Springer Old Growth Forests - Chapter 7 ppt

Biosphere–Atmosphere Exchangeof Old-Growth Forests: Processes and Pattern Alexander Knohl, Ernst-Detlef Schulze, and Christian Wirth 7.1 Introduction Forests are important agents of the

Biosphere–Atmosphere Exchange

of Old-Growth Forests: Processes and Pattern

Alexander Knohl, Ernst-Detlef Schulze, and Christian Wirth

7.1 Introduction

Forests are important agents of the global climate system in that they absorb and reflect solar radiation, photosynthesise and respire carbon dioxide and transpire water vapour to the atmosphere (Jones 1992) Through these functions, forests act

as substantial sinks for carbon dioxide from the atmosphere (Wofsy et al 1993; Janssens et al 2003) and sources of water vapour to the global climate system (Shukla and Mintz 1982) Since old-growth forests differ in age, structure and composition from younger or managed forests (see Chap 2 by Wirth et al., this volume) the question arises whether these characteristics also result in differences

in the biosphere atmosphere exchange of carbon, water, and energy of old-growth forests

This chapter reviews studies using two contrasting experimental approaches: the eddy covariance technique, and paired catchment studies The eddy covari-ance technique is a micrometeorological standard method to directly quantify the exchange of trace gasses between forest ecosystems and the atmosphere by mea-suring up- and down-drafts of air parcels above the forest (Baldocchi 2003) Fluxes

of scalars such as carbon dioxide, water vapour as well as sensible heat can be inferred from the covariance between scalar and vertical wind speed (Aubinet et al 2000) The advantages of this approach are that no disturbances or harvests are needed to assess fluxes and that the eddy flux tower typically integrates over a flux source area of approximately 1 km2 This approach, however, assumes that the underlying surface, i.e the forest, is horizontally homogeneous, which is typically the case over managed, even-aged forests Old-growth forests, however, are often characterised by a dense and structured canopy including canopy gaps and a diverse range of tree heights (see Chap 2 by Wirth et al., this volume; Parker et al 2004) Additionally, in many parts of the world, old-growth forests occur mainly in complex often sloped terrain of mountain ranges, which are less favourable or accessible for anthropogenic land use [see Chaps 15 (Schulze et al.) and 19 (Frank

et al.), this volume] This raises the question of how these characteristics of old-growth forests affect the direct measurement of biosphere atmosphere exchange of

C Wirth et al (eds.), Old ‐Growth Forests, Ecological Studies 207, 141 DOI: 10.1007/978 ‐3‐540‐92706‐8 7, # Springer‐Verlag Berlin Heidelberg 2009

carbon, water, and energy With the second approach, i.e paired catchment studies, only water exchange is quantified This is done by comparing the streamflow of two catchments that are similar with respect to soil, topography and climate but differ in land use or vegetation cover (Andre´assian 2004) The method is suited to the study of differences in evapotranspiration and water yield between contrasting land-use types, forest developmental stages, and management strategies Topo-graphic complexityper se does not pose a problem However, this comes at the expense of a lower temporal resolution and the need for multi-year calibration periods

In this chapter, we summarise results from studies in old-growth forests across the globe in order to (1) describe structural characteristics of old-growth forests relevant for biosphere atmosphere exchange (Sect 7.2); (2) show how these characteristics influence net ecosystem carbon fluxes (Sect 7.3); (3) investigate the interplay between canopy structure, water, and energy fluxes (Sect 7.4); and (4) study the absorption of radiation, particularly of diffuse radiation in old-growth forests (Sect 7.5)

7.2 Characteristics of Old-Growth Forests Relevant

for Biosphere–Atmosphere Exchange

When forest ecosystems advance in age they typically undergo changes in their structural properties (see Chap 2 by Wirth et al., this volume) Old and large trees are more at risk to external forces such as disturbance by wind or by rotting of the heartwood due to fungal attack (Dhoˆte 2005; Pontailler et al 1997) As a conse-quence, individual trees, or parts of trees, sporadically die resulting in small scale canopy gaps (Spies et al 1990) These gaps then supply light to lower parts of the canopy that were previously in shade With this light supply, individuals previously limited by light are able to enhance their growth and finally close the canopy gap In old-growth forests gaps are typically very dynamic, leading to ongoing changes in canopy structure, light environment, and hence species composition (see Chap 6 by Messier et al., this volume) The spatial extent of canopy gaps and speed of canopy closure is likely to depend on species, site conditions and disturbance intensity, and varies greatly among biomes For old-growth forests in the Pacific Northwest of the United States canopy gaps were reported to remain open for decades (Spies et al 1990) Even in cases where canopy gaps in old-growth deciduous forests caused by, e.g., storms were closed within a few years, the light quantity and quality reaching understorey vegetation may remain dynamic for decades or even longer (see Chap

6 by Messier et al., this volume) As a consequence of these gap-phase dynamics, old-growth forests typically form a canopy consisting of diverse age classes and also varying heights of individual trees and canopy parts Older and tall trees may act as shelter for younger trees The 450-year-old Douglas fir/Western hemlock forest at the Wind River Canopy Crane Research Facility (WRCCRF) consists of

an extremely complex outer canopy surface due to high and narrow crowns and numerous larger and smaller gaps (Parker et al 2004) As a result, the surface area

of the canopy reaches more than 12 times that of the ground area The outer shape of the canopy strongly influences the permeability to solar radiation and the coupling

of environmental conditions such as air temperature and humidity with the atmo-sphere Since the top canopy consists of narrow crowns, a large part of leaf area is distributed to lower parts of the canopy, hence allowing solar radiation to penetrate deeply into the canopy resulting in a high efficiency in trapping light and hence low surface reflectance (Weiss 2000)

Along with processes leading to canopy gaps, coarse woody detritus, either standing or lying on the ground, accumulates and may account for a substantial fraction of the carbon pool within in an ecosystem The amount and decay rates of coarse woody debris vary among biomes and environmental conditions (see Chap 8 by Harmon et al., this volume.) At the WRCCRF forest about 25% of aboveground biomass is dead, resulting in large carbon pools contributing to heterotrophic respiration (Harmon et al 2004) Also, old-growth forests often contain large aboveground biomass stocks (see Chap 15 by Schulze, this volume) for temperate and boreal biomes Pregitzer and Euskirchen (2004) show a consistent increase in biomass carbon pools with age for boreal, temperate and tropical ecosystems Similarly, soil carbon pools are also often large due to carbon accumulation during stand development since the last disturbance (Harmon et al 2004; Pregitzer and Euskirchen 2004)

All these structural features typical of old-growth forests are expected to influ-ence biosphere atmosphere exchange of such forests In this chapter we will focus

on structural features of old-growth, i.e the fact that old-growth forests tend to be uneven-aged, horizontally and vertically structured forests, which at high age show gap dynamics and contain large amounts of woody detritus In general, we concen-trate on forests located in the temperate zone, but also include some examples from the boreal and tropical zones

7.3 Exchange of Carbon Dioxide

Old-growth forests are often considered to be insignificant as carbon sinks since it is assumed that they are in a state of dynamic equilibrium (Odum 1969; Salati and Vose 1984) where assimilation is balanced by respiration as a forest stand reaches

an old stage of development (Jarvis 1989; Melillo et al 1996) This hypothesis is based on studies showing a decline with stand age in net primary productivity at stand level (Yoder et al 1994; Gower et al 1996; Ryan et al 1997) and in photosynthesis

at tree level (Hubbard et al 1999; and see Chap 4 by Kutsch et al., this volume) and the general idea that ecosystem respiration increases with stand age (Odum 1969) Potential mechanisms such as increasing respiration costs and nutrient or hydraulic limitation are critically discussed by Kutsch et al (Chap 4, this volume) and Ryan et al (2004) Recent studies find carbon uptake rates in old-growth

forests indicating a small-to-moderate carbon sink (Phillips et al 1998; Carey et al 2001), sometimes even comparable to younger forests in the same region (Anthoni

et al 2004) Data for coniferous forests show that, even when old, some forests can retain their capacity to absorb carbon from the atmosphere, as shown for a 450-year old Douglas fir/Western hemlock site in Washington (Paw et al 2004), a 250-yearold ponderosa pine site in Oregon (Law et al 2001), a 300-year oldNothofagus site in New Zealand (Hollinger et al 1994), and 200- to 250-year old boreal forests (Roser et al 2002) This is supported by results from studies in mixed and deciduous forests that remained significant carbon sinks even when at high age, such as a 250-year old uneven-aged mixed beech forest in Germany (Knohl

et al 2003), a 200-year old mixed forest in China (Guan et al 2006; Zhang et al 2006), and a 350-year old uneven-aged mixed forest in the United States (Desai et al 2005)

In this book, Kutsch et al (Chap 4) and Schulze et al (Chap 15; and see Luyssaert et al 2008) argue that structure not age determines the capacity of forest ecosystems to absorb carbon from the atmosphere, and hence old forests may remain carbon sinks even at high age The argumentation is based on a global dataset of net primary productivity, biomass, stand density and net ecosystem exchange measurements (Luyssaert et al 2007) showing that a decline in productivity is more strongly related to leaf area index than to stand age, and that it only occurs when stand density drops below 330 trees ha–1 in temperate forest and 690 trees ha–1in boreal forest, independent of tree age This finding is supported by recent grafting studies showing that leaf level decline in photosynthe-sis is also related not to age, but to tree structure (Mencuccini et al 2007; Vanderklein et al 2007) Moreover, we also find that even 211-year old Pinus sylvestris trees have the ability to maintain high growth rates, as seen by an increase

in radial growth by factor of five immediately after thinning This indicates that these trees have been limited not by an age-related effect but by competition for resources (Fig 7.1) Once resources became more abundant again due to exclusion

of competitors, even old trees increase their growth Individuals with previously high growth rates responded more strongly to thinning than individuals with smaller growth rates These findings are supported by a study in the temperate zone Tall 140-year old Norway spruce trees in southern Germany showed an increase of about 50% in annual stem volume increment after stand thinning via harvest (Mund

et al 2002)

A global compilation of net ecosystem exchange data from eddy covariance (Luyssaert et al 2007) reveals that there are several old-growth forests (older than

200 years) that are net carbon sinks (Fig 7.2) It is important to note that the global coverage of eddy covariance flux measurements is strongly biased towards younger and managed forests Only very few flux towers are located in old-growth forests Additionally, some of these old-growth forests are ecosystems where factors other than just age play an important role A chronosequence of boreal forests in Canada shows following classical theory a decrease in net ecosystem produc-tivity with age, with the oldest forests (aged around 160 years) being close to carbon neutral (Amiro et al 2006) However, a more detailed study from the same

old-growth forest reveals that the low net ecosystem productivity at this site is determined mainly by a combination of low stand density and large heterotrophic respiration due to peat decomposition depending on changes in water table depth (Dunn et al 2007) Midday carbon uptake rates of this old-growth forest, however, are not lower than at other much younger ecosystems (Goulden et al 2006) Similarly, a recent study of eddy covariance measurements across five chronose-quences in Europe showed a strong age-related pattern of net ecosystem exchange, where young forests are carbon sources, intermediate forests carbon sinks and the only older forests in this study was close to carbon neutral (Magnani

et al 2007) However, when looking more closely at the oldest forest in that study, a boreal coniferous forest in Sweden, it seems likely that factors other than just age are important such as horizontal advection of CO2 (A Lindroth, personal communication)

There has been a recent controversial discussion over whether the eddy covari-ance technique can be used to accurately measure the exchange of carbon between forest and atmosphere in terrain typical of old-growth forests, i.e mountainous regions or tall and dense canopies (Kutsch et al 2008) Advection, i.e a non-turbulent transport of scalars such as CO2, has been observed at several sites across the globe, often in dense forests, even at sites with only a minor slope (Staebler and Fitzjarrald 2004; Aubinet et al 2003, 2005; Feigenwinter et al 2008; Kutsch et al 2008) Measuring advection directly is technically challenging since it requires

Fig 7.1 Radial stem increment of 211 year old Pinus sylvestris trees (n = 9) in Central Siberia The stand was thinned via harvest in 1983 resulting in a strong increase in radial growth Error bars Standard error

additional tower measurements on a horizontal gradient and hence has so far only been done at a few selected sites Advection often occurs at night during conditions

of low turbulent mixing and hence results in a loss of CO2from the ecosystem not measured by the eddy covariance system Most studies, however, correct empiri-cally for non-turbulent conditions using the so-called u*-correction, where all flux data with a friction velocity (u*) value below a certain threshold are replaced by

an empirical model (Goulden et al 1996) Recent studies, however, question the validity of this correction (Kutsch et al 2008) Furthermore, in tall and dense forests, such as tropical forests, the choice of u* threshold may lead to very divergent annual sums of net carbon exchange Miller et al (2004) show that a u*-correction turns the closed tropical forest at the FLONA Tapajo´s km 83 tower site (Brazil ) from a large sink of approximately 400 g C m–2year–1into a carbon source of 50 100 g C m–2 year–1 (cf Chap 17 by Grace and Meir, this volume) Since old-growth forests are often characterised by tall and dense cano-pies with heterogeneity in their horizontal and vertical structure, and since they are often located at least in Central Europe in less accessible, often mountain-ous, terrain, there is a risk that advection may play a significant role in the carbon exchange of such forests Therefore, annual sums of net ecosystem exchange in old-growth forests may carry an uncertainty or even biases larger

Fig 7.2 Net ecosystem exchange (NEE) vs stand age for coniferous and deciduous forests in temperate and boreal biomes NEE is derived from eddy covariance measurements and compiled

in a global database (Luyssaert et al 2007) Positive values carbon sink, negative values carbon source

than the 30% typically given for eddy covariance measurements (Baldocchi 2003; Loescher et al 2006)

More interesting than just the question of whether old-growth forest are carbon sinks or not, is the understanding of the processes controlling carbon dynamics in old-growth forests Net ecosystem exchange is the balance of assimilation and respiration Since both are expected to be high in old-growth forest due to high biomass and large carbon pools (Pregitzer and Euskirchen 2004), small changes in the control of assimilation and respiration may shift the balance between them, leading to day-to-day and year-to-year variability Guan et al (2006) showed for a 200-year-old temperate mixed forest in north-eastern China that assimilation and ecosystem respiration are both close to 10 g C m–2 day–1 during the summer Depending on cloud cover, overcast and sunny conditions, this ecosystem switches between being a sink or source on a day to day basis A similar sensitivity to environmental conditions is observed on an annual time scale for the oldest forest being studied with the eddy covariance technique, the 450-year-old conifer-ous forest at the Wind River Canopy Crane Research Facility (WRCCRF) This forest switches between being a carbon sink or a carbon source depending on the timing of key transitions periods during the course of the year (Falk 2005, 2008) Net carbon uptake occurs mainly during the wet and cool period in spring, while the ecosystem releases carbon during the dry and hot summer The timing

of the transition from wet and cool to dry and hot determines the annual carbon balance (Falk et al 2005, 2008)

In summary, we need to extend the simplified picture concerning net carbon exchange of forests along ecosystem development where old-growth forests are considered to be carbon neutral (Odum 1969; Salati and Vose 1984; Jarvis 1989; Melillo et al 1996) More than forest age, forest structure seems to determine the capacity of forest ecosystems to absorb carbon from the atmosphere (Fig 7.3) Young forests typically carry the legacy of a previous disturbance They may act as carbon sources over years to decades depending on how fast decomposable carbon such as coarse woody detritus and exposed soil carbon is respired, and how rapidly new active biomass develops (see also Chap 8 by Harmon, this volume) Common disturbances include harvest (Giasson et al 2006), fire (Amiro 2001), wind-throw (Knohl et al 2002), and insects (Schulze et al 1999) The initial respiration component will depend on how much carbon remains at the site after disturbance Including the effect of disturbances in the assessment of carbon uptake by forests is essential since disturbances typically lead to a rapid release of large amounts of carbon that have been accumulated over a long period of time (Ko¨rner 2003) Once net assimilation of active biomass exceeds respiration from plants, coarse woody debris, and soil, forests act as carbon sinks The duration of this period is expected

to depend on site conditions, species, and disturbance history When stand density falls below a critical threshold at which canopy closure is not fully sustained (see Chap 15 by Schulze et al., this volume), when photosynthesis declines due to structural changes in tree morphology (Martinez-Vilalta et al 2007; Vanderklein

et al 2007; and Chap 4 by Kutsch et al., this volume), and when the amount of respiring carbon increases compared to photosynthetic active biomass, then forest

ecosystems may become close to carbon neutral Depending on the amount of carbon accumulated as coarse woody debris on the forest floor or as soil organic matter in the soil (see Chap 11 Gleixner et al., this volume) and lost as dissolved organic carbon old-growth forests may, however, never reach carbon balance, and continue to accumulate carbon at a low rate This stage needs to be seen as highly dynamic Small climatic variations may switch the ecosystem from being a carbon sink to a carbon source and vice versa (Falk et al 2005, 2008) Similarly, small-scale disturbances and regeneration lead to changes in growth rates of individual trees, both remaining tall trees and young rejuvenating trees Even though there is a correlation between structural development and stand age, we expect that this varies strongly from biome to biome and from site to site depending on site quality, soil properties, climate, nitrogen deposition and competition

Fig 7.3 Changes in carbon dynamics and stand properties with structural development of forest ecosystem

7.4 Exchange of Water and Energy

Water and energy exchange in forest ecosystems is strongly controlled by surface reflectance, the partitioning of available energy into latent and sensible heat, and stomatal conductance controlling transpiration (Jones 1992) At many sites across the globe, it has been observed that old and taller trees exhibit a lower stomata conductance and hence show lower transpiration rates (Ryan and Yoder 1997) Potential mechanisms are that older and taller trees are hydraulically limited due to increased resistance along the extended hydraulic path length and due to higher gravitational potential opposing the upward transport of water in tall trees (see Sect 4.3, in Chap 4 by Kutsch et al., this volume) As a result, stomata of old and tall trees may show a stronger response to high vapour pressure deficit than of younger trees, resulting in lower transpiration rates (Hubbard et al 1999) The available data, however, do not all support the hydraulic limitation hypothesis (see also Sect 4.3.3 in Chap 4 by Kutsch et al., this volume) In a 450-year-old Douglas fir stand (60 m tree height) in the Pacific Northwest (United States) leaf level stomatal conductance did not differ in stands of 20 years (15 m tree height) and

40 years (32 m tree height) of age during summer time measurements even though carbon isotope measurements suggested that the older trees were hydraulically limited during spring (McDowell et al 2002) Similarly, ponderosa pines stands

in Oregon show smaller canopy conductance for old (250 years) than for younger (25 years and 90 years) stands as long as water is not limited During summer, however, when soil dries out, the younger stands show a strong decline in transpi-ration while the old stand maintains high transpitranspi-ration rates due to access to ground water (Irvine et al 2004) At the ecosystem scale, however, evapotranspiration was controlled by available energy and hence both old and young stands had almost identical evapotranspiration flux rates Old-growth forests may even have higher evapotranspiration, i.e latent heat fluxes, than younger forests due to an albedo (surface reflectance) effect At a series of Douglas fir stands in the Pacific North-west evapotranspiration was highest at the 450-year-old stand (Chen et al 2004) Surface net radiation measurements revealed that these high fluxes were driven by high surface net radiation, i.e the difference between incoming and outgoing long and short wave radiation The increase in net radiation was caused by lower surface reflectance (albedo ) at the old stand compared to the younger stands This decline

in albedo, however, is not necessarily related to stand age, but to surface roughness, here called surface rugosity, and describing canopy complexity (Ogunjemiyo et al 2005) Remote sensing data showed a linear decline in albedo with surface rugosity

in the vicinity of the WRCCRF site (Ogunjemiyo et al 2005) Young stands absorbed about 79% of incoming radiation, while older stands absorbed 89%, an increase of about 12.7% in available energy resulting in a net radiation larger than

650 W m–2for the old-growth stand (Ogunjemiyo et al 2005) In order to maintain

a physiologically acceptable leaf temperature, the old-growth stands need to increase transpiration, resulting in high water fluxes As with the exchange of carbon dioxide, structure, i.e tree height, canopy rugosity and root depth, rather

than age per se, controls the exchange of water and energy between old-growth forests and the atmosphere as measured by eddy covariance

Paired catchment studies provide a longer-term and larger-scale picture of water exchange in response to forest structure In these studies, precipitation and runoff is monitored in two catchments (control and treatment) which have to be broadly similar with respect to soil, topography, climate and (initially) vegetation cover (Andre´assian 2004; Brown et al 2005) The target variable is usually the streamflow or, if expressed as a fraction of precipitation, the water yield Water-shed evapotranspiration can also be estimated as the difference between precipita-tion and streamflow, assuming that the storage change term is small (Brown et al 2005) After a multi-year calibration period, the ‘treatment catchment’ is subject to

an experimental manipulation, e.g complete or partial deforestation or just thin-ning To control for climate variability, the treatment effect is then estimated as the difference between two regression lines relating the target variable of the control and treatment catchment before and after the manipulation, respectively Existing catchment studies tend to focus on rather drastic land-use changes such as the conversion from forest to non-forest vegetation The need for a common calibration period precludes the comparison of vegetation attributes that require a long time to develop, such as structural or compositional changes with stand age Thus, catch-ment chronosequence studies do not exist and the only way of studying the effect of stand age is to follow experimental manipulations over time with the longest observation periods being in the order of 50 years In the following discussion,

we will focus on two key results emerging from existing meta-analyses of catch-ment studies with respect to the effect of (1) deforestation; and (2) differences in forest structure and composition

Deforestation of primary forests and, here especially, old-growth forests is

a global phenomenon [see Chaps 18 (Achard et al.) and 19 (Frank et al.), this volume] and thus of particular relevance for the topic of our book For the temperate zone, existing reviews found unequivocally that the short-term response to defor-estation despite considerable variability is an increase in water yield (Hibbert 1967; Bosch and Hewlett 1982; Sahin and Hall 1996) This increase was propor-tional to the fracpropor-tional reduction in forest cover and to the mean annual rainfall This general response was explained by the circumstance that forests exhibit higher rates of evapotranspiration than grasslands, which usually replace forests after deforestation (Zhang et al 2001) The magnitude of the deforestation response differed between forest types (see below) In the subtropics the effect of deforesta-tion on streamflow during the dry season depended on how deforestadeforesta-tion changes the infiltration opportunities (Bruijnzeel 1988) If infiltration is reduced, quick surface runoff during the wet season will lead to a reduced water yield during the dry season If infiltration remains constant, deforestation leads to an increase in water yield as was the case for temperate forests One consequence of increased water yield is an increased propensity for floods to occur In his review of paired catchment studies, Andre´assian (2004) concluded that deforestation indeed increased the frequency of flood peaks by about 40% (range 18% to 200%)