Integrated Land Ecosystem - Atmosphere Process Study

Issue No. 14- April 2014

Extreme Events and Environments

www.ileaps.org1 A Editorial 4 16 4 Extreme events and environments - complementary components of global change research

B Science 6 Permanently extreme environments and the need to observe the two poles of the carbon cycle

9 Exploiting plants from extreme environments: have we both world enough and time? 12 Challenges in quantifying global carbon cycle extremes

16 Measuring and modelling global vegetation dynamics in response to chronic and extreme climate changes 21 ‘The great, big, broad land ’way up yonder’: biogeochemical cycling and in the high latitudes

C Activities 26 Societal adaptation to extreme events and 21 environments D People 28 26 E Regional iLEAPS 30

iLEAPS Newsletter Executive Editor ISSN Printed version 1796-0363 Tanja Suni ISSN On-line version 1796-0401 Newsletter Coordinator For submissions and subscriptions, please contact [email protected] Magdalena Brus Publisher Hardcopy circulation iLEAPS International Project Office (IPO) 3500 Erik Palmenin aukio PO Box 48 iLEAPS IPO is sponsored by the University of Helsinki, the Finnish FI-00014 University of Helsinki Meteorological Institute, and the Ministry of Education, Finland. Tel: +358 (0)9 191 50571 [email protected] www.ileaps.org

2 News and Science Highlights

Observations of increased Future Earth forming its final Annual global carbon emissions tropical rainfall preceded by air structures estimated to have reached record passage over The nominations for the Future Earth Engage- 36 billion tonnes in 2013 In a new Nature publication, Spracklen ment Committee and the selection for the et al. used satellite remote-sensing data permanent, globally distributed secretariat Global emissions of carbon dioxide from burn- of tropical precipitation and vegetation take place during spring 2014, and the offi- ing fossil fuels were set to have risen again in combined with simulated atmospheric cial website is ready to go live in the spring as 2013, reaching a record high of 36 billion tonnes transport patterns to assess whether for- well. The new Future Earth blog can be found - according to new figures from the Internation- ests actually have an influence on tropi- at http://www.futureearth.info. al Geosphere-Biosphere Programme’s Global Permanently extreme environments and the need cal rainfall. They found that for more than Intended to be a home for innovative new ide- Carbon Project. to observe the two poles of the carbon cycle 60% of the tropical land surface, air that as and essential reading for everyone engaged More information on the global carbon budgets had passed over extensive vegetation in in global sustainability, this online magazine and trends can be found on website of Global Exploiting plants from extreme environments: the preceding few days produced at least will be a showcase and discussion forum for Carbon Project: http://www.globalcarbonpro- have we both world enough and time? twice as much rain as air that has passed the latest ideas and developments in research ject.org/carbonbudget/index.htm over little vegetation. in this area, both in the projects that form part The article can be found here: http:// of Future Earth’s network and beyond. www.nature.com/nature/journal/v489/ n7415/full/nature11390.html

New research project endorsed under Climate extremes can negate the expected in- ‘The great, big, broad land ’way up yonder’: iLEAPS umbrella crease in terrestrial carbon uptake biogeochemical cycling and climate change in the high latitudes “Methane loss from Arctic: towards an annual budg- The terrestrial biosphere is a key component of the glob-

et of CH4 emissions from ecosystems across - a latitudinal gradient” is a new interesting research enced by climate. Continuing environmental changes project endorsed by iLEAPS. This project will be areal carbon thought cycle to increase and its carbon global balanceterrestrial is stronglycarbon uptake. influ But evidence is mounting that climate extremes such as

full annual budget of both CH4 and CO2 net emissions droughts or storms can lead to a decrease in regional eco- fromamong three the firsttundra efforts ecosystems toward acrossthe estimation a transect of ina system carbon stocks and therefore have the potential to Arctic . Of primary importance is the quanti- negate an expected increase in terrestrial carbon uptake.

4 emissions because The article by M. Reichstein et al. explores the mecha- of the potentially large impact on overall climatic nisms and impacts of climate extremes on the terrestrial effects.fication This of non-summertime information has CHhistorically been very carbon cycle, and propose a pathway to improve our un- - derstanding of present and future impacts of climate ex- mote monitoring stations. Recent advances in meas- tremes on the terrestrial carbon budget. difficult to collect because of severe weather and re urement technology will make these studies feasible The article can be found here: http://www.nature.com/ in remote locations and under extreme weather con- nature/journal/v500/n7462/full/nature12350.html ditions.

New phase for the European Alli- on local conditions. As Future Earth ance of Global Change Research aims not only at advancing understand- Committees ing of global change and sustainability is- sues but also at contributing to solutions, In the 5th meeting of the European Alli- it will require understanding of local eco- ance on 3-4 Dec 2013 in Helsinki, Fin- nomic and political systems and cultures land was elected as the new chair coun- and an access to local decision-makers try and a host for the secretariat for and stakeholders that are instrumental 2014-2016. in implementing the solutions. This can National committees have always been only be achieved through national bod- an integral part of the GEC programmes ies such as global change national com- but now, under Future Earth, they will mittees. become important instruments for co- The 6th meeting of the European Alliance design and co-production of knowl- will take place on 28-29 October 2014 in edge. Solutions to sustainability prob- Switzerland. www.euroalliance-globalchange.org 3 lems are local and depend significantly A Editorial

Markus Reichstein1, Dan Yakir2, Michael Bahn3, Francesco Loreto4 1. Biogeochemical Model-Data Integration Group, Max-Planck Institute for Biogeochemistry, Jena 2. Weizmann Institute of Science, Rehovot, Israel 3. Institute of Ecology, University of Innsbruck, Austria 4. The National Research Council of Italy - Department of Biology, Agriculture and Food Sciences (CNR-DISBA), Rome, Italy Extreme events and environments - complementary components of global change research

The need to make informed guesses mary production have been estimated and structural adjustments [6] and bio- (that is, projections & scenarios) with to be on the order of the global land logical adaptations (Cheeseman), yield- respect to the response of the land bio- carbon sink [2]. On the other hand, the ing organisms and ecosystems that can sphere to the global changes current- biosphere as a whole has been vigor- survive under extreme conditions. Fur-

ly underway is a challenge that moti- ously taking up CO2 during the last dec- thermore, although various extreme vates much of the research in the IGBP/ ades [3-5], indicating that the effects of environments, such as deserts, exhib- Future-Earth and iLEAPS contexts. In- extreme events may be reversible and it low productivity and carbon stock, creasingly, a distinction is made be- may be balanced by the biosphere at a Schimel and Cox point out that in oth- tween the possible impact of changes global scale. However, if certain thresh- er cases very warm or very cold regions in the frequency and intensity of ex- olds are passed (McDowell & Cham- may have relatively low productivity, ber, this issue), extreme events could but they can accumulate large carbon heat and cold waves, insect outbreaks), lead to irreversible changes in organ- stocks because, for example, primary andtreme changes events in (such the mean as droughts, climate flood, with isms and ecosystems, and to mortality. productivity is enhanced relative to de- possible shifts in climatic zones where As Mahecha et al. point out, the thresh- composition, or decomposition is de- environmental conditions are more ex- old is not necessarily a change in one pressed relative to productivity. Such treme. In extreme environments, such variable only, for instance a hot or dry aspects of extreme environments may changes can expose new resources for spell, but rather a multivariate bound- therefore contribute to the resilience of human exploitation (such as new ar- ary line, for instance hot and dry, or a the biosphere. able soil associated with receding ice surface representing different combi- Finally, ecosystems and species cover; see Williams, this issue) with im- nations of variables and their intensi- in extreme environments, while ful- plications for productivity, and sustain- ties within the environmental space ly adapted to such environments, may ability of fragile environments. Both, considered. also operate near their limits (Cheese- changes in climate extremes, and cli- In addition, continuously extreme man) and are therefore still threatened mate change in extreme environments, environments (such as deserts, alpine both by chronic climate changes and pose distinct challenges for research and polar regions, salt-marshes or hot - springs; Williams, Cheeseman) have Fig. 1. Contrasting aspects of extreme events tent, meeting these challenges also re- led to a wide range of ecophysiological versus extreme environments quiresand society that we (Fig. improve 1). To oura significant observation ex systems (Schimel and Cox, this issue), our dynamic vegetation and other mod- Extreme events Long-term extreme els (Mahecha et al., this issue), and our environments ability to reduce human pressure on the • Adaptation difficult • Highly adapted organisms natural system (Cheeseman, this issue). Climate extremes have been shown • Disturbance-recovery • Narrow niches to be able to undo several years of car- • Resilience important • Stress resistance important bon dioxide uptake by ecosystems [1] • Threshold-type response • Recovery often slow – and globally integrated negative ex- tremes in photosynthesis or gross pri- low resilience

4 by unprecedented extreme events. In References such cases, ecosystems in extreme en- 1. Ciais P et al. 2005. Europe-wide reduction vironments may go extinct, but this can in primary productivity caused by the heat and drought in 2003. Nature 437, 529-533. potentially be accompanied by com- 2. Reichstein M et al. 2013. Climate extremes pensating shifts of other extreme envi- and the carbon cycle. Nature 500, 287-295. ronments into milder ones. 3. Ballantyne A, Alden C, Miller J, Tans P & Overall, recent research has clear- White J 2012. Increase in observed net carbon ly shown that both extreme events dioxide uptake by land and oceans during the past 50 years. Nature 488, 70-72. and environments must be consid- 4. Gloor M et al. 2010. What can be learned ered when predicting the response of about carbon cycle climate feedbacks from the biosphere to climate change. How CO2 airborne fraction? ACP 10, 7739-7751. strongly climate extremes will change 5. Gurney KR et al. 2011. Regional trends in the land biosphere remains largely elu- terrestrial carbon exchange and their seasonal signatures. Tellus, 63B, 328-339. sive. It will depend on the rate of chang- 6. Rotenberg E & Yakir D 2010. Contribution of es, both in mean climate and frequen- semi-arid forests to the climate system. cy of extreme events, and on the rate Science, 327, 451-454. of adjustment and adaptation. The bot- tom line is that research in extreme en- vironments can continue to tell us how adaptation can work, while unprece- dented extreme events tell what hap- pens when environmental conditions (suddenly) exceed the range organisms and ecosystems have adapted to.

5 B Science

David Schimel1 and Peter Cox.2 1. Past Co-Chair, AIMES (Analysis, Integration and Modelling of the Earth System) Jet Propulsion Lab, California Institute of Technology, Pasadena, CA, USA 2. Co-Chair, AIMES (Analysis, Integration and Modelling of the Earth System) College of Engineering, Mathematics and Physical Sciences, University of Exeter, Exeter, UK Permanently extreme environments and the need to observe the two poles of the carbon cycle

of temporary climate extremes on the carbon cycle is attracting more and more attention. InThe recent influence years, heat waves and droughts have caused observable changes in regional and global carbon

- biningfluxes [1]. in situ However, and global climate-related analyses, we disturbances are increasingly also include and insect outbreaks [2, 3]. By com Furthermore, we now know that climate variations provideable to quantify vital insights the influence into the of sensitivity such extreme of the events. car- bon cycle to anthropogenic climate changes [4]. Such “Emergent Constraints” provide a bridge that links - jects such as iLEAPS (Integrated Land Ecosystem – At- David Schimel is currently a Senior Peter Cox is co-chair of the IGBP mospherethe measurement Processes of short-term Study) to the flux modelling variations of in long- pro Research Scientist at the Jet Propul- (International Geosphere-Bio- term climate-carbon cycle feedbacks in projects such sion Lab, leading research focused sphere Programme) project AIMES on carbon-cycle climate interactions, (Analysis, Integration and Model- as AIMES. combining models and observa- ling of the Earth System), and Pro- The role of permanently extreme environments tions. For the previous five years, Dr fessor of Climate System Dynam- (such as drylands, deserts, regions with extreme rain- Schimel led the National Ecological ics at the University of Exeter in fall or ) is less obvious but equally im- Observatory Network (NEON) pro- the UK. His personal research has portant. As the climate changes, environments con- ject, was responsible for the top- focused on interactions between sidered extreme relative to current conditions may level science design, site selection the land-surface and climate, in- become more or less common. For instance, the ex- and observing system simulations. cluding the first climate projections - Before this, Dr Schimel worked at to include vegetation and the car- fects on carbon storage if the zone expands into cur- the National Center for Atmospheric bon cycle as interactive elements. rentlypansion productive, of the arid high zone carbon could ecosystems have significant that then ef Research (NCAR) and the Max Planck Peter Cox is a Lead Author on the begin to lose carbon. However, woody and Institute for Biogeochemistry in Jena, (IPCC) 4th and 5th Assessment Re- Germany. He served as convening port, and is also a member of the carbon storage are also increasing in many arid eco- Lead Author for the first Intergov- Science Advisory Group for the UK systems. As another example, zones of extreme rain- ernmental Panel on Climate Change Department of Energy and Climate fall may shift geographically as a result of the inten- (IPCC) assessment of the carbon cy- Change. cle, and has since served as an Coor- regional-to-global storage of carbon directly, through dinating Lead Author four times, and temperatesification of and the tropical hydrological rainforest cycle. biomass, This could and affect indi- as a Lead Author twice. rectly through altered geomorphic processes. In today’s environment, most of the world’s car- bon storage occurs in environments that have ex- tremes of temperature (low and high), rainfall and insolation (Fig. 1)! Tropical rainforests store large

6 amounts of carbon because the long growing season, high rainfall, insola- tion and temperature allow high rates of plant growth and high carbon stor- age in biomass. High latitude ecosys- tems have low rates of plant growth, as a result of temperature, insolation Carbon storage in and growing season length, but have terrestial ecosystems even lower respiration rates, leading (Tonnes per ha) to high carbon storage in detrital ma- 0 to 10 10 to 20 terial. These two zones are the north 20 to 50 50 to 100 and south “poles” of the terrestrial car- 100 to 150 150 to 200 bon cycle and any change in their area 200 to 300 300 to 400 400 to 500 global carbon cycle. More than 500 or conditionsThe tropical is forestslikely to have influence the ability the

cle present a great challenge to car- Figure 1. Mapped terrestrial carbon stocks, “As the climate changes, bon scientists. The bulk of our carbon showing the concentration of high storage observing system is in the northern per unit area in high and low (tropical) lati- environments considered hemisphere mid-latitudes. For exam- tude regions. These regions are crucial to fu- extreme relative to current ple about 85% of all eddy covariance ture carbon-climate coupling yet are poorly understood and poorly observed (http:// sites are located between 30o and 50o conditions may become www.grida.no/publications/rr/natural-fix/ N; similar statistics apply to most other page/3724.aspx). more or less common.“ carbon-related observations [9]. While the current network of ecosystem re- search sites spans climate space (tem- will dominate 21st century carbon cycle to adjust carbon storage quickly since perature and precipitation) fairly well feedbacks. they have high rates of plant produc- [10], the highest photosynthesis and Our current observations in the tivity, but the resulting carbon is large- carbon storage regions are sampled ex- world’s extremes of temperature and ly stored in plant biomass () that tremely sparsely compared to the mid- rainfall are inadequate to constrain can decline or be lost quickly [5]. If dis- latitudes. The most diverse regions models of the future, which continue to turbances increase oxidation of bio- of the world, such as the humid trop- diverge [12]. Redoubled effort to sam- - ics, where carbon responses to climate ple these regions in situ is crucial (see, tially instantaneous. The combination could be the most variable between for instance, http://above.nasa.gov/) ofmass high through productivity fire, losses and can short be essen resi- species are the worst sampled! as are creative new remote sensing ap- dence times of storage in biomass al- The temperate mid-latitudes may proaches (http://science.jpl.nasa.gov/ low the tropics to respond quickly to have played a disproportionate role projects/CARVE/). Hopefully this is- climate shifts. in recent carbon uptake [11], but they sue’s highlighting of the challenges of The polar regions are extreme in have relatively low stocks of carbon their climate, but also in their combi- compared to the tropics and high lati- nation of insolation and temperature. tudes and so are not forecast by models “In today’s environment, The region of extreme winter cold is to play a major role in future carbon- almost certain to decrease with warm- climate feedbacks. The tendency of sci- most of the world’s carbon er climates, but as noted in Williams entists and funding agencies in the de- storage occurs in environ- (this issue), much else is less clear. As veloped world to prefer local projects ments that have extremes high-latitude ecosystems warm, the po- has limited the acquisition of data in tential for respiration of the vast soil the crucial regions. But this is not the of temperature, rainfall carbon reservoirs to increase is large, whole story. The cost and effort to ob- and insolation.” while the potential for GPP to increase serving carbon dynamics in the high may be limited by light and short sun- and low latitudes is formidable, with lit season length [8]. This is one of the limited access, to power and other in- understanding both temporal and geo- most worrisome tipping points in the frastructure restricting the amount of graphic extremes will lead to increased Earth System, as it creates an environ- data collection (for instance: http:// focus in the iLEAPS and broader com- ment where photosynthesis is limited to www.zottoproject.org/, http://earth- munity of the gaps in the global sci- a shorter period than respiration, in the observatory.nasa.gov/Features/LBA/), ence program and innovative efforts to presence of large reservoirs of carbon. yet these extreme environments dom- address them. IGBP has tabled a plan The two poles of the carbon cy- that includes some of these elements,

inate carbon fluxes and storage and 7 including a focus on under-sampled References 1. Reichstein, M., Bahn, M., Ciais, P., Frank, D., tions 23:1743–1744. Schuur EAG, Abbott BW, regions, careful coupling of in situ and Mahecha, M.D., Seneviratne, S. I., Zscheischler, Bowden WB et al. 2013. Expert assessment of remote observations and new part- J., Beer, C., Buchmann, N., Frank, D. C., Papale, vulnerability of permafrost carbon to climate nerships to reduce sampling bias out- D., Rammig, A., Smith, P., Thonicke, K., van der change. Climatic Change, 119:359-374. side the developed world. The effort, Velde, M., Vicca, S., Walz, A. and Wattenbach, 7. Kattge L, Diaz S Lavorel S et al. 2011. TRY – a M. (2013) Climate extremes and the carbon global database of plant traits. Global Change dubbed the “Merton Initiative” [13], is cycle. Nature, 500, 287-295. Biology, 17: 2905–2935 an effort to focus the international sci- 2. 8. Jung M et al. 2009. Towards global em- ence community’s attention on the op- factor. Nature 420, 29-30 pirical upscaling of FLUXNET eddy covari- portunities resulting from a more inte- 3. SchimelMoore, D,D Nand Trajan, Baker P D Wilkes, 2002. TheT Quaife, B ance observations: validation of a model grated approach to global observations Stephens, K Elder, A Desai, J Negron and R ensemble approach using a biosphere model. Monson. Persistent reduced ecosystem respi- Biogeosciences, 6, 2001-2013. ration after insect disturbance in high eleva- 9. Schimel DS 1995. Terrestrial ecosystems leading role in realising this vision! tion forests. 2013. Ecology Letters (2013) 16: and the carbon cycle. Global Change Biology and field projects. iLEAPS can play a 731–737 1: 77-91. [email protected] 4. Cox PM et al. 2013. Sensitivity of tropical 10. Anav A et al. 2013. Evaluating the Land [email protected] carbon to climate change constrained by car- and Ocean Components of the Global Carbon bon dioxide variability. Nature 494, 341–344. Cycle in the CMIP5 Earth System Models. J. Cli- 5. Lewis SL et al. 2011. The 2010 Amazon mate, 26, 6801–6843. drought. Science, 331: 554. 11. Schimel D and Cox P 2012. The Merton 6. McGuire, D., L. D. Hinzman, J. Walsh, J. Hob- Initiative (a Word document downloadable bie, and M. Sturm 2013. Trajectory of the Arctic from internet with Google query “the Merton as an integrated system1. Ecological Applica- initiative”)

12th AsiaFlux workshop Bridging Atmospheric Flux Monitoring to National and International Climate Change Initiatives International Rice Research Institute, College, 4030 Los Baños, Laguna, Philippines

Important dates: 30 April: Deadline of abstract submission 15 May: Notification of abstract acceptance, Deadline for business display application 22 May: Registration opens 30 June: Deadline for early registration and payment of early registration fees 18-19 August: Pre-conference training course (optional) 20-22 August: Conference (including field visits to IRRI site and facilities) 23 August: Field trip (optional)

Further information: www.asiaflux.net/asiafluxws2014/

8 B Science

John M Cheeseman University of Illinois, Urbana IL, USA

Exploiting plants from extreme environments: have we both world enough and time?

The increases in global temperature goal of mitigating at least some of them, over the last four decades, coupled one approach is to develop new plant with more severe droughts in many ar- genetic models based on species al- eas and more generally extreme weath- ready adapted to the more extreme con- er variability are seriously challenging ditions. Advances in genomic research many marginal, already stressed eco- have shown such promise, based espe- systems [1]. Unfortunately, one of the cially on comparative genomics of hal- main lessons already available from such environments (such as deserts and saline ecosystems) is that, with “The challenge for society the proper attention (or lack of it), any- thing can be killed. is to produce more food Plants in extreme environments are and other agricultural John Cheeseman is professor emeritus in the often already operating at their lim- commodities at least sus- Department of Plant Biology at the Univer- its. For example, mangroves and salt sity of Illinois. He is an organismal biologist, marshes, adapted to living with fre- tainably enough to pre- having begun his career as an environmen- quent seawater inundations, are nev- vent the total collapse of tal physiologist and expanded his studies to ertheless both vulnerable to sea level the world’s ecosystems.“ both the molecular and the ecosystem levels. rise. And although they have always His major research has been on halophytes weathered severe storms, they are at (salt-adapted plants), especially mangroves. risk as the violence of those storms in- ophytes (salt-adapted plants, such as His work has taken him to both terrestri- creases [2, 3]. Both systems are also the Thellungiella spp.) and their non- al and coastal ecosystems throughout the US, in the Caribbean, Western Australia and very slow to recover if degraded. More- adapted relatives (Arabidopsis thali- Queensland, New Zealand, Tanzania, and over, their degradation results in sig- ana). Modern sequencing techniques Saudi Arabia. are setting the stage for understand- mobilised through decay of their high- ing how genomes have evolved to deal lynificant organic release soils [4]. of greenhouse gasses, with extreme environmental condi- Similarly, in deserts, despite adap- tions. Such “extremophiles” may both tation of endemic species to generally show what plants can actually do un- warm, dry conditions, they can be made der sever conditions and how, and un- vulnerable by prolonged drought, even veil genetic traits that may be usefully if it is not apparently prolonged or se- transferred to crops [6]. vere enough to kill them. Then, given another similar event even a long time principles” may be an attractive ap- (in human terms) afterwards, they suc- proachWhile for workingmolecular from geneticists, such “first bi- cumb [5] (Fig. 1). ochemists and physiologists, exactly To understand the potential future what this means in practise is still far effects of changing climates, with the from clear. This is true in particular

9 could be useful. However, he noted that projected increase in the next 36 years - the current trend is unquestionably for is nearly equal to the current combined erance”,about some the requirementscentral but poorly for salt defined man- the major cereal crops to become even populations of China and India [10]. No agementconcepts, bysuch the as plants the definition themselves, of “tol or more major and for lesser crops to be- the mechanisms of water acquisition come even more “lesser”. on by climate change will be reduced Nevertheless, attention to halo- byecological a higher or population. social difficulties brought phytes in particular has increased over Already in the current climate, “Both natural and engi- the last 20 years, associated with in- the UNDP estimates that 44% of the neered plants with ex- creasing, often total, losses of agricul- world’s cultivated systems are dry- treme stress tolerance tural lands to salinisation. Potential bi- lands. These are home to 2.3 billion omass production by some halophytes people in 100 countries, including half must be part of the solu- rivals or exceeds that of non-halophyt- of the world’s poor [11]. Further, near- tion. It only remains to be ic crops, especially those used for for- ly 40% of irrigated agricultures depend seen how – not if – we will age or oil production. For example, the on ground water reserves that are also seeds of Salicornia bigelovii contain being depleted and degraded. Nearly do this.” up to 30% oil and 35% protein, and its one third of the world’s irrigated are- productivity under saline conditions as are affected by salinity or salinity-re- and transport in dry soil. It requires - lated problems, mostly associated with not only understanding the mecha- rent world average for other major oil nisms by which tolerant plants are tol- can be significantly higher than the cur- Figure 1. An ancient Utah juniper (Juniperus erant, but also those by which intoler- line conditions [9] (Fig. 2). osteosperma) in Arches National Park, Utah, crops, such as sunflower, under non-sa ant plants are sensitive [7]. USA. The oldest known individual in Utah is Alternatively, in this new world, “But at my back I always hear more than 1275 years old. While they can some already adapted species may be Time’s winged chariot hurrying near; grow at high altitudes, on very thin, nutrient useful directly in mitigation or in crop And yonder all before us lie poor, rocky soil in severely water-limited con- substitution. There are a large number Deserts of vast eternity.” ditions, they are also highly sensitive to the of extremely tolerant plants currently - Andrew Marvel “correct” duration and repetition of drought used for food which could potential- [5]. As is indicated by the amount of dead ly be cultivated much more widely or Clearly, large-scale exploitation of ei- wood on this tree, one of their mechanisms for survival is to grow slowly and, during which could potentially tolerate even ther native or engineered plants will droughts, to decrease the amount of living future changing climates. Harlan [8], take time. Time, however, may be in material that needs to be supported with nu- for example, while noting that a mere short supply: at least as important as trients and water. Nevertheless, plants such handful of crops form the basis of the climate change itself and too often ig- as these harbor a wealth of genetic informa- world’s diets, provided an 11-page nored in discussing its effects is the in- tion, including a genetic history of having “short list” of other world crops that creasing human population. This can- dealt with climate change for centuries. not be left out of the equation. The

10 Figure 2. Salicornia europaea (saltwort, glass- wort, samphire and other common names) is one of a number of Salicornia species being in- vestigated for fodder and oil production using seawater irrigation. The plants themselves are edible, but the seeds are the primary target for halophyte agriculture, being high in oil and protein. The oil is suitable for biodiesel pro- duction. The plants shown here were growing on the side of a 5M NaCl stream at the base of a nearly pure NaCl mountain at the K&S potash mine near Gransee, Germany. Production tri- als are currently underway in the Middle East (Kuwait, Saudi Arabia), Eritrea, and Mexico.

2. Blasco F, Saenger P, Janodet E 1996. Man- groves as indicators of coastal change. Catena, 27, 167-178. 3. Kirwan ML et al. 2010. Limits on the adapt- ability of coastal marshes to rising sea level. Geophysical Research Letters, 37, L23401. 4. Donato DC et al. 2011. Mangroves among the most carbon-rich forests in the tropics. Nature Geoscience, 4, 293-297. 5. Breshears DD et al. 2005. Regional vegeta- tion die-off in response to global-change-type drought. Proceedings of the National Academy of Sciences of the United States of America, 102, 15144-15148. poor irrigation practices [12]. This in- potatoes). Moreover, were they avail- 6. Oh D et al. 2012. Life at the extreme: les- cludes both developing and developed able, if conversion to those crops were sons from the genome. Genome Biology, 13, countries, the two most affected be- limited to, or seemingly forced upon, 241-249. 7. Cheeseman JM 2013. The integration of ac- ing India and the US. Layered on top developing countries, yet another gap tivity in saline environments: problems and of this is the competition between ag- between rich and poor nations would perspectives. Functional Plant Biology, 40, riculture and development, a competi- be created. The sociological aspects of 759-774. tion that is increasing because of both this may well make their adoption dis- 8. Harlan JR 1992 Crops & Man. (2nd ed), climate change and shifts in population tasteful or even impossible, as it has for Madison, WI: American Society of Agronomy. 284 pp. from the countryside to cities. The net GMO crops already [13]. 9. Glenn EP, Brown JJ, Blumwald E 1999. Salt effect, a decline in water available for tolerance and crop potential of halophytes. growing food and an even greater de- history and future of our planet, Home, Critical Reviews in Plant Sciences, 18, 227-255. crease in the available land per capita concludedIn 2009, with a documentary the statement film that on the“it 10. United Nations, Department of Economic can only be exacerbated by accelerating is too late to be a pessimist” [14]. The and Social Affairs, Population Division (2013). World Population Prospects: The 2012 Revi- land and water degradation. problems of food supply and security sion, Highlights and Advance Tables. Working The challenge for society is clearly facing the earth’s entire human popu- Paper No. ESA/P/WP.228. to produce more food and other agri- lation are daunting, but the alternative 11. UNDP. Sustainable Land Management. cultural commodities on less land with to solving them is unspeakable. Both 2013 [Accessed November 17, 2013]; Avail- less water, and to do so at least sustaina- natural and engineered plants with ex- able from: http://www.undp.org/content/ undp/en/home/ourwork/environmentand- bly enough to prevent the total collapse treme stress tolerance must be part of energy/focus_areas/sustainable_landman- of the world’s ecosystems. Whether the solution. It only remains to be seen agement.html. this can be done for even the current how – not if – we will do this. 12. FAO. 2011. The state of the world’s land population is doubtful. It is equally and water resources for food and agriculture [email protected] (SOLAW) – Managing systems at risk. Food and Agriculture. Organization of the United based on plants adapted to extreme en- Nations, Rome and Earthscan, London. doubtful that a global scale techno-fix vironments is possible. On the global References 13. Belay M, Guri B. 2013. Why African farm- scale, there are no potential new crops - 1. Fischlin A et al. 2007. Ecosystems, their ers do not want GMOs. [Accessed January 30, properties, goods, and services, in Climate 2014]; Available from: http://truth-out.org/ pre-adapted to stress or manipulable to Change 2007: Impacts, Adaptation and Vul- increase yield while preserving stress news/item/20058-why-african-farmers-do- nerability. Contribution of Working Group II not-want-gmos. to the Fourth Assessment Report of the Inter- tolerance - that have the potential to 14. Arthus-Bertrand Y. 2009. Home. [Accessed produce the calories and proteins con- governmental Panel on Climate Change, Parry November 24, 2013]; Available from: http:// sumed from our current major crops ML et al. (eds.) Cambridge University Press, www.homethemovie.org/ Cambridge, UK. (such as rice, wheat, maize, pulses and

11 B Science

Miguel D. Mahecha1 Jakob Zscheischler1,2, Anja Rammig3 1. Max Planck Institute for Biogeochemistry, Jena, Germany 2. Max Planck Institute for Intelligent Systems, Tübingen, Germany 3. Potsdam Institute for Climate Impact Research (PIK), Potsdam, Germany

Challenges in quantifying global carbon cycle extremes

The question of how frequencies, intensities, and spa- tial extents of climate extremes change in the wake of climate change is increasingly in the focus of climate science and emphasized by the IPCC [1]. Multiple lines of evidence further suggest that these pulses of ex- treme climate conditions can affect the terrestrial car- bon cycle of most ecosystem types [2]. In particular, a series of well-documented case studies report that carbon-sequestering ecosystems (“carbon sequestra- tion” is the process of capture and long-term storage of atmospheric carbon dioxide) forfeit their accumula- tion potential under climate extremes (recall the stud- Miguel D. Mahecha studied geo- Jakob Zscheischer obtained his de- ies on the European heat wave 2003 [3], the 2000- ecology at the University of Bay- gree in mathematics from the Hum- 2004 US drought [4], or the 2005 and 2010 Amazon reuth. He leads the research group boldt University in Berlin in 2010. drought [5]). Hence, we have good reasons to expect on “Empirical Inference in the Earth Since then he has been pursuing additional releases of land carbon if certain climate ex- System” at the Max Planck Insti- his PhD at the interface between tremes intensify over the coming decades. One of the tute for Biogeochemistry. He has geoscience and machine learning at immediate goals of our studies is therefore to investi- pursued research into ecosystem- the two Max Planck Institutes: Bio- gate to what extent extreme events constitute a posi- atmosphere interaction and bio- geochemistry and Intelligent Sys- tive feedback to global change. diversity patterns. He is particu- tems. His research focuses on the To reach this goal we have to carefully analyse the larly interested in exploring hidden detection and quantification of- ex ecological consequences of extreme climate events [6]. ecological patterns and processes treme events in variables describ- Moreover, it is particularly important to understand and in heterogeneous data sources by ing the state of the ecosystem and means of new data mining tools their attribution to climate condi- quantify where climate extremes have led to extreme and time series analytics. tions and other drivers. changes in the carbon cycle, that is, in changes of carbon

Anja Rammig is a scientist the heat wave 2003, most available monitoring data indi- stocks or fluxes [2]. For example, during the European “Earth System Analysis” depart- ment at Potsdam Institute for Cli- land carbon stocks [3]. Though one could deduce from mate Impact Research (PIK), Ger- studiescated clear of this changes kind thatin carbon we are fluxes well implyingequipped depleted from an many. Her research focuses on observational perspective, the current data archives are ecosystem responses to cli- often sparse in space and fragmented in time. mate change, in particular on - ex This remark is in no way meant to downgrade treme events. She is interested in the obvious merits of the steadily growing global car- developing process-based dynamic bon cycle data archives. Especially eddy covariance global vegetation models for site- specific applications and testing models with observational data. FLUXNET network (in tandem with regional initia- tives)measurements have substantially of ecosystem improved carbon our fluxes process from under the-

12 approach it has been shown that a few extremes explain most of the interannu- al variability of GPP [11]. In Fig. 1, we show the integrated decrease in GPP during the 100 largest extremes over time. Clearly, the areas around the black sea, Germany and France are particular- ly prone to dramatic impacts on GPP. A key question is what type of cli- mate extremes has the strongest ten- dency to cause pronounced impacts on the carbon cycle. Hence, in a second ex- ample, we detect coincidences between extreme decreases of in net primary

production (NPP; net CO2 assimilated by the vegetation, i.e. GPP- autotroph- ic respiration) as simulated by the LP- Figure 1. Average decrease in GPP caused by tremes in empirically upscaled gross JmL dynamic vegetation model [12] the 100 largest spatiotemporal extremes af- primary production (GPP; the total and climate extremes during the grow- fecting the terrestrial biosphere. amount of carbon assimilated via pho- - tosynthesis into vegetation) over three ductions of NPP are driven by hot sum- decades in Europe [9]. Given that we mersing season. as well We as find by thatdry substantialsummers, butre standing. The main problem is that the cannot rely on long time series, we ex- the largest decreases in NPP are found probability of observing the impact of ploit the highly correlated spatial in- when hot phases coincide with dry pe- climate extremes is much lower than formation to robustly identify extreme riods (Fig. 2). the occurrence probability of extremes impacts. Using a three-dimensional The second example analysis gives anywhere in the world. To put it in oth- (latitude x longitude x time) segmen- a hint that compound events may have er words: we are plagued by sparse tation approach [10] results in spati- the strongest effect on NPP (at least in sampling and short time series. Statisti- otemporally connected extremes that cal frameworks built to estimate, for in- can be sorted by their spatial extents, Figure 2. A modelling experiment: Central stance, the recurrence probabilities of length, or integral effect on GPP. The European growing season NPP (net primary idea is that even if the data are noisy production) during normal growing seasons to carbon cycle research. Alternatives and locally uncertain, we can identify (black) compared to years with extremely low 100-year floods are not (yet) applicable are, for instance, long-term data on tree extreme impacts by analysing where NPP and coinciding extreme warm summers, ring width [7] or annual crop yields [8], the extreme behaviour affects large ar- extreme dry summers, as well as dry and hot both of which allow, however, only indi- eas for longer time periods. With this summers. rect assessments of extreme impacts on the carbon cycle. Another option is to rely on either gridded, empirically derived or pro- et al. [9] have shown that machine-learn- cess-based modelled flux fields. Jung to learn the nonlinear relation between ing methods are sufficiently powerful local CO2 sensing information. This integration can be exploited flux observations to generate and spatiotem remote- porally continuous reconstructions of - ellite era. Results from process-based modelsland-atmosphere allow us to fluxes draw a over longer-term the sat picture of the impacts of climate ex- tremes on the carbon cycle, for instance by differentiating types of extreme cli- matic constellations and investigating coincidences of these with extremes in ecosystem variables. As an example, we explore here ex-

13 Figure 3. Definition of univariate extremes (external panels and straight lines in central plot) and of multivariate (compound) events (central panel, enveloped line). The triangle enveloped by the threshold of univariate ex- tremes and multivariate extreme delineates the region of a (bivariate) ex- treme is not captured by a one-dimensional detection approach. The size of this region depends on the selected quantile and increases if a more ex- treme quantile is chosen.

References 1. IPCC 2012: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. A Special Report of Work- ing Groups I and II of the Intergovernmental Panel on Climate Change [Field CB et al. (eds.)]. Cambridge University Press, Cambridge (UK), New York (USA). 2. Reichstein M et al. 2013. Climate extremes and the carbon cycle. Nature 500, 287-295. 3. Ciais P et al. 2005. Europe-wide reduction in primary productivity caused by the heat and drought in 2003. Nature 437, 529-533. 4. Schwalm CR et al. 2012. Reduction in car- bon uptake during turn of the century drought in western North America. Nature Geoscience 5, 551-556. 5. Phillips OL et al. 2009. Drought sensitivity of the Amazon rainforest. Science 323, 1344-1347. 6. Reyer C et al. 2012. A plant’s perspective of extremes: terrestrial plant responses to perception is of a more general na- only die once” [6]). In this context, one changing climatic variability. Global Change Biology 19, 75-89. ture.our specificFor instance, model Leonard run). Indeed, et al. [13] this alsospecific has droughtto understand tolerances to what (“You extent can 7. Babst F et al. 2012. 500 years of regional show that in situations where various ecophysiological adaptations at eco- forest growth variability and links to climatic variables are extreme (“compound ex- system level can attenuate the impacts extreme events in Europe. Environmental Re- treme”), we can expect the most severe of climate extremes, e.g. via changes in search Letters 7, Art. 045705. impacts. However, non-extreme drivers stand structure. The ultimate hope is to 8. Wu X et al. 2013. Climate-mediated spati- can also cause extreme impacts espe- derive more general conclusions from otemporal variability in the terrestrial pro- ductivity across Europe. Biogeosciences Dis- cially if rare and unfavourable constel- - cussions 10, 17511-17547. lation of climatic drivers occur. While pacity in anticipating extreme impacts 9. Jung M et al. 2011. Global patterns of this can be easily imagined in a two- onfield the observations global terrestrial to improve carbon ourcycle. ca variable system (Fig. 3), it rapidly gets How robust are the results from latent heat, and sensible heat derived from land-atmosphere fluxes of carbon dioxide, challenging as the number of consid- terrestrial biosphere models when it eddy covariance, satellite, and meteoro- logical observations. Journal of Geophysical ered variables increases. Clearly, future comes to the analysis of carbon-cycle Research - Biogeosciences 116, G00J07, studies need to provide us with a better extremes? So far, most model bench- doi:10.1029/2010JG001566. understanding of the consequences of mark systems emphasise the perfor- 10. Zscheischler J et al. 2013. Detection and compound extreme events for the ter- mance of models for reproducing mean attribution of large spatiotemporal extreme restrial biosphere. events in Earth observation data. Ecological Informatics 15, 66-73. Besides analysing compound trends [15] rather than evaluating 11. Zscheischler J et al. 2014. Few extremes events, yet another important aspect thefluxes, tails seasonal of the data cycles, distributions and long-term (ex- dominate global interannual variability in currently discussed by climate scien- treme events). Our objective, howev- gross primary production. Environmental Re- tists is the question of “unprecedent- er, is to interpret these models under search Letters 9, in press. ed climate extremes” [14]. In the con- circumstances that were not in the fo- 12. Sitch S et al. 2003. Evaluation of ecosystem dynamics, plant geography and terrestrial car- text of land-atmosphere interactions, cus of the initial developments. Polem- bon cycling in the LPJ dynamic global vegeta- one can think of an intensity of an ex- ically speaking, we are in a situation tion model. Global Change Biology 9, 161-185. treme event that has not been expe- where unintentional model use meets 13. Leonard M et al. 2013. A compound event rienced since the establishment of a observation scarcity. Consequently, framework for understanding extreme im- - our branch of science needs to consid- pacts. Wiley Interdisciplinary Reviews: Cli- mate Change, doi:10.1002/wcc.1252. portant to study the impacts of unprec- er the role of extremes when it comes 14. Anderson A and Kostinski A 2010. Revers- edentedspecific ecosystem.spatial extents Likewise, or durations it is im of to structural model-developments or ible record breaking and variability: temperature extreme events. Our immediate ques- model-data integration exercises. distributions across the globe. Journal of Applied tion here is whether it makes a sub- Meteorology and Climatology 49, 1681-1691. [email protected] stantial difference to ecosystems if cer- 15. Luo YQ et al. 2012. A framework for bench- marking land models. Biogeosciences 9, 3857- tain thresholds are passed, e.g. plant 3874.

14 One of the biggest challenges of current climate research is to analyse, understand, and pre- dict the regional effects of global climate change, especially to describe consequences and impacts of climate change on a relevant scale for society. The Climate Initiative REKLIM (Regionale Klimaänderungen / Regional Climate Change) (www.reklim.de), a consortium of nine research centres in the Helmholtz-Association and the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research are pleased to invite you to the International Conference “Our climate – our future, Regional perspec- tives on a global challenge”, bringing together scientists from all over the world to present their results and discuss current and future perspectives of a challenging issue, the regional aspect of global climate change. The scientific programme will offer a broad and interdis- ciplinary spectrum of current international and national research work on regional climate change. The REKLIM conference 2014 will be held from 6 - 9 October, 2014 at the Umweltforum in Berlin, Germany. For more information, the session programme and the registration proce- dure please visit our conference webpage: www.reklim-conference-2014.de

15 B Science

Nate G. McDowell1, Jeffrey Q. Chambers2

1. Earth and Environmental Sciences Division, Los Alamos National Laboratory, Los Alamos, NM, USA 2. Earth Sciences Division, Climate Sciences Department, Lawrence Berkeley National Laboratory, Berkeley, CA, USA Measuring and modelling global vegetation dynamics in response to chronic and extreme climate changes

Why must we understand, monitor, and simu- late global vegetation disturbances? Terrestrial vegetation is among the most important yet vulnerable climate regulators on Earth. Terrestri- al ecosystems sequester ~30% of anthropogenic fos- sil emissions annually and regulate precipitation and temperature. However, these services are already di- minishing through both chronic and abrupt events that lead to ecosystem disturbance [1,2]. Dynamic global vegetation models (DGVMs), the essential land- surface boundary component of climate models, re- quire accurate treatment of disturbances to simulate terrestrial climate forcing and improve climate change Nate McDowell has been a staff Jeff Chambersis an Associate Pro- predictions over the 21st century. Here, we highlight scientist in the Earth and Environ- fessor in the Geography Depart- the importance of improving global measurements mental Sciences Division at Los ment, University of California, and simulations of disturbances in response to both Alamos National Laboratory (LANL) Berkeley, and a Faculty Scientist in chronic and extreme disturbance drivers. since 2004. He received his PhD the Climate Sciences Department Multiple are exhibiting accelerating rates in tree physiology at Oregon State at Lawrence Berkeley National of vegetation mortality at continental scales in associ- University in 2002. He has pub- Laboratory. He received his PhD ation with warming surface temperature (Fig. 1), and lished approximately 80 papers in ecology from the University of major regional “die-offs” have been associated with since 2000, and advised approxi- California, Santa Barbara, in 1998. mately 20 graduate students and He has authored and co-authored periods of low precipitation or wind storms super- 25 postdocs. He was awarded about 50 peer-reviewed manu- imposed upon this warming [3,4,5,6,7]. These are not LANL and the US Department of scripts, including first-authored pa- just increases (for instance, from 1% to 2% mortali- Energy‘s Distinguished Mentor pers in Nature, Science, and PNAS, ty), but accelerations (for instance, increasing by ~1% Awards for advising undergradu- and has advised and co-advised each decade) [3, 6]. These increases in background ates in 2008 and 2010 respective- over 30 postdocs and graduate stu- mortality are insidious because they are relatively un- ly, and testified before Congress dents. He served as co-Director for noticed by the casual observer, yet they cause long- regarding DOE’s climate change DOE’s National Institute for Climate term reductions in ecosystem services such as carbon research in 2009 and before the Change Research (NICCR) Coastal storage. For example, a modelled doubling of the mor- House of Representatives in 2012 Center, and has led a number of in- tality rate from ~1 to 2% for a Central Amazon forest regarding climate impacts on for- ternational science projects funded caused a > 50% reduction in above-ground biomass ests of the intermountain west. by the DOE and NASA. with a lag-time exceeding 50 years [8]. In more dra- matic fashion, some regions have experienced mas- sive, widespread “die-offs” recently [5,9]. Thus, both chronically increasing background mortality as well as extreme “die-off” events should cause large shifts

16 out increasing extreme precipitation or wind events. Firstly, rising temperature has di- rect, negative effects on plant carbon storage because it increases respira- tion; it also affects the plant indirect- ly by increasing the vapour pressure

a higher likelihood of hydraulic failure, carbondeficit (VPD),starvation, which, and in insect turn, and leads path to- ogen attack [12,17]. This results in an exponential relationship between for- est mortality resulting from biotic at- tack and VPD, and an ominous forecast for future forest survival (Fig. 2B) [7]. The negative impacts of increasing VPD are particularly concerning be- cause VPD is rising faster than surface temperature and because it depends exponentially on temperature (despite Figure 1. Background mortality rates are in- in part through evaluation against eco- rising humidity). This suggests that creasing throughout much of North America, system-scale climate manipulations DGVM simulations currently under- regardless of regional climate, species, or [13,14,15] (Fig. 3), although consid- predict mortality (compare Figs. 2A land management history (reproduced from erable work remains. As a result, an and 2B; note 2A also does not include [3] and [6]). untenably wide range of future car- bon storage forecasts exist, making conservative). These hypotheses must in regional carbon and energy balances. befire tested simulation, by experiments making it even designed more A weakening of the terrestrial carbon all forecasts predict a positive climate to unravel cause-and-effect (Fig. 3). sink is therefore expected if mortality warmingour confidence feedback weak due at to best a decreasing (although Strong support for the negative impact and decomposition of forest necromass terrestrial carbon sink [12,16]). of temperature and VPD on plant sur- (mass of dead plant material) is a signif- vival has now been demonstrated in ex- icant global process. Disturbance mechanisms periments with many species (numer- Predictions consistently suggest decreasing forest survival in the future. Clarifying if increasing background For example, both empirical models mortality rates are anomalous and un- Figure 2. a) DGVM simulations of forest loss and DGVMs predict that North Ameri- derstanding the climate factors un- from NW and SW North America show SW can forests will lose >50% of their cur- derlying increasing mortality are chal- USA will have lost all of its current conifer for- rent distribution before 2100AD [7,10] lenging because baselines for historic ests by 2050, with NW following closely be- (Fig. 2). Despite the consistency be- hind [10]. b) Forest mortality always occurred tween observations and simulations, length to partition natural versus an- in SW USA when the unitless forest growth index (based on tree rigns) reached -1.4 (grey - thropogenicmortality rates driven are rarelyimpacts. of sufficientTheoret- bars indicate severe droughts). Rising temper- mains weak because of a paucity of ical and empirical understanding of our confidence in DGVM accuracy re ature along with minor changes in forecasted knowledge regarding the physiological temperature impacts on physiology, precipitation suggest all SW USA forests will and abiotic mechanisms of disturbance however, support the concept that ris- have exceeded the megadrought threshold by [11,12]. Simulation of DGVM mortali- ing temperature is underlying increas- 2050 [7]. c) A recently dead old-growth Pinus ty mechanisms has recently improved ing mortality rates, both with and with- edulis tree in New Mexico, USA (next page).

a b

17 ly, key measurements such as agents c of mortality are often not measured in

occur at an annual time step required forforest DGVM field evaluation,inventories, and few inventories are limited to regions of particular in-

A near-term solution lies with val- idationterest or ofaffluence remote [reviewed sensing byproducts 11,12]. across regions where we have data and then application of evaluated al- gorithms to global remote sensing datasets [20]. A globally comprehen- sive disturbance and mortality moni- toring system would allow substan- tial increases in our knowledge of the spatial and temporal patterns of mor- tality, thereby enabling analyses of cli- matic and anthropogenic drivers that precede mortality events. In addition, a global vegetation mortality monitor- ous examples can be found in [12,13]. bon from live to dead pools across re- These mechanistic experiments that kill gions [18], a chronically stronger wind Figure 3. Examples of experimental manipula- are highly valuable for testing the can also increase mortality in a more tions needed for testing DGVM process in re- underlying theory regarding the accel- subtle manner: an increase in natural lation to chronic and extreme events. A) Eco- erating rates of mortality, and are essen- “background” tree mortality via snap- system scale precipitation reductions, such tial for DGVM evaluation [13,14] (Fig. 3). ping, uprooting by wind, or partial can- as this one shown in a moderately wet tropi- Secondly, the atmospheric energy opy loss [19]. cal forest in Caxiuanã, Brazil, allow investiga- - Ground-based observations are at tion into the impacts of a sustained drought mate. This increases the intensity of cy- least as important as mechanistic ex- on forest survival and mortality [14]. B) The cycleclones intensifies and convective with stormsthe warming and leads cli periments (Fig. 3) for understanding addition of heat to drought manipulations al- to an increase in wind-driven tree mor- mechanisms driving vegetation mor- lows examination of the interactive impacts tality, creating another pathway for tality because they can also be used of multiple drivers of environmental stress on increasing disturbance regimes [4]. for evaluating remote sensing observa- forest survival and mortality. Pictures in panel A and B are from Patrick Meir and Josh Smith, While storms such as hurricanes cause tions and DGVMs at appropriate scales respectively. huge and immediate transfers of car- (Lichstein et al. in press). Unfortunate-

a b

18 ing system is absolutely necessary for to individual trees, often with a patchy vances are required to move the predic- DGVM benchmarking, parameterisa- distribution across the landscape at a tive science of vegetation mortality for- tion, and structural improvements to scale that may be below the resolution ward. We need broad coverage of the mortality algorithms. Currently, bench- of most remotely sensed products (for - marking for land-surface models in- instance 250 m MODIS, 30 m Landsat). ventory plots that provide a) improved - Recent developments suggest 30-m and statisticalEarth’s terrestrial samples surface of the with state field of rein- 1-m-resolution imagery can be used to gional vegetation, b) annual census, cludes (among other parameters) flux capture the bulk of mortality and dis- and c) tallying of agents of mortality in “A global vegetation mor- turbance suggesting this challenge can - be surmounted [18,22,23,24]. stand patterns and drivers, and evalu- tality monitoring system Secondly, attributing the remote- atefield DGVMs notes, [11,12,23].so that we canAn observation then under- is absolutely necessary ly sensed signal to mortality per se, as al network of plots that for DGVM benchmarking, opposed to canopy litterfall or die-back includes all six forested continents cou- that is (or is not) recovered in subse- pled with high-resolution remote sens- parameterisation, and for ing data would be extremely valuable structural improvements extremely important to predicting fu- for establishing baselines and then de- to mortality algorithms.” turequent vegetation years [19], that is difficult is critical but forcan acbe- tecting shifts in mortality regimes that curate DGVM simulations. Active re- result in climate-relevant feedbacks to mote sensing such as LIDAR and radar the Earth system. This network will es and inventory-based biomass; both can alleviate some of the challenges by be essential to test and calibrate glob- of these are extremely valuable to as- providing information to changes in al standardised high-resolution (30 m sure that models are capturing criti- ecosystem structure (such as canopy or less) remote sensing datasets [20]; height). Overall, attribution of cause(s) however, this simultaneously requires and energy balance [21]. However, DG- of changes in ecosystem structure may that the remote sensing products are VMscal processes simulate that migration define carbon,of plant water, func- be the single largest challenge to a broadly available to the research com- tional types based on survival, mortal- globally comprehensive vegetation dy- munity. ity, dispersal and regeneration, and are namics detection system. We must improve links among the run at regional or global scales, thus modelling (Fig. 2), experimental (Fig. these processes must also be provid- Preparing for the future 3), and observational (Figs. 1, 4) sci- ed as benchmarks at the appropriate ence communities to ensure that ad- scales for DGVM evaluation. vances in process-based understanding - climate-terrestrial impacts and feed- - lenges with any observational sys- backsIt is extremely because difficultof the myriad to predict complexi future- dence in our understanding of forest temThere of vegetation are multiple disturbance. significant Firstchal- ties involved and because of the lack of dynamicsare relevant and to their DGVMs. climate Gaining dependen confi- ly, drought-induced mortality occurs evaluated global datasets. Multiple ad- cies and in our ability to accurately sim-

Figure 4. The first user-friendly, interactive web-based tool that allows examination of the global patterns of forest loss and gain since year 2000 via a 30-m resolution Landsat analysis constitutes a major breakthrough. Below is a series of scenes of forest gain and loss (blue is gain, red is loss, green is forest extent) [20]), zooming from A) the Amazon basin, to B) the Manaus region, to C) the ZF2 research site north of Manaus, and D) evalu- ation of (C) against field-validated mortality observations also using Landsat imagery from Negron-Juarez (2010). Comparing panels (C) and (D) highlights that while the new analysis is a great step forward, mortality flux datable with Landsat is missing at finer scales. E) Pan-tropical patterns of forest loss and gain [20] highlight strong regional differences, for instance, high disturbance rates (largely an- thropogenic fire and ) in Brazil, Malay- sia, and Indonesia versus lower disturbance impact in the Congo basin and Papua New Guinea.

19 ulate vegetation survival and mortality 5. Carnicer J et al. 2011. Widespread crown 16. Arora VK et al. 2013. Carbon-concentra- in relation to chronic and extreme dis- condition decline, food web disruption, and tion and carbon-climate feedbacks in CMIP5 - Earth system models. Journal of Climate. - mate change-type drought. Proceedings of the 17. Breshears DD et al. 2013. The critical role Nationalamplified Academy tree mortality of Sciences with 108, increased 4: 1474- cli of increasing atmospheric moisture demand continuesturbances to could change have and a leads significant to anom in- 1478. on tree mortality and associated regional die- alousfluence effects on policy on society. as the Earth’s climate 6. Peng CZ et al. 2011. A drought-induced per- off. Frontiers in Plant Science, 4, 266-269. vasive increase in tree mortality across Cana- 18. Negrón-Juárez RI, Chambers JQ, Guimaraes [email protected] da’s boreal forests. Nature Climate Change 1,9: G et al. 2010. Widespread Amazon forest tree [email protected] 467-471. mortality from a single cross-basin squall line 7. Williams AP, Meko D, Woodhouse C et al. event. Geophysical Research Letters 37, 16: 2013. Temperature as a potent driver of re- L16701. Acknowledgments: Writing of this gional forest drought stress and tree mortal- 19. Saatchi S et al. 2013. Persistent effects of manuscript was supported by the US ity. 2013. Nature Climate Change 3: 292-297 a severe drought on Amazonian forest canopy. - 8. Chambers JQ et al. 2004. Response of tree bi- Proceedings of the National Academy of Sci- omass and wood litter to disturbance in a Cen- ences 110.2: 565-570. ical and Environmental Research. Sup- Department of Energy, Office of Biolog tral Amazon forest. Oecologia 141, 4: 596-611. 20. Hansen MC, Potapov PV, Moore R et al. port for the inspiration of this manu- 9. Phillips OL et al. 2011. Drought–mortality 2013. High-Resolution Global Maps of 21st- script was provided by the European relationships for tropical forests. New Phy- Century Forest Cover Change. Science 342, Carbo-Extreme project and the Inte- tologist 187, 3: 631-646. 6160: 850-853. grated Network for Terrestrial Ecosys- 10. Jiang XS et al. 2013. Increasing risk of 21. Luo YQ, Randerson JT, Abramowitz G et al. tree mortality in western North America 2012. A Framework for Benchmarking Land tem Research on Feedbacks to the At- towards the end of the 21st century. Jour- Models. Biogeosciences, 9(10): 3857–3874. mosphere and Climate (INTERFACE) nal of Geophysical Research, 10.1174/JCLI- 22. Garrity SR et al. 2013. Quantifying project. D-12-00430.1 drought-induced tree mortality in a piñon- 11. Allen CD et al. 2010. A global overview of juniper woodland using multitemporal high drought and heat-induced tree mortality re- spatial resolution satellite imagery. Remote References: veals emerging climate change risks for for- Sensing of Environment, 129:54-65. 1. IPCC, 2013. Summary for Policymakers. In: ests. and Management 259: 23. Chambers JQ, Negron-Juarez RI, Magna- Climate Change 2013: The Physical Science Ba- 660-684. bosco D et al. 2013. The steady-state mosaic sis. Contribution of Working Group I to the Fifth 12. McDowell NG et al. 2011. Interdependence of disturbance and succession across an old- Assessment Report of the Intergovernmental of mechanisms underlying climate-driven veg- growth Central Amazon forest landscape. Pro- Panel on Climate Change [Stocker, T.F., D. Qin, etation mortality. Trends in Ecology and Evo- ceedings of the National Academy of Sciences G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung, lution, 26, 523-532. 110, 10: 3949-3954. A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. 13. McDowell NG, Fisher RA, Xu C et al. 2013. 24. Meddens AJH et al. 2013. Evaluating meth- Cambridge University Press, Cambridge, Unit- Evaluating theories of drought-induced veg- ods to detect bark beetle-caused tree mortal- ed Kingdom and New York, NY, USA. etation mortality using a multi-model-exper- ity using single-date and multi-date Landsat 2. Reichstein M, Bahn M, Ciais P et al. 2013 iment framework. New Phytologist 200(2), imagery. Remote Sensing of Environment 132: Climate extremes and the carbon cycle. Nature 304-321. 49-58. 500, 7462: 287-295. 14. Powell TL et al. 2013. Confronting model 3. van Mantgem PJ, Stephenson NL, Byrne JC et al. 2009. Widespread increase of tree mor- of Amazon forests subjected to experimental tality