover without dynamic properties only resultsin an average cooling of around 05 K This suggests that

(a) bryophytes and lichens have a significant impact on soiltemperature in high-latitude ecosystems and (b) a process-based description of their thermal properties is necessaryfor a realistic representation of the cooling effect The ad-vanced land surface scheme including a dynamic bryophyteand lichen model will be the basis for an improved futureprojection of landndashatmosphere heat and carbon exchange

1 Introduction

Vegetation cover on the ground consisting of bryophytes(mosses liverworts and hornworts) and lichens may play asignificant role for the energy and carbon balance of high-latitude ecosystems since it reduces the exchange of heat be-tween atmosphere and soil (Soudzilovskaia et al 2013) Thisinsulating effect causes a decrease in soil temperature andthereby protects permafrost (Jorgenson et al 2010 Turetskyet al 2010) Consequently a reduction in the surface cov-erage of bryophytes and lichens may result in increased soiltemperature accompanied by thawing of permafrost (Rinkeet al 2008) This might lead to an increased release of car-bon from the soil to the atmosphere which would accelerateglobal warming (Schuur et al 2015) although this may bepartially compensated by increased productivity eg fromshrubs in northern ecosystems It is therefore crucial to de-termine whether the bryophyte and lichen cover will expandor recede in response to a warming climate and elevated at-mospheric CO2 Depending on the direction of this response

Published by Copernicus Publications on behalf of the European Geosciences Union

2292 P Porada et al Insulation by bryophytes and lichens

the effect of bryophytes and lichens at high latitudes on theglobal climate might represent either a positive or a negativefeedback

As a first step to analyse this feedback we focus in thisstudy on the effect of the bryophyte and lichen cover on soiltemperature at high-latitudes At the local to regional scaleexperimental as well as modelling studies have assessed theeffect of the bryophyte and lichen cover on soil temperatureBlok et al (2011) show that the experimental removal ofmoss leads to increased ground heat flux and increased soilevaporation Gornall et al (2007) observe higher soil tem-peratures in summer with decreasing thickness of the mosscover on the ground and lower soil temperatures in winter Ina modelling study Bonan (1991) perform simulations witha local energy balance model for 20 forest stands in centralAlaska to estimate the effect of moss removal on soil tem-perature They find an increase in soil temperature due to thereduction in moss ground cover Jorgenson et al (2010) usea local soil temperature model to quantify the effect of sev-eral factors on permafrost and find that vegetation on groundstrongly promotes permafrost stability

The insulating effect of the bryophyte and lichen groundcover on soil temperature is strongly modulated by bryophyteand lichen moisture content The thermal conductivity of thecover increases by around 1 order of magnitude from dryto water-saturated bryophytes and lichens while the heatcapacity even increases by around 2 orders of magnitude(OrsquoDonnell et al 2009 Soudzilovskaia et al 2013) More-over thermal conductivity increases strongly during the tran-sition from liquid to frozen water inside the bryophyte andlichen cover while heat capacity is reduced (Hinzman et al1991) Therefore a large and important seasonality of the in-sulating effect is expected in winter and to a lesser extentin autumn and spring the insulating effect of bryophytes andlichens will be small due to the high thermal conductivityof ice and water In summer however a dry bryophyte andlichen cover may strongly insulate the ground leading to anet cooling effect for the whole year The relation betweenthermal properties of bryophytes and lichens and their wa-ter saturation is very similar for a large range of differentspecies (Soudzilovskaia et al 2013) However the speciesshow considerable differences in water-holding capacity andother properties which affect their water status Thereforedifferent species may exhibit differing water saturation un-der similar climatic conditions

Several global land surface models which attempt to simu-late permafrost soil at high latitudes include a bryophyte andlichen ground cover while others approximate near-surfacevegetation using grass (eg Zhu et al 2015) Alternativelythe insulating effect is attributed solely to the carbon-richuppermost soil layer (Koven et al 2009) However the dy-namic nature of bryophyte and lichen surface coverage andthe dependence of the insulating effect on moisture contenthave not been explicitly considered so far in global landsurface models Beringer et al (2001) for instance extend

the land surface model NCAR-LSM (National Center forAtmospheric Research Land Surface Model) by lichen ormoss topsoil parameterisations for an arctic tundra site butthey assume constant thermal conductivity and surface cov-erage of the organic layer Ekici et al (2014) use the globalland surface model JSBACH ((Jena Scheme for BiospherendashAtmosphere Coupling in Hamburg)) to simulate permafrostat high latitudes and thereby include an organic layer abovethe soil which reduces heat exchange with the atmosphereThe layer is assumed to cover the whole land surface uni-formly and is set to constant thermal properties RecentlyChadburn et al (2015a) added a representation of moss tothe land surface model JULES (Joint UK Land EnvironmentSimulator) While the surface coverage of the moss layer isprescribed by a static map the thermal conductivity of themoss layer is dynamic Instead of being related to moss wa-ter content however the thermal conductivity is determinedby the moisture content of the uppermost soil layer Subse-quently Chadburn et al (2015b) extend JULES by an em-pirical scheme which determines moss health as a functionof climate variables and consequently allows for the com-putation of a dynamic moss ground cover In summary theuse of constant thermal properties and empirical relations incurrent global land surface models makes it difficult to pre-dict the impact of future bryophyte and lichen functions onpermafrost ecosystems

Here we present a process-based model which simulatesproductivity and dynamic surface coverage of the bryophyteand lichen ground cover at the global scale The model rep-resents the organismsrsquo water and ice content and thereby ac-counts for dynamic thermal properties of the cover The basisfor this model is the stand-alone dynamic non-vascular veg-etation model LiBry which has been successfully applied toquantify global productivity by bryophytes and lichens (Po-rada et al 2013) as well as estimating their contribution toglobal biogeochemical cycles (Porada et al 2014) LiBry isfully integrated into the land surface scheme JSBACH of theMax Planck Institute Earth system model JSBACH simu-lates the carbon balance of vascular plants and the soil at theglobal scale it includes a water and energy balance as well asa representation of permafrost Based on the thermal proper-ties of the near-surface vegetation layer we use JSBACH toquantify the effect of bryophytes and lichens on heat transferbetween atmosphere and soil and therefore on soil temper-ature Since JSBACH does not include a scheme for wetlandhydrology the formation of peatlands cannot be simulatedHence LiBry in JSBACH mainly represents bryophyte andlichen growth on upland forest floor sites

We compare the soil thermal regime of a JSBACH simu-lation including a dynamic bryophyte and lichen layer to asimulation where the thermal properties and surface cover-age of this layer are set to constant and to another simulationwith the bryophyte and lichen cover switched off In this waywe can assess quantitatively the impact of the bryophyte andlichen ground cover on soil temperature at high latitudes In a

The Cryosphere 10 2291ndash2315 2016 wwwthe-cryospherenet1022912016

P Porada et al Insulation by bryophytes and lichens 2293

next step our process-based approach can be used to predictthe role of bryophytes and lichens for high-latitude ecosys-tems under climatic change

2 Methods

21 Model description

For our analysis we use the global land surface model JS-BACH 30 (Raddatz et al 2007 Brovkin et al 2009)which is a part of the Max Planck Institute Earth systemmodel (MPI-ESM 11) JSBACH uses a process-based ap-proach to simulate both physical and biochemical ecosystemfunctions such as the exchange of energy and water at theland surface and carbon fluxes between atmosphere vegeta-tion and soil which are determined by photosynthesis andrespiration The soil part of JSBACH includes a soil car-bon model (Goll et al 2015) and a five-layer soil hydrologyscheme (Hagemann and Stacke 2015) It has been extendedby Ekici et al (2014) to take into account both several dy-namic snow layers and the latent heat of fusion associatedwith freezing and thawing which is the basis for simulatingpermafrost extent and active layer thickness

To represent bryophyte and lichen ground cover theprocess-based non-vascular vegetation model LiBry de-scribed in detail by Porada et al (2013) is integrated intoJSBACH LiBry is a stand-alone dynamic global vegetationmodel that uses climate data such as radiation temperatureand precipitation to predict photosynthesis respiration andgrowth The model combines approaches used in many dy-namic vegetation models such as the Farquhar photosynthe-sis scheme with lichen- and bryophyte-specific processeseg the adaptation of the metabolism to changing levels ofwater content

LiBry uses the Farquhar scheme (Farquhar and von Caem-merer 1982) to calculate photosynthesis Thus both CO2and photosynthetically active radiation (PAR) are limitingfactors The availability of PAR on the ground depends on theshading by trees and it is consequently inversely related to theleaf area index (LAI) of the vegetation above the bryophyteand lichen cover In JSBACH the fraction of PAR availableat the ground can be directly computed from the simulatedLAI of the overlying vegetation (Knorr 2000) and it is thenused as an input for LiBry To estimate the availability ofCO2 for bryophytes and lichens the model computes diffu-sion of CO2 from the atmosphere into the organisms TheCO2 diffusivity is decreasing with increased water content ofthe organisms due to narrowing of diffusion pathways andformation of water films (Cowan et al 1992) In addition toCO2 and PAR the rate of modelled photosynthesis dependson surface temperature which is calculated by JSBACH fromthe surface energy balance Furthermore the photosynthesisrate is related to the level of metabolic activity of bryophytesand lichens which is controlled by their water content and

which ranges from inactive for dry organisms to fully activefor water-saturated organisms This is an important functionfor predicting the response to climate change as precipitationpatterns are projected to change (Pachauri et al 2014)

The water content of LiBry in JSBACH is computed fromthe balance of water input and loss at a given water stor-age capacity For bryophytes and lichens water input cor-responds to rainfall or snowmelt and water loss takes placein the form of evaporation While JSBACH provides fluxesof rainfall and snowmelt it does not compute the evaporationflux from ground-based vegetation In JSBACH evapotran-spiration is partitioned only into transpiration from vegeta-tion bare soil evaporation and evaporation from the intercep-tion reservoir The latter includes water evaporating freelyfrom vegetation surfaces in the canopy and on the groundSince it is impractical to modify this scheme we approxi-mate evaporation from bryophytes and lichens by evapora-tion from the interception reservoir This potentially neglectsthe morphological control of the organisms on evaporationHowever free evaporation is more suitable than transpira-tion by vascular plants to describe water loss of bryophytesand lichens (Nash III 1996 Proctor 2000) Hence the waterbalance of the simulated bryophytes and lichens is coupled tothe interception reservoir in JSBACH This is done by addingthe water storage capacity of bryophytes and lichens to thesize of the interception reservoir and setting the water satura-tion of the bryophytes and lichens equal to the saturation ofthe interception reservoir

In addition to processes involving liquid water freezingand melting of water are taken into account in JSBACH Con-sequently water inside the bryophyte and lichen cover maybe partly or completely frozen Since frozen water occupiespore space the size of the interception reservoir is reducedby the amount of ice inside the bryophyte and lichen coveras long as ice occurs

The net primary productivity (NPP) of bryophytes andlichens is obtained in LiBry by subtracting respiration fromgross photosynthesis where respiration is simulated as afunction of temperature and metabolic activity Turnover ofbiomass associated with mortality is also considered and it ismodelled as a function of the protein content of biomass Thenet growth of bryophytes and lichens is then determined bythe balance of NPP and turnover By multiplying net growthof simulated bryophytes and lichens with their specific areathe growth in surface coverage is derived in the model With-out processes that reduce the bryophyte and lichen cover thesurface coverage would increase to 100 in all areas wherenet growth is larger than zero in steady state Hence a dis-turbance cycle is included into LiBry which periodically setsback the surface coverage to a small initial value The surfacecoverage in steady state is then obtained by averaging over awhole disturbance cycle

LiBry in JSBACH is designed to predict the dynamics ofthe cover in transient scenarios of climate change Conse-quently the steady-state calculation of the surface coverage

wwwthe-cryospherenet1022912016 The Cryosphere 10 2291ndash2315 2016

from the original LiBry is replaced by a dynamic schemeThis also accounts for potential changes in disturbance fre-quency To avoid unrealistic fluctuations of the bryophyte andlichen cover the dynamic disturbance scheme of LiBry inJSBACH accumulates losses of cover over a month Surfacecoverage is then updated on a monthly basis the simulatedbryophyte and lichen cover is reduced by the accumulatedlosses due to disturbance such as fire and it is increased bypositive net growth accumulated over a month Net growthis translated into coverage via the specific area of the organ-isms Negative net growth is subtracted from the cover Thedynamic scheme also accounts in a simple way for dispersaland establishment of bryophytes and lichens (eg Bohn et al2011) the absolute increase in cover per area of ground isboth limited by the already existing cover which generatesthe new biomass as well as the free area that is still avail-able for the growth of new cover The transition of new coverfrom the existing cover to the free area may take place in theform of spores or via vegetative growth (Grime et al 1990Rogers 1990) The efficiency of these processes is sum-marised in the dynamic scheme by an ldquoexpansion efficiencyrdquo(see Fig 1) Equation (1) shows the change of bryophyte andlichen cover 1A1t in one month

1A1t = EminusRGminusRD (1)

where E is the expansion of the bryophyte and lichen coverdue to positive net growth which means that accumulatedgross photosynthesis exceeds accumulated respiration andbiomass turnover RG is the reduction of cover due to neg-ative net growth and RD stands for reduction of cover by dis-turbance Expansion of cover E is calculated as follows

min (GA(10minusA)ηE 10minusA) if Gge 00 if Glt 0 (2)

where A is the surface coverage of bryophytes and lichensin [m2 cover mminus2 ground] G is net growth in [m2 newcover mminus2 cover] accumulated over a month and ηE is a di-mensionless expansion efficiency which is set to 085 Thisvalue is taken from the global stand-alone version of Li-Bry which has also been updated to a dynamic disturbancescheme It was selected to obtain a realistic global distribu-tion of surface coverage The minimum statement ensuresthat the cover increment cannot exceed the available areaThe term A(10minusA) describes limitation by both existingcover and free area available for growth Reduction of coveris calculated as

0 if Gge 0min (minusGA A) if Glt 0 (3)

RD =DA (4)

where D is the fraction of cover lost due to disturbanceIt is set to 000083 per month This value corresponds to

Bryophytes amp lichensFree area

Disturbance

Unit area

Potential new bryophyte amp lichen area

G bull A bull η

(1 - A)A

Figure 1 Change in surface coverage A of bryophytes and lichensA is reduced by disturbance and increases due to net growthG Newbryophyte and lichen cover per unit area is limited by the generat-ing area A the expansion efficiency η and the available free area(1minusA)

a fire return interval of 100 years which is characteristicfor the boreal forest (Bonan and Shugart 1989 Beer et al2006 Mouillot and Field 2005) The minimum statement inEq (3) ensures that the cover reduction cannot lead to nega-tive cover

Not only the scheme for computing surface coverage butalso the representation of the organismsrsquo physiological vari-ation had to be modified to integrate LiBry into JSBACH Inthe original version of the model a Monte Carlo approachis used to sample broad ranges of possible parameter valuesFrom that artificial ldquospeciesrdquo are generated that represent thefunctional diversity of bryophytes and lichens The speciesperform differently under a given climate Those which can-not maintain a cover larger than zero cannot ldquosurviverdquo in themodel and the remaining ones are used to compute the pro-ductivity This species-based approach is conceptually differ-ent from the traditional plant functional type (PFT) approachused in many land surface models and in JSBACH To adaptLiBry to the PFT design of JSBACH the number of artificialspecies is set to one This species is then parameterised tocorrespond to boreal ground-based bryophytes and lichens

In general PFTs in JSBACH are not able to coexist inthe same place since they are represented in distinct non-overlapping tiles The tiles specify which fraction of a modelgrid cell is occupied by a certain PFT thereby representingsubgrid-scale heterogeneity The bryophyte and lichen PFThowever should be able to grow in combination with variousdifferent PFTs It is therefore not limited to a specific tilebut instead it is implemented as an additional layer on top ofthe soil This layer is available on tiles which contain treesor grasses while it is excluded from tiles covered by cropsor glaciers The surface coverage of the bryophyte and lichenPFT may differ between the tiles of a given grid cell Thiscan result from the differential influences of the tilesrsquo vege-tation types on bryophyte and lichen growth for instance dueto differing LAI between the tiles

0 01 02 03 04 05 06 07 08 09 1

ndash2

Relative moisture content [ ]

(a)Unfrozen

50 frozenFrozen

Constant

500 000

15e+06

25e+06

35e+06

0 01 02 03 04 05 06 07 08 09 1

ndash3

(b)Unfrozen

50 frozenFrozen

Constant

ndash1

Figure 2 Dependence of (a) thermal conductivity and (b) heat capacity of the bryophyte and lichen cover on relative moisture content Thesolid blue curves show the relation between thermal properties and moisture content for liquid water the magenta dashed curves correspondto a mixture of 50 liquid and 50 frozen water and the light blue dashed-dotted curves correspond to frozen water The black dotted linesshow the thermal properties of the organic layer from the previous model version (Ekici et al 2014) for comparison They are constant sincethey do not depend on water content

Defining the bryophyte and lichen PFT as a layer in JS-BACH has implications for the representation of the organ-ismsrsquo hydrological properties The original version of LiBrycomputes the water content of the organisms through the bal-ance of water input and loss where the specific storage ca-pacity for water per biomass is sampled for each species froma range of possible values from the literature These valuesare not related to the geometry of the bryophyte or lichenIn JSBACH however the water storage capacity of a layeris determined by multiplying layer thickness by the poros-ity of the layer Hence to be consistent with JSBACH thewater storage capacity of the bryophyte and lichen PFT iscomputed from the thickness and porosity of the bryophyteand lichen layer Based on the study by Soudzilovskaia et al(2013) we set thickness to 45 cm which corresponds tothe median of the measured values The measurements arebased on green and undecomposed brown tissue of bryophytemats Porosity is not directly measured in Soudzilovskaiaet al (2013) but they provide values of maximum volumetricmoisture Since we are actually interested in the water stor-age capacity of the bryophyte and lichen layer we set theeffective porosity to 80 which is at the higher end of themeasured values of volumetric moisture

In JSBACH two thermal properties of the bryophyte andlichen layer have to be known to derive their influence on soiltemperature thermal conductivity and heat capacity Bothare strongly dependent on the relative moisture content ofthe bryophyte and lichen layer Moreover they depend onthe state of matter of the water in the bryophytes and lichenswhich can be liquid or frozen in various relative amountsIn analogy to the soil layers in JSBACH (Ekici et al 2014)we write the dependence of the thermal conductivity of thebryophyte and lichen layer κ on water content as

κ = κ(10minus(vw+vi))o κvw

w κvii Ke+ (10minusKe)κd (5)

where κo is the thermal conductivity of organic matter whichis set to 025 [W Kminus1 mminus1] (Beringer et al 2001) κw is

the thermal conductivity of liquid water κi is the ther-mal conductivity of ice and κd is the thermal conductiv-ity of the dry bryophyte and lichen cover which is set to005 [W Kminus1 mminus1] according to OrsquoDonnell et al (2009)vw and vi are the volumetric moisture and ice contents of thebryophyte and lichen cover which are calculated by dividingthe absolute water or ice content in [m] by the thickness ofthe cover Ke is the Kersten number which is calculated forthe linear regime as

Ke=min(

10vw+ vi

where ε is the porosity of the bryophyte and lichen coverThe dependence of the heat capacity of the bryophyte andlichen cover C on water content is written as

C = (10minus ε)Co+ vwCwρw+ viCiρi (7)

where Co is the heat capacity of organic matter which is setto 25times 106 [J mminus3 Kminus1] (Beringer et al 2001) Cw is theheat capacity of liquid water Ci is the heat capacity of iceand ρw and ρi are the densities of liquid water and ice

Figure 2 shows the thermal conductivity as well as the heatcapacity of the modelled bryophyte and lichen cover in rela-tion to the water saturation of the cover and the state of mat-ter of the water Both the curves for thermal conductivity andheat capacity of the cover compare well with measured val-ues from eg Soudzilovskaia et al (2013) Jorgenson et al(2010) and OrsquoDonnell et al (2009) This suggests that ourchosen average values for thickness and porosity result in re-alistic thermal properties of the bryophyte and lichen cover

To simulate the influence of the bryophyte and lichen layeron soil temperature the layer and its thermal properties areincluded in the vertical heat transfer scheme of JSBACH Theoriginal heat transfer scheme for permafrost soil which isdescribed in detail in Ekici et al (2014) already contains anorganic layer which represents bryophyte and lichen groundcover This layer however has constant thermal properties

since it does not consider the water content of bryophytesand lichens The thermal conductivity of this layer is set to025 [W Kminus1 mminus1] and its heat capacity is set to 25times 106

[J mminus3 Kminus1] These values are shown in Fig 2 for compari-son The constant layer is replaced by the new layer whichexplicitly simulates water and ice content of bryophytes andlichens and the associated dynamic thermal conductivity andheat capacity (Fig 2)

The heat transfer scheme subdivides the soil column intoseveral layers with the bryophyte and lichen layer on topAdditionally a set of snow layers is simulated above theother layers in case snow is present Thermal conductivityand heat capacity of each layer are then used to determine thevertical temperature profile by solving the equation for heatconduction for all layers Thereby the scheme determines thetemperature of a layer in a given grid cell by first calculatingthe temperature profile for all tiles of the grid cell separatelyand then averaging the temperatures in each layer weightedby the area fraction of the tiles The lower boundary condi-tion of the scheme is set by assuming zero heat flux at thebottom of the soil column The upper boundary condition issurface temperature which is calculated from the surface en-ergy balance using radiative forcing and the ground heat fluxfrom the previous time step The scheme also considers theinfluence of phase change of water on the temperature of alayer thereby allowing it to compute freezing and melting ofwater in the bryophyte and lichen layer

Another important difference between the old and the newversion of the heat transfer scheme is dynamic surface cov-erage of bryophytes and lichens While the surface coverageof the old organic layer is 100 everywhere the coverageof the new bryophyte and lichen layer varies between 0 and100 between the grid cells of JSBACH (Eq 1) This meansthat each tile in a grid cell has a part where the bryophyte andlichen layer is present and another part where the layer is ab-sent The vertical heat transfer scheme however requires aconstant number of layers for each tile Hence in the newversion of the scheme the temperature profile is calculatedtwice for each tile once for all layers including the bryophyteand lichen layer and again leaving out the bryophyte andlichen layer Subsequently the average temperature of eachlayer of the respective tile is obtained by weighting the twoprofiles with their associated surface coverage However thetwo profiles have a different number of layers not only due tothe bryophyte and lichen layer but also because the numberof snow layers may differ between the parts Consequentlywe fill up the ldquoemptyrdquo layers on top by using the surface tem-perature for the averaging procedure according to Fig 3

We implemented two further changes to the heat transferscheme described in Ekici et al (2014) we increased thenumber of soil layers from five to seven thereby extendingthe layer boundaries at 65 32 123 413 and 983 cm depthby boundaries at 2304 and 5318 m depth This ensures thatbiases due to the zero heat flux lower boundary conditionsare negligible and temperature of the lowest layer is not fluc-

Surface

Bryophytesamp lichens

Figure 3 Averaging scheme for the vertical temperature profile ofa tile the left column corresponds to the part of the tile where nobryophyte and lichen layer is present the right column correspondsto the part covered by bryophytes and lichens The columns consistof seven soil layers (brown) an optional bryophyte and lichen layer(green) up to five snow layers (grey) and several optional emptylayers (white) The black horizontal bars connect layers which areaveraged with regard to their temperature The surface fractions ofthe two parts are used as weights for the averaging The empty lay-ers are assumed to have surface temperature

tuating after spin-up Moreover we complement the snowscheme by dynamic formulations of snow density (Verseghy1991) and thermal conductivity (Goodrich 1982) which takeinto account the effect of changes in snow density with timeon the thermal conductivity of the snow layer By consid-ering the dynamic nature of snow thermal conductivity wemake the snow layers consistent with the other dynamic lay-ers of the vertical heat transfer scheme in JSBACH Note thatcompaction of the bryophyte and lichen layer by snow is notconsidered here since the effect of this process on overallthermal conductivity is probably small

22 Model set-up

The advanced JSBACH model is forced by observation-based gridded climate fields at 05 pixel size and dailyresolution during the period 1901ndash2010 following Beeret al (2014) The grid cells have a size of approximately30times 30 km at 60 N and they are divided into four tiles ac-cording to the four most dominant plant functional types ofthis grid cell This vegetation coverage is assumed to stayconstant over the time of simulation In the model simula-tions used in this study we apply new soil parameters Hy-drological parameters have been assigned to each soil textureclass following Hagemann and Stacke (2015) according tothe percentage of sand silt and clay at 1 km spatial resolutionas indicated by the Harmonized World Soil Database (FAOIIASA ISRIC ISS-CAS and JRC 2009) Thermal parame-ters have been estimated as in Ekici et al (2014) at the 1 kmspatial resolution Then averages of 05 grid cells have beencalculated Soil depth until bedrock follows the map used inCarvalhais et al (2014) based on Webb et al (2000) Thetime step of JSBACH is set to 30 min In our simulation weonly consider the region north of 50 N since most of the

minus9

[g C mminus2 aminus1]

Surface coverage

minus9

[m2 mminus2]

Figure 4 Large-scale patterns of (a) NPP and (b) surface coverage of bryophytes and lichens simulated in JSBACH The values in the gridcells are calculated by averaging the last 15 years of a 1901 to 2010 dynamic simulation g C stands for ldquogram carbonrdquo

worldrsquos permafrost soils are located in this region and it isthus possible to save a substantial amount of computationtime

We use a stepwise approach to run JSBACH into a steadystate before we continue the simulation with a transient runleading to current climatic conditions First a 50-year hy-drothermal spin-up is performed where freezing and thaw-ing of water is switched off so that only liquid soil wateroccurs and develops into a steady state with climate Thisis done to prevent water in the deep soil layers from freez-ing before the water content had time to deviate from theinitial conditions since the initial values may not necessar-ily be realistic Since we do not want to include the influ-ences of a climate change in the spin-up these first 50 yearsare run with WATCH climate data randomly selected fromthe time period 1901 to 1930 Atmospheric CO2 is set to apreindustrial value of 285 ppmv Subsequently the spin-uprun is continued with a 100-year simulation which is usedto generate a steady state of soil ice water and temperatureTherefore freezingthawing is switched on this time Againthis run uses random climate data from 1901 to 1930 andpreindustrial CO2 The hydrothermal spin-up is followed bya 5000-year simulation of the CBALANCE model whichis a simplified version of JSBACH that simulates only theslow carbon pools of soil and vegetation as a function ofthe NPP of the vegetation By using the NPP output of thelast 30 years of the hydrothermal spin-up and repeating it for5000 years in CBALANCE soil carbon is run into a steadystate After that the simulation is continued with a 50-yearrun from 1851 to 1900 where the climate data are againrandomly selected from the 1901 to 1930 period but atmo-spheric CO2 content is transient Finally the simulation iscompleted by a fully transient run from 1901 to 2010 whereclimate data are taken from Beer et al (2014) The model isforced with atmospheric CO2 concentrations during 1851ndash2010 following Meinshausen et al (2011)

We perform spin-up and transient runs for three dif-ferent configurations of the new JSBACH model version

(a) the standard configuration which includes the process-based bryophyte and lichen layer (b) a configuration inwhich the bryophyte and lichen layer has 100 surface cov-erage and constant thermal properties corresponding to theorganic layer in the old permafrost version of JSBACH and(c) a configuration where no bryophyte and lichen layeris simulated These three simulations will be referred to asldquoDynamicrdquo ldquoConstantrdquo and ldquoWithoutrdquo throughout the textBy comparing the three simulations we can assess the im-pact of the dynamic coverage and thermal properties of thebryophyte and lichen layer on soil temperature

To assess how uncertainty in the parameterisation of thebryophyte and lichen PFT affects our estimated difference insoil temperature between the Dynamic and Without simula-tions we run a sensitivity analysis We test several bryophyteand lichen parameters which can affect productivity andconsequently surface coverage and the associated thermalproperties of the bryophyte and lichen layer A detailed de-scription of the sensitivity analysis can be found in Sect Ain the Appendix

To evaluate our modelling approach we compare simu-lated bryophyte and lichen surface coverage and NPP av-eraged over the study region to field measurements Sincelarge-scale observations are not available to our knowledgewe estimate characteristic values of surface coverage andNPP for the study region based on small-scale measurements(see Sect B in the Appendix for details)

3 Results

Effects of bryophytes and lichens

Large-scale patterns of net primary productivity (NPP) andsurface coverage of the bryophyte and lichen ground coversimulated in JSBACH are shown in Fig 4 The maps arebased on average values over the last 15 years of a transientDynamic JSBACH simulation from 1901 to 2010 where dy-

namic thermal properties of the bryophyte and lichen layerare taken into account (see Sect 22 for details)

The spatial pattern of NPP by bryophytes and lichens inJSBACH (Fig 4a) can be structured into areas with highproductivity of over 100 g mminus2 aminus1 of carbon areas of in-termediate productivity of 40 to 100 g mminus2 aminus1 and areasof low productivity of less than 40 g mminus2 aminus1 It should benoted however that also in areas of low productivity NPPusually exceeds 20 g mminus2 aminus1 Highly productive areas arefound in northern Scandinavia the northern European part ofRussia north-western Siberia Kamchatka parts of westernAlaska and parts of eastern Canada Areas of intermediateproductivity extensively surround the highly productive onesin general Less productive areas are found in the southernparts of Russia in middle and eastern Siberia and in parts ofcentral Canada This spatial pattern may be explained by acombination of rainfall and available photosynthetically ac-tive radiation (PAR) on the ground (see Fig B2 in the Ap-pendix) regions such as northern Scandinavia north-westernRussia Kamchatka Alaska and eastern Canada exhibit rela-tively high precipitation but simulated above-ground vegeta-tion is sparse enough to allow for moderate levels of light onthe ground This promotes high productivity of bryophytesand lichens Southern Scandinavia parts of eastern EuropeanRussia or the region south of the Hudson Bay however showrelatively high precipitation together with little availability ofPAR on the ground which limits productivity Large regionsin northern Canada and eastern Siberia in turn exhibit a highPAR availability on the ground due to sparse vegetation butprecipitation is very low there which again limits bryophyteand lichen growth

The bryophyte and lichen layer is excluded fromJSBACH-tiles containing cropland (Sect 21) Consequentlyproductivity of bryophytes and lichens is strongly limitedby available area in regions with extensive agriculture suchas southern Russia and southern Canada This explains whyNPP is low in these regions in spite of favourable climaticconditions for growth

The spatial distribution of simulated bryophyte and lichensurface coverage (Fig 4b) correlates in general with the spa-tial pattern of NPP This is due to the uniform disturbanceinterval of 100 years prescribed in our simulation for thewhole study region With constant disturbance variation inthe growth of the bryophyte and lichen cover results onlyfrom variation in the organismsrsquo NPP (see Eq 1) Howeversome areas in Russia and Canada show a high surface cov-erage in spite of only intermediate NPP whereas other areasshow less coverage for even higher values of NPP This isdue to surface coverage being plotted on a grid-scale basisIf a significant part of the grid cell is covered by lakes orglaciers coverage cannot reach 100 although NPP mightbe very high

Figure 5 shows the effect of the bryophyte and lichen layeron soil temperature averaged over the whole study regionbut limited to the biomes boreal forest and tundra as defined

by Olson et al (2001) (see also Fig B2c) We compare three1901 to 2010 transient simulations with differing propertiesthe bryophyte and lichen layer with dynamic surface cov-erage and dynamic thermal properties (Dynamic) an organiclayer with constant uniform coverage and thermal properties(Constant) and a set-up with no bryophyte and lichen layer(Without see Sect 22 for details) The annual mean tem-perature of the uppermost soil layer is shown in Fig 5a forthe last 15 years of the simulations The dynamic bryophyteand lichen layer leads to a considerable decrease in topsoiltemperature of 27 K in the model (Table 1) compared to thesimulation without bryophyte and lichen layer The effect ofthe organic layer with constant coverage and thermal prop-erties is much smaller only around 05 K compared to theWithout simulation

In Fig 5b the annual amplitude of topsoil temperature isshown averaged over the last 15 years of the simulationsThe annual decreasing effect of the bryophyte and lichenlayer on soil temperature (Fig 5a) results from a strong de-crease in summer that overrules the slight increase in soiltemperature in winter Both the dynamic and the constantbryophyte and lichen layer show this pattern The organiclayer with constant properties however should actually haveno overall effect on annual soil temperature since it damp-ens the heat flux between atmosphere and soil uniformlythroughout the year Hence to explain this outcome the ad-ditional effect of the snow layer on vertical heat transfer inJSBACH needs to be considered in winter the snow layerlimits heat transfer already to such an extent that the furtherdampening effect of the organic layer is secondary Interest-ingly the dampening effect of the dynamic bryophyte andlichen layer is not uniform throughout the year In summerthe dynamic layer is a more efficient insulator than the con-stant layer which can be explained by the additional reduc-tion in thermal conductivity due to low moisture content Inwinter however the dynamic layer is less efficient in insu-lating the soil than the constant one due to the occurrenceof ice inside the layer which strongly increases its thermalconductivity (see also Fig 2)

Figure 5c and d display the vertical soil temperature pro-files for July and January for the seven soil layers simulatedin JSBACH In summer the insulating effect of the bryophyteand lichen layer leads to a strong decrease in soil temperaturein the topsoil which becomes less pronounced with increas-ing soil depth In winter insulation by the bryophyte andlichen layer results in warmer topsoil temperatures Howeverbelow 1 m depth the seasonal dampening effect is not visibleanymore and the overall decreasing effect of the bryophyteand lichen layer on soil temperature prevails

In Fig 6a and b annual average topsoil temperature andactive layer thickness estimated by JSBACH are shown av-eraged for the years 1996 to 2010 of the Dynamic simulationSoil temperature decreases with increasing latitude and thelowest values outside of Greenland are found in north-easternSiberia particularly in the eastern Siberian mountains The

Table 1 Net primary productivity (NPP) and surface coverage of the simulated dynamic bryophyte and lichen layer as well as differencesbetween the Dynamic and Without simulations with regard to topsoil temperature and active layer thickness The values are based on the last15 years of 1901 to 2010 transient simulations and the region considered is limited to boreal forest and tundra g C stands for ldquogram carbonrdquo

Variable Average Range Unit

Bryophyte and lichen NPP 49 0ndash192 g C mminus2 aminus1

Bryophyte and lichen surface coverage 069 0ndash10 m2 mminus2

Difference in soil temperature minus27 0ndashminus57 KDifference in active layer thickness minus107 0ndashminus224 cm

1996 1998 2000 2002 2004 2006 2008 2010

(a)Without bryophyteslichensConstantDynamic

2 4 6 8 10 12

Months

(b)Without bryophyteslichensConstantDynamic

100-30 -25 -20 -15 -10 -5 0

Soil temperature January [degC]

Without bryophyteslichensConstantDynamic

100-15 -10 -5 0 5 10 15

Soil temperature July [degC]

(d)Without bryophyteslichensConstantDynamic

Figure 5 Temperature of the uppermost soil layer (a b) and the whole simulated soil column (c d) averaged over the study region for thebiomes boreal forest and tundra The dots stand for (a) annual average values (b) monthly average values or (c d) values at the mid-depthof a soil layer in JSBACH The mid-depths of the layers are at 003 019 078 268 698 164 and 3811 m depth Without denotes atransient simulation from 1901 to 2010 without a bryophyte and lichen layer Constant denotes an equivalent simulation but including anorganic layer with constant coverage and thermal properties and Dynamic stands for a bryophyte and lichen layer with dynamic coverageand thermal properties

spatial pattern of active layer thickness is similar to soil tem-perature Large parts of north-eastern Siberia and northernCanada exhibit shallow active layer thickness of less than 1 mwhile the regions adjacent to the south show a deeper activelayer of up to 3 m By definition we did not include regionswith more then 3 m of active layer thickness in our analy-sis of permafrost soil These regions are coloured in white inFig 6b

Figure 6c and d show the large-scale pattern of differencesin topsoil temperature and active layer thickness betweenthe Dynamic and the Without simulation The general pat-tern of soil temperature difference is similar to surface cov-erage which makes sense since it is the bryophyte and lichenground cover which reduces soil temperature However in

mountainous regions the reducing effect of the moss andlichen cover on soil temperature seems to be stronger thanin flat terrain indicating a non-linear response of the insu-lating effect to the climate forcing or to soil properties Thespatial differences in soil temperature span a range of 0 to57 C

The reduction in active layer thickness resulting from theinsulating effect of the bryophyte and lichen layer is sub-stantial particularly at the centre of the permafrost area inSiberia and northern Canada However also in the southernregions the reduction in soil temperature is large enough todecrease active layer thickness below the 3 m-threshold forpermafrost soil For this reason the areal extent of permafrostsoil in Fig 6b is considerably larger than the coloured area

Soil temperature (layer 1)

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

Active layer thickness

minus90˚

Soil temperature difference

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

Difference in active layer thickness

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

Figure 6 The maps display the spatial patterns of (a) topsoil temperature (b) active layer thickness (c) the temperature difference in theuppermost soil layer and (d) the difference in active layer thickness based on average values over the last 15 years of transient simulationsfrom 1901 to 2010 The differences are calculated by subtracting output of the Without simulation with no bryophyte and lichen layer fromthe Dynamic one

in Fig 6d Estimated permafrost area as defined by an activelayer thickness of less than 3 m increases due to the insulat-ing effect of the bryophyte and lichen layer by 32 from148 to 195 million km2 This new result is much more com-parable to the 22 million km2 permafrost area reported for theNorthern Hemisphere (French 2007 p 95) On average ac-tive layer thickness is reduced by 107 cm for the study regionwith a spatial range of 0 to 224 cm

Average values over the study region are listed in Table 1They show a significant impact of bryophytes and lichenson soil temperature and active layer thickness with a largespatial variation

The outcomes of the sensitivity analysis are shown in Ta-ble A1 in Sect A in the Appendix The general result is thatthe impact of the simulated bryophyte and lichen layer onthe temperature difference between the Dynamic and With-out simulations is robust under a large range of parametervalues of the bryophyte and lichen PFT Moreover if severalparameters are varied at the same time their combined effecton the soil temperature difference is less than the sum of allindividual effects

The results of the model evaluation are shown in Sect Bin the Appendix Field measurements of NPP (Table B1)

show a large variation but the characteristic value of NPPfor the study region is likely higher than 10 and lower than100 g mminus2 aminus1 of carbon Hence our JSBACH estimate of49 g mminus2 aminus1 of carbon (Table 1) compares reasonably wellwith the values from the field Observations of surface cover-age of bryophytes and lichens for the study region also showsome variation (Table B2) but most likely the characteris-tic value lies in the range of 06 to 08 This suggests thatour estimated average surface coverage of 069 is realistic Acomparison of simulated subsoil temperature and active layerthickness to observation-based maps for the region of Yaku-tia shows that JSBACH still underestimates subsoil tempera-ture as previous model versions did This finding is expectedsince the bryophyte and lichen layer in JSBACH reduces soiltemperature and consequently cannot reduce the cold biasThe estimate of active layer thickness however is improvedin general since this variable was overestimated in the previ-ous model version The corresponding maps for Yakutia areshown in Fig B1

4 Discussion

In this study we quantified the reducing effect of thebryophyte and lichen ground cover on soil temperature athigh latitudes at the large scale For this purpose we esti-mated dynamic surface coverage and thermal properties ofbryophytes and lichens with a process-based model and weintegrated this model into the global land surface model JS-BACH

We estimated an average decrease in temperature of theuppermost soil layer of 27 K with a spatial range of 0 to57 K This is a substantial effect it has a similar size as theprojected increase in global near-surface air temperature un-der the RCP45 warming scenario of the IntergovernmentalPanel on Climate Change (Pachauri et al 2014) Our find-ing is consistent with various field experiments as well asmodelling studies at site level which confirm the importantrole of bryophytes and lichens for reducing heat exchange be-tween atmosphere and soil at high latitudes (Beringer et al2001 Gornall et al 2007 Jorgenson et al 2010) Our re-sults suggest that the insulating effect of the bryophyte andlichen ground cover should be taken into account in large-scale modelling studies which focus on feedbacks betweenpermafrost soil and atmospheric CO2 under climatic change

Moreover we showed that representing the dynamics ofboth surface coverage and thermal properties of bryophytesand lichens is crucial for estimating their insulating effectUsing a simple organic layer with constant coverage and ne-glecting the influence of water or ice content on thermal con-ductivity and heat capacity of the layer likely results in un-derestimating the decrease in soil temperature

Alternatively to introducing a bryophyte and lichen layerwith dynamic thermal properties one might ask if calibrat-ing a constant organic layer would also be sufficient to sim-ulate the insulating effect In Fig 7 we show that there arequalitative differences between varying thermal conductiv-ity of a constant organic layer and the dynamic bryophyteand lichen layer if constant thermal conductivity is cali-brated to reproduce soil temperature under a bryophyte andlichen layer in summer the temperature in winter (and con-sequently annual average temperature) is too warm To sim-ulate an average annual soil temperature similar to that underthe bryophyte and lichen layer the constant thermal conduc-tivity has to be strongly reduced since the warming effect inwinter has to be compensated This leads to an unrealisticallyweak seasonal amplitude of soil temperature

We want to point out that the bryophyte and lichen layershould not be confused with an organic soil layer from amodelling perspective So far JSBACH does not include anorganic soil layer Differences between these two types oflayers are the relatively low thickness of the bryophyte andlichen layer compared to the organic layer which can beseveral tens of centimetres thick due to the accumulation ofpeat This means that the organic layer may have a signifi-cantly higher heat capacity when saturated with water or ice

0 2 4 6 8 10 12

Months

Without bryophyteslichensConstant 025Constant 008Constant 003Dynamic

Figure 7 Temperature of the uppermost soil layer of a model gridcell at 5530prime N 9830primeW which is also used for the sensitivityanalysis (Sect A in the Appendix) The dots stand for monthly aver-age values for the period 1996 to 2010 Without denotes a transientsimulation from 1901 to 2010 without a bryophyte and lichen layerldquoConstant 025rdquo denotes an equivalent simulation but including anorganic layer with constant coverage and thermal properties Thethermal conductivity of this layer is set to 025 [W Kminus1 mminus1] as inEkici et al (2014) ldquoConstant 008rdquo corresponds to an organic layercalibrated to reproduce topsoil temperature under the bryophyte andlichen layer in summer ldquoConstant 003rdquo corresponds to an organiclayer calibrated to reproduce average annual topsoil temperature un-der the bryophyte and lichen layer Dynamic stands for a bryophyteand lichen layer with dynamic coverage and thermal properties

Therefore the reducing effect of high thermal conductivityon insulation can be counteracted by the large amount of en-ergy required for a temperature change of the layer whichincreases insulation Furthermore the higher residual wa-ter content of the peat layer compared to the bryophyte andlichen layer may result in a lower variability of thermal con-ductivity These two differences may explain why eg Atch-ley et al (2016) find a relatively small influence of water con-tent on the insulating effect at the surface since their modelapproach includes an organic layer

Although the focus of our study is on soil temperature theprocess-based bryophyte and lichen scheme in JSBACH alsoprovides an estimate of the organismsrsquo net primary produc-tivity (NPP) The average simulated NPP of 49 g mminus2 aminus1 ofcarbon for the biomes boreal forest and tundra correspondsto roughly 10 of average boreal forest NPP (Gower et al2001) This is a lower fraction than stated in the study ofTuretsky et al (2010) which estimate a contribution of 14 to58 of moss to total ecosystem NPP for various boreal for-est sites The study by Goulden and Crill (1997) howeverestimates a lower contribution of around 10 by mossesto black spruce forest NPP The variation in the fraction offorest NPP attributed to bryophytes and lichens can be ex-plained by differences in hydrological conditions and vege-tation structure between the sites High values of productivityare mainly found on wetland sites where high water and lightavailability sustain a productive Sphagnum-cover JSBACH

however is mainly designed to simulate upland soils of theboreal forest Wetlands are not yet included consequentlythe high productivity of mosses in these areas is not reflectedin our estimate Implementing a scheme for wetland hydrol-ogy and the associated additional supply of water for mossesfrom below would be a useful extension of JSBACH in thefuture

The size of the insulating effect depended on the relationsbetween the thermal properties of the bryophyte and lichencover and its water or ice content These relations are well es-tablished by field measurements and theory (see Fig 2) Fur-thermore our estimate of soil temperature was sensitive tothe surface coverage of bryophytes and lichens which in ourmodel largely depends on the simulated NPP of the organ-isms To assess how well JSBACH is able to represent cur-rent bryophyte and lichen NPP and surface coverage at highlatitudes we compared our estimates to field measurementsHowever due to large variation in the field studies and a lackin up-scaled estimates it was difficult to define characteris-tic values of NPP and surface coverage for the study region(see Tables B1 and B2) Given the considerable uncertaintyin the measurements our estimates of bryophyte and lichenNPP and surface coverage agreed well with field observa-tions The enhancement of JSBACH due to representationof near-surface vegetation state and functions will improvethe reliability of future projections of northern ecosystem re-sponses to environmental change as well as climate-carboncycle feedbacks in future studies

We compared modelled subsoil temperature and activelayer thickness to observations for the region of Yakutia Ourestimated active layer thickness matches reasonably well toobservations and it is improved over the previous version ofJSBACH (Ekici et al 2014) This suggests that the dynamicbryophyte and lichen layer leads to a more realistic repre-sentation of vertical heat transfer in the model JSBACH stillunderestimates subsoil temperature compared to the previousmodel version (Ekici et al 2014) The cold bias is most pro-nounced for the eastern Siberian mountains where bryophyteand lichen cover is relatively low (Fig 4b) Hence it seemslikely that the reason for the cold bias is not directly relatedto the bryophyte and lichen layer A probable explanation forthe cold bias is the relatively strong sensitivity of annual av-erage soil temperature to the parameterisation of the snowlayer scheme in JSBACH Improvements in the representa-tion of snow may improve the simulated soil temperature

Our modelling approach is designed to capture the mostimportant properties of the bryophyte and lichen groundcover which influence heat exchange between atmosphereand soil These properties include the dependency ofbryophyte and lichen heat conduction on moisture and icecontent as well as the relation of productivity and conse-quently surface coverage to various environmental factorsMoreover indirect effects of the bryophyte and lichen coveron heat exchange are represented such as cooling of the sur-face due to evaporation from the organismsrsquo thalli However

there are several other relations between the bryophyte andlichen cover and the environment that may affect heat ex-change at the surface which we did not consider in this studyfor simplicity It has been shown by Bernier et al (2011) forinstance that conversion of black spruce forests into lichenwoodlands in Canada results in atmospheric cooling due tothe higher albedo of lichens compared to forest Effects oflichens and bryophytes on albedo and consequently on sur-face temperature have also been shown by Stoy et al (2012)who discuss species-specific differences in these effects InJSBACH we did not include the effect of bryophytes andlichens on surface albedo since the organisms vary consid-erably in their colour Instead of assigning an arbitrary valuefor albedo to the bryophyte and lichen PFT in JSBACH wethus assume that the albedo of the PFT is similar to the soilalbedo The evaporative cooling effect of the bryophyte andlichen cover on surface temperature may be modulated bywater uptake from Larch trees which are able to root intothe near-surface vegetation layer

The bryophyte and lichen layer is represented by one sin-gle PFT in JSBACH for reasons of consistency with vascu-lar vegetation in JSBACH and for computational efficiencyHowever this lack of diversity may have consequences forthe estimated effect of the bryophyte and lichen layer on soiltemperature Given similar climatic conditions bryophyteand lichen species may differ in their degree of water satura-tion and consequently in their thermal properties Lichensfor instance have in many cases a lower porosity and wa-ter storage capacity than bryophytes particularly mosses(Schulze and Caldwell 1995) This means that lichens maybe less sensitive to water content than bryophytes regard-ing their thermal properties However both groups includespecies which do not match this pattern If large regions dif-fer in their dominant bryophyte or lichen species this mayaffect our estimated patterns of soil temperature and activelayer thickness Additionally bryophyte and lichen speciesdiffer in their thickness while thickness within a species isrelatively constant (Soudzilovskaia et al 2013) Since thebryophyte and lichen PFT in JSBACH has a constant thick-ness we may underestimate further spatial effects on the soilthermal regime

Another simplifying assumption in JSBACH is the uni-form disturbance interval of 100 years for the whole studyregion This value represents the average fire return intervalin the boreal forest where fire is the dominant process fordisturbance (Mouillot and Field 2005) The occurrence offire however may vary between regions Fires may be muchless frequent in parts of northern Canada and north-easternSiberia for instance (Bonan and Shugart 1989) which couldlead to larger spatial differences in simulated bryophyte andlichen surface coverage Under climatic change shifts in veg-etation distribution and increased temperatures may lead tochanges in the fire interval at high latitudes Consequentlyincluding a more dynamic representation of disturbance inour approach would be particularly beneficial for modelling

future bryophyte and lichen cover under scenarios of climatechange

5 Conclusions

Here we present an new version of the global land surfacemodel JSBACH that estimates NPP surface coverage and dy-namic thermal properties of bryophytes and lichens througha process-based scheme We apply JSBACH to quantify theimpact of the bryophyte and lichen ground cover on the soilthermal regime at high latitudes Thereby we estimate a con-siderable average cooling effect of the bryophyte and lichencover of 27 K (0 to 57 K) for the uppermost soil layerFurthermore we find that the strength of the cooling effectlargely depends on an accurate representation of dynamiccoverage and thermal properties of bryophytes and lichens

These results suggest that the reducing effect of thebryophyte and lichen ground cover on soil temperatureshould be accounted for in studies which aim at quantifyingfeedbacks between permafrost soil temperature and atmo-spheric CO2 due to climate change Since our process-basedapproach also allows for predicting future bryophyte andlichen surface coverage at high-latitude ecosystems a poten-tial next step is to simulate the future impact of bryophytesand lichens on active layer thickness and permafrost extentunder a transient scenario of climate change

6 Data availability

The land surface model JSBACH used in this study is in-tellectual property of the Max Planck Society for the Ad-vancement of Science Germany The JSBACH source codeis distributed under the Software License Agreement of theMax Planck Institute for Meteorology and it can be ac-cessed on personal request The steps to gain access are ex-plained under the following link httpwwwmpimetmpgdeensciencemodelslicense

The climatic fields used in this study as forcing data forthe JSBACH model are available upon registration underthe following link (the tag ldquoGeocarbonrdquo has to be selected)httpswwwbgc-jenampgdegeodbprojectsHomephp

The map of soil temperature and active layer thick-ness for the region of Yakutia which is used as a part ofour model evaluation is available under the following linkdoi101594PANGAEA808240

JSBACH output data which are presented as maps in thisstudy are available as netCDF files from the authors on re-quest

Appendix A Sensitivity analysis

We perform a sensitivity analysis to assess how uncertaintyin the parameter values chosen for the bryophyte and lichenPFT affects our estimates In the following we describe sev-eral parameters which are characteristic for the bryophyteand lichen PFT and which may affect NPP surface coverageor thermal properties of the PFT

ndash The parameter porosity ε of the bryophyte and lichensurface layer is set to 80 in the model This value isvaried by around 20 in each direction from a min-imum value of 65 to a maximum of 95 Poros-ity shows large natural variation in bryophytes andlichens from 10 in some lichen species (Valladareset al 1993) to 99 in some moss species (OrsquoDonnellet al 2009) We are interested however in uncertaintyconcerning the average porosity of the bryophyte andlichen ground cover for large regions Hence we do notvary ε for the full range since it is unlikely that the av-erage porosity is close to an extreme value

ndash The thickness z of the simulated bryophyte and lichenlayer is set to 45 cm in the model It is defined as theundecomposed living and dead parts of a bryophyte orlichen mat Observed values of thallus thickness show alarge variation between species ranging from less thana millimetre to several tens of centimetres (Nash III1996 Bell and Hemsley 2011) Here we vary z from1 to 10 cm which is a slightly larger range than deter-mined by Soudzilovskaia et al (2013) for 18 bryophytespecies We do not test extreme values of z since we areinterested in bryophyte and lichen species that are actu-ally able to form a macroscopic ground cover We donot for instance consider flat lichen crusts on bouldersurfaces since they are not likely to play a significantrole for large-scale heat exchange between soil and at-mosphere

ndash The diffusivity of the water-saturated bryophyte orlichen thallus for CO2 DCO2sat affects CO2-uptakeand consequently NPP and surface coverage ofbryophytes and lichens in the model We set DCO2satto a value of 001 [mol mminus2 sminus1] based on Williams andFlanagan (1998) and vary this value by multiplying itby the factors 05 and 20 We choose this form of varia-tion since DCO2sat shows relatively large natural varia-tion from around 5times 10minus4 to 2times 10minus2 [mol mminus2 sminus1](Williams and Flanagan 1998 Cowan et al 1992)consequently a linear variation would not be adequate(Porada et al 2013) It should be noted that varia-tion in DCO2sat represents an extension to the originalbryophyte and lichen model described in Porada et al(2013)

ndash The model parameter critical water saturation2crit de-termines which value of saturation is necessary for the

bryophyte and lichen layer to reach full metabolic activ-ity The increase from zero saturation and no activity to2crit is assumed to be linear (Porada et al 2013) Here2crit is set to 30 which corresponds to the lower endof the range of possible values (Porada et al 2013)and reflects the relatively fast activation of common bo-real forest floor mosses (Williams and Flanagan 1998)Hence we vary 2crit by setting it closer to the upperbound to a value of 60

ndash The specific maintenance respiration rate Rmain variesover 3 orders of magnitude between different bryophyteand lichen species (Porada et al 2013) For Li-Bry in JSBACH we chose an intermediate value of15times 10minus6 [mol mminus2 sminus1] and due to the large rangeof possible values we vary Rmain by multiplication bythe factors 05 and 20 In the model Rmain is relatedto photosynthetic capacity and biomass turnover rate ofthe bryophyte and lichen layer via the parameter ldquoRatioof Rubisco content to maintenance respirationrdquo 8RRwhich describes the tradeoff between photosynthetic ca-pacity and respiration (Porada et al 2013) LiBry hasbeen shown to be sensitive to this parameter so we varythe standard value of 5 [s] by 20 in each direction

ndash In addition to photosynthetic capacity LiBry contains4 parameters to calculate photosynthesis which varybetween different species (Porada et al 2013) the en-zyme activation energies EaKc and EaKo which con-trol the temperature response of the MichaelisndashMentenconstants of the carboxylation and oxygenation reac-tions of Rubisco and the molar carboxylation and oxy-genation rates of Rubisco VMC and VMO For LiBryin JSBACH EaKc is set to 120 000 [J molminus1] which isclose to the upper bound of observed values and EaKois set to 10 000 [J molminus1] which is a relatively lowvalue regarding the observations (Porada et al 2013)Hence we vary EaKc by setting it to the lower bound of30 000 [J molminus1] and EaKo by setting it close to the up-per bound of 55 000 [J molminus1] The molar carboxylationrate of Rubisco VMC varies by 2 orders of magnitude(Porada et al 2013) We therefore vary the intermediatevalue of 15 [sminus1] used here by multiplying it by the fac-tors 05 and 20 The molar oxygenation rate of RubiscoVMO is set here to 22 [sminus1] close to the observed max-imum value We vary it by setting it to 05 [sminus1] closeto the observed minimum

ndash Photosynthesis and respiration in LiBry are related tosurface temperature via the two parameters optimumtemperature of photosynthesis Topt and the Q10 valueof respiration Q10 Here Topt is set to 22 C and it isvaried in both directions by 5 C Q10 is set to 24 andit is varied by 03 in both directions

ndash Furthermore we test how sensitive the effect of thebryophyte and lichen layer on soil temperature reacts to

the disturbance interval τD set in the model We there-fore vary the standard value of 100 years for τD by mul-tiplying it with the factors 05 and 20

ndash Finally we vary the expansion efficiency ηE which isset to 085 by 20 in each direction

For each varied parameter we run a stepwise simulationconsisting of a steady-state spin-up and a transient run asdescribed above For reasons of computational speed we runthe model only on a single grid cell The location of thegrid cell is 5530prime N 9830primeW which roughly corresponds tothe northern study area of the BOREAS project (eg Goweret al 1997) We then compare the difference in soil temper-ature of the uppermost soil level between the standard sim-ulation and each of the simulations with varied parametersto quantify the impact of parameter uncertainty on our over-all estimates Moreover we run additional simulations withmore than one varied parameter to assess their combined ef-fect on our estimates

Many of the tested parameters have no strong effect onNPP and surface coverage of the simulated bryophyte andlichen layer Thus they do not affect the difference in topsoiltemperature resulting from the dynamic layer The follow-ing parameters however lead to changes in the temperaturedifference

ndash The porosity ε of the bryophyte and lichen layer affectsthe temperature difference in both directions Interest-ingly a reduction in porosity leads to a reduced cover-age in spite of slightly increased NPP The reason forthis is that lower porosity results in a smaller specificarea of the bryophyte and lichen PFT Since net growthof cover (Eq 1) is calculated by multiplying NPP withspecific area the effect of reduced specific area mayoverrule an increase in NPP This leads to less cover ex-pansion The inverse effect occurs for higher porosity

ndash The thickness z affects the temperature difference toa moderate extent The slight increase in coverage fora lower value of z mainly results from increased spe-cific area Comparing the parameters in Table A1 theslightly higher coverage cannot explain by itself theincrease in temperature difference It is probably thedecrease in heat capacity associated with a thinnerbryophyte and lichen layer which leads to a warmerand consequently drier layer in summer which acts asa more efficient insulator The opposite then happensfor a thicker layer This means that the reducing effectof low moisture content on thermal conductivity of thebryophyte and lichen layer overrules the increasing ef-fect of higher thickness in the model

ndash Reducing the expansion efficiency ηE also reduces sur-face coverage and consequently the soil temperaturedifference However the model is not very sensitive to

this parameter ηE would have to be quite low to signif-icantly affect our estimates

ndash Increasing critical water saturation 2crit has a relativelystrong decreasing effect on the temperature differencesince the associated slower activation of bryophyte andlichen photosynthesis significantly reduces NPP andsurface coverage

ndash The specific maintenance respiration rate Rmain has thestrongest influence on the temperature difference com-pared to the other parameters but only in one direc-tion A doubling of Rmain significantly reduces the tem-perature difference whereas halving Rmain does nothave any effect The reason for this is that our standardvalue of Rmain is close to optimal meaning that NPPdecreases in both directions The associated biomassturnover however increases linearly with Rmain and ithas a strong reducing impact on net growth and coverexpansion This means that low turnover compensatesfor low NPP at low Rmain but high turnover has an ad-ditional reducing impact at high Rmain The ldquoRatio ofRubisco content to maintenance respirationrdquo 8RR hasonly a minor impact on the temperature difference

ndash Halving the molar carboxylation rate of Rubisco VMCreduces significantly the temperature difference due tostrongly decreased NPP and coverage Doubling VMChowever does not have any effect on temperature due tothe non-linear response of coverage on increased NPPThe activation energy EaKc has a moderate reducingimpact on NPP coverage and thus on the temperaturedifference

ndash Reducing the optimum temperature of photosynthesisTopt by 5 C results in significantly lower NPP cover-age and temperature difference while increasing Topthas no significant effect The reason for the asymmet-ric response of NPP to Topt is the exponential depen-dence of respiration on the in this case mostly negativedifference between surface temperature and Topt SinceTopt is already relatively high a further increase leadsonly to a small reduction in respiration Lowering Topthowever results in a strong increase in respiration andconsequently reduced NPP

To assess the effect of combined varied parameters we se-lect four parameters which significantly affect the estimateddifference in topsoil temperature between the Dynamic andWithout simulations (see Table A1) The outcome of thisanalysis is shown in Fig A1 It can be seen that the combinedeffect of parameters is not additive but becomes weaker witheach additional parameter

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

Figure A1 Effect of model parameters on simulated difference intopsoil temperature between the Dynamic and Without simulationsThe individual effects of four varied parameters are shown whichsignificantly reduce the temperature difference These are the spe-cific maintenance respiration rate Rmain the optimum temperatureof photosynthesis Topt the molar carboxylation rate of RubiscoVMC and the porosity ε of the bryophyte and lichen layer (see Ta-ble A1 for details) Also shown are the combined effects of Rmainand Topt Rmain Topt and VMC as well as Rmain Topt VMC andε The effect on the temperature difference is shown in percentageof reduction

Table A1 Sensitivity of the estimated topsoil temperature difference due to the addition of a dynamic bryophyte and lichen layer to theparameterisation of the bryophyte and lichen PFT See Sect 22 for a description of the parameters 1T is the temperature difference in theuppermost soil layer between the Dynamic simulation and the Without one (Sect 22) NPP is in g mminus2 aminus1 of carbon 1P stands for achange in a model parameter either by multiplication with a factor (lowast symbol) percentage change or by setting the parameter to the upperboundary (ub) or lower boundary (lb) 1T 1P is the percentage change in temperature difference (1T ) due to a change in a modelparameter (1P ) compared to the control run The sensitivity analysis is performed on a single grid cell of JSBACH (see Sect 22 for details)

Parameter Value 1T Coverage NPP 1P 1T 1P

DCO2sat 0005 [mol mminus2 sminus1] minus27 081 308 05lowast 0

τD 200 [a] minus27 082 321 05lowast 0 τD 50 [a] minus27 082 321 2lowast 0 ηE 07 minus26 078 321 minus20 minus4 ηE 10 minus27 084 322 +20 0 ε 65 minus20 068 327 minus20 minus26 ε 95 minus29 086 302 +20 +7 EaKc 30 000 [J molminus1

] minus24 075 256 lb minus11 EaKo 50 000 [J molminus1] minus27 081 302 ub 0 Rmain 75times 10minus7

[mol mminus2 sminus1] minus27 080 256 05lowast 0

Rmain 30times 10minus6[mol mminus2 sminus1

] minus16 053 259 2lowast minus41 Q10 21 minus27 080 295 minus125 0 Q10 27 minus27 083 341 +125 0 8RR 40 minus26 080 263 minus20 minus4 8RR 60 minus27 083 367 +20 0 2crit 60 minus20 067 169 ub minus26 Topt 17 [C] minus22 070 243 minus5 K minus19 Topt 27 [C] minus27 083 324 +5 K 0 VMC 075 [sminus1

] minus20 066 141 05lowast minus26 VMC 30 [sminus1

] minus27 084 451 2lowast 0 VMO 05 [sminus1

] minus27 082 321 lb 0 z 10 mm minus30 083 368 lb +11 z 100 mm minus24 080 288 ub minus11

Control run ndash minus27 082 321 ndash ndash

Appendix B Model evaluation

To evaluate our modelling approach we compare the JS-BACH estimates of bryophyte and lichen surface coverageand NPP to field observations However to our knowledgethere are no field measurements of bryophyte and lichen sur-face coverage and NPP which cover the global scale There-fore we compile a list of available small-scale observationsfrom various studies which are largely taken from overviewsin Turetsky et al (2010) and Bona et al (2013) We do notattempt to create a comprehensive review of measurementsof surface coverage and NPP in high-latitude regions How-ever our list is sufficient to establish characteristic valuesof surface coverage and NPP for the region north of 50 NThereby we constrain our analysis to the biomes boreal for-est and tundra (after Olson et al 2001 see also Fig B2c)Since JSBACH is not primarily designed to simulate peat-lands we exclude studies which measure surface coverageand NPP in peatlands Several studies contain measurementsboth on peatland and upland sites In this case we do notconsider values from peatlands In many cases the originalstudies provide NPP measurements in grams of biomass Weconvert these values to grams of carbon using a factor of 045(Bauer et al 2009)

We compare the characteristic values of measured cover-age and NPP to the average bryophyte and lichen surfacecoverage and NPP simulated by JSBACH also constrainedto boreal forest and tundra Furthermore we compare sim-ulated soil temperature and active layer thickness to large-scale observation-based maps for the region of Yakutia

Table B1 shows studies which provide measurements ofbryophyte and lichen NPP for the biomes boreal forest andtundra To obtain a characteristic value for NPP we calculatethe median NPP of each study and from this set of values wetake again the median which is then shown in the last row ofTable B1 together with the range of the median values fromall studies Given the large variation in these median valuesof 2 orders of magnitude it is difficult to constrain NPP in thestudy region based on field measurements The average valueprobably lies somewhere between 10 and 100 g mminus2 aminus1 ofcarbon

In Table B2 several studies are listed which measure sur-face coverage of bryophytes and lichens for boreal forest andtundra As with NPP we show the range and median of allstudiesrsquo median values in the last row to obtain a charac-teristic surface coverage for the region The range of ob-served surface coverage is large but half of the (rounded)values lie between 06 and 08 We want to point out that thestudy by Rapalee et al (2001) provides a large-scale estimateof surface coverage based on remote sensing They coverthe whole BOREAS region which has an area of approxi-mately 500 000 km2 and arrive at a moss surface coverage of057 This value agrees well with our estimate given that theBOREAS region has a slightly lower than average simulatedbryophyte and lichen coverage (Fig 4)

Figure B1 compares the spatial patterns of subsoil tem-perature and active layer thickness estimated by JSBACH toobservations from Beer et al (2013) for the region of Yaku-tia JSBACH underestimates subsoil temperature to a similarextent as the version presented in Ekici et al (2014) Poten-tial reasons for the mismatch are discussed in Ekici et al(2014) and include a reduced spatial heterogeneity of the ob-servational map or biases in the representation of climatesoil properties or snow depth in JSBACH The estimate ofactive layer thickness however is improved in general com-pared to the previous version of JSBACH with the exceptionof the eastern Siberian mountain range The underestimationof active layer thickness there likely follows from the strongunderestimation of soil temperature in this area The overesti-mation of active layer thickness in the southern part of Yaku-tia may result from the patchy structure of the permafrostarea in these regions as discussed by Ekici et al (2014)

Table B1 Studies which measure NPP in the biomes boreal forest and tundra Median and range of all values listed in a study are shown stands for the number of values NPP is in g mminus2 aminus1 of carbon BOREAS N or S stands for the northern and southern site of the BOREASproject (eg Gower et al 1997) BOREAS N is located in Manitoba Canada and BOREAS S is located in Saskatchewan Canada The lastrow shows median and range of the values in the first column

Median Range Study Location

21 20ndash26 4 Camill et al (2001) BOREAS N35 0ndash144 7 Bond-Lamberty et al (2004) BOREAS N11 0ndash16 7 Mack et al (2008) central Alaska42 40ndash44 2 Oechel and Van Cleve (1986) central Alaska30 29ndash31 2 Ruess et al (2003) central Alaska14 13ndash17 4 Vogel et al (2008) AlaskaSaskatchewan22 12ndash60 3 Schuur et al (2007) central Alaska30 2ndash68 4 Shaver and Chapin III (1991) northern Alaska25 ndash 1 Bisbee et al (2001) BOREAS S12 ndash 1 Gower et al (1997) BOREAS NS104 ndash 1 Swanson and Flanagan (2001) BOREAS S13 5ndash20 4 Lange et al (1998) northern Alaska39 ndash 1 Oechel and Collins (1976) northern Alaska2 ndash 1 Uchida et al (2006) Svalbard7 ndash 1 Uchida et al (2002) Svalbard7 1ndash21 6 Koshurnikova (2007) western Siberia35 16ndash53 2 OrsquoConnell et al (2003) BOREAS S50 ndash 1 OrsquoDonnell et al (2011) central Alaska78 ndash 1 Frolking et al (1996) Manitoba93 ndash 1 Benscoter and Vitt (2007) Alberta60 ndash 1 Skre and Oechel (1979) central Alaska68 ndash 1 Hermle et al (2010) Quebec25 ndash 1 Trumbore and Harden (1997) BOREAS N

27 (2ndash104)

Table B2 Studies which measure surface coverage in the biomes boreal forest and tundra Median and range of all values listed in a studyare shown stands for the number of values The BOREAS N study area is located in Manitoba Canada and BOREAS S is located inSaskatchewan Canada (Gower et al 1997) The last row shows median and range of the values in the first column

08 ndash 1 Bisbee et al (2001) BOREAS S095 ndash 1 Swanson and Flanagan (2001) BOREAS S065 013ndash095 8 Bond-Lamberty and Gower (2007) BOREAS N093 077ndash099 13 Camill et al (2001) BOREAS N05 ndash 1 Street et al (2012) northern Sweden06 ndash 1 Kolari et al (2006) southern Finland05 ndash 1 Wahren et al (2005) Alaska057 ndash ndash Rapalee et al (2001) BOREAS region

063 (05ndash095)

Soil temperature (simulated)

60˚ 60˚

70˚ 70˚

minus250 minus225 minus200 minus175 minus150 minus125 minus100 minus75 minus50 minus25 00 25 50

[degC]

Active layer thickness (simulated)

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

Soil temperature (observed)

60˚ 60˚

70˚ 70˚

[degC]

Active layer thickness (observed)

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

Soil temperature (simulated ‒ observed)

60˚ 60˚

70˚ 70˚

minus150 minus125 minus100 minus75 minus50 minus25 00 25 50 75 100 125 150

[degC]

Active layer thickness (simulated ‒ observed)

60˚ 60˚

70˚ 70˚

minus15 minus12 minus09 minus06 minus03 00 03 06 09 12 15

Figure B1 The maps show the spatial patterns of (a) simulated subsoil temperature and (b) active layer thickness (c) observed subsoiltemperature and (d) active layer thickness as well as the differences between simulation and observations regarding (e) subsoil temperatureand (f) active layer thickness for the region of Yakutia The maps (a) and (b) are based on average values over the last 15 years of a transientDynamic simulation from 1901 to 2010 Subsoil temperature is calculated by averaging the temperatures of the third and fourth soil layer ofJSBACH The observation-based maps in (c) and (d) are taken from Beer et al (2013)

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

[micromol mminus2 sminus1]

Biome mask

minus90˚

Figure B2 Additional model output averaged over the last 15 years of a JSBACH transient simulation from 1901 to 2010 Spatial patternsof (a) precipitation and (b) photosynthetically active radiation (PAR) on the ground Note that PAR on ground is a variable of LiBry inJSBACH it is therefore only larger than zero where bryophytes and lichens exist in the model (c) shows the biome mask based on Olsonet al (2001) which is used to constrain our estimates to boreal forest and tundra

Author contributions Philipp Porada developed the model codeand prepared the manuscript with contributions from all co-authorsPhilipp Porada and C Beer designed the simulation set-up and per-formed the simulations

Acknowledgements This work has been supported by the PAGE21project grant agreement number 282700 funded by the EC seventhFramework Programme theme FP7-ENV-2011 and the CARBOP-ERM project grant agreement number 03G0836B funded by theBMBF (German Ministry for Science and Education) The authorsthank the Bolin Centre for Climate Research for financial supportStockholm University and the National Supercomputer Centreat Linkoumlping University provided computational resources Theauthors thank three anonymous referees for their thorough andhelpful reviews

Edited by C J WilsonReviewed by three anonymous referees

References

Atchley A Coon E Painter S Harp D and Wilson CInfluences and interactions of inundation peat and snow onactive layer thickness Geophys Res Lett 43 5116ndash5123doi1010022016GL068550 2016

Bauer I Bhatti J Swanston C Wieder R and Preston COrganic Matter Accumulation and Community Change at thePeatlandndashUpland Interface Inferences from 14C and 210PbDated Profiles Ecosystems 12 636ndash653 doi101007s10021-009-9248-2 2009

Beer C Lucht W Schmullius C and Shvidenko A Small netcarbon dioxide uptake by Russian forests during 1981ndash1999Geophys Res Lett 33 L15403 doi1010292006GL0269192006

Beer C Fedorov A and Torgovkin Y Permafrost temperatureand active-layer thickness of Yakutia with 05-degree spatial res-olution for model evaluation Earth Syst Sci Data 5 305ndash310doi105194essd-5-305-2013 2013

Beer C Weber U Tomelleri E Carvalhais N Mahecha Mand Reichstein M Harmonized European Long-Term ClimateData for Assessing the Effect of Changing Temporal Variabilityon LandndashAtmosphere CO2 Fluxes J Climate 27 4815ndash4834doi101175JCLI-D-13-005431 2014

Bell P and Hemsley A (Eds) Green Plants ndash Their Origin and Di-versity 2nd Edn Cambridge University Press New York 2011

Benscoter B and Vitt D Evaluating feathermoss growth achallenge to traditional methods and implications for the bo-real carbon budget J Ecol 95 151ndash158 doi101111j1365-2745200601180x 2007

Beringer J Lynch A Chapin F Mack M and BonanG The Representation of Arctic Soils in the Land SurfaceModel The Importance of Mosses J Climate 14 3324ndash3335doi1011751520-0442(2001)014lt3324TROASIgt20CO22001

Bernier P Desjardins R Karimi-Zindashty Y Worth DBeaudoin A Luo Y and Wang S Boreal lichen wood-lands A possible negative feedback to climate change in

eastern North America Agr Forest Meteorol 151 521ndash528doi101016jagrformet201012013 2011

Bisbee K Gower S Norman J and Nordheim E Environmen-tal controls on ground cover species composition and produc-tivity in a boreal black spruce forest Oecologia 129 261ndash270doi101007s004420100719 2001

Blok D Heijmans M Schaepman-Strub G van Ruijven J Par-mentier F Maximov T and Berendse F The Cooling Capac-ity of Mosses Controls on Water and Energy Fluxes in a SiberianTundra Site Ecosystems 14 1055ndash1065 doi101007s10021-011-9463-5 2011

Bohn K Dyke J Pavlick R Reineking B Reu B and Klei-don A The relative importance of seed competition resourcecompetition and perturbations on community structure Biogeo-sciences 8 1107ndash1120 doi105194bg-8-1107-2011 2011

Bona K Fyles J Shaw C and Kurz W Are Mosses Requiredto Accurately Predict Upland Black Spruce Forest Soil Carbon inNational-Scale Forest C Accounting Models Ecosystems 161071ndash1086 doi101007s10021-013-9668-x 2013

Bonan G A biophysical surface energy budget analysis of soiltemperature in the boreal forests of interior Alaska Water Re-sour Res 27 767ndash781 doi10102991WR00143 1991

Bonan G and Shugart H Environmental Factors and Ecologi-cal Processes in Boreal Forests Ann Rev Ecol Syst 20 1ndash281989

Bond-Lamberty B and Gower S Estimation of stand-levelleaf area for boreal bryophytes Oecologia 151 584ndash592doi101007s00442-006-0619-5 2007

Bond-Lamberty B Wang C and Gower S Net primary pro-duction and net ecosystem production of a boreal black sprucewildfire chronosequence Global Change Biol 10 473ndash487doi101111j1529-881720030742x 2004

Brovkin V Raddatz T Reick C Claussen M and Gayler VGlobal biogeophysical interactions between forest and climateGeophys Res Lett 36 L07405 doi1010292009GL0375432009

Camill P Lynch J Clark J Adams J and Jordan BChanges in Biomass Aboveground Net Primary Productionand Peat Accumulation following Permafrost Thaw in the Bo-real Peatlands of Manitoba Canada Ecosystems 4 461ndash478doi101007s10021-001-0022-3 2001

Carvalhais N Forkel M Khomik M Bellarby J Jung MMigliavacca M Mu M Saatchi S Santoro M ThurnerM Weber U Ahrens B Beer C Cescatti A RandersonJ and Reichstein M Global covariation of carbon turnovertimes with climate in terrestrial ecosystems Nature 514 213ndash217 doi101038nature13731 2014

Chadburn S Burke E Essery R Boike J Langer M Heiken-feld M Cox P and Friedlingstein P An improved representa-tion of physical permafrost dynamics in the JULES land-surfacemodel Geosci Model Dev 8 1493ndash1508 doi105194gmd-8-1493-2015 2015a

Chadburn S E Burke E J Essery R L H Boike J LangerM Heikenfeld M Cox P M and Friedlingstein P Impactof model developments on present and future simulations ofpermafrost in a global land-surface model The Cryosphere 91505ndash1521 doi105194tc-9-1505-2015 2015b

Cowan I Lange O and Green T Carbon-dioxide exchange inlichens Determination of transport and carboxylation character-istics Planta 187 282ndash294 doi101007BF00201952 1992

Ekici A Beer C Hagemann S Boike J Langer M and HauckC Simulating high-latitude permafrost regions by the JSBACHterrestrial ecosystem model Geosci Model Dev 7 631ndash647doi105194gmd-7-631-2014 2014

FAO IIASA ISRIC ISS-CAS and JRC Harmonized worldsoil database (version 11) FAO Rome Italy and IIASALaxenburg Austria httpwwwfaoorgnrlandsoilsharmonized-world-soildatabaseen (last access 10 Octo-ber 2011) 2009

Farquhar G and von Caemmerer S Modelling of Photosyn-thetic Response to Environmental Conditions in Encyclopediaof Plant Physiology vol 12B edited by Lange O Nobel POsmond C and Ziegler H Springer Heidelberg 1982

French H ed The periglacial environment 3rd Edn John Wi-ley amp Sons Chichester 2007

Frolking S Goulden M Wofsy S Fan S-M Sutton DMunger J Bazzaz A Daube B Crill P Aber J Band LX W Savage K Moore T and Harriss R Modelling tem-poral variability in the carbon balance of a sprucemoss borealforest Global Change Biol 2 343ndash366 1996

Goll D Brovkin Vand Liski J Raddatz T Thum T and Todd-Brown K Strong dependence of CO2 emissions from anthro-pogenic land cover change on initial land cover and soil car-bon parametrization Global Biogeochem Cy 29 1511ndash1523doi1010022014GB004988 2015

Goodrich L The influence of snow cover on the ground ther-mal regime Can Geotech J 19 421ndash432 doi101139t82-0471982

Gornall J Joacutensdoacutettir I Woodin S and Van der Wal R Arc-tic mosses govern below-ground environment and ecosystemprocesses Oecologia 153 931ndash941 doi101007s00442-007-0785-0 2007

Goulden M and Crill P Automated measurements of CO2 ex-change at the moss surface of a black spruce forest Tree Physiol17 537ndash542 doi101093treephys178-9537 1997

Gower S Vogel J Norman J Kucharik C Steele S and StowT Carbon distribution and aboveground net primary productionin aspen jack pine and black spruce stands in Saskatchewanand Manitoba Canada J Geophys Res 102 29029ndash29041doi10102997JD02317 1997

Gower S Krankina O Olson R Apps M Linder S and WangC Net primary production and carbon allocation patterns of bo-real forest ecosystems Ecol Appl 11 1395ndash1411 2001

Grime J Rincon E and Wickerson B Bryophytes andplant strategy theory Bot J Linnean Soc 104 175ndash186doi101111j1095-83391990tb02217x 1990

Hagemann S and Stacke T Impact of the soil hydrology schemeon simulated soil moisture memory Clim Dynam 44 1731ndash1750 doi101007s00382-014-2221-6 2015

Hermle S Lavigne M Bernier P Bergeron O and Pareacute DComponent respiration ecosystem respiration and net primaryproduction of a mature black spruce forest in northern QuebecTree Physiol 30 527ndash540 doi101093treephystpq002 2010

Hinzman L Kane D Gieck R and Everett K Hydrologicand thermal properties of the active layer in the Alaskan Arc-

tic Cold Reg Sci Technol 19 95ndash110 doi1010160165-232X(91)90001-W 1991

Jorgenson M Romanovsky V Harden J Shur Y OrsquoDonnellJ Schuur E Kanevskiy M and Marchenko S Resilienceand vulnerability of permafrost to climate change Can J For-est Res 40 1219ndash1236 doi101139X10-060 2010

Knorr W Annual and interannual CO2 exchanges of the terrestrialbiosphere process-based simulations and uncertainties GlobalEcol Biogeogr 9 225ndash252 2000

Kolari P Pumpanen J Kulmala L Ilvesniemi H Nikinmaa EGroumlnholm T and Hari P Forest floor vegetation plays an im-portant role in photosynthetic production of boreal forests ForestEcol Manage 221 241ndash248 doi101016jforeco2005100212006

Koshurnikova N Annual Production of Moss Layer in Dark Conif-erous Forests of Ket-Chulym Forest District (by the Exam-ple of Moss Hylocomium splendens) Biol Bull 34 532ndash536doi101134S1062359007050184 2007

Koven C Friedlingstein P Ciais P Khvorostyanov D KrinnerG and Tarnocai C On the formation of high-latitude soil car-bon stocks Effects of cryoturbation and insulation by organicmatter in a land surface model Geophys Res Lett 36 L21501doi1010292009GL040150 2009

Lange O Hahn S Meyer A and Tenhunen J Upland Tundra inthe Foothills of the Brooks Range Alaska USA Lichen Long-Term Photosynthetic CO2 Uptake and Net Carbon Gain ArctAlp Res 30 252ndash261 doi1023071551972 1998

Mack M Treseder K Manies K Harden J Schuur E Vo-gel J Randerson J and Chapin III F Recovery of above-ground plant biomass and productivity after fire in mesic and dryblack spruce forests of interior Alaska Ecosystems 11 209ndash225doi101007s10021-007-9117-9 2008

Meinshausen M Smith S Calvin K Daniel J Kainuma MLamarque J-F Matsumoto K Montzka S Raper S RiahiK Thomson A Velders G and van Vuuren D The RCPgreenhouse gas concentrations and their extensions from 1765to 2300 Climatic Change 109 213ndash241 doi101007s10584-011-0156-z 2011

Mouillot F and Field C Fire history and the global carbonbudget a 1times 1 fire history reconstruction for the 20th cen-tury Global Change Biol 11 398ndash420 doi101111j1365-2486200500920x 2005

Nash III T (Ed) Lichen Biology 2nd ed Cambridge UniversityPress New York 1996

OrsquoConnell K Gower S and Norman J Comparison of Net Pri-mary Production and Light-Use Dynamics of Two Boreal BlackSpruce Forest Communities Ecosystems 6 236ndash247 2003

OrsquoDonnell J Romanovsky V Harden J and McGuire A TheEffect of Moisture Content on the Thermal Conductivity of Mossand Organic Soil Horizons From Black Spruce Ecosystems inInterior Alaska Soil Sci 174 646ndash651 2009

OrsquoDonnell J Harden J McGuire A Kanevskiy M Jorgen-son M and Xu X The effect of fire and permafrost inter-actions on soil carbon accumulation in an upland black spruceecosystem of interior Alaska implications for post-thaw carbonloss Global Change Biol 17 1461ndash1474 doi101111j1365-2486201002358x 2011

Oechel W and Collins N Comparative CO2 exchange patterns inmosses from two tundra habitats at Barrow Alaska Can J Bot54 1355ndash1369 doi101139b76-148 1976

Oechel W and Van Cleve K The role of bryophytes in nutri-ent cycling in the taiga in Forest ecosystems in the Alaskantaiga A synthesis of structure and function edited by Van CleveK Chapin III F Flanagan P Vierect L and Dyrness CSpringer New York 122ndash137 1986

Olson D Dinerstein E Wikramanayake E Burgess N Pow-ell G Underwood E DrsquoAmico J Itoua I Strand HMorrison J Loucks C Allnutt T Ricketts T KuraY Lamoreux J Wettengel W Hedao P and KassemK Terrestrial Ecoregions of the World A New Map ofLife on Earth BioScience 51 933ndash938 doi1016410006-3568(2001)051[0933TEOTWA]20CO2 2001

Pachauri R Allen M Barros V Broome J Cramer W ChristR Church J Clarke L Dahe Q Dasgupta P et al ClimateChange 2014 Synthesis Report in Contribution of WorkingGroups I II and III to the Fifth Assessment Report of the Inter-governmental Panel on Climate Change IPCC Geneva Switzer-land 2014

Porada P Weber B Elbert W Poumlschl U and Kleidon AEstimating global carbon uptake by lichens and bryophyteswith a process-based model Biogeosciences 10 6989ndash7033doi105194bg-10-6989-2013 2013

Porada P Weber B Elbert W Poumlschl U and Kleidon AEstimating Impacts of Lichens and Bryophytes on GlobalBiogeochemical Cycles of Nitrogen and Phosphorus and onChemical Weathering Global Biogeochem Cy 28 71ndash85doi1010022013GB004705 2014

Proctor M The bryophyte paradox tolerance of desiccation eva-sion of drought Plant Ecol 151 41ndash49 2000

Raddatz T Reick C Knorr W Kattge J Roeckner E SchnurR Schnitzler K-G Wetzel P and Jungclaus J Will the trop-ical land biosphere dominate the climatendashcarbon cycle feedbackduring the twenty-first century Clim Dynam 29 565ndash574doi101007s00382-007-0247-8 2007

Rapalee G Steyaert L and Hall F Moss and lichen cover map-ping at local and regional scales in the boreal forest ecosystemof central Canada J Geophys Res-Atmos 106 33551ndash33563doi1010292001JD000509 2001

Rinke A Kuhry P and Dethloff K Importance of a soil organiclayer for Arctic climate A sensitivity study with an Arctic RCMGeophys Res Lett 35 L13709 doi1010292008GL0340522008

Rogers R Ecological strategies of lichens Lichenologist 22 149ndash162 doi101017S002428299000010X 1990

Ruess R Hendrick R Burton A Pregitzer K Sveinbjorns-son B Allen M and Maurer G Coupling fine root dynamicswith ecosystem carbon cycling in black spruce forests of inte-rior Alaska Ecol Monogr 73 643ndash662 doi10189002-40322003

Schulze E-D and Caldwell M M (Eds) Ecophysiology ofphotosynthesis Springer Science amp Business Media Springer-Verlag Berlin Heidelberg doi101007978-3-642-79354-71995

Schuur E Crummer K Vogel J and Mack M Plant SpeciesComposition and Productivity following Permafrost Thaw and

Thermokarst in Alaskan Tundra Ecosystems 10 280ndash292doi101007s10021-007-9024-0 2007

Schuur E McGuire A Schadel C Grosse G Harden J HayesD Hugelius G Koven C Kuhry P Lawrence D Natali SOlefeldt D Romanovsky V Schaefer K Turetsky M TreatC and Vonk J Climate change and the permafrost carbon feed-back Nature 520 171ndash179 doi101038nature14338 2015

Shaver G and Chapin III F Production Biomass relationshipsand element cycling in contrasting arctic vegetation types EcolMonogr 61 1ndash31 doi1023071942997 1991

Skre O and Oechel W Moss production in a black spruce Piceamariana forest with permafrost near Fairbanks Alaska as com-pared with two permafrost-free stands Ecography 2 249ndash254doi101111j1600-05871979tb01296x 1979

Soudzilovskaia N van Bodegom P and Cornelissen J Domi-nant bryophyte control over high-latitude soil temperature fluc-tuations predicted by heat transfer traits field moisture regimeand laws of thermal insulation Funct Ecol 27 1442ndash1454doi1011111365-243512127 2013

Stoy P Street L Johnson A Prieto-Blanco A and Ew-ing S Temperature Heat Flux and Reflectance of Com-mon Subarctic Mosses and Lichens under Field ConditionsMight Changes to Community Composition Impact Climate-Relevant Surface Fluxes Arct Antarct Alp Res 44 500ndash508doi1016571938-4246-444500 2012

Street L Stoy P Sommerkorn M Fletcher B Sloan VL HillT and Williams M Seasonal bryophyte productivity in the sub-Arctic A comparison with vascular plants Funct Ecol 26 365ndash378 doi101111j1365-2435201101954x 2012

Swanson R and Flanagan L Environmental regulation ofcarbon dioxide exchange at the forest floor in a borealblack spruce ecosystem Agr Forest Meteorol 108 165ndash181doi101016S0168-1923(01)00243-X 2001

Trumbore S and Harden J Accumulation and turnover ofcarbon in organic and mineral soils of the BOREAS north-ern study area J Geophys Res-Atmos 102 28817ndash28830doi10102997JD02231 1997

Turetsky M Mack M Hollingsworth T and Harden J The roleof mosses in ecosystem succession and function in Alaskarsquos bo-real forest Can J Forest Res 40 1237ndash1264 doi101139X10-072 2010

Uchida M Muraoka H Nakatsubo T Bekku Y Ueno TKanda H and Koizumi H Net Photosynthesis Respirationand Production of the Moss Sanionia uncinata on a Glacier Fore-land in the High Arctic Ny-Aringlesund Svalbard Arct AntarctAlp Res 34 287ndash292 doi1023071552486 2002

Uchida M Nakatsubo T Kanda H and Koizumi H Estimationof the annual primary production of the lichen Cetrariella deliseiin a glacier foreland in the High Arctic Ny-Aringlesund SvalbardPolar Res 25 39ndash49 doi103402polarv25i16237 2006

Valladares F Wierzchos J and Ascaso C Porosimetric Study ofthe Lichen Family Umbilicariaceae Anatomical Interpretationand Implications for Water Storage Capacity of the Thallus AmJ Bot 80 263ndash272 doi1023072445349 1993

Verseghy D CLASS ndash A Canadian land surface schemefor GCMs I Soil model Int J Climatol 11 111ndash133doi101002joc3370110202 1991

Vogel J Bond-Lamberty B Schuur E Gower S Mack MOrsquoConnell K Valentine D and Ruess R Carbon allocation

in boreal black spruce forests across regions varying in soil tem-perature and precipitation Global Change Biol 14 1503ndash1516doi101111j1365-2486200801600x 2008

Wahren C Walker M and Bret-Harte M Vegetation responsesin Alaskan arctic tundra after 8 years of a summer warming andwinter snow manipulation experiment Global Change Biol 11537ndash552 doi101111j1365-2486200500927x 2005

Webb R Rosenzweig C and Levine E Global Soil Textureand Derived Water-Holding Capacities (Webb et al) ORNLDistributed Active Archive Center Oak Ridge Tennesseedoi103334ORNLDAAC548 2000

Williams T and Flanagan L Measuring and modelling environ-mental influences on photosynthetic gas exchange in Sphagnumand Pleurozium Plant Cell Environ 21 555ndash564 1998

Zhu D Peng S Ciais P Viovy N Druel A KageyamaM Krinner G Peylin P Ottleacute C Piao S Poulter BSchepaschenko D and Shvidenko A Improving the dynam-ics of Northern Hemisphere high-latitude vegetation in the OR-CHIDEE ecosystem model Geosci Model Dev 8 2263ndash2283doi105194gmd-8-2263-2015 2015

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability



Author contributions

Acknowledgements

References

the effect of bryophytes and lichens at high latitudes on theglobal climate might represent either a positive or a negativefeedback

As a first step to analyse this feedback we focus in thisstudy on the effect of the bryophyte and lichen cover on soiltemperature at high-latitudes At the local to regional scaleexperimental as well as modelling studies have assessed theeffect of the bryophyte and lichen cover on soil temperatureBlok et al (2011) show that the experimental removal ofmoss leads to increased ground heat flux and increased soilevaporation Gornall et al (2007) observe higher soil tem-peratures in summer with decreasing thickness of the mosscover on the ground and lower soil temperatures in winter Ina modelling study Bonan (1991) perform simulations witha local energy balance model for 20 forest stands in centralAlaska to estimate the effect of moss removal on soil tem-perature They find an increase in soil temperature due to thereduction in moss ground cover Jorgenson et al (2010) usea local soil temperature model to quantify the effect of sev-eral factors on permafrost and find that vegetation on groundstrongly promotes permafrost stability

The insulating effect of the bryophyte and lichen groundcover on soil temperature is strongly modulated by bryophyteand lichen moisture content The thermal conductivity of thecover increases by around 1 order of magnitude from dryto water-saturated bryophytes and lichens while the heatcapacity even increases by around 2 orders of magnitude(OrsquoDonnell et al 2009 Soudzilovskaia et al 2013) More-over thermal conductivity increases strongly during the tran-sition from liquid to frozen water inside the bryophyte andlichen cover while heat capacity is reduced (Hinzman et al1991) Therefore a large and important seasonality of the in-sulating effect is expected in winter and to a lesser extentin autumn and spring the insulating effect of bryophytes andlichens will be small due to the high thermal conductivityof ice and water In summer however a dry bryophyte andlichen cover may strongly insulate the ground leading to anet cooling effect for the whole year The relation betweenthermal properties of bryophytes and lichens and their wa-ter saturation is very similar for a large range of differentspecies (Soudzilovskaia et al 2013) However the speciesshow considerable differences in water-holding capacity andother properties which affect their water status Thereforedifferent species may exhibit differing water saturation un-der similar climatic conditions

Several global land surface models which attempt to simu-late permafrost soil at high latitudes include a bryophyte andlichen ground cover while others approximate near-surfacevegetation using grass (eg Zhu et al 2015) Alternativelythe insulating effect is attributed solely to the carbon-richuppermost soil layer (Koven et al 2009) However the dy-namic nature of bryophyte and lichen surface coverage andthe dependence of the insulating effect on moisture contenthave not been explicitly considered so far in global landsurface models Beringer et al (2001) for instance extend

the land surface model NCAR-LSM (National Center forAtmospheric Research Land Surface Model) by lichen ormoss topsoil parameterisations for an arctic tundra site butthey assume constant thermal conductivity and surface cov-erage of the organic layer Ekici et al (2014) use the globalland surface model JSBACH ((Jena Scheme for BiospherendashAtmosphere Coupling in Hamburg)) to simulate permafrostat high latitudes and thereby include an organic layer abovethe soil which reduces heat exchange with the atmosphereThe layer is assumed to cover the whole land surface uni-formly and is set to constant thermal properties RecentlyChadburn et al (2015a) added a representation of moss tothe land surface model JULES (Joint UK Land EnvironmentSimulator) While the surface coverage of the moss layer isprescribed by a static map the thermal conductivity of themoss layer is dynamic Instead of being related to moss wa-ter content however the thermal conductivity is determinedby the moisture content of the uppermost soil layer Subse-quently Chadburn et al (2015b) extend JULES by an em-pirical scheme which determines moss health as a functionof climate variables and consequently allows for the com-putation of a dynamic moss ground cover In summary theuse of constant thermal properties and empirical relations incurrent global land surface models makes it difficult to pre-dict the impact of future bryophyte and lichen functions onpermafrost ecosystems

Here we present a process-based model which simulatesproductivity and dynamic surface coverage of the bryophyteand lichen ground cover at the global scale The model rep-resents the organismsrsquo water and ice content and thereby ac-counts for dynamic thermal properties of the cover The basisfor this model is the stand-alone dynamic non-vascular veg-etation model LiBry which has been successfully applied toquantify global productivity by bryophytes and lichens (Po-rada et al 2013) as well as estimating their contribution toglobal biogeochemical cycles (Porada et al 2014) LiBry isfully integrated into the land surface scheme JSBACH of theMax Planck Institute Earth system model JSBACH simu-lates the carbon balance of vascular plants and the soil at theglobal scale it includes a water and energy balance as well asa representation of permafrost Based on the thermal proper-ties of the near-surface vegetation layer we use JSBACH toquantify the effect of bryophytes and lichens on heat transferbetween atmosphere and soil and therefore on soil temper-ature Since JSBACH does not include a scheme for wetlandhydrology the formation of peatlands cannot be simulatedHence LiBry in JSBACH mainly represents bryophyte andlichen growth on upland forest floor sites

We compare the soil thermal regime of a JSBACH simu-lation including a dynamic bryophyte and lichen layer to asimulation where the thermal properties and surface cover-age of this layer are set to constant and to another simulationwith the bryophyte and lichen cover switched off In this waywe can assess quantitatively the impact of the bryophyte andlichen ground cover on soil temperature at high latitudes In a

2 Methods

RD =DA (4)

Disturbance

Unit area

G bull A bull η

(1 - A)A

0 01 02 03 04 05 06 07 08 09 1

ndash2

(a)Unfrozen

50 frozenFrozen

Constant

500 000

15e+06

25e+06

35e+06

0 01 02 03 04 05 06 07 08 09 1

ndash3

(b)Unfrozen

50 frozenFrozen

Constant

ndash1

Ke=min(

10vw+ vi

Surface

22 Model set-up

minus9

Surface coverage

minus9

[m2 mminus2]

3 Results

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

2 Methods

RD =DA (4)

Disturbance

Unit area

G bull A bull η

(1 - A)A

0 01 02 03 04 05 06 07 08 09 1

ndash2

(a)Unfrozen

50 frozenFrozen

Constant

500 000

15e+06

25e+06

35e+06

0 01 02 03 04 05 06 07 08 09 1

ndash3

(b)Unfrozen

50 frozenFrozen

Constant

ndash1

Ke=min(

10vw+ vi

Surface

22 Model set-up

minus9

Surface coverage

minus9

[m2 mminus2]

3 Results

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

RD =DA (4)

Disturbance

Unit area

G bull A bull η

(1 - A)A

0 01 02 03 04 05 06 07 08 09 1

ndash2

(a)Unfrozen

50 frozenFrozen

Constant

500 000

15e+06

25e+06

35e+06

0 01 02 03 04 05 06 07 08 09 1

ndash3

(b)Unfrozen

50 frozenFrozen

Constant

ndash1

Ke=min(

10vw+ vi

Surface

22 Model set-up

minus9

Surface coverage

minus9

[m2 mminus2]

3 Results

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

0 01 02 03 04 05 06 07 08 09 1

ndash2

(a)Unfrozen

50 frozenFrozen

Constant

500 000

15e+06

25e+06

35e+06

0 01 02 03 04 05 06 07 08 09 1

ndash3

(b)Unfrozen

50 frozenFrozen

Constant

ndash1

Ke=min(

10vw+ vi

Surface

22 Model set-up

minus9

Surface coverage

minus9

[m2 mminus2]

3 Results

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

Surface

22 Model set-up

minus9

Surface coverage

minus9

[m2 mminus2]

3 Results

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

minus9

Surface coverage

minus9

[m2 mminus2]

3 Results

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

1996 1998 2000 2002 2004 2006 2008 2010

2 4 6 8 10 12

Months

100-30 -25 -20 -15 -10 -5 0

100-15 -10 -5 0 5 10 15

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

minus90˚

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus5

[degC]

minus90˚

minus50

minus45

minus40

minus35

minus30

minus25

minus20

minus15

minus10

minus05

[degC]

minus90˚

minus250

minus225

minus200

minus175

minus150

minus125

minus100

minus75

minus50

minus25

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

4 Discussion

0 2 4 6 8 10 12

Months

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

5 Conclusions

6 Data availability

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

ain x 2

Topt - 5 K

C x 05

ain Topt

ain Topt V

Reduction in s

rence [

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

27 (2ndash104)

063 (05ndash095)

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

00 03 06 09 12 15 18 21 24 27 30

60˚ 60˚

70˚ 70˚

[degC]

60˚ 60˚

70˚ 70˚

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

Precipitation

minus90˚

[mm aminus1]

PAR on ground

minus90˚

Biome mask

minus90˚

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

Abstract

Introduction

Methods

Model description

Model set-up

Results

Discussion

Conclusions

Data availability




Acknowledgements

References

Documents

Effects of bryophyte and lichen cover on permafrost soil ... · simulates bryophyte and lichen cover on upland sites. Wet-lands are not included. We take into account the dynamic