Climate change and energy policy: The importance - Keck Science
Post on 25-Feb-2022
1 Views
Preview:
Transcript
ARTICLE IN PRESS
0360-5442/$ - se
doi:10.1016/j.en
�CorrespondE-mail addr
Energy 32 (2007) 262–268
www.elsevier.com/locate/energy
Keynote Paper
Climate change and energy policy: The importanceof sustainability arguments
Roland Clift�
Centre for Environmental Strategy, University of Surrey, Guildford, Surrey GU2 7XH, UK
Received 1 November 2005
Abstract
It is the stated policy of the UK government to reduce emissions of carbon dioxide by 60% by 2050. This policy, which goes far beyond
commitments under the Kyoto agreement, was originally advocated by the Royal Commission on Environmental Pollution, of which the
author was a member. Its acceptance was seen by many as a surprising development, possibly reflecting the strength of the underlying
case. The target was developed by a three-legged argument which reflects the three components of sustainability:
Environmental constraints: limits on emissions to avoid risk of major climate change;
Social equity: equal per-capita allocation of emissions;
Techno-economic: the feasibility and cost of reduction on this scale.
Assessment of techno-economic feasibility shows that the target can be achieved economically if the efficiency of energy use is
improved to achieve reduction in demand, combined with a shift to lower-carbon energy sources. The greatest scope for demand
reduction lies in improving the building stock, combined with providing low-grade heat from sources such as biomass. On the supply
side, the principle questions are how much controllable electricity generation is needed, and whether this capacity should be nuclear or
fired by fossil fuels with the carbon dioxide formed sequestered in geological strata. Increased use of biomass is a key part of the shift to a
lower carbon economy; the barriers which have retarded the development of biomass in the UK are explored.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Energy policy; Climate change; Energy scenario
1. Introduction
The author was, until 2005, a member of the RoyalCommission on Environmental Pollution (RCEP), auniquely UK institution appointed to advise on matters,
both national and international, concerning the pollution of
the environment; on the adequacy of research in this field;
and the future possibilities of danger to the environment. TheRCEP has been in continuous existence but with rotatingmembership since 1970. It is constituted as a body ofindependent experts, standing outside the political main-stream but intended to provide advice to guide long-termenvironmental policy. An important role for the RCEP isto present scientific evidence and its implications whichmay be obvious to the scientific community but politically
e front matter r 2006 Elsevier Ltd. All rights reserved.
ergy.2006.07.031
ing author. Tel.: +441483 689271; fax: +44 1483 686671.
ess: r.clift@surrey.ac.uk.
inconvenient. Some of the RCEP’s reports have had asignificance which extends outside the UK; for example,the concepts of Integrated Pollution Control and BestPracticable Environmental Option originated with theRCEP.In a report published in 2000 [1], the Royal Commission
addressed the issue of climate change and its relationship toenergy supply. The principal recommendation of theRCEP’s analysis was that carbon dioxide emissions fromhuman activities in the UK must be reduced by 60% below1998 levels by 2050. The year 2050 was itself chosencarefully: even the most enthusiastic advocates of nuclearfusion did not consider that it would be available before2050, so that the target would have to be met by knownenergy technologies.1 At the time, this target appeared to
1RCEP also recommended a decrease of 80% by 2100, but this longer-
term target is not discussed in this paper.
ARTICLE IN PRESSR. Clift / Energy 32 (2007) 262–268 263
be too radical to be politically acceptable; for example, itgoes way beyond negotiations under the Kyoto protocol.However, it was accepted, is now UK Government policyand progress towards the target is currently under review.The present paper reviews the argument which supportedits acceptance and some of the implications of this policy.
2. The context: sustainable development
The concept of sustainable development is now suffi-ciently well known that it will not be rehearsed here.However, a particular interpretation of sustainable devel-opment which has proved useful in teaching the concept toengineers and scientists [2] will be reviewed, because itillustrates and supports the argument developed by theRCEP [1].
The basic idea is that human activities are limited bythree sets of long-term constraints, summarised by theVenn diagram in Fig. 1. Eco-centric concerns represent theconstraints imposed by the fact that the earth is, inthermodynamic terms, a closed system. Thus, energy flux isreceived from the sun, but the ‘‘capital’’ resources availableto us on a global scale are finite, as is the capacity of thebiosphere to absorb or adapt to the emissions from humanactivities. Techno-centric concerns represent the constraintsimposed by finite human abilities: the technology which weare able to deploy and the economic system within whichwe deploy that technology. This lobe represents thetraditional scope of engineering. Socio-centric concerns
represent human expectations: the need to provide a better
quality of life for everyone, now and in the future (to quotefrom the UK Government’s interpretation of sustainabledevelopment). This lobe incorporates the principles ofinter- and intra-generational equity which are central to theconcept of sustainable development. Sustainable develop-ment involves moving towards complying with all threesets of constraints, not of trading off one set of objectivesagainst another. Sustainability is to be found in the centralarea of Fig. 1, which meets all the sets of constraints, while
ECO-CENTRICCONCERNS
Natural resourcesand ecological
capacity
Techno-economicsystems
TECHNO-CENTRICCONCERNS
SOCIO-CENTRICCONCERNS
Human capital andsocial expectations
Fig. 1. Sustainability expressed as long-term constraints [2].
sustainable development is a process of moving towardsthat region. While it is recognised that Fig. 1 is a simplisticrepresentation of a very rich concept, it has neverthelessproved to be useful as an educational device.Although not stated explicitly in the RCEP report, it is
implicit that future use of fossil hydrocarbons will beconstrained not by their availability but by the capacity ofthe biosphere to adapt to the emissions resulting from theiruse, specifically carbon dioxide. In other words, we alreadyknow the whereabouts of more fossil fuels than can beburned without risking disastrous climatic impacts; hydro-carbons in particular will continue to be available, albeit atprices well above historical levels (and possibly abovecurrent levels). To support this argument, current crude oilprices make it not only economic but very profitable toexploit oil sands; the quantity of ‘‘synthetic crude’’available in Alberta is of the same order as the oil reservesin Saudi Arabia—a simple fact which is sometimesoverlooked in discussions over the extent of remaining oilreserves. The driver for change in energy technology istherefore the effect of the emissions, not limited supply; asSheikh Yamani famously put it, ‘‘The stone age did not endbecause we ran out of stones’’. While the conventionalmarket system can deal with scarcity of supply throughrising prices which make new reserves economic, it isnecessary to invent a new mechanism to deal with scarcityof ‘‘carrying capacity’’ to absorb emissions. It remains tobe seen whether emission trading systems like that beingintroduced in Europe will constitute an effective market,whether the price will be sufficiently high to influenceenergy use and whether carbon prices will be sufficientlystable for effective long-term planning and restructuring ofthe energy system.The Royal Commission sidestepped the continuing
arguments over whether the effects of carbon dioxideemissions, can be represented in terms of a simpleeconomic damage cost (or ‘‘externality’’) by addressingenergy policy and climate change in a different way usingthe approach, introduced above, of recognising and tryingto estimate the constraints.
3. Policy recommendations
The Royal Commission’s report starts with an analysisof the evidence that emissions of climate-forcing ‘‘green-house gases’’ from human activities are causing changes inthe global climate and regional weather patterns. In effect,it endorsed the conclusions of the UN IntergovernmentalPanel on Climate change (IPCC). To quote the RCEPreport [1], the world is now faced with a radical challenge of
a totally new kind which requires an urgent responsey By
the time the effects of human activities on the global climate
are clear and unambiguous it would be too late to take
preventive measures. This statement may not be news to thescientific community in Europe, but was necessary at thetime of publication (2000) to underpin the Commission’srecommendations; although the UK government now
ARTICLE IN PRESSR. Clift / Energy 32 (2007) 262–268264
stresses its belief that climate change represents a real andserious threat, this commitment has emerged since pub-lication of the RCEP report.
Of more interest here is the analysis which followed,which had three principal steps:
1.
If the concentration of carbon dioxide in the atmospheregets too high, then there will be a risk of reallycatastrophic climate change associated, for example,with changes in ocean circulation patterns. RCEPproposed 550 ppm as the ceiling, a constraint whichmust not be breached. Some would prefer a lowerlimit—at 550 ppm, the predicted rise in sea level is stillbad news for low lying Pacific islands, for example—andinformation on the effects of climate change which hasbecome available since 2000 arguably supports a lowerfigure. However, 550 ppm already represents a majordeparture from ‘‘business as usual’’ [3] and there wasalso a measure of political realism behind the figure (seebelow).2.
Working back from the 550 ppm constraint, RCEPestimated the total tolerable CO2 emissions, and thendivided them by total global population to get the percapita ration. Multiplying by the population of the UKgave about 40% of current emissions—hence therecommended target of a 60% reduction below 1998levels. This is known as the ‘‘contract-and-converge’’principle. It simply ignores economists’ attempts tocalculate an optimal level of CO2 emissions, on the basisthat the damage cost or ‘‘externality’’ figures mean littlewhen the climatic system is so complex and non-linear(and therefore chaotic) that detailed predictions ofclimate change and its effects, as distinct from estimat-ing thresholds or constraints, are meaningless.
3.
2Off-peak ‘‘night storage’’ heating is still used in the UK. It has been
argued, principally by advocates of nuclear power, that this is to be
encouraged, on the basis that future energy systems will be dominated by
low-carbon fixed-output electricity generation. RCEP did not subscribe to
this view, rejecting its implication that new buildings should be equipped
with electrical heating to fit hypothetical energy system which might
possibly exist at some future date. A subsequent study by the House of
Lords [4] endorsed this conclusion.
To explore the practicality of the 60% reduction target,RCEP developed representative scenarios to illustratehow it might be achieved (see below). Based on thesescenarios, the cost of shifting to a low-carbon energyeconomy was subsequently estimated as about 2% ofannual GDP. The cost is sometimes described as‘‘enormous’’ but, as an economist will always ask,compared to what? The economists who produced theseestimates also project 4% annual growth in GDP. Sotaking action to limit the potentially catastrophic effectsof global climate change would merely slow down therise in consumer spending in developed countries. Giventhat most forms of consumer spending in industrialisedcountries lead to carbon dioxide emissions, there is anargument that this in itself could have some effect inconstraining the growth of climate-forcing emissions.
It is interesting that, although not explicitly formulatedin terms of the approach to sustainability summarised inFig. 1, the three steps in the argument map exactly onto thethree sets of constraints: ecological, societal and techno-economic. Perhaps this illustrates the strength of thisapproach to sustainability analysis. To quote the RCEP
report again [1], y the UK could cut its carbon dioxide
emissions by 60% by 2050. Achieving this will require vision,
leadership, and action which begins now. Acceptance of thetarget of 60% reduction by 2050 has, so far at least,survived subsequent discussions over commitments underthe Kyoto process and softening of UK government targetsfor shorter term reductions. In fact, the UK governmenthas been advocating international acceptance of the 2050target, and the Swedish government has also adopted it asa basis for energy policy.
4. Energy scenarios
In order to explore the implications of a 60% reductionin carbon dioxide emissions, the RCEP constructed fourscenarios to illustrate measures by which it might beachieved. These scenarios are only illustrations, notintended to be interpreted as predictions or projections,but they enabled rough estimates of the cost of the change(see above). Scenarios 1 and 4 were deliberately selected asextremes—respectively, ‘‘techno-fix’’ and ‘‘ecological liv-ing’’ —with scenarios 2 and 3 intended to represent morerealistic energy futures. They had a number of commonfeatures. They were based on known technologies, assum-ing that electrical generation by nuclear fusion is notavailable by 2050 (see above). They took as their startingpoint the 1998 pattern of energy use in the UK,summarised in Table 1. Fig. 2 shows energy use updatedto 2003, while Fig. 3 shows how UK carbon dioxideemissions have changed since 1990. Compared to con-tinental European countries at comparable latitudes,domestic energy use is high because of the relatively poorenergy performance of the UK building stock and thealmost complete absence of heating systems distributing‘‘low-grade’’ heat; the principal fuels for space and waterheating are natural gas and electrical resistive heating.2 Forobvious geographical reasons, the electricity grid in the UKis separate from that of continental Europe (apart fromlimited imports via a cross-channel link from France).Most of the decline in carbon dioxide emissions since 1990is attributable to replacement of coal by natural gas inelectricity generation (Fig. 3) although coal use has risenagain since 1999. However, the proportion of renewablesources in the UK electricity system is still low byEuropean Standards: only just over 1% of total inlandenergy use even in 2004. These features of the UK energysector present possible savings in carbon intensity whichare not available in other European countries, and which
ARTICLE IN PRESS
Table 1
Final energy consumption in 2050 by end-use (annual averages in GW) [1]
End-use category Scenario
1 (as 1998) 2 3 4
High-grade heat 87 44 44 30
Electricity 32 24 24 21
Low-grade heat 16 12 12 11
Transport 70 53 53 47
Total 205 132 132 109
Transport
Domestic
Industry
Services (including agriculture)
12.3%
21.9%
30.1%
35.5%
Fig. 2. Final energy consumption in the UK, 2003 [5].
60
50
40
30
20
10
0
Mill
ion
tonn
es o
f ca
rbon
1990 1993 1996 1999 2003Power stations
Transport
Services and agriculture Other sectors
Domestic
Industrial combustion
Fig. 3. UK carbon dioxide emissions by source [5].
Table 2
Outputs from energy sources in 2050 (annual averages in GW) [1]
Source 1998 Scenario
1 2 3 4
Intermittent renewables
On-shore wind 0.10 6.5 3.3 0.2 3.3
Off-shore wind 11.4 11.4 11.4 5.7
Photovoltaic 10.0 5.0 0.5 0.5
Wave 3.75 3.75 3.75 3.75
Tidal stream 0.25 0.25 0.25 0.25
Tidal barrage 2.2 2.2 0.0 2.2
Total 0.10 34.1 25.9 16.1 15.7
Other renewables
Hydro 0.61 0.89 0.89 0.89 0.79
Energy crops — 10.2 10.2 1.8 1.8
Bio-waste 0.04 5.7 5.7 5.7 1.2
MSW 0.15 1.9 1.9 0 0
Total renewable 0.90 52.8 44.6 24.5 19.5
Low-C large-scale
electrical generation
11.4 52 0 19 0
Other fossil 266 106 106 106 106
3The role of air transport was considered in a subsequent RCEP study
[6], which has also triggered an intense debate over the need to constrain
growth in demand.
R. Clift / Energy 32 (2007) 262–268 265
just might enable economic growth to be decoupled fromcarbon dioxide emissions.
The final energy demands in the RCEP’s four scenariosare summarised in Table 1. The corresponding energysupply systems are summarised in Table 2, in terms ofoutput of electricity, heat and intermediate energy carriers(primarily hydrogen for use as a transport fuel). Inaddition, ‘‘high-grade’’ heat, used primarily for energy-intensive industrial processes, is supplied in all scenarios bynatural gas. Industrial, commercial and domestic demandfor electrical energy is reduced in some scenarios throughimprovements in the efficiency of appliances and machines,but it is recognised that such reductions would represent a
reversal of current trends. Low grade heat is used primarilyfor space- and water-heating in domestic and commercialbuildings. The reductions in some scenarios are assumed tobe achieved primarily by improving the energy efficiency ofthe UK building stock, with further reductions in carbonemissions achieved by using biomass and waste as fuels inheat-only and CHP plants. Transport is seen as continuingto be dependant on fossil hydrocarbons, although thecarbon efficiency of transport is assumed to improvethrough a combination of incremental technologicaldevelopment, larger shifts such as use of hydrogenproduced from fossil fuels but with higher well-to-wheelcarbon efficiency, and modal shifts (primarily from road torail) to carry both people and goods with lower carbondioxide emissions.3
Scenario 1 represents a ‘‘technology can fix it’’ approach.It assumes that final demand returns to the 1998 levels andthat the 60% reduction is achieved solely by technologicalchange with the maximum deployment of renewables andmaximum use of electrical generation using nuclear sourcesor fossil fuels. In this scenario, energy crops, agriculturaland forestry waste and municipal solid waste (MSW) areused as fuels in CHP plants or heat-only plants generatinglow-grade heat output.
Scenario 2 and 3 assume that energy demand can bereduced by 36% below the 1998 level. The largest reductionis in low-grade heat (see Table 1) brought about primarilyby improving the energy efficiency of the building stock. Inscenario 2 the reduced demand is met by a combination ofrenewable sources and fossil fuels, while scenario 3 is based
ARTICLE IN PRESSR. Clift / Energy 32 (2007) 262–268266
on renewable sources plus large controllable generatingplants using nuclear fission or fossil fuel with sequestrationof carbon dioxide.
Scenario 4 represents the most extreme case: a very largereduction in demand (corresponding, inter alia, to repla-cing or retrofitting most of the UK building stock toNordic standards) supplied from renewable sources withno large generating stations. However, the reduction indemand means that the energy provided by renewablesources can still be less than in scenario 1.
The scenario analysis underlines the importance ofimproving the efficiency of energy use in the UK, andhighlights the need to reduce energy use in buildings and touse appropriate fuels such as biomass to provide ‘‘low-grade’’ heat [4]. Agricultural waste, energy crops and MSWrepresent significant resources which are currently essen-tially unused in the UK; their significance is discussedfurther below. This emerges as being at least as significantas the more obvious public debate over the future ofnuclear power for electricity generation.
The scenarios also differ in their assumptions over theenergy sources used for electricity. The use of renewables isbased on assessments of the cost-effective availableresource in the UK [7]. There is limited hydro-electricgeneration capacity available, and the scenarios assumeonly modest expansion. On- and off-shore wind is assumedto expand, a development which has already started in theUK. Wave generation is assumed to have developed tomake a modest contribution by 2050, with a further smallcontribution from tidal-stream turbines. Some scenariosalso assume construction of tidal barrages; there issignificant technical scope for developing this resourcebut it may not prove to be environmentally or sociallyacceptable.4
Even with maximum use of renewable sources, there iscontinuing debate over the proportion of embedded andpossibly intermittent generation which can be accommo-dated in the system for distribution and supply of electricalenergy. Fast-response high-density energy storage is clearlyimportant to smooth out short-term imbalance betweensupply and demand. In the UK, this is currently achievedby pumped storage but the scope for expansion is at bestlimited; thus there is a clear need for new energy storagetechnologies.
Scenarios 1 and 3 assume that such new technologies donot emerge, so that intermittent renewable generation ofelectricity must be ‘‘backed up’’ by a proportion ofcontrollable generating stations. In the kind of lowercarbon energy systems exemplified by these scenarios, thefuel options for large-scale generation are nuclear fission oralternatively fossil fuels with the carbon dioxide captured
4One of the areas which has been considered for a tidal barrage but
rejected on environmental grounds is the Bristol Channel in the South
West of England. However, it is possible that this project could be
revisited, on the argument that it would actually provide environmental
protection against some of the effects of rising sea level.
and sequestered in geological strata, primarily salineaquifers; the estimated capacity in the UK sectors of theNorth and Irish seas is sufficiently large to accommodateemissions way beyond 2050 [1]. ‘‘Other fossil’’ in Table 2refers to uses from which the carbon dioxide is emitted tothe atmosphere. The scenarios also implicitly assume thathydrogen used in transport is produced from hydrocarbonsbut in dedicated reforming plants from which the carbondioxide is sequestered.Thus the scenarios illustrate that the significant techno-
logical choice, at least for the UK, is not, as presented inthe mass media, between renewable and nuclear power.Rather, the choice is between nuclear power and fossil-firedgeneration with carbon sequestration, with the proportionof either of these sources dependant on the availability ofstorage technology. The feasibility of carbon dioxidesequestration should be a central concern, along with thequestion of how much generating capacity is needed to‘‘back up’’ intermittent renewable sources.The significance of this analysis is that, with the possible
exception of scenario 4, the 60% reduction is achievablewithout totally novel technologies and without completechange in social and economic structures or lifestyles. Asnoted above, these scenarios provided the basis forassessment of the economic cost of switching to a lower-carbon economy. There is little doubt that the assessmentof the cost of the transition was an important considerationleading to the political acceptance of the RCEP’s ‘‘head-line’’ recommendation of a 60% reduction in carbondioxide emissions by 2050.
5. Use of biofuels
A feature of the RCEP report [1], not usual in policydocuments, is frequent reference to the laws of thermo-dynamics. The primary reason for this is that UK energypolicy has concentrated on the electricity sector, almost asif heat is not a form of energy. Results of this focus onelectrical generation can be seen in the relatively slowdevelopment of the Combined Heat and Power (CHP)sector in the UK, the general resistance of local authoritiesand the construction sector to contemplate anything otherthan single-dwelling space and water heating, and coolingtowers—those obvious symbols of energy waste—atelectrical generating stations. The RCEP study tries tobring heat provision more centrally into energy policy.As a further result of the policy neglect of heat,
biological energy resources remain essentially untapped inthe UK: agricultural wastes, energy crops and MSW.5
A more recent RCEP report [8] examined biomass, toinvestigate its possible contribution to energy supply in the
5For these purposes, MSW is treated as a renewable fuel, on the basis
that it is available anyway but currently consigned mainly to landfill.
Public resistance to recovering energy from solid waste is a UK
peculiarity, not discussed here.
ARTICLE IN PRESS
Fig. 4. Supply and demand of wood in Great Britain (i.e. England, Wales
and Scotland) [8].
R. Clift / Energy 32 (2007) 262–268 267
UK and why the sector has been so slow to developcompared to other European countries.
Austria has the most developed biomass sector inEurope. Starting from small-scale heating plants, the sectorin Austria has grown over some 20 years to the point wherebiomass provides around 15% of primary energy. This hasenabled certain Austrian provinces to achieve one of theholy grails of sustainable development: decoupling eco-nomic growth from carbon to the point where GDP hasincreased but carbon dioxide emissions have decreased. InDenmark, biomass is now widely used, often co-fired withcoal, primarily in CHP plants. These are usually built fortheir heat output, for example into community heatingsystems, with electrical output used to back-up intermittentrenewables, primarily wind. In Sweden, biomass hasdeveloped more recently but rapidly, initially for commu-nity heating and then CHP plants operated, as inDenmark, over a range of heat/power ratio. It is thereforeanomalous that, in the UK, renewables still only accountfor just over 1% of total inland energy use, of whichbiomass is only a small fraction.
The RCEP diagnosed one of the reasons for this failureas the complex and confused government support forbiomass, exacerbated by the focus on generating electricalpower with no corresponding incentives for renewable heatproduction. The focus on power generation seems to havebeen amplified by the notion that ‘‘high technology’’processes can be developed for exploitation of biomassleading to possible export markets. One result of thisapproach has been the conspicuous failure of the ARBREproject in South Yorkshire, an electricity generating plantfuelled by short-rotation coppiced (SRC) willow and usingSwedish biomass gasification technology. The plant failedto progress beyond start-up to enter service—for reasonswhich are hotly discussed.
The RCEP argued that, rather than being seen as apossible fuel for power generation, biomass should be seenprimarily as a local fuel for heating or CHP. Subsequentstudies have confirmed this conclusion [9,10]. Local supplychains must therefore be developed and, until the long-term demand is clear, it is unreasonable to expect farmersto plant energy crops even when planting grants areavailable. Co-firing of biomass with coal in electricalgenerating stations has been promoted as an interimmeasure to develop a market for biomass (although RCEPwere critical of some restrictions imposed by the UKregulator which greatly increase the cost of co-firing atexisting power stations) by attracting the financial creditsavailable for renewable electricity. However, while creditsare available to promote the development of renewableelectricity supply, there has been a marked reluctance onthe part of the UK government to extend a similarapproach to renewable heat supply—a further manifesta-tion of the neglect of heat as a significant part of the totalnational energy budget. The Biomass Task Force, set up bythe Department of the Environment Food and RuralAffairs (Defra) following the RCEP, argued that, if
biomass is in competition with hydrocarbon fuels atcurrent prices, the economics should favour biomass soclearly that renewable heat credits are not needed [9].The detailed analyses of the availability of bio-energy
broadly confirmed the earlier estimates summarised inTable 2. Substantial quantities of unused biomass arealready available from poorly managed forest, mainly ofbroad-leaf trees which give a product with relatively highcalorific value. Furthermore, significant quantities ofbiomass will become available over the next 20 years fromforest plantations, primarily of Sitka spruce, developed toprovide pulpwood based on projections of demand whichnow appear to have been too high. Fig. 4 shows estimatesby the Forestry Commission for the availability of wood inthe UK. They show that there is already a surplus of supplyover demand, and it appears that these figures are actuallyunderestimates of the available resource [11]. The surplusbecomes much larger if agricultural residues—primarilystraw—are included. This persuaded the RCEP of the needto develop a market to stimulate demand for the existingresource before there is any point in encouraging plantingof energy crops.However, energy crops will also be needed if the biomass
sector develops in the UK as it has elsewhere in Europe.Based on assessments of energy yield per hectare, SRCsalix (willow) appears to be generally the highest yieldingcrop but other plants such as miscanthus (elephant grass)may be preferred where favoured by local climate and soilconditions. Fig. 5 shows the way in which the biomasssector could develop in the UK, using the estimated currentsupplies of wood from Fig. 4 together with agriculturalresidues as a basis for developing the market for energycrops, up to the level of primary energy supply in 2050needed to provide the delivered energy required byscenarios 1 and 2 in the earlier RCEP report [1]. A phased
ARTICLE IN PRESS
Fig. 5. Scenario to achieve 16GW of energy from forestry, straw and
energy crops [8].
R. Clift / Energy 32 (2007) 262–268268
approach to the development of bioenergy in the UK wasforeseen:
Phase 1 (2004–2012): Increasing use of both agriculturaland forestry resources, and of set-aside land for energycrops.
Phase 2 (2012–2018): Area producing energy cropsincreases to include all set-aside land.
Phase 3 (2018–2025): Energy crops become establishedas accepted main crops.
Phase 4 (2025–2050): The area of land under energycrops becomes a significant proportion of total availableagricultural land.
It is clear that competition for land use will be aconstraint on the development of the bioenergy sector ifanything like the scenario in Fig. 5 develops. The RCEPanalysis leads to the conclusion that the focus should be onbiofuels for heat or CHP rather than for transport. Only ifsurplus land is available once the heat demand is metshould transport fuels be considered, because the energyyield and carbon reduction per hectare are lower than forsalix or miscanthus; the RCEP analysis suggests that thiswill never be the case, providing a post hoc justification forthe assumption in Tables 1 and 2 that transport willcontinue to represent the primary demand for hydro-carbons. This conclusion runs counter to the EU’s policy ofpromoting biofuels for transport, and therefore begs thequestion of why transport fuels, with their less efficient useof agricultural land, have developed ahead of fuels for heatand CHP. One possible reason, which seems plausible tomany involved in this sector, is that biofuels for transporthave provided a way to support agricultural activities inspite of the Common Agricultural Policy. More pragma-tically, the market for liquid fuels such as biodiesel has
developed more rapidly because the supply chain forvegetable oils was already in place. This serves to underlinethe importance of developing the supply chain for otherbioenergy crops.
6. Concluding remarks
The work of the Royal Commission on EnvironmentalPollution reviewed here provides both positive andnegative examples of success in using scientific analysis tounderpin policy recommendations: positive in the sensethat the ‘‘headline’’ recommendation on carbon dioxideemissions has been accepted by the UK Government;negative in the sense that more specific recommendations,such as actions to promote the development of biomass asa renewable energy source, have not as yet beenimplemented. However, the overall message, that climatechange is a real threat and that action to mitigate it ispossible, does now appear to be widely accepted. The keyquestion is not whether new energy technologies can bedeveloped; on the contrary, the RCEP analysis shows thatthe necessary reductions in carbon dioxide emissions couldbe achieved using known technology. The real question iswhether the political will can be found to take the necessaryaction.
References
[1] Royal Commission on Environmental Pollution Energy—the chan-
ging climate. London: The Stationery Office, 2000.
[2] Mitchell CA, Carew AL, Clift R. The role of the professional
engineer and scientist in sustainable development. In: Azapagic A,
Perdan S, Clift R, editors. Sustainable development in practice—case
studies for engineers and scientists. Chichester: Wiley; 2004. p. 29–55
[Chapter 2].
[3] Clift R, Hoskins BJ. Energy and climate change—mission possible.
The Chemical Engineer, August Issue 2005. p. 21–5.
[4] House of Lords: Energy Efficiency, Science and Technology
Committee, 2nd Report of Session 2005-6 (HL Paper 21-I), The
Stationery Office, London, 2005.
[5] Department of Trade and Industry Digest of UK Energy Statistics
(DUKES), DTI, 2004.
[6] Royal Commission on Environmental Pollution. The environmental
effects of civil aircraft in flight. London: RCEP, 2002.
[7] Energy Technology Support Unit (ETSU). New and renewable
energy: prospects in the UK for the 21st century—supporting
analysis, ETSU, Marwell, 1999.
[8] Royal Commission on Environmental Pollution. Biomass as a
renewable energy source, RCEP, London, 2004.
[9] Biomass Task Force. Report to Government, Department of the
Environment, Food and Rural Affairs (Defra), 2005.
[10] Paul Arwas Associates. Biomass sector review for the Carbon Trust,
The Carbon Trust, London, 2005.
[11] Davis, J. The Forestry and Timber Association (FTA), Pers. Comm,
2006.
Ecology, 87(8), 2006, pp. 1907–1914� 2006 by the the Ecological Society of America
GOING WITH THE FLOW: USING SPECIES–DISCHARGE RELATIONSHIPSTO FORECAST LOSSES IN FISH BIODIVERSITY
MARGUERITE A. XENOPOULOS1
AND DAVID M. LODGE
Department of Biological Sciences, University of Notre Dame, Notre Dame, Indiana 46556 USA
Abstract. In response to the scarcity of tools to make quantitative forecasts of the loss ofaquatic species from anthropogenic effects, we present a statistical model that relates fishspecies richness to river discharge. Fish richness increases logarithmically with discharge, anindex of habitat space, similar to a species–area curve in terrestrial systems. We apply thespecies–discharge model as a forecasting tool to build scenarios of changes in riverine fishrichness from climate change, water consumption, and other anthropogenic drivers thatreduce river discharge. Using hypothetical reductions in discharges (of magnitudes that havebeen observed in other rivers), we predict that reductions of 20–90% in discharge would resultin losses of 2–38% of the fish species in two biogeographical regions in the United States(Lower Ohio–Upper Mississippi and Southeastern). Additional data on the occurrence ofspecific species relative to specific discharge regimes suggests that fishes found exclusively inhigh discharge environments (e.g., Shovelnose sturgeon) would be most vulnerable toreductions in discharge. Lag times in species extinctions after discharge reduction provide awindow of opportunity for conservation efforts. Applications of the species–discharge modelcan help prioritize such management efforts among species and rivers.
Key words: climate change; freshwater biodiversity; Lower Ohio–Upper Mississippi River Basin;riverine fish; scenarios; Southeastern USA; species–discharge curves; stream flows; water consumption.
INTRODUCTION
The best global evidence suggests that freshwater
biodiversity is more threatened than terrestrial biodiver-
sity by global changes (Ricciardi and Rasmussen 1999,
Jenkins 2003, UNESCO 2003). According to the IUCN
(International Union for the Conservation of Nature),
for example, four freshwater groups (freshwater mussels,
crayfishes, amphibians, and freshwater fish) occupy the
top spots when taxonomic groups are ranked in
declining order of the proportion of U.S. species at risk
of extinction. However, global patterns and determi-
nants of many freshwater taxonomic groups are poorly
known compared to those for terrestrial species (Chapin
et al. 2000, Sala et al. 2000). In particular, quantitative
information on the large-scale responses of freshwater
taxa to anthropogenic environmental changes is largely
lacking. Thus, increased efforts are needed to gather
better quantitative data on global freshwater biodiver-
sity patterns and how they respond to environmental
change. Simultaneously, it is essential for the scientific
community to fully use existing data to guide increas-
ingly urgent conservation efforts for freshwater ecosys-
tems, including attempts to anticipate biodiversity
responses to human impacts and management of fresh-
water ecosystems.
Freshwater ecosystems experience many widespread
human impacts like climate change, as well as more
intense anthropogenic effects because human popula-
tions are concentrated near lakes and rivers (Postel
2000). Waterways are used for navigation, recreation,
irrigation, hydropower generation, sewage disposal, and
human consumption of water, all of which affect habitat
quantity or quality (Table 1). Most studies of regional
and global effects of anthropogenic environmental
changes on freshwater biodiversity have been largely
qualitative (see reviews by Sala et al. 2000, Lodge 2001,
Poff et al. 2001). Although the need for quantitative
models relating freshwater biodiversity to environmental
conditions is great, few such algorithms exist. Thus,
there has been no concerted effort to develop large-scale
quantitative forecasts of changes in freshwater biodiver-
sity based either on known trends in environmental
drivers or on scenarios of environmental change.
Despite the acknowledged scarcity of broad-scale data
on most components of freshwater biodiversity and
most environmental drivers, we draw here on two sets of
available data to provide an example of a quantitative
model, which we apply hypothetically in a forecasting
mode. We hope this effort will stimulate improvements
in assembly and analysis of existing data, foster new
collections of data on both freshwater biodiversity and
environmental drivers, and prompt the development of
better models or better applications of similar models.
Accepting the limitations of easily available data, we
Manuscript received 1 December 2004; revised 2 September2005; accepted 30 September 2005; final version received 9November 2005. Corresponding Editor: O. E. Sala. For reprintsof this Special Feature, see footnote 1, p. 1875.
1 Present address: Department of Biology, Trent Univer-sity, Peterborough, Ontario K9J 7B8 Canada; E-mail:mxenopoulos@trentu.ca
1907
SPECIALFEATURE
narrow our focus to rivers and fishes. We use river
discharge to forecast the loss of species of riverine fishes
under hypothetical (but realistic) scenarios of water loss
such as often occur from many anthropogenic drivers
(Table 1).
Climate change and water withdrawal for irrigation,
for example, often reduce flow rates (Vorosmarty et al.
2000, Vorosmarty and Sahagian 2000, Alcamo et al.
2003). Flow rates in turn determine many of the habitat
characteristics that are critical for aquatic plants and
animals (O’Brien 1987, Matthews 1998), including water
quality (e.g., temperature, solute concentrations), sub-
strate (e.g., rocky, sandy, or silty substrates), and
physical features (e.g., pools, riffles, or runs). Thus, in
both evolutionary and ecological time, larger rivers
provide a greater diversity of conditions and resources
than those with low discharge, enhancing the evolution
of species, increasing immigration relative to extinction
rates, and increasing the potential for species coexistence
within a river ecosystem (Hocutt and Wiley 1986,
Hugueny 1989).
Accordingly, positive relationships between species
richness and river discharge (or its correlate catchment
area) have been documented for fishes and a variety of
aquatic invertebrate taxa (Sepkoski and Rex 1974,
Welcomme 1979, Livingstone et al. 1982, Strayer 1983,
Bronmark et al. 1984, Eadie et al. 1986, Hugueny 1989,
Oberdorff et al. 1995, Guegan et al. 1998, Poff et al.
2001). Species–discharge curves are similar to terrestrial
species–area curves in the sense that richness numbers
increase logarithmically with discharge. Below we
develop regression models relating fish species to river
discharge for two large river basins, and explore the use
of these algorithms to forecast biodiversity loss.
DECLINING DISCHARGE AS A MAJOR ENVIRONMENTAL
DRIVER OF BIODIVERSITY LOSS
Many of the anthropogenic drivers affecting fresh-
water biodiversity exert their impact in part via
reductions in river discharge (Table 1). For example,
all recent scenarios from the Intergovernmental Panel
for Climate Change (IPCC) forecast declining water
availability for some regions of earth (Nakicenovic et al.
2000, IPCC 2001, Xenopoulos et al. 2005). In these
scenarios, reduction in river discharge result from a
combination of reductions in precipitation and temper-
ature-induced increases in evapotranspiration.
In addition to these possible future climate induced
changes, numerous examples exist of past and ongoing
reductions in river discharge in many parts of the globe
(Postel and Richter 2003). For several African rivers,
reductions in downstream discharge from dams (and the
consequent increases in evaporation) are estimated to be
13–26% (Vorosmarty and Sahagian 2002). The effects of
these drivers combined with other impacts (e.g., climate
change, diversions, irrigation, water consumption) can
cause much greater discharge reductions (Vorosmarty et
al. 2000). Although discharge measurements are not
made regularly on many such rivers (especially in lesser
developed countries), reductions in discharge are well
quantified in some cases (see Fig. 1).
Despite the known examples of declines in discharge
(Fig. 1; Vorosmarty et al. 2000, Alcamo et al. 2003) and
the documented positive relationship between species
number and discharge (e.g., Oberdorff et al. 1995),
population reductions or extirpations of fishes as a result
of reduced discharge are surprisingly poorly docu-
mented. In most cases, overlapping temporal sequences
of data on both fish and discharge do not exist, even in
wealthy countries. However, the little existing data is
consistent with the expectation of declines in fishes (or
other taxa) in rivers undergoing reduced discharge (for a
review, see Lake 2003). One example of altered flows
and fish vulnerability is that of the Colorado pike-
minnow (Ptcyhocheilus lucius), the populations of whichhave been much reduced by flow reductions from
damming (O’Brien 1987). The extinction of several
other fishes worldwide can be attributed in part to
reductions in river discharge from a variety of human
activities (Table 2). Since extinction most likely lags by
TABLE 1. Environmental drivers linked with reductions of freshwater biodiversity.
Environmental driver Recent reference
Drivers that can change river discharge
Climate change Findlay et al. (2001)Dams, channelization, siltation, and reservoirs Degerman et al. (2001)Increasing human population size, increased demand for potable water, irrigation Postel and Richter (2003)Physical habitat alteration, poor land-use practices, deforestation, degradation of wetlands Alin et al. (1999)
Drivers that are intensified by reduced discharge
Agricultural pollutants, fertilizer, pesticides, and other intrusive agricultural practices Hites et al. (2004)Eutrophication Hillebrand and Sommer (2000)Exotic species introductions Kolar and Lodge (2002)Acidification Vinebrooke et al. (2003)Industrial pollution, PCBs, dioxins, and others Metcalfe-Smith et al. (1998)Mining Niyogi et al. (2002)Ultraviolet radiation Xenopoulos and Frost (2003)
Other drivers
Over fishing and commercial exploitation Bradford and Irvine (2000)
MARGUERITE A. XENOPOULOS AND DAVID M. LODGE1908 Ecology, Vol. 87, No. 8
SPECIALFEATURE
an unknown but substantial time behind reductions in
discharges (Minckley et al. 2003), many more fishes are
currently vulnerable and may be headed towards
extinction without restoration of flow or other manage-
ment interventions.
MODEL DEVELOPMENT AND HYPOTHETICAL APPLICATION
Given the need to quantify ongoing losses of aquatic
taxa and forecast potential future losses of freshwater
biodiversity under different scenarios of human–envi-
ronment interaction (e.g., IPCC 2001, Millennium
Ecosystem Assessment 2003), we offer here a prelimi-
nary step toward developing the existing understanding
of the species–discharge relationship into a statistical
forecasting tool for biogeographic regions (for a global
analysis, see Xenopoulos et al. 2005). To illustrate this
approach, we use existing data on fish and discharge in a
regression model for two selected biogeographical
FIG. 1. Annual discharge through time for six rivers selected to illustrate anthropogenic reductions in discharge: (a) Syr Darya,Tyumen-Aryk, Kazakhstan; (b) Firat, Keban, Turkey; (c) Colorado River, Lees Ferry, Arizona, USA; (d) Murray River, BadenPowell WTR, Australia; (e) Niger, Ke-Macina, Malawi; (f) Apalachicola River, Sumatra, Florida, USA. Maximum annualdischarge data in panels (a)–(e) are from the Global Runoff Data Centre hhttp://grdc.bafg.dei and mean annual discharge data inpanel (f ) are from the U.S. Geological Survey hhttp://waterdata.usgs.gov/nwisi.
TABLE 2. Examples of riverine fish species whose extinction resulted at least in part from reduced water flows caused by humanactivities (including water diversion, dams, and irrigation).
Fish River
Syr-Darya shovelnose (Pseudoscaphirhynchus fedtschenkoi) Syr Darya River basin, UzbekistanDwarf sturgeon (Pseudoscaphirhynchus hermanni) Amu Darya River basin, UzbekistanSnake River sucker (Chasmistes muriei) Snake River, Wyoming, USADurango shiner (Notropis aulidion) Rio Tunal, MexicoPhantom shiner (Notropis orca) Rio Grande and Pecos River, USALas Vegas dace (Rhinichthys deaconi) Las Vegas Creek, Nevada, USACachorrito del la Presa (Cyprinodon inmemriam) Ojo de Agua La Presa, MexicoWhiteline topminnow (Fundulus albolineatus) Big Spring and Spring Creek, Alabama, USA
Note: Data are from the Committee on Recently Extinct Organisms hhttp://creo.amnh.org/index.htmli.
August 2006 1909DETERMINANTS OF BIODIVERSITY CHANGE
SPECIALFEATURE
regions in the United States as described in Mayden
(1992): (1) Lower Ohio–Upper Mississippi and (2)
Southeastern. For each biogeographic region, we apply
hypothetical reductions in discharge to generate hypo-
thetical reductions in fish species. We limit our analysis
to fishes because data at a comparable scale for other
taxa are not available. Some of the southeastern rivers
are, in fact, already suffering declining discharge from
increasing irrigation and human water consumption
(Elfner and McDowell 2004; Fig. 1f), which is expected
to increase in severity under rapidly growing human
populations (e.g., the city of Atlanta) and some future
scenarios of climate (IPCC 2001, Alcamo and Henrichs
2002).
We obtained current fish species richness numbers
(native species only) by river basin from Hocutt and
Wiley (1986). Discharge data (from the gauging station
closest to the river mouth) were downloaded for 1960–
2000 from USGS and averaged annually (data available
online).2 Although some human influence may already
be incorporated in the species and discharge data that
are available, our application of the model assumes that
such affects are minimal. In other words, we assume that
our species number is in ecological equilibrium with the
measured discharge.
As expected, fish species richness increases logarithmi-
cally with discharge for both biogeographical regions
(Fig. 2). The magnitude of the coefficients of determi-
nation (r2¼ 0.47, 0.61) indicate that roughly half of the
variation in fish species number is explained by river
discharge. Residual variation is likely related to many
other drivers, including historical factors and ecosystem
productivity. Although caution is necessary in applying
these regressions in a forecasting mode (Prairie 1996),
we modeled hypothetical fish extinctions, given hypo-
thetical reductions of 20–90% in river discharge (Table
3). The hypothetical percentage of fish species becoming
extinct ranged among rivers and flow reduction scenar-
ios from 2% to 30% in the Lower Ohio–Upper
Mississippi basin and 3% to 38% in the Southeastern
U.S. biogeographic region (Table 3).
At this stage in the development of this forecasting
approach, it is difficult to draw conclusions about
particular species. However, a reasonable inference is
that fish species limited to high discharge river reaches
would be among the species most certainly committed to
extinction in the absence of flow restoration or other
conservation efforts. For example, the list of obligate
high discharge fishes (large river fishes) in the Lower
Ohio–Upper Mississippi basin and the Southeastern
U.S. region would constitute the minimum number of
fish species that would be most vulnerable to a reduction
in discharge (Table 4). Applying this logic more
specifically to the Mississippi River proper, which has
137 native fishes (Table 3) and supports 20 of the 22
large river species in the entire basin (Table 4, all except
cisco and lake whitefish according to Hocutt and Wiley
[1986]), we might forecast that a reduction in river
discharge between 50% (loss of 14 total species) and 75%
(loss of 26 total species; Table 3) would imperil most of
the large river species in the basin (Table 4). A high
proportion of these large river species are already
endangered because of hydrological alterations to their
habitat (Table 4) and/ or other drivers (habitat
fragmentation and loss) that are expected to have a
greater effect on species loss. For example, the Plains
minnow (Hybognathus placitus) is in decline and
extirpated in some rivers because of reduced discharge
(Nature Serve 2004). More certain species-specific
forecasts would require a detailed analysis of the habitat
requirements of individual species, and how river
discharge interacts with them.
CURRENT LIMITATIONS ON THE USE OF SPECIES–DISCHARGE
CURVES AND RESEARCH PRIORITIES
Species–discharge models are less applicable when riverdischarge increases.—Approximately 50–70% of global
river systems experience increases in discharge under
FIG. 2. Number of fish species (data from Hocutt and Wiley[1986]) plotted against discharge at the river mouth (annualmeans from USGS) for the Lower Ohio–Upper MississippiUSA basin (top panel) and the Southeastern USA basin(bottom panel). Curved upper and lower lines denote the 95%CI.
2 hhttp://waterdata.usgs.gov/nwisi
MARGUERITE A. XENOPOULOS AND DAVID M. LODGE1910 Ecology, Vol. 87, No. 8
SPECIALFEATURE
most current future climate scenarios (e.g., IPCC 2001).
Increases in discharge, though, would not necessarily
lead to increases in global or local fish species richness,
because species evolution typically operates on a longer
time scale than the forecast climate effects. Rapid
increases in local species richness are unlikely because
even if natural trans-basin migration of species is
possible, it would happen more slowly than the
discharge changes in the scenarios. However, if humans
aid introductions of nonindigenous species, increases in
discharge would likely create ecological space favorable
to some such species. Because of the highly uncertain
consequences of increased discharge, however, we
suggest that species–discharge curves should only be
used to forecast reductions in biodiversity that are
expected from decreased discharge.
The magnitude of time lags of extinctions resulting
from reduced discharge are unknown.—Species losses do
not occur instantaneously with reductions in discharge
(Minckley et al. 2003). Rather, populations often decline
slowly, depending on life cycle length and other species
characteristics (e.g., Salvelinus leucomaensis; Morita and
Yamamoto 2002). Research should focus on estimating
such lag times from data on past extinctions and on-
TABLE 3. Forecast of the percentage of fish species becoming extinct if at river mouth dischargewere reduced by 20%, 50%, 75%, and 90% in selected rivers of two biogeographic regions in theUnited States: the Lower Ohio–Upper Mississippi and the Southeast.
River n
Fish species extinct (%) with river discharge reduction of
20% 50% 75% 90%
Lower Ohio–Upper Mississippi
Mississippi 137 4 10 19 30Ohio 98 3 9 18 28Illinois 116 3 7 14 22Wisconsin 115 2 6 11 17
Southeastern United States
Alabama 103 5 13 25 38Apalachicola 68 5 13 24 37Flint 62 4 11 20 31Yellow 66 3 8 15 22
Note: n ¼ number of native fish species.
TABLE 4. Fishes designated by Hocutt and Wiley (1988) as exclusively ‘‘large river’’ species.
Lower Ohio–Upper Mississippi Southeastern United States
Lake sturgeon (Acipenser fulvescens)� Chestnut lamprey (Ichthyomyzon castaneus)Pallid sturgeon (Scaphiryhnchus albus)� Sea lamprey (Petromyzon marinus)Shovelnose sturgeon (Scaphiryhnchus platorynchus)� Shortnose sturgeon (Acipenser brevirostrum)�Paddlefish (Polyodon spathula)� Lake sturgeon (Acipenser fulvescens)�Spotted gar (Lepisosteus oculatus) Atlantic sturgeon (Acipenser oxyrhinchus)Shortnose gar (Lepisosteus platostomus) Shovelnose sturgeon (Scaphirhynchus platorynchus)�Alligator gar (Atractosteus spatula) Paddlefish (Polyodon spathula)�Alabama shad (Alosa alabamae)� Alligator gar (Atractosteus spatula)Skipjack herring (Alosa chrysochloris) Mooneye (Hiodon tergisus)Goldeye (Hiodon alosoides) Chain pickerel (Esox niger)Mooneye (Hiodon tergisus) Silverside shiner (Notropis candidus)Cisco (Coregonus artedi) Quillback (Carpiodes cyprinus)Lake whitefish (Coregonus clupeaformis) Highfin carpsucker (Carpiodes velifer)Lake chub (Couesius plumbeus) Spotted sucker (Minytrema melanops)Western silvery minnow (Hybognathus argyritis) Silver redhorse (Moxostoma anisurum)Plains minnow (Hybognathus placitus) River redhorse (Moxostoma carinatum)Sturgeon chub (Hybopsis gelida)� Black redhorse (Moxostoma duquesnei)Flathead chub (Hybopsis gracilis) Golden redhorse (Moxostoma erythrurum)Sicklefin chub (Hybopsis meeki) White catfish (Ictalurus catus)River shiner (Notropis blennius) Blue catfish (Ictalurus furcatus)Ghost shiner (Notropis buchanani) Atlantic needlefish (Strongylura marina)Silverband shiner (Notropis shumardi) Channel catfish (Ictalurus punctatus)Blue sucker (Cycleptus elongatus) Yellow bass (Morone mississippiensis)Blue catfish (Ictalurus furcatus) Striped bass (Morone saxatilis)
Goldine darter (Percina aurolineata)�,�
Notes: Because these species occur exclusively or primarily in high discharge environments, they would likely be the species mostdirectly affected by reductions in river discharge. Species that have been extirpated in some rivers within the basin are shown inboldface type. Data are from Nature Serve (2004).
� Species on the IUCN endangered or vulnerable species list.� Endemic species.
August 2006 1911DETERMINANTS OF BIODIVERSITY CHANGE
SPECIALFEATURE
going population declines because these delays consti-
tute the window of opportunity for restoration efforts to
prevent local or global extinction.
Data limitations are severe, especially for biodiver-
sity.—River discharge data—at least snapshot data—are
available for most of the world’s major rivers, but good
temporal sequences of discharge data are not available
for many rivers of interest. Species richness numbers are
not readily available for most rivers of the world,
especially for invertebrate taxa (e.g., mollusks, crusta-
ceans, zooplankton, phytoplankton). In many cases,
such data may not exist at all, especially in temporal
sequences spanning changes in discharge. Data limita-
tions thus make impossible any rigorous tests of species–
discharge models. A critical research need is to assemble
for analysis the data that do exist (often in hard to access
local publications or technical reports), gather data in
the field for understudied waterways (especially those
that have already undergone substantial discharge
reduction), and test the sort of model we have presented
here. Such tests would foster the development of
improved and more robust models of aquatic species–
habitat relationships, which are urgently needed for
freshwater conservation management.
Better models would likely include more biologically
relevant metrics of the river hydrograph.—Preliminary
analyses suggest that species–discharge models con-
structed with average annual discharge explain more
variation in species number than those built with annual
maximum or annual minimum discharge (M. Xenopou-
los and D. Lodge, unpublished analyses). However,
humans often affect other metrics (e.g., maximum
annual discharge, Fig. 1) more strongly than mean
discharge, and the survival of many aquatic organisms is
linked to temporal variations in flow (Olden and Poff
2003). Rivers undergo seasonal periods of floods and
droughts, such that temporary local extirpations are
common, and sections of streams and rivers are often
recolonized annually. Models that explain a greater
proportion of species numbers will likely include
discharge metrics that reflect a more complete under-
standing of the biology of fishes and other organisms.
The addition of parameters other than river discharge
might improve the predictive power of models.—Although
the hydrograph is clearly important for fishes, many
other natural and anthropogenic factors (e.g., temper-
ature, habitat diversity not related to discharge, evolu-
tionary history, productivity, trophic interactions, and
pollutants) can strongly influence species diversity
(Hocutt and Wiley 1986, Oberdorff et al. 1995). For
example, reduced discharge also means increased
temperature (Matthews and Zimmerman 1990), and
concentrated nutrients and pollutants. Eutrophication
associated with low water levels and higher water
temperatures facilitates the establishment of parasites
and fosters infections in fish (Steedman 1991). Some of
these factors could be included in local, regional, or
global models, although data limitations and absence of
simple algorithms relating many of these factors to
species richness make this a challenging research goal.
The current models forecast loss of river-specific
biodiversity.—Overlap in species among watersheds
within a biogeographic region, combined with uncer-
tainty about which species would be lost, make
forecasting global extinctions more challenging than
forecasting local extinctions. The assembly of additional
species lists that distinguish endemic species, and
improvements in ability to forecast the fate of specific
species, would address this need. Even without such
advances, however, local loss of species is often of great
importance to humans because it represents a loss of
ecosystem services, especially if species are harvested or
are forage for harvested species (as is the case with many
fishes).
The scale dependence of species–discharge models
should be evaluated.—The species–discharge models
presented here have many parallels with terrestrial
species–area relationships (SAR). Just as terrestrial
SAR are scale dependent (e.g., the curves for islands,
mainlands, and provinces of Rosenzweig [1995]), aquatic
species–discharge parameters (e.g., slope and intercept)
should be evaluated for scale dependence and any
systematic variation related to different taxa or biogeo-
graphic regions. The robustness of forecasts could
probably be improved after such analysis.
For example, since rivers are considered to be
biogeographical islands (Hugueny 1989) the lower slope
(z value) for the Lower Ohio–Upper Mississippi species–
discharge curve is consistent with greater potential for
local colonization compared to the Southeastern U.S.
region (sensu Rosenzweig [1995] ‘‘less remote islands’’
and ‘‘isolated islands’’). Indeed, the Lower Ohio–Upper
Mississippi region is composed of tributaries that allow
for colonization whereas the Southeastern U.S. region
contains only rivers that drain into the ocean, where
colonization rates might approach zero. Accordingly,
fish losses calculated from the Lower Ohio–Upper
Mississippi region are much lower per unit reduced
discharge. Clearly additional work needs to address this
possibility.
CONCLUSION
Because the statistical approach introduced here
relates species richness to river discharge, it is a
promising tool for evaluating the regional and global
imperilment of freshwater biodiversity. Such tools are
urgently needed to prioritize management and restora-
tion efforts among the many rivers that are experiencing
or will experience losses of water from anthropogenic
activities. Given the lack of rigorous testing of the
models, and the many possible improvements suggested
above, current applications of the species–discharge
models should be cautious. Nevertheless this approach is
the best quantitative tool available. At the global scale
we predicted fish losses up to 75% using climate and
water consumption scenarios from the IPCC (Xenopou-
MARGUERITE A. XENOPOULOS AND DAVID M. LODGE1912 Ecology, Vol. 87, No. 8
SPECIALFEATURE
los et al. 2005). Given the many anthropogenic impacts
on flowing waters that often interact in ways that
increase negative effects on aquatic biota, this approach
is likely to provide a conservative estimate of future
species losses. At the very least, it can provide a relative
index of the threat of river basin-specific extirpation of
fish species for rivers suffering reductions in discharge.
The lag times between reductions in flow and species
extinctions provide an opportunity for biologists to offer
advice on how to target conservation efforts to slow or
prevent the loss of freshwater species.
ACKNOWLEDGMENTS
M. A. Xenopolous was financially supported by a postdoc-toral fellowship from Canada’s Natural Sciences and Engineer-ing Research Council (NSERC), the University of Notre Dame,the Integrated Systems for Invasive Species (ISIS) projectfunded by the National Science Foundation (to D. M. Lodge;DEB 02-13698), the Millennium Ecosystem Assessment (MA),and an NSERC University Faculty Award (to Trent Univer-sity). We thank the MA and our WaterGAP colleagues: J.Alcamo, M. Marker, and K. Schulze for fruitful discussions onthe use of this approach.
LITERATURE CITED
Alcamo, J., P. Doll, T. Henrichs, F. Kaspar, B. Lehner, T.Rosch, and S. Siebert. 2003. Development and testing of theWaterGAP 2 global model of water use and availability.Hydrological Sciences Journal 48:317–337.
Alcamo, J., and T. Henrichs. 2002. Critical regions: A model-based estimation of world water resources sensitive to globalchanges. Aquatic Sciences 64:352–362.
Alin, S. R. et al. 1999. Effects of landscape disturbance onanimal communities in Lake Tanganyika, East Africa.Conservation Biology 13:1017–1033.
Bradford, M. J., and J. R. Irvine. 2000. Land use, fishing,climate change, and the decline of Thompson River, BritishColumbia, coho salmon. Canadian Journal of Fisheries andAquatic Sciences 57:13–16.
Bronmark, C., J. Herrman, B. Malmqvist, C. Otto, and P.Sjostrom. 1984. Animal community structure as a function ofstream size. Hydrobiologia 112:73–79.
Chapin, F. S., E. S. Zavaleta, V. T. Eviner, R. L. Naylor, P. M.Vitousek, H. L. Reynolds, D. U. Hooper, S. Lavorel, O. E.Sala, S. E. Hobbie, M. C. Mack, and S. Diaz. 2000.Consequences of changing biodiversity. Nature 405:234–242.
Degerman, E., J. Hammar, P. Nyberg, and G. Svardson. 2001.Human impact on the fish diversity in the four largest lakesof Sweden. Ambio 30:522–528.
Eadie, J. M., T. A. Hurly, R. D. Montgomerie, and K. L.Teather. 1986. Lakes and rivers as islands: species arearelationships in the fish faunas of Ontario. EnvironmentalBiology of Fishes 15:81–89.
Elfner, M. A., and R. J. Mcdowell. 2004. Water conversation inGeorgia: Bringing efficiency into mainstream thinking.Journal American Water Works Association 96:136–142.
Findlay, D. L., S. E. M. Kasian, M. P. Stainton, K. Beaty, andM. Lyng. 2001. Climatic influences on algal populations ofboreal forest lakes in the Experimental Lakes Area.Limnology and Oceanography 46:1784–1793.
Guegan, J. F., S. Lek, and T. Oberdorff. 1998. Energyavailability and habitat heterogeneity predict global riverinefish diversity. Nature 391:382–384.
Hillebrand, H., and U. Sommer. 2000. Diversity of benthicmicroalgae in response to colonization time and eutrophica-tion. Aquatic Botany 67:221–236.
Hites, R. A., J. A. Foran, D. O. Carpenter, M. C. Hamilton, B.A. Knuth, and S. J. Schwager. 2004. Global assessment oforganic contaminants in farmed salmon. Science 303:226–229.
Hocutt, C. H., and E. O. Wiley. 1986. The zoogeography ofNorth American freshwater fishes. John Wiley and Sons,Toronto, UK.
Hugueny, B. 1989. West African rivers as biogeographicislands: species richness of fish communities. Oecologia 79:236–243.
IPCC (Intergovernmental Panel on Climate Change). 2001.Climate change 2001: impacts, adaptation, and vulnerability.Cambridge University Press, Cambridge, UK.
Jenkins, M. 2003. Prospects for biodiversity. Science 302:1175–1177.
Kolar, C. S., and D. M. Lodge. 2002. Ecological predictionsand risk assessment for alien fishes in North America. Science298:1233–1236.
Lake, P. S. 2003. Ecological effects of perturbation by droughtin flowing waters. Freshwater Biology 48:1161–1172.
Livingstone, D. A., M. Rowland, and P. E. Bailey. 1982. On thesize of African riverine fish faunas. American Zoologist 22:361–369.
Lodge, D. M. 2001. Lakes. Pages 277–314 in O. E. Sala, F. S.Chapin, III, and E. Huber-Sannwald, editors. Globalbiodiversity in a changing environment. Springer, New York,New York, USA.
Matthews, W. J. 1998. Patterns in freshwater fish ecology.Chapman and Hall, New York, New York, USA.
Matthews, W. J., and E. G. Zimmerman. 1990. Potential effectsof global warming on native fishes of the Southern GreatPlains and the Southwest. Fisheries 15:26–32.
Mayden, R. L. 1992. Systematics, historical ecology, and NorthAmerican freshwater fishes. Stanford University Press,Stanford, California, USA.
Metcalfe-Smith, J. L., S. K. Staton, G. L. Mackie, and N. M.Lane. 1998. Changes in the biodiversity of freshwater musselsin the Canadian waters of the lower Great Lakes drainagebasin over the past 140 years. Journal of Great LakesResearch 24:845–858.
Millennium Ecosystem Assessment. 2003. Ecosystems andhuman well-being: a framework for assessment. Island Press,New York, New York, USA.
Minckley, W. L., P. C. Marsh, J. E. Deacon, T. E. Dowling, P.W. Hedrick, W. J. Matthews, and G. Mueller. 2003. Aconservation, plan for native fishes of the Lower ColoradoRiver. BioScience 53:219–234.
Morita, K., and S. Yamamoto. 2002. Effects of habitatfragmentation by damming on the persistence of stream-dwelling charr populations. Conservation Ecology 16:1318–1323.
Nakicenovic, N. et al. 2000. Special report on emissionscenarios. Intergovernmental Panel on Climate Change.Cambridge University Press, Cambridge, UK.
Nature Serve. 2004. NatureServe Explorer: an online encyclo-pedia of life. Version 4.0. Nature Serve, Arlington, Virginia,USA. hwww.natureserve.org/exploreri
Niyogi, D. K., W. M. Lewis, and D. M. McKnight. 2002.Effects of stress from mine drainage on diversity, biomass,and function of primary producers in mountain streams.Ecosystems 5:554–567.
Oberdorff, T., J.-F. Guegan, and B. Hugueny. 1995. Globalscale patterns of fish species richness in rivers. Ecography 18:345–352.
O’Brien, J. S. 1987. A case study of minimum streamflow forfishery habitat in the Yampa river. Pages 921–946 in C. R.Thorne, J. C. Bathurst, and R. D. Hey, editors. Sedimenttransport in gravel-bed rivers. John Wiley and Sons, NewYork, New York, USA.
Olden, J. D., and N. L. Poff. 2003. Redundancy and the choiceof hydrologic indices for characterizing streamflow regimes.River Research and Applications 19:101–121.
August 2006 1913DETERMINANTS OF BIODIVERSITY CHANGE
SPECIALFEATURE
Poff, N. L., P. L. Angermeier, S. D. Cooper, P. S. Lake, K. D.Fausch, K. O. Winemiller, L. A. K. Mertes, M. W. Oswood,J. D. Reynolds, and F. J. Rahel. 2001. Fish diversity instreams and rivers. Pages 315–349 in O. E. S. Chapin, III andE. Huber-Sannwald, editors. Global biodiversity in achanging environment: scenarios for the 21st century.Springer, New York, New York, USA.
Postel, S. L. 2000. Entering an era of water scarcity: thechallenges ahead. Ecological Applications 10:941–948.
Postel, S., and B. Richter. 2003. Rivers for life: managing waterfor people and nature. Island Press, Washington, D.C., USA.
Prairie, Y. T. 1996. Evaluating the predictive power ofregression models. Canadian Journal of Fisheries andAquatic Sciences 53:490–492.
Ricciardi, A., and J. Rasmussen. 1999. Extinction rates ofNorth American freshwater fauna. Conservation Biology 13:1220–1222.
Rosenzweig, M. L. 1995. Species diversity in space and time.Cambridge University Press, New York, New York, USA.
Sala, O. E. et al. 2000. Biodiversity: global biodiversityscenarios for the year 2100. Science 287:1770–1774.
Sepkoski, J. J., and M. A. Rex. 1974. Distributions offreshwater mussels: coastal rivers as biogeographic islands.Systematic Zoology 23:165–188.
Steedman, R. J. 1991. Occurrence and environmental correlatesof black spot disease in stream fishes near Toronto, Ontario.Transactions of the American Fisheries Society 120:494–499.
Strayer, D. 1983. The effects of surface geology and stream sizeon freshwater mussel (Bivalvia, Unionidae) distribution insoutheastern Michigan, U.S.A. Freshwater Biology 13:253–264.
Vinebrooke, R. D., D. W. Schindler, D. L. Findlay, M. A.Turner, M. Paterson, and K. H. Milis. 2003. Trophicdependence of ecosystem resistance and species compensa-tion in experimentally acidified lake 302S (Canada). Ecosys-tems 6:101–113.
Vorosmarty, C. J., P. Green, J. Salisbury, and R. B. Lammers.2000. Global water resources: vulnerability from climatechange acid population growth. Science 289:284–288.
Vorosmarty, C. J., and D. Sahagian. 2000. Anthropogenicdisturbance of the terrestrial water cycle. BioScience 50:753–765.
Welcomme, R. L. 1979. Fisheries ecology of floodplain rivers.Longman, London, UK.
UNESCO. 2003. The United Nations world water developmentreport: water for people, water for life. UNESCO andBerghahn Books, Paris, France.
Xenopoulos, M. A., and P. C. Frost. 2003. Ultravioletradiation, phosphorus and their combined effects on thetaxonomic composition of phytoplankton in a boreal lake.Journal of Phycology 39:291–302.
Xenopoulos, M. A., D. M. Lodge, J. Alcamo, M. Marker, K.Schulze, and D. P. van Vuuren. 2005. Scenarios of freshwaterfish extinctions from climate change and water withdrawal.Global Change Biology 11:1557–1564.
MARGUERITE A. XENOPOULOS AND DAVID M. LODGE1914 Ecology, Vol. 87, No. 8
SPECIALFEATURE
Ecology, 87(10), 2006, pp. 2500–2510� 2006 by the Ecological Society of America
CLIMATE EFFECTS ON FIRE REGIMES AND TREE RECRUITMENTIN BLACK HILLS PONDEROSA PINE FORESTS
PETER M. BROWN1
Rocky Mountain Tree-Ring Research, Inc., 2901 Moore Lane, Fort Collins, Colorado 80526 USA
Abstract. Climate influences forest structure through effects on both species demography(recruitment and mortality) and disturbance regimes. Here, I compare multi-centurychronologies of regional fire years and tree recruitment from ponderosa pine forests in theBlack Hills of southwestern South Dakota and northeastern Wyoming to reconstructions ofprecipitation and global circulation indices. Regional fire years were affected by droughts andvariations in both Pacific and Atlantic sea surface temperatures. Fires were synchronous withLa Ninas, cool phases of the Pacific Decadal Oscillation (PDO), and warm phases of theAtlantic Multidecadal Oscillation (AMO). These quasi-periodic circulation features areassociated with drought conditions over much of the western United States. The oppositepattern (El Nino, warm PDO, cool AMO) was associated with fewer fires than expected.Regional tree recruitment largely occurred during wet periods in precipitation reconstructions,with the most abundant recruitment coeval with an extended pluvial from the late 1700s toearly 1800s. Widespread even-aged cohorts likely were not the result of large crown firescausing overstory mortality, but rather were caused by optimal climate conditions thatcontributed to synchronous regional recruitment and longer intervals between surface fires.Synchronous recruitment driven by climate is an example of the Moran effect. The presence ofabundant fire-scarred trees in multi-aged stands supports a prevailing historical model forponderosa pine forests in which recurrent surface fires affected heterogenous forest structure,although the Black Hills apparently had a greater range of fire behavior and resulting foreststructure over multi-decadal time scales than ponderosa pine forests of the Southwest thatburned more often.
Key words: Atlantic Multidecadal Oscillation; crossdating; dendrochronology; El-Nino–SouthernOscillation; fire history; fire regime; Pacific Decadal Oscillation; precipitation variation.
INTRODUCTION
Projected future climate changes will likely affect
complex biogeographic shifts as a result of effects on
both species demography (recruitment and mortality)
and disturbance regimes (Overpeck et al. 1990, Dale et
al. 2001, Burkett et al. 2005). The ability of species to
tolerate changes in temperature and moisture regimes
will undoubtedly be confounded by alterations in
disturbances (Burkett et al. 2005). For example, recent
global-change-type drought coupled with widespread
bark beetle outbreaks has caused massive die-off of
pinon pine (Pinus edulis) and other tree species across
large portions of the Colorado Plateau (Breshears et al.
2005, Mueller et al. 2005). Empirical and modeling
studies have largely focused on species responses to
drought and drought-related impacts on disturbances,
such as increased fire frequency, fire severity, or insect
outbreaks. However, precipitation changes are expected
to be spatially variable, with some areas probably seeing
increased moisture while others become dryer (e.g.,
Intergovernmental Panel on Climate Change 2001).
Better understanding of potential response to increased
moisture is as critical for predicting future vegetation
patterns as is understanding responses to drought.
Prediction of vegetation change in response to climate
change can benefit from paleoecological studies that
document longer-term responses of plant communities
to varying climate and disturbance regimes. Here, I
compare chronologies of regional fire years and tree
recruitment in ponderosa pine (Pinus ponderosa var.
scopulorum) forests of the Black Hills of southwestern
South Dakota and northeastern Wyoming to recon-
structions of precipitation and global circulation indices
over the past five centuries. My objectives are to
evaluate temporal linkages between climate variations,
fire years, and tree recruitment patterns, and to infer the
role of climate in both fire occurrences and formation of
regional forest structure over time. The climate recon-
structions document large annual to multi-decadal
variations over the time period encompassed by fire-
year and tree-recruitment chronologies, and include
both severe droughts and extended pluvial periods with
which to develop inferences about climatic effects on
long-term forest structural change.
This study also addresses important questions about
the disturbance ecology and historical variability of
Black Hills ponderosa pine forests. The dominant
Manuscript received 19 December 2005; accepted 3 April2006. Corresponding Editor (ad hoc): D. D. Breshears.
1 E-mail: pmb@rmtrr.org
2500
disturbance regime in ponderosa pine forests across
North America prior to Euro-American settlement in
the 19th century consisted of recurrent surface fires (e.g.,
Swetnam and Betancourt 1998, Allen et al. 2002,
Heyerdahl et al. 2002, Swetnam and Baisan 2003, Hessl
et al. 2004, Brown and Wu 2005). However, early
accounts from the Black Hills in the late 19th century
document large areas of trees killed by fire (Graves 1899,
Dodge 1965). Broad expanses of even-aged forest also
were present at settlement and are thought to have been
the result of widespread crown fires in the 1700s and
early 1800s (Graves 1899, Shinneman and Baker 1997).
Shinneman and Baker (1997) assert that a historical fire
regime dominated by surface burning does not apply
across much of the Black Hills landscape. This assertion
has raised critical questions for both understanding
ecological dynamics in Black Hills forests and for
guiding forest management in this region. If crown fires
and denser forest structure were fundamental features of
the historical Black Hills forest, landscape-scale restora-
tion of open, multi-aged forests and surface fire regimes
as has been proposed for ponderosa pine forests in other
regions (e.g., Allen et al. 2002) may be inappropriate.
METHODS
Study area
The Black Hills are a relatively isolated mountain
range that rises over 1000 m above the surrounding
mixed-grass prairies of the northern Great Plains. The
main part of the range is in southwestern South Dakota
with a smaller extension, the Bear Lodge Mountains, in
northeastern Wyoming (Fig. 1). Elevations range from
1050 to 1350 m on the margins with the Great Plains to
the highest point at 2207 m. The Black Hills support
extensive ponderosa pine forests (Shepperd and Batta-
glia 2002). White spruce (Picea glauca) and aspen
(Populus tremuloides) are occasional co-dominants of
higher and wetter forests in the northern and central
Hills. In most areas ponderosa pine is the only tree
species present. Annual precipitation declines from
FIG. 1. Landscapes sampled for tree-recruitment data (shaded) and fire-scar site locations (numbered) in the Black Hills. Theboundary of the Black Hills National Forest is marked by the dashed line.
October 2006 2501CLIMATE, FIRE, AND TREE RECRUITMENT
about 740 mm in the north to about 480 mm in the
south.
Euro-American settlement began after 1874. Prior to
this, the area was home to the Lakota Sioux who
apparently used the Black Hills only sporadically after
the Prairie horse culture arose in the early to mid-18th
century (Hassrick 1964). Less is known of the Native
American history prior to this time. Intensive logging
that began with Euro-American settlement has resulted
in large areas of second-growth forest (Graves 1899,
Shepperd and Battaglia 2002). The Black Hills National
Forest Reserve (today the Black Hills National Forest)
was the first federal forest preserve established in the
United States in 1897, partly as a response to intensive
and often wasteful timber practices up to that time
(Graves 1899). Timber production is still a major use of
much of the forest.
Tree-ring data
Tree-ring evidence is central to documenting historical
fire regimes in ponderosa pine forests throughout its
range (e.g., Heyerdahl et al. 2002, Swetnam and Baisan
2003, Hessl et al. 2004, Sherriff 2004, Brown and Wu
2005). These reconstructions rely on proxy evidence of
fire timing and behavior recorded in long-lived trees.
Fire history is typically reconstructed using two types of
tree-ring evidence: (1) fire scars created during surface
burning, and (2) recruitment dates of trees that
potentially postdate crown-opening fires (Brown and
Wu 2005).
For this study, I collected both fire-scar and tree-
recruitment data. ‘‘Tree recruitment’’ refers to trees that
established in the overstory and have persisted to the
present, either as living trees or remnant trees (stumps,
logs, or snags) that are able to be sampled. To
reconstruct regional patterns of tree recruitment, trees
were sampled from 37 plots across three landscapes on
the Limestone Plateau, an area of rolling hills and
canyons on the western margins of the main range (Fig.
1; Brown 2003). The Limestone Plateau has been cited as
an area of extensive even-aged forest structure thought
to have resulted from widespread fires (Graves 1899,
Shinneman and Baker 1997). Landscapes were de-
lineated on a precipitation gradient from wet to dry in
the northern, middle, and southern portions of the
Limestone Plateau (Fig. 1). Plots within each landscape
were randomly chosen latitude/longitude coordinates
and located in the field using a handheld GPS unit. N-
tree distance sampling methods (Jonsson et al. 1992)
were used to collect data from the nearest 30 presettle-
ment trees to each plot center within a maximum plot
radius of 40 m (;0.5-ha circular plot; Brown and Cook
2006, Brown and Wu 2005). Most plots were �0.25 ha
in size (Brown 2003). Presettlement trees were defined as
all remnant trees and living trees that either were not
‘‘blackjacks’’ (younger trees with primarily dark bark) or
that were �30 cm dbh. Based on prior age sampling of
ponderosa pine in the Black Hills, trees tend to have
mainly dark bark (blackjacks) until about 75–100 years
of age, after which it progressively changes to a buff or
orange color. I assumed all blackjack trees and trees
,30 cm dbh established post-settlement. Increment
cores or cross sections were removed from 10 cm height
on all trees in each plot. Because of past harvest in all
but one plot in the northern landscape, many of the trees
sampled were stumps (Brown and Cook 2006).
To reconstruct regional fire years, I compiled dates of
fire scars found on the plot trees and additional data
from fire-scarred trees collected at 27 other sites (Fig. 1;
Brown and Sieg 1996, 1999, Brown et al. 2000, Brown
2003). These sites were selected to span elevational and
moisture gradients present across the Black Hills and
Bear Lodge Mountains. Older trees exhibiting multiple
fire scars were selected at each site to develop the longest
records of fire years (sensu Swetnam and Baisan 2003).
All cores and cross sections were dendrochronologi-
cally crossdated using locally developed master chronol-
ogies. Visual matching of ring characteristics and
correlated measured ring widths were used to assure
crossdating. If crossdating could not be determined,
samples were not used in subsequent analyses. After
crossdating of tree rings was completed on fire-scarred
cross sections, dates were assigned to fire scars. On
increment cores and cross sections that did not include
pith but inside ring curvature was visible, pith dates were
estimated using overlaid concentric circles of varying
diameters that take into account both average inside ring
widths and an estimated distance to pith. Plot and site
chronologies were compiled using program FHX2, an
integrated package for graphing and statistical analyses
of fire and forest history data (Grissino-Mayer 2001).
Evaluating tree recruitment, fire year,
and climate relationships
Tree pith dates from Limestone Plateau plots were
combined to document regional patterns of tree recruit-
ment from AD 1500. Pith dates at 10 cm height were
first corrected to recruitment dates by subtracting five
years, the average time estimated for seedlings to grow
from germination to 10 cm height. This correction is
based on height-growth measurements on open-grown
ponderosa pine in the Front Range of central Colorado
(P. M. Brown, W. D. Shepperd, and M. W. Kaye,
unpublished manuscript) and estimation from nodal
growth on seedlings in the Black Hills. A regional tree-
recruitment chronology was developed from 5-yr sums
of annual recruitment dates.
A regional fire-year chronology was developed from
51 site- and plot-level composite fire-year chronologies.
Fire years included in each composite chronology are
those recorded on �2 trees. Fire years recorded on only
one tree were excluded since there may be false positives,
scars not caused by fire but assumed to be fire scars
(Falk 2004, Brown and Wu 2005), when determining fire
dates. I assume that use of scar dates recorded on two or
more trees in a stand minimizes the likelihood of false
PETER M. BROWN2502 Ecology, Vol. 87, No. 10
positives since it is less likely that other possible scarring
mechanisms (e.g., lightning, fell-tree abrasions) affected
more than one tree in an individual stand during the
same year (Brown and Wu 2005). The regional fire-year
chronology was made up of years recorded in �10% of
the 51 site- and plot-level composite chronologies.
Percentages of sites recording fire scars is assumed to
be reflective of relative spatial scales of burning across
the Black Hills during individual fire years.
Regional tree-recruitment and fire-year chronologies
were graphically and statistically compared with two
independently derived tree-ring based reconstructions of
precipitation, two proxy indices of the El Nino-Southern
Oscillation (ENSO), an index of the Pacific Decadal
Oscillation (PDO), and an index of the Atlantic Multi-
decadal Oscillation (AMO). ENSO is a coupled atmos-
phere/ocean feature of the equatorial Pacific. PDO is an
index of sea surface temperature (SST) anomalies in the
North Pacific basin. Both ENSO and PDO have been
associated with fire occurrence across the western
United States through synoptic control of annual
droughts that synchronize fire timing (Heyerdahl et al.
2002, Hessl et al. 2004, Brown and Wu 2005, Taylor and
Beaty 2005, Sibold and Veblen 2006). AMO is an index
of North Atlantic basin (0 to 708 N) SST anomalies that
has recently been associated with droughts (McCabe et
al. 2004, Sutton and Hodson 2005) and fire occurrence
(Sibold and Veblen 2006) across the central and western
United States. I used superposed epoch analysis (SEA;
e.g., Brown and Wu 2005) to compare average annual
climate conditions for the set of regional fire years to
climate for the entire period of record. SEA also was
used to compare climate during years prior to fire years
to assess antecedent climate conditions that may be
important for fire occurrence. Significant climate
anomalies were determined using bootstrapped con-
fidence intervals based on average annual climate values
with the same number of years as regional fire-year data
sets. Five climate reconstructions were used in SEA: (1)
percentage of August-to-July mean precipitation from
instrumental stations in the Black Hills and northern
Great Plains (the reconstruction spans the period from
1596–1990; Stockton and Meko [1983], updated by D.
M. Meko, personnel communication); (2) annual precip-
itation for the Bighorn Basin in north-central Wyoming
(1260–1998; Gray et al. [2004a]); (3) Southern Oscil-
lation Index (SOI; 1706–1977; Stahle et al. [1998]); (4)
Nino3 SST (1408–1978; Cook [2000]); and (5) PDO
(1661–1991; Biondi et al. [2001]). SOI is a commonly
used measure of ENSO and is the difference in surface
air pressure between Darwin, Australia, and Tahiti. The
Nino3 index is average SST from mid-tropical Pacific
recording stations, the region with the largest SST
variability on ENSO (3 to 4 yr) time scales. I also tested
for significant contingent states of reconstructed AMO
(Gray et al. 2004b) with PDO and SOI on fire years
recorded in �2 of the composite site- and plot-fire-scar
chronologies. A chi-square test of expected vs. observed
fire years was used to test each of eight annual
combinations of anomalies (positive/negative) of the
three climate indices (sensu Sibold and Veblen 2006). A
larger set of fire years was used for this analysis to
increase the number of observations available for the
chi-square statistic. Finally, correlations were used to
assess significant temporal agreement between 5-yr
means of the two precipitation reconstructions to the
regional tree-recruitment chronology.
RESULTS
Tree-recruitment and fire-scar chronologies
I crossdated a total of 644 trees (of 720 trees sampled)
from 24 plots in the middle and southern landscapes
(Fig. 2). Plots were largely multi-aged, with at least a few
older trees found with younger trees in almost every
plot. An additional 110 mainly living trees (of 372 trees
sampled) were dated from the northern landscape.
However, I was not able to crossdate enough of the
remnant trees from the northern plots to develop plot-
level chronologies as in Fig. 2 (Brown 2003). The
northern area is more mesic than either of the other
landscapes and remnant trees did not have enough ring
variability to crossdate with confidence. Although
reliable pith dates were obtained on living trees from
the northern landscape, results from the middle and
southern landscapes showed that inclusion of data from
stumps and other remnants is required for reconstruc-
tion of presettlement patterns of tree recruitment (Fig. 2;
Brown 2003). Because of harvest of larger (and, thus,
mainly older) trees from almost all stands, abundant
living trees dating from the late 1700s to mid-1800s
make the contemporary forest appear to be even-aged
even though it was not at the time of initial harvest
(Brown and Cook 2006).
All trees sampled were ponderosa pine except for
three aspen from a single plot in the middle landscape.
Many of the randomly selected plot trees, mostly
stumps, recorded fire scars (Fig. 2). Heartwood of larger
remnant ponderosa pine trees may last a very long time
although erosion or burning of heartwood surfaces was
evident on older logs or snags (almost no stumps had
been burned because of fire cessation after harvest; see
Plate 1). An additional five trees that dated before 1500
from plots 201, 203, and 207 are not shown in Fig. 2. Of
these trees, three logs had pith dates of 1190, 1192, and
1206 and are the oldest known tree-ring dates from the
Black Hills.
Composite fire chronologies from an additional 276
trees collected from the 27 fire-scar sites (Fig. 1) and fire
scars recorded on trees from the 24 plot chronologies
from the middle and southern landscapes are summar-
ized in Fig. 3a. The most extensively recorded regional
fire year was 1785. There was variability in frequency of
regionally synchronous fires throughout the record.
Regionally synchronous fires increased in the late
1800s, and fire scars were generally absent after circa
1890.
October 2006 2503CLIMATE, FIRE, AND TREE RECRUITMENT
Climate, fire, and tree recruitment
Regional tree recruitment largely occurred during
distinct episodes that were coeval with pluvial periods in
the northern Great Plains and Bighorn Basin precip-
itation reconstructions (Fig. 3b). Recruitment dates
occurring in the combined 120 years between 1520–
1549, 1610–1639, 1770–1799, and 1830–1859 account for
;80% of all recruitment dates during the 401-yr period
between 1500 and 1900. The most abundant pulse of
establishment in the late 1700s occurred during the
wettest extended period in the precipitation records and
followed the most intense drought in the Great Plains
record from circa 1752 to 1762. The tree-recruitment
FIG. 2. Tree-recruitment chronologies from plots sampled in the (a) middle and (b) southern landscapes. Numbers in eachpanel refer to plot designations from Brown (2003). Horizontal lines mark time spans of individual trees, with dates of fire scarsmarked by inverted triangles. Dashed lines are the estimated number of years to 10-cm height pith dates. Vertical lines to the left arepith dates, with inside dates (i.e., unknown number of years to pith) marked by slanted lines. Vertical lines to the right are barkdates (death dates in remnant trees), with outside dates (i.e., unknown number of years to death date) marked by slanted lines.Vertical gray bars in each panel mark the period from 1770 to 1850 when a majority of regional tree recruitment occurred.
PETER M. BROWN2504 Ecology, Vol. 87, No. 10
chronology was significantly correlated with 5-yr aver-
ages of both the Great Plains precipitation reconstruc-
tion (1596–1900; N ¼ 61, r ¼ 0.32, P ¼ 0.015) and the
Bighorns Basin reconstruction (1500–1900; N ¼ 80, r ¼0.31, P ¼ 0.006).
SEA documents that regional fire years were signifi-
cantly dry on average (Fig. 4a, b). No significant lagged
relationships were seen between years prior to fire years
and precipitation variability. Fire years were on average
La Nina years (high SOI, low SST; Fig. 4c, d). Fire
years, coupled with one year prior and two years
following, also were on average years of significantly
cool-phase PDO (Fig. 4e). There was a strong associa-
tion of fires with contingent phase combinations of
climate indices, with significantly more fires from 1706
to 1900 occurring during years of combined warm
(positive) AMO, cool (negative) PDO, and warm
(positive) SOI (Table 1). Significantly less fires occurred
during years of the opposite pattern (Table 1).
DISCUSSION
Climate effects on regional fires
This study extends to the Black Hills the importance
of drought and an emerging recognition of the central
FIG. 3. (a) Composite fire-scar chronologies from 51 sites and plots. Horizontal lines mark time spans of chronologies. Longvertical lines mark regional fire years recorded on �10% of composite chronologies. Short vertical lines mark fire years recorded on,10% of composite chronologies. Regional fire years are listed at the bottom of the chronologies. (b) Regional tree recruitmentchronology (histograms) with Great Plains (solid line) and Bighorn Basin (dotted line) precipitation reconstructions. Theprecipitation reconstructions have been smoothed with 10-yr smoothing splines to emphasize multi-annual variations and areplotted as standard deviation departures from the mean to show relative wet and dry anomalies. The distribution of treerecruitment dates after the late 1800s is uncertain since only presettlement trees were sampled.
October 2006 2505CLIMATE, FIRE, AND TREE RECRUITMENT
role of both Pacific and North Atlantic SST anomalies
on regional fire occurrence across western North
America (Fig. 4, Table 1; Heyerdahl et al. 2002, Hessl
et al. 2004, Brown and Wu 2005, Taylor and Beaty 2005,
Sibold and Veblen 2006). Regional fire years in the
Black Hills were strongly associated with droughts in
both the Great Plains and Bighorn Basin precipitation
reconstructions (Fig. 4a, b). No lagged relationships
with prior moisture conditions were seen in SEA, which
suggests that buildup of grasses and herbaceous fuels
during antecedent wet years was not necessary for
regional fire occurrence. This is in contrast to wet/dry
oscillations typically associated with large fire years in
ponderosa pine forests of the Southwest (e.g., Brown
and Wu 2005). Fire frequency was generally less in the
Black Hills than in the Southwest and it is likely that
longer intervals between fires led to enough fine fuel
accumulation that sufficient fuel drying was the only
condition needed for widespread burning to have
occurred (Brown 2003).
Regional fires also were strongly associated with
synoptic climate features that affect drought occurrence
across the western United States. Regional fires were
associated with both La Nina years (positive SOI,
negative Nino3 SST; Fig. 4c, d) and cool phases of the
PDO (Fig. 4e). La Ninas are associated with summer-
dry conditions in the Northern Plains Region (Bunkers
et al. 1996), and droughts in the central United States
are magnified when La Ninas occur during periods of
negative PDO (McCabe and Dettinger 1999). Additional
contingent states of climate system effects on fire
occurrence are seen with the inclusion of AMO. The
strongest signal for total fire occurrence between 1706
and 1900 was the combination of La Ninas, cool phases
of the PDO, and warm phases of the AMO (Table 1).
Warm AMO, especially when coupled with negative
PDO, contributed to broad-scale summer droughts over
the central US (McCabe et al. 2004, Sutton and Hodson
2005). Both the PDO and AMO have recently been
identified as probable low-frequency (multi-annual to
multi-decadal) drivers of fire occurrence across western
North America (Hessl et al. 2004, Taylor and Beaty
2005, Sibold and Veblen 2006), and results from the
Black Hills confirm findings from these other studies.
The opposite combination of cool AMO, warm PDO,
and El Ninos (low SOI) led to fewer fire years than
expected, probably because of increased moisture over
this region. Because these circulation indices are quasi-
periodic and their effects on regional climate generally
predictable, advance forecasting of fire years and their
likely severity across the Black Hills may be possible
based on findings reported here.
Climate and fire effects on regional tree recruitment
Regional pulses of tree recruitment from 1500 to 1900
were largely coeval with pluvial periods in the northern
Great Plains and Bighorn Basin precipitation recon-
structions (Fig. 3b). Extended wet conditions with fewer
FIG. 4. Superposed epoch analyses (SEA) of averageprecipitation and circulation anomalies for regional fire yearslisted in Fig. 3a. (a) Great Plains precipitation (Stockton andMeko 1983; updated by D. M. Meko, personal communication);(b) Bighorn Basin precipitation (Gray et al. 2004a); (c) SOI(Southern Oscillation index; Stahle et al. 1998); (d) Nino3 SST(sea surface temperatures; Cook 2000); and (e) PDO (PacificDecadal Oscillation; Biondi et al. 2001). Horizontal lines ineach panel mark significant departures based on bootstrappedconfidence intervals (dotted, P , 0.05; dashed, P , 0.01; solid,P , 0.001). SOI is reversed in terms of moisture relative toother series (dry years are high SOI, or La Nina years). Fireyear lag 0 is the average climate anomaly for regional fire years,with antecedent conditions indicated by negative lags; y-axisunits are relative anomalies from mean for each index.
PETER M. BROWN2506 Ecology, Vol. 87, No. 10
droughts would have promoted seedling establishment,
less seedling mortality from drought stress, and faster
tree growth. Much of the presettlement forest dates from
an extended pluvial that lasted from circa 1770 to 1850,
the wettest period in both the Great Plains and Bighorn
Basin reconstructions (see also Pederson et al. 2004).
Cohorts in the Black Hills during this period also were
contemporaneous with abundant recruitment in ponder-
osa pine forests in the Bighorn Mountains (P. M.
Brown, unpublished data) and in the Bighorn Basin
(Wentzel 2005) to the west of the Black Hills in
Wyoming. The presence of synchronous recruitment
over such a broad region is an example of the Moran
effect (Moran 1953), in which widespread synchroniza-
tion of population dynamics is affected by a strong
exogenous factor, in this case optimal pluvial conditions
in typically dry ponderosa pine forests.
Abundant synchronous tree recruitment affected by
optimal climate forcing is probably the reason for
extensive stands of even-aged forests in the Black Hills,
rather than widespread crown fires as postulated by
Shinneman and Baker (1997). Shinneman and Baker
base their contention of extensive crown fires on age
data and conclusions previously presented by Graves
(1899). Graves, in an early survey of timber resources in
the Black Hills, compiled an unknown amount of non-
crossdated age data and ring counts to fire scars found
on stump tops to propose that extensive crown fires
occurred in the late 1700s and early 1800s, the largest of
which he suggested occurred around 1790. The most
widespread fire year found by this study was 1785 (Fig.
3a). This is likely the same fire date as estimated by
Graves, and the extensive even-aged cohorts he found
were undoubtedly the same found by this study dating
from the late 1700s and early 1800s. However, cross-
dated tree-ring data document that abundant tree
recruitment synchronous with pluvial conditions began
before 1785 (Figs. 2 and 3b) and, thus, cannot be the
result of crown opening from burning during 1785.
Furthermore, crossdated data document that almost all
TABLE 1. Expected vs. observed numbers of fire years recorded in more than two composite fire-scar chronologies (Fig. 3a) from1706 to 1900 (N ¼ 44) for eight phase combinations of the AMO (Atlantic Multidecadal Oscillation; Gray et al. 2004b), PDO(Pacific Decadal Oscillation; Biondi et al. 2001), and SOI (Southern Oscillation Index; Stahle et al. 1998).
Case
AMO � AMO � AMO � AMO þ AMO þ AMO þ AMO þ AMO �PDO � PDO � PDO þ PDO � PDO þ PDO þ PDO � PDO þSOI � SOI þ SOI � SOI � SOI þ SOI � SOI þ SOI þ
Observed 5 6 3 5 4 6 11 4Expected 5.4 6.1 8.8 3.2 4.1 5 6.1 5.4Difference �0.4 �0.1 �5.8 1.8 �0.1 1.0 4.9 �1.4
Note: Bold numbers are significantly different (P , 0.1) in the chi-square test.
PLATE 1. (Left) A ponderosa pine stump with a ‘‘cat face’’ formed from repeated fire scars (visible as distinct ridges in the catface). Bark and sapwood have eroded from the outside of the stump, and only the heartwood is left. The stump was cut with acrosscut saw, placing the date of cutting in the late 19th or early 20th centuries before widespread use of motorized chainsaws.(Right) Cross-section sample from a fire-scarred ponderosa pine snag (standing dead tree). The tree died in 1916 from a bark beetleattack as evidenced by bluestain in the sapwood. Photo credit: P. M. Brown.
October 2006 2507CLIMATE, FIRE, AND TREE RECRUITMENT
stands with trees that recruited during the 1700s and
early 1800s pluvial contained trees that predated this
period (Fig. 2), indicating that even if the cohort
established in response to a severe fire event there was
not complete canopy kill within plot boundaries (most
of which were �.25 ha in size; Brown 2003, Brown and
Cook 2005).
Trees existing at the time of cohort establishment
suggest that stands were open enough for seedling
recruitment to occur during the late 1700s–early 1800s
pluvial (Fig. 5). Variation in tree ages in multi-aged
stands would have resulted from continuous, long-term
patch dynamics that included both individual and
clumped (in both space and time) tree mortality by
any number of factors, only one of which may have been
lethal fire but also including drought stress or other
disturbances such as insects, pathogens, windthrow, or
lightning (Brown and Wu 2005). In the case of the late
1700s–early 1800s pluvial, this period followed the most
intense drought in the Great Plains reconstruction from
circa 1752 to 1762 (Fig. 3b). It is probable that this
drought contributed to open conditions in many stands
via both direct tree mortality from drought stress and
possibly drought-related disturbance effects, such as
mortality of individual trees or small patches of trees
during surface fires or from outbreaks of mountain pine
beetles (Dendroctonus ponderosae). Mountain pine bee-
tles have been a major disturbance agent in the Black
Hills during the 20th century, often causing widespread
tree mortality (Shepperd and Battaglia 2002; see Plate
1). However, any trees that may have died during the
1750s drought did not contain sapwood because of
decay (Fig. 2), and the presence of evidence such as
bluestain indicating the possible role of bark beetles in
widespread mortality during this period cannot be
verified.
Variations in fire timing and frequency also likely
contributed to increased survivorship of trees during
climatically optimal recruitment episodes such as the
late 1700s and early 1800s (Brown and Wu 2005).
Longer intervals between surface fires are generally
associated with cohort establishment in individual
stands (Fig. 2), and regionally synchronous cohorts
largely occurred during periods between regional fire
dates (Fig. 3). Most seedlings and smaller trees are killed
by surface fires, and longer intervals between fires would
have permitted more trees to survive to reach canopy
status. Trees in many southwestern ponderosa pine
forests date to fire-quiescent periods in the 1700s and
early 1800s (e.g., Swetnam and Betancourt 1998, Brown
and Wu 2005). It is likely that variations in fire
frequency and timing were more critical to structuring
the historical forest than variations in fire severity as
hypothesized by Shinneman and Baker (1997).
Disturbance ecology in Black Hills ponderosa pine forests
Abundant fire scars recorded on trees in all stands
provide evidence that surface fires were common
disturbances in Black Hills ponderosa pine forests prior
to Euro-American settlement (Figs. 2 and 3a; Brown
and Sieg 1996, 1999, Brown et al. 2000, Brown 2003).
Mature ponderosa pine trees are well adapted to survive
surface burns, with thick bark that generally protects
vascular cambium from mortality and high crowns that
lessen the possibility of crown scorch. The main effect of
recurrent surface fires was to kill a majority of tree
regeneration, limiting the number of seedlings that
ultimately were able to reach maturity. Surface fires,
because they were primarily driven by climatic varia-
FIG. 5. A model for regional scale, multi-decadal, process-pattern linkages in Black Hills ponderosa pine forests. Heterogenous,mostly open forest structure is promoted and maintained by recurrent surface fires, droughts, and other disturbances that affectboth episodic and continuous mortality on spatiotemporally varying scales. Open stands allow for abundant new trees to establishwhen climate forcing affects episodic, broadscale recruitment opportunities. Successful synchronous recruitment results in morehomogenous, closed-canopy forest conditions for some length of time until patchy mortality results in open canopy conditionsagain. To put specific dates on each segment, the top condition prevailed prior to the extended pluvial of the late 1700s and early1800s (box to the right). The bottom condition is what largely prevailed at the time of Euro-American settlement in the late 1800s.Natural disturbance processes likely would have returned the forest to the heterogeneous, open condition except for disruption byland use changes, including fire suppression and timber harvest, that accompanied Euro-American settlement beginning in 1874.
PETER M. BROWN2508 Ecology, Vol. 87, No. 10
tions (Fig. 4) and not by overstory tree density, acted as
a density-independent mechanism for population regu-
lation in ponderosa pine forests (Brown and Wu 2005).
Historic accounts and early settlement photographs of
ponderosa pine forests both in the Black Hills and
throughout the West document often ‘‘park-like,’’ open,
multi-aged, forest stands, composed of large trees
scattered across grassy understories (e.g., Graves 1899,
Allen et al. 2002, Grafe and Horstead 2002). Individual
seedlings or patches of seedlings were occasionally able
to survive to reach maturity because of spatiotemporal
heterogeneity in burning. However, even more impor-
tant to regional forest structure were broad-scale
variations in climate regimes that resulted in conditions
optimal for seedling survival and growth and that
resulted in fewer surface fires (Fig. 3; Brown and Wu
2005).
Much of the historical fire regime and resulting forest
structure that has been reconstructed from the Black
Hills fits with a dominant model of ponderosa pine
ecosystems largely developed from the Southwest (e.g.,
Allen et al. 2002). In this model, recurrent surface fires
resulted in broad areas of typically open, multi-aged
forest stands. However, the Black Hills apparently
contained a coarser-scale mosaic of dense to open
stands and a greater range of variation in landscape
structure over multi-decadal time scales than what was
present in the Southwest, largely related to the abundant
tree recruitment that occurred during the late 1700s and
early 1800s (Fig. 5). These dense stands were still present
at settlement (Brown and Cook 2005), and likely
contributed to extensive patches of crown fire noted by
early explorers and scientists during the late 1800s
(Dodge 1965, Graves 1899). More extensive patches of
denser trees would have contributed to ladder and
canopy fuel conditions that permitted more sustained
crown fire behavior across larger areas, especially given
the increase in regional fire frequency in the late 1800s
(Fig. 3a). However, over longer time scales, the
dominant fire regime largely consisted of recurrent
surface fires that contributed to and maintained hetero-
genous, multi-aged forest structure at both stand and
regional scales (Figs. 2 and 3).
The pervasive cessation of fires that began in the early
20th century (Fig. 3a) corresponds to patterns found in
virtually all ponderosa pine forests across western North
America (Swetnam and Baisan 2003, Swetnam and
Betancourt 1998, Allen et al. 2002, Hessl et al. 2004,
Brown and Wu 2005). Fire cessation resulted from
changes in land use that accompanied settlement,
including livestock grazing that reduced grass and
herbaceous fuels through which surface fires spread
and, later in the twentieth century, active fire suppres-
sion by land management agencies. And as in ponderosa
pine forests of other regions, fire cessation coupled with
other impacts such as timber harvest has led to denser
and more homogeneous forest structure across the Black
Hills landscape (Brown and Cook 2005). Because of this
shift in landscape patterns, restoration of open forest
structure and multi-aged stands across large portions of
the Black Hills is supported by all available historical
evidence. Dense stands will likely always be present in
any future forest, but what is largely missing are mosaics
of open stands of variable density. However, forest
management in this and other regions also must
recognize that future climate change will likely lead to
complex population dynamics and shifts in disturbance
regimes that will complicate ecological restoration
efforts (e.g., Burkett et al. 2005). Perhaps the best that
forest management can do at this time is to restore as
much ecosystem resiliency as possible such that ecosys-
tem structure and function can be retained in the face of
future climate change. In ponderosa pine forests
throughout its range in western North America, these
efforts must include restoration of surface fire regimes
and resulting heterogenous forest structure.
ACKNOWLEDGMENTS
I especially thank Carolyn Hull Sieg of the Rocky MountainResearch Station and Claudia Regan of U.S. Forest ServiceRegion 2 for their support of this work. I thank Craig Allen,William Baker, Daniel Binkley, William Laurenroth, Jeff Lukas,William Romme, Carolyn Sieg, F. W. (Skip) Smith, ThomasSwetnam, Thomas Veblen, and an anonymous reviewer for theircomments on this and previous versions of this paper. ElizabethBauer, Baxter Brown, Chris Brown, Anthony Caprio, MarkLosleben, Jeff Lukas, Daniel Manier, James Riser, Ronald Sieg,and Connie Woodhouse provided field assistance. Tree-recruit-ment and fire-scar data are available through the InternationalMultiproxy Paleofire Database (available online).2 Funding wasprovided by the National Park Service and U.S. Forest Service.
LITERATURE CITED
Allen, C. D., M. Savage, D. A. Falk, K. F. Suckling, T. W.Swetnam, T. Schulke, P. B. Stacey, P. Morgan, M. Hoffman,and J. T. Klingel. 2002. Ecological restoration of southwest-ern ponderosa pine ecosystems: a broad perspective. Eco-logical Applications 12:1418–1433.
Biondi, F., A. Gershunov, and D. R. Cayan. 2001. NorthPacific decadal climate variability since AD 1661. Journal ofClimate 14:5–10.
Breshears, D. D., et al. 2005. Regional vegetation die-off inresponse to global-change-type drought. Proceedings of theNational Academy of Sciences (USA) 102:15144–15148.
Brown, P. M. 2003. Fire, climate, and forest structure inponderosa pine forests of the Black Hills. Dissertation.Colorado State University, Fort Collins, Colorado, USA.
Brown, P. M., and B. Cook. 2006. Early settlement foreststructure in Black Hills ponderosa pine forests. ForestEcology and Management 223:284–290.
Brown, P. M., M. G. Ryan, and T. G. Andrews. 2000.Historical fire frequency in ponderosa pine stands inResearch Natural Areas, central Rocky Mountains andBlack Hills, US. Natural Areas Journal 20:133–139.
Brown, P. M., and C. H. Sieg. 1996. Fire history in interiorponderosa pine forests of the Black Hills, South Dakota,USA. International Journal of Wildland Fire 6:97–105.
Brown, P. M., and C. H. Sieg. 1999. Historical variability in fireat the ponderosa pine–Northern Great Plains prairie ecotone,southeastern Black Hills, South Dakota. Ecoscience 6:539–547.
2 hhttp://www.ncdc.noaa.gov/paleo/impd/paleofire.htmli
October 2006 2509CLIMATE, FIRE, AND TREE RECRUITMENT
Brown, P. M., and R. Wu. 2005. Climate and disturbanceforcing of episodic tree recruitment in a southwesternponderosa pine landscape. Ecology 86:3030–3038.
Bunkers, M. J., J. R. Miller, and A. T. DeGaetano. 1996. Anexamination of El Nino–La Nina-related precipitation andtemperature anomalies across the northern plains. Journal ofClimate 9:147–160.
Burkett, V. R., D. A. Wilcox, R. Stottlemyer, W. Barrow, D.Fagre, J. Baron, J. Price, J. L. Nielsen, C. D. Allen, D. L.Peterson, G. Ruggerone, and T. Doyle. 2005. Nonlineardynamics in ecosystem response to climate change: casestudies and policy implications. Ecological Complexity 2:357–394.
Cook, E. R. 2000. Nino3 index reconstruction. InternationalTree-Ring Data Bank. IGBP PAGES/World Data Center-Afor Paleoclimatology Data Contr. Ser. #2000-052. NOAA/NGDC Paleoclimatology Program, Boulder Colorado, USA.
Dale, V. H., et al. 2001. Climate change and forest disturbances.BioScience 51:723–734.
Dodge, R. I. 1965 [reprint]. The Black Hills. Ross and Haines,Minneapolis, Minnesota, USA.
Falk, D. A. 2004. Scaling rules for fire regimes. Dissertation.University of Arizona, Tucson, Arizona, USA.
Grafe, E., and P. Horstead. 2002. Exploring with Custer: the1874 Black Hills Expedition. Golden Valley Press, Custer,South Dakota, USA.
Graves, H. S. 1899. The Black Hills Reserve. Nineteenth annualreport of the survey, 1897–1898. Pages 67–164 in Part V.Forest Reserves. U. S. Geological Survey, Washington, D.C.,USA.
Gray, S. T., C. L. Fastie, S. T. Jackson, and J. L. Betancourt.2004a. Tree-ring-based reconstruction of precipitation in theBighorn Basin, Wyoming, since 1260 A. D. Journal ofClimate 17:3855–3865.
Gray, S. T., L. J. Graumlich, J. L. Betancourt, and G. D.Pederson. 2004b. A tree-ring based reconstruction of theAtlantic Multidecadal Oscillation since 1567 A. D. Geophys-ical Research Letters 31:L12205. 10.1029/2004GL019932.
Grissino-Mayer, H. D. 2001. FHX2: software for analyzingtemporal and spatial patterns in fire regimes from tree rings.Tree-Ring Research 57:115–124.
Hassrick, R. B. 1964. The Sioux. University of Oklahoma Press,Norman, Oklahoma, USA.
Hessl, A. E., D. McKenzie, and R. Schellhaas. 2004. Droughtand Pacific Decadal Oscillation linked to fire occurrence inthe Pacific Northwest. Ecological Applications 14:425–442.
Heyerdahl, E. K., L. B. Brubaker, and J. K. Agee. 2002. Annualand decadal climate forcing of historical fire regimes in theinterior Pacific Northwest, USA. Holocene 12:597–604.
Intergovernmental Panel on Climate Change. 2001. Climatechange 2001: the scientific basis. Cambridge University Press,New York, New York, USA.
Jonsson, B., S. Holm, and H. Kallur. 1992. A forest inventorymethod based on density-adapted circular plot size. Scandi-navian Journal of Forest Research 7:405–421.
McCabe, G. J, and M. D. Dettinger. 1999. Decadal variationswith the strength of ENSO teleconnections with precipitationin the western United States. International Journal ofClimatology 19:1399–1410.
McCabe, G. J., M. A. Palecki, and J. L. Betancourt. 2004.Pacific and Atlantic Ocean influences on multidecadal
drought frequency in the United States. Proceedings of theNational Academy of Sciences (USA) 101:4136–4141.
Moran, P. A. P. 1953. The statistical analysis of the Canadianlynx cycle. II. Synchronization and meteorology. AustralianJournal of Zoology 1:291–298.
Mueller, R. C., C. M. Scudder, M. E. Porter, R. T. Trotter, C.A. Gehring, and T. G. Whitham. 2005. Differential mortalityin response to severe drought: evidence for long-termvegetation shifts. Journal of Ecology 93:1085–1093.
Overpeck, J. T., D. Rind, and R. Goldberg. 1990. Climateinduced changes in forest disturbance and vegetation. Nature343:51–53.
Pederson, G. T., D. B. Fagre, S. T. Gray, and L. J. Graumlich.2004. Decadal-scale climate drivers for glacial dynamics inGlacier National Park, Montana, USA. Geophysical Re-search Letters 31:L12203. 10.1029/2004GL019770.
Shepperd, W. D., and M. A. Battaglia. 2002. Ecology,silviculture, and management of Black Hills ponderosa pineforests. General Technical Report RMRS-GTR-97. USDAForest Service, Rocky Mountain Research Station, FortCollins, Colorado, USA.
Sherriff, R. L. 2004. The historic range of variability ofponderosa pine in the northern Colorado Front Range: pastfire types and fire effects. Dissertation. University ofColorado, Boulder, Colorado, USA.
Shinneman, D. J, and W. L. Baker. 1997. Nonequilibriumdynamics between catastrophic disturbances and old-growthforests in ponderosa pine landscapes of the Black Hills.Conservation Biology 11:1276–1288.
Sibold, J. S., and T. T. Veblen. 2006. Relationships of subalpineforest fires in the Colorado Front Range to interannual andmulti-decadal scale climate variations. Journal of Biogeog-raphy 33:833–842.
Stahle, D. W., et al. 1998. Experimental dendroclimaticreconstruction of the Southern Oscillation. Bulletin of theAmerican Meteorological Society 79:2137–2152.
Stockton, C. W., and D. M. Meko. 1983. Drought recurrence inthe Great Plains as reconstructed from long-term tree-ringrecords. Journal of Climate and Applied Climatology 22:17–29.
Sutton, R. T., and D. L. R. Hodson. 2005. Atlantic Oceanforcing of North American and European summer climate.Science 309:115–118.
Swetnam, T. W., and C. H. Baisan. 2003. Tree-ring recon-structions of fire and climate history in the Sierra Nevada andsouthwestern United States. Pages 158–195 in T. T. Veblen,W. Baker, G. Montenegro, and T. W. Swetnam, editors. Fireand climatic change in temperate ecosystems of the westernAmericas. Ecological Studies Volume 160. Springer, NewYork, New York, USA.
Swetnam, T. W., and J. L. Betancourt. 1998. Mesoscaledisturbance and ecological response to decadal climaticvariability in the American southwest. Journal of Climate11:3128–3147.
Taylor, A. H., and R. M. Beaty. 2005. Climatic influences onfire regimes in the northern Sierra Nevada Mountains, LakeTahoe Basin, Nevada, USA. Journal of Biogeography 32:425–438.
Wentzel, C. R. 2005. Temporal patterns of expansion of anisolate stand of ponderosa pine in the Bighorn Basin,Wyoming. Thesis. University of Wyoming, Laramie, Wyom-ing, USA.
PETER M. BROWN2510 Ecology, Vol. 87, No. 10
top related