Artic/es

rie Wetland Com as u a

C CI

W. CARTER JOHNSON, BRETT WERNER , GLENN R. GUNTENSPERGEN, RICHARD A. VOLDSETH. BRUCE MILLETT.

DAVID E. NAUGLE. MtRELA TUlBURE , ROSEMARY W. H. CARROLl. JOHN TRACY, AND CRAIG OLAWSKY

The wetll/IU! (Omplf.~ is Ihe jimCliollal c(oiogiCilI unit of lilt' pm;ric potlrole '('Sioll (PPR) of (I'll/rill North AmeriC/l, Diverse complexes of wel/IllU/5 (oll/ rilmll' higlr spmill/ /II/d fempoml ellvirOIIIIH'/l/(Illu:rcrogfl1eil)', producti,·i!f. allli bilXliwfsity /0 these gitlriMt'" pmiri/' Imu/s{(lpN. C/imlltr­Wtlflllillg sil/lilialious mh'g tht' 111.'11' mOfIel I'IETUNDSCAPE ( 1Vl.S) projrct /tIl/jer uduc/iollS iI, "'liter Imume, 5/",,/,,"illg of h)'dropt'riods. IIIIIi less-dYllamic 1'I.'gt'la/ iOll for pmiri .. we/IUlui (omplexes. The \ilLS modr! por/mys ,//1: fillllrt' PPR liS II II1l1ch less resiNI'm fC'OS)'S/('II1: Tire I<'I'SI .. m PPR will be 100 (Ifr ami Ilrl' eaSll'm PPR will Irm'e 100 [I'll' fllllCfimlll1 ..... flall/ls and IINling IlIIbilm 10 SIIpporllrisloric 11'1'1'15 of ... aterfowl mId ollJt'r wel/tlnd-l/epl'nc/I'nl spl'Cies, A'fllillfilitrillS {'(osrSlem goods mlll st'rl'iet's (/I (1I rr"I1I/cI'd s ill (/ WllmH'r dimille will be II mtljor c/1(IIlellg{' fo r Ihe collserl'll/ioll COlllllllmil}',

Keywords: pmirie pol/JO/C M'I/mllis. hydr% sy. IIIlmetic lIIodds, dill/n le d rmlgt, waterfowl

Freshwater wetlands worldwide are projected to be

particularly vulnerable to climate change (Kundzewicz et al. 2007 ). Their shallow depths and rapid evapora­tion rates contribute most to this vulnerabili ty, In Nort h America, wetlands in dry climates, such as playas and prairie potholes, have especially labile surface water-most dries up seasonally in all but the wellest years. In the future, a warmer climate without compensatory increases in precipitation is expected to significantly degrade and reduce wetland 'Heas, alTecting their ability to provide ecosystem goods and ser­viccs worldwide (M EA 2005).

Climate-change analyses for prairie pothole wetlands have focused on the semipermanent class- \\'etlands that hold wat('r throughout most rears and have more complex veg­etation zonation ,mel dynamics (Poiani and Johnson 199 1, Poiani et al. 1996, ClI rroll et al. 2005, Johnson et al. 2005). Yet the large majority of Ihe 5 million to 8 million \\'et land basins embedded in the glaciated landscapes of the prairie pothole region (PPR) in cenlral North America (figure I ) arc of the temporary and seasonal classes. Temporary wet ­lands generally occur as small, shallow basins that maintain surface water (or on I)' one to IWO months, whereas seasonal wetlands are gener.rlly hrrger and deeper, holding wllter lon­ger (two to three months; Stewart and Kantrud 1978).

Wetland clusters of these diverse permanence types com­prise a wetl:lIld complex (Weller 1988 ). j\'lcmbers of the complex, e\'en if distant from each ot her, are often hrdro­logicllll), connected by surface or groundlvater (Winter and Rosenbcrq' 1995, Murkin et al. 2000). Organisms move among members of the wetland complex seeking food. water, and cover (Naugle et al. 2001 ).

Wetland ecologists have recognized the contribution of the PPR weiland complex 10 ecosystem goods and services at the 13ndsca pe SCale (llrown and Dinsmore 1986. Fairbairn and Dinsmore 200 I, Swanson et al. 2003 ), There arc five key wet­land funCl ions that provide important services: flood abate­ment , water quality improvement, biooi\'Crsit}' enhance­melli, carbon manag('rnent, and aquifer recha rge (Gleason et a1. 2008 ), although support for wildlife habitat and the susta inability of waterfowl and other \vater-dependent populations have received the mosl attention.

Yet li ttle research has examined how climate change may alter prairie wetland complexes. Temporary and seasonal wetlands promote biological activit ), in the early spring becausc thei r dry basins recharge with snowmelt runoff when nearby semipermanent wetlands, which carryover wat('r in I110St winters as thick icc. arc still thllwing. Swan ­son and colleagues ( 1985) reported that the availability and

8ioScil'I/« 60: 128-140. ISSN 0006· 3S68. d eClronic tSSN t SH-32<44. C 2010 b)' American tn~lit ute of lIiotogicaJ Sciences. All rights rt"Sc:'T\-ro. Request

permission 10 photocopy or reproduce arlick coruml at the Uniwrsil)' of California I'ress's Rights Jnd l'crmissions Web site al \\~,~~. ucpressjoumu/s.coml rrprimill/o.llSp. doi: I O. 1 525/bio.20 I 0.60.2. 7

t 28 BioScience ' Febrrwry 2010 I Vol. 60 No.2 IVlvw.bioscieIJUlIIllg.org

Articles

Twentieth-century PPR climate The climate of the PPR. became warmer and wet­ter during the 20th (enlurr, but these changes were not sl':ltially uniform (Millett ct aL 2009). Minimum daily temperatures warmed by 1.0 de­gTC!.' Celsius (0C), while maximum daily temper.l­lures cooled by 0.15°C. Minimum temperatures \\'aTmcd more in winter than in sumlll(,T, whereas maximum tem peratures fell in summer and rose in winter. Mean annual precipitation increased by 49 millimeters (mm) (9%). The strong east-west gradient across the PPR of decreasing moisture toward the rain shadow of the Rocky Mountains steepened during the 20th century, with a small number of stations in the western PPR (paniell­larly in the western Cana(\i<ln prairies) becoming effectively drier, and a larger number of stations in the eastern PPR becoming effectively wetter.

D Prairie pothole region ,,-, i~~~~~'::'::~O"~' " ==~,~.:_:,:,,~,~_j_ . o 100 200 '" OV\I 800 .,

The PPR warmed during the 20th century at a level simililr to the global average; more warming is expected during the 21 st century. The Intergov­ernmental Panel on Climate Ch ange (lPCC 2007 ) projected that the mean temperature of Earth's atmosphere will increase by 1.8°C to 4.0°C by the year 2100. A suite of models has produced isocline maps for the approximate PPR (lPCC 2007), pro-

• Weather stations •

Figllre I. Location of tile prairie pOlllole region witllill Norlll Amer­ica, alld Ille names alld locations of long-term weatller stations II se(/ ill t/ris (lIIalysis.

abundance of aquatic irwertebr,\Ies, such as those found very early in the breed ing season in temporary and season::ll wetlands, increased waterfowl nest ing success. Later in sum­mer, when the more ephemeral wetlands arc often dry, semi­permanent wetlands provide habitat for waterfowl broods, molting ad ult ducks, and for other wetland vertebrates such <IS amphibiiUls that require relatively long hydroperiods to complete their life cycles.

Because of the v<lriability of water conditions over se,lsons and years, wetland complexes are more likely to ha\'e at le::lst some wetlands in a water and plant regime favora ble to a given species, thus ensuring diverse species representation in wetland landscapes (\'Veller 1999). \'Vater bi rds often build thei r local habitat units around a wetland complex that pro­vides various needs and also may act as a b;lckup in the ("vent of catastrophic change (Weller ! 999). As il result, wetland complexes support higher species richn("ss compared with single, isolated wetlands of comparable tOtal su rface ,m:a (Naugle et al. 1999).

A key unanswered question is the extent to which climate change might disproportionately affect the understudied, more labile members of the wetland complex, ul timately reducing its overall resilience and ability 10 support his­torically high levels of biodiversit y and ecosystem services. Winter (2000) proposed that sem ipermanent wetlands gen­erally found in lower topographic positions, especially those underlain by highly permeable deposits such as sand and gravel. should be less vulnerable than tempOT<lry or seasonal wetlands because they are supported by groundwater.

www.biosciellccIIUlg.org

jecting temperature increases near 4.0°C accompa­nied by small shifts in precipitation ( - 5% to 10%).

An intensified hydrologic cycle is also anticipated that will cause increased frequency of both drought and deluge (Ojima et a1. 2002, 10hnson et al. 2004 ).

Modeling approach This artide examines the cons("quences of a changing di ­mate on wetland comple.xes in PPR landscapes using a new simulation model. WETLANDSCA PE (WLS) is a dirnate­driven, process-based. deterministic simulation model. Its predecessor. \VETSIM, W,IS the backbone of our previous climate-change research (e.g., Poiani and Johnson 1993, Poiani ("t al. 1996, Johnson ("t al. 2005), but it modeled only serniperrnanell t wet lands. The development ofWLS allowed for a more comprehensive analysis of the climate-change issue across the northern prairies because it simul taneously simulates wetland surface water, groundwater, and vegeta­tion dynamics of the wetland complex, including multiple wetland basins of semipermanent, seasonal, and temporary permanence types, in addit ion to overflows between basins.

The key differences between WLS and WETS Il\'1 (ver­sions 1.0-3.1) are (a) the simulation of a \I'etland complex rather than a single, semi permanent basin; (b) the incorpo­ration of a "double-bucket" model structure, allowi ng for separate Gllculation of surface water depth and volume and depth to groundwater to model the very different hydrol ­ogy of the three wetland permanence types (figure 2); (c) a change from a daily to a 10-day time step; (d) the lise of STE LLA rather than MATHEMATICA (Wolfralll 1999) as

FebrullfY 2010 / 1'01. 60 NO.2 · BioSci(,llce 129

Articles

• • ', ••• • ,. p .. " . " ", ' , rectpla on • . .", • • • o· , I, ", 'I Iree D on upland

D precipitation on wetland

~ET Dry basin groundwater ET

Wetland surface bucket

~ ---~ ---

Maximum depth of

loca' groundwater

---- -----

10 Groundwater surface

----_'!'_---

-- --~---7'--

/ ------~

Water table

Local groundwater bucket

Figllre 2. Generalizer/ wel/and waler bl/dget witll double-bucket submodeillserl ill tile simulation model WETLA NDSCAPE. £T, eV(lpotrallspir(l,iot, .

the modeling software platform (use of trade or product nallle does not imply endorsement by the US government ); lind (e) use of the Blaney-Criddle equation to calculate potential evapotranspiration used in WETSIM 2.0 (Poian i ct al. 1996), rillher than the modified Hargr{'3 \'cs equa­tion used in WET$ IM 3.1 (Joh nson CI 31. 2005). We chose the Blaney-Criddle equa tion because it produced a bettef fit than did the Hargreaves equation with field data from our Orchid lvlcadows site. (Refer to Carroll and colleagues 120051 for a full description of the I Q-day time-step method using avcrages of daily inputs of precipitation, minimum and maximum temperature, and the technical aspects of implementing the Blaney-Criddle equation. )

WLS was parameterized and tested using long-term monitoring data from 10 wetlands (3 temporaries, 3 season­als, and 4 semi permanents) and 40 groundwater wells at our Orchid Meadows field site ncar Clear Lake, South Dakota (Johnson et al. 2004 ). Model setup required the determi ­nation of basin morphometry and overfl ow connect ivity among wetlands. Calibration of the model used 13 years of data for all wetlands (1993-2005), wi th three add itional years of observations (1987-1989) for some basins. The monitoring data included weather ex tremes of deluge (the early to mid 199Os) and drought (the late 1990s and ea rly 2000s; Johnson ct al. 2004). Wctland watersheds were 111

permanent grass cover with occasional grazing.

130 BioScience ' f ... brtlary lOlO / Vol. 60 No. 2

WLS successfu ll)' modc\cd thc kcy hrdrological properties that distinguished each wetland permanence typc, including spring rise, summer drawdo, ... n, hydropcriod, and depth to groundwater (figure 3). An exact match between modeled and observed conditions was neither expected nor attain­able through calibration because of the mismatch between the to-day model time step and the two-week field sampling schedule and dependcnce on off-site primary \\'eathcr sta­tion data for winter precipitation.

Driving wetland models with weather data from many sta­tions has proven to be an instructi"e proxy for understanding both the temporal and spatial variability of wetland dynamics, and the sensitivity and geographic complexity of the response of prairie wetlands to climate change (Johnson ct al. 2005). The range of basin morphometry of the Orchid Meadows wetland complex, however, ,"'as less than the range across the PPR.

In summary, our analytical approach \Vas to isolate, to the extent possible, the effect of climate on the conditiOn of the wetland com plex by driving WLS, calibrated at Orchid Meadows, with weather data from 19 other PPR stat ions (fi gure I), each with 100 years of data (1905-2004 ). To improve geographic dispersion of weather stations in our previous analysis (Johnson et al. 2005), we added two sta­tions (Aberdeen, South Dakota, and Regina, Saskatchewan) and dropped one (Clark, South Dakota ). \Ve assessed the effects of fllture climate conditions by adopting three

11'11" 1'. b jose; ... nUll/a g. org

--III 528.0 E E 527.0

~ 526.0 .--• > 525.0 • -• .& 524.0

~ 523.0

524.0 --~

E 523.5 • .§. 523.0 c o :.: 522.5 • ~ 'i 522.0 ~

•

•

2 521.5

~ - Modeled wetland

521 .0

_ 525.4 -E 525.2 (IJ 525.0 E _ 524.8 c 524.6 0 . -- 524.4 • > .!! 524.2 • 524.0 ~ • - 523.8 ~ 523.6

51 -Modeled

SPf'

93 94 95 96 97

level

, . , •

• • •

• •

98 99 00 01 02 03 04 05

Year

Figure 3. Calibratiollllydrogmpits for (I temporary (Tl), seasOllal (S I), and semipermflllelll (51'4) wetlalUl at till! Orcllid MetulolVs, 50111/1 Dakota, fi eld site over a 1 Z-year period (1993- 2004). Horizollfal/ille is tlte wet/mitt bOllom. Elevation is ill meters above meall sea level.

climate scena rios (+ 2°e, + 4°C, +4°C + 10% precipitation) on the basis of projections from global circulation models (Ipee 2007).

Measures of wetland response to climate The biodiversity and productivity of wetland complexes are affected by exogenous forces, such as climate, and endogenous forces, such as the mix of permanence types, su rfici.11 geology, water regimes, wetland juxtaposition, and vegetation (Weller 1994, 1999, Weller and Fredrickson 1974, Swanson et al. 2003,

11'\\ ·w. b i osci" /Ice /I r (I g. 0 rg

Articles

van der Valk 2005 ). 'We quantified the model response to a range of climates using two primary measures or indices: wet­land hydroperiod (time of inundation) for each permanence type, and a measure of the dynamism of wetland vegetation for semipermanent wetlands. These modeled responses of­fer three novel contributions to scientific understanding of climate-wetland interactions: First, they refine and reinforce earlier research on semipermanent basins; second, they fill the research gap on the effects of climate change on more labile wetlands; and third, they simulate the potential effect of climate on wetland complexes-information that was previously unavailable or largely conjcrtural.

Hydroperiod. Many wetland species require a minimum time of inundation either within a home wetland or across a wetland complex to complete their life cycles. For example, most d<lbbling ducks (e.g., mallard. teal) require a mini­mum of approx imately 80 to 110 days of surface water for their young to Aedge and for breeding adults to complete molting (Bcllrose 1980, Austin <I!ld Miller 1995). WLS computed the hydroperiod across years for e<lch wetland of the complex to evaluate the effect of historic and alternati\'e future climates on life-history requirements and thresholds for key wetland species or guilds.

Vegetation cover-cycle indel. The vegetation cover cycle of semipermanent prairie wetlands has been well studied and modeled . [n brief, the drought and deluge frequen ­cies associated wi th a given climate determine the speed of the nutrient and vegetation cycles (Weller and Spatcher 1965, rvlurkin et al. 2000). Prolonged high water produces a "lake" wetland with little emergent cover and few nu­trients in detritus, whereas persistent low wa ter produces heavy emergent cover and high nutrient sequestering in plant material. The occurrence of both extremes during a weat her cycle causes plant population turnover (maintain­ing biological diversity) and nutrient mobilization . These events have been described as a wetland cover cycle that includes four stages: ( I) a dry stage with dense emergent cover and little or no standing water; (2) a regenerating

stage with germination from a diverse seed bank, reflood­ing, and vegeta tive propagation; (3) a degenerating stage when emergen t plants sta rt to decline; and (4) the lake stage with high water and lillIe emergent vegetation (van der Val k and Davis 1978) .

The speed of the cover cycle (return time) and the num ­ber of switches between cover-cycle stages over a period of time ,Ire strongly correlated to productivity and biodiversity (van der Valk and Da\·is 1978, S\\'anson el al. 2003 ). Long return times (one cycle per century) or extended periods without switches produce wetlands "stuck" in either the lake stage or the dry stage with stable, but relatively unproduc­tive, conditions. Weller and Fredrickson (1974 ) noted that stable water levels produce orni thologica!ly "dead" ma rshes characterized by a cen trally open marsh with a perimeter of dense emergent vegetation. Short return times (e.g., three to

FebruM}' 20101 Vol. 60 No.2 · BioScience 131

Articles

Table 1. Watertleptll (md duratiOIl rilles for cover-cycle stage swilcilcs. Source: Modified from Poia"i and Joim soll (1993).

Current stage New stage

Lake marSh Hemi-marsh

Hemi·marsh lake marsh

Hemi·marsh Dry marsh

Dry marsh Heml-marsh

Dry marSh lake marsh

fOllT complete cycles per century), however, creale the tem­poral scale environmental heterogeneity necessa ry for high productivity. \VetJands with short return times marc often produce the hcmi-marsh conditions desired by wetland managers (approximately equal proportions of emergent co\'cr and open waler, resulting from a combination of regenerating and degenerating stages).

We developed a cover-cycle index (CCI. sec the equation beloll') of climate favo rability 011 the basis of two variables of equal weight: ( I) the proportion of time, averaged across the three semipermanent WLS wetlands, spen t in the hemi ­marsh stage during a lOO-year simulat ion; and (2) the average number of cover-cycle state changes (i,e., swi tches) over the same time period. \Ve used a transition probability model to estimate the proportion of tim(' that model I.,.et ­lands spent in each of the three cover-cycle stages (lake, dry, and hemi-m'lrsh). Transitions between stages were based on water depth classes (table 1) modified from Poiani and Johnson (1993). For the second variable of the index, we r('cord('d and summed ('ach tra nsition that occurred during the simulation between any of the cover stages.

CCI = (HMIHM ' + SH'ISW')/2

In the equation above, HM = time in hemi-marsh, HM ' = maximum percentage of time in hemi -marsh across weather stations; SW = the number of switches, and SW' = the maximum number of sl"itches across weather stations. The maximum number of switches across all simulations was 21 (Academy, South Dakota- historic), whereas the maxi­mum time spent in the hemi-marsh stage was 51% (Water­town, South Dakota-2°C). The highest CCI value was 0.86 at Academy, South Dakota; Algona and Webster CilY (Iowa) tied for the lowest score of 0.04. This index is offered as a starting point in quantifying the relationship between climate (and climate change) and prairie wetland productivity.

Modeling labile wetlands and the wetland complex WLS captured the contrasting water regimes among moni­tored wetlands of the Orchid Meadows complex (figure 4). We eval uated three representative wetlands, one from each permanence type, which all exhibited a simultaneous spring rise when the snowpack melted and rain ran off fTolen or saturated soil. The semipermanent wetland carried over water from the previous growing season, whereas the more ephem-

t 32 BioScience ' Fc/mlllry 20 10 I Vol. 6(} No.2

Maximum depth (meters) Duration

< 0.5 May- July

> 0.75 2 years

< 0.1 May- July

Between 0.4 and 1.0 1.5 years

> 0.75 2 years

I.' II SP3

1.2 . 53

-E 1.0 _ n

-~ a. 0.8

• ~ 0.' -• ~ • 0.' ;: 0.2

0.0 J F M A M J J A S 0 N D

1998

Figure 4. Modele,1 liydrogmpll compnrisolls llmOllg (l tem­pornry (T 1), seasotw/ (53), mul semipermallent (SP3) wet­/nlld by tile simulatioll //Jodel WETLA ND5CAPE ( WLS) from 1998 weatller "aW for tile Orchid Meadows field site.

eral permanence types were dry or nearly dry in the fall and winter, but filled from spring runoff. As the weather warmed, the temporary wetland dried up first, because it was the shal ~

lowest and leaked to the groundwater (recharge wetland). The deeper and less-leaky seasonal wetland had a considerably longer hydro period, especially the particular example we eval­uated, with a closed basin and no active outlet (figure 4). The semipermanent weIland had a strong summer drawdown, but carried water over into the next year.

In the historic period simulated by WLS, temporary wet­lands had the shortest average hydroperiods (proportion of time wet ), ranging from approximately 0.[5 to 0.55 among weather slat ions (figure 5) . Expressed as times of inunda­tion,lhese a\'erage hydroperiods ranged from approximately one to IwO weeks to two months. The shortest hydroperi­ods occurred at the semi-arid stations in the western PPR, including ~'Iontana, Alberta, and Saskatchewan (figure 5), wher('as the longest hydroperiods were at the subhumid stations in the southeastern PPR of Iowa and western Min­nesota. Seasonal wetlands produced a wider range of hydro­periods than the temporary wetlands, from approximately 0.15 to 0.80; the longest hydropcriods were associated with semipermanent wetlands (figure 5). Water depth exhibited a sim ilar geographic pattern.

1V1Y1I'.iJ lose If' 1« IlIlIg. () rg

wetlands 1.0 ,--__ ,-,-,,-,:..::.:.:c -"-'-:=.::-=------0.8

0.6

0.4

0.2

0.0

Seasonal wetlands 1.0 ,----....::::=====-----

0.8

0.6

0.4

0.2

0.0

Semipermanent wetlands 1.0 ,------'======"--0.8

0.6

0.4

0.2

0.0

Figure 5. Simulate,1 mean Ilydroperiotl patterns (lcross tile prairie podlole region for COmp01lell ts of fi, e wel/(llul COlll ­plex at tile Orchid Meadows field site usillg Ille sillllliatiol/ model WETLANDSCAPE. Sltmdard error oflile metm sl/Owl/ above eacll bar. See figure J for well/I,er stllliotl foca liolls.

The contrasts in hydroperiod among the permanence types were very strong; at the large majority of stations, semipermanent wetlands would have historically held deeper standing water for more than half of the open-water season. whereas the standing water for seasonal and temporary wetlands was present for less than half of the open -water season at most stations. Overall, there was a strong geo­graphically based contrast between the simulated hydro· logical conditions of the wetland complex's basin members. H)'droperiods were much lower on the western, more arid boundaries of the PPR, and much higher on th(' ('astern, more humid fringes.

Results of c limate-change simulations We explain the resul ts of the WLS simul'ltions for the PPR in the following sections.

lVII''''. It i () sdc 11 (fill II g. 0 rg

Articles

I Muell't.r, SK: 165 (19%) I I Reglna, SK: 178 (13%) I

• Popl.,. MT: 200 (17%) • • AIgOll., IA: 216 114%) I

• AbtirdMII, SO: 204 (:it%) • • Ac.demy, SO: 222 (16%) •

I J311 I FeblMar lAP' IMay l JU II I JUt lAug l sep loet lNov lOK I

Figure 6. Simulated meatl length (usillg WETLAND­SCAPE) of tlJe open-water seasoll at six prairie pOlllole region weatller stations (olle frolll eaell ecoregioll) for historic ( J 905- 2004; witltll of ligllt blue bar), 2 tlegrees Celsill s (0C) warmer scenario (yellow + ligllt bille btlr), U/u/4°C warmer scenarios (red + yellow + liglll bllte btlrs). Nlllllber of open-water dtlys reported for eaell sla­tion, alollg witl, percentage illcrease ill opell-water tlays between Ili storic alld 4°C sccnllrio. lee-free setlSO llllir tempertllitre IlIresllOlds in sprillg and fall (l O-day tJlrerage of mean daily air tempert/tllre of J0C).

Open·water season. Earlier icc-out and later freeze dates during the past 150 rears ha\'e been reported for Northern Hemi­sphere lakes (Magnuson et al. 2000). Longer ice-free periods are projected in the future with conti nued climate warming opec 2007). The WLS simulation projected that a surface temperature increase of 4°C dist ributed even I)' across months would extend the ice· free season for pr;lirie wetlands by 13% to 26% across weather stations (figure 6). WLS used a 10·dar threshold mean daily air temperature of Joe in the spring and fall to estimate the start and end of the open-water season each year. These temperature thresholds were based on ice data at our Orchid rvleadows site (Johnson et a1. 2004).

Our results suggest that clim,lIe warming may produce threshold or nonlinear effects (Burkett et al. 2005) in some PPR climates. Each 2°C increment in temperature extended the icc-free season about one model time stcp (10 days). Spring W;lS advanced more than fall was extended. At Academy, Sou th Dakota. near the southern boundary of the PPR, snowmelt runoff was complete in late March, on average, under a 4°C scenario, rather than mid-April under the historic cl imate. At Muenster, Saskatchewan, ne:lr the northern boundary of the PPR, snowmelt runoff would first occur around 1 May instead of mid-May under the historic climate. The percentage increase in the length of the ice-free season was similar among stations (13% to 19%) except for Aberdeen, South Dakota, which exhibited a considerably lengthened ice-free season (26%), most of which occurred between the zoe and 4°e temperature scenarios (figure 6).

Pennanence types and water regime, The complexi!}' of Ihe response of the model wetlands to climate-change scena rios

February 2010 I Vol. 6Q No.2 ' BioScience 133

Articles

1.4

12 ~ .s 1.0 ~ - 0.8 "-• ~ " 0.6 • -• !; 0.4

02

0.0

1.4

12 ~

S 1.0 ~ - 0.8 "-• ~ " 0 6 • -• !; 0.4

02

0.0

1.4

1.2 ~

§. 1.0 ~ - 0.8 "-• ~ " 0.6 • -• !; 0.4

0.2

Seasonal

Semipermanent

Histonc Plus 2·C Plus 4"C

Historic Plus 2' C Plus 4' C

Historic Plus 2°C Plus 4' C

Figure 7. TII ree mUllin/ WETLA NDSCAPE ( WLS) /I),dro­gmplls ( 1986- 1989) slww;Ilg ti,e effect of 2 degrees Celsius (OC) (l mI4°C warming scellarios for tl,e deptlt (it! melers {m}) of a temporary, seasonal, ami semipermallent wet­land from tile Orchid Mem/oll's wetlalld complex.

was part icularly evident during periods wi th variable weather (e.g., 1986- 1989). The ma in consequences of a warmer cl imate were ea rlier snow-pack mci ting, reduced wetland water depths and volumes, shorter hydropcriods, faSler dr:J.wdoWIl, and reduced snowmelt and rain fall peaks (figure 7). The response of WLS to specific precipitation CVCnlS, howevcr, was quitc complcx. Modd water budgets fo r each of the simulation yca rs revealed that the primary cause of reduced pcaks was great ly reduccd ru noff because of drier soi ls in the watershed under a warmer climate. This was a product of greater evapot ra nspiration in the uplands

t34 BioScicncc ' Fr/lnlllry 20101 Vol. 60 No. 2

during the previous year, and of the extended fall and eilrlier spring waler seasons. \·\le program med runoff curve num­bers (Voldseth et a1. 2007) in \OVL$ 10 va rr with calculated soil moisture values.

The annual simulations revealed that the effect of higher surface tem peratures increased with increasi ng wetland permanence (figure 7). For example, weathcr during thc years 1986 (cspecially) and 1987 produced the strongest differences among scenarios fo r the temporary and seasonal wctlands, whereas the water regime of the semi permanent wetland was greatlr al tered by the scenarios in all fo ur years of the simulation . The scmipcrma ncn t wetland dried up each year under the 4°C scenario, a behavior more typical of seasonal wetlands in the cur rent climate.

Semipennanent cover-cycle dynamics. Geographic trends in wetland cover-cycle dynamics, as expressed by the CO, fol­lowed a strong longit ud inal gradient across the PPR during thc historic period (figure 8). Simulations indica ted that the most favo rable clima te fo r high wetland productivity (high­est CCI scores of 0.44 to 1.0; dark green in figure 8) would have covered nearly half of the PPR (47%; table 2) and included virtually all of thc eastern Dakotas and portions of southwestern Manitoba. southeastern Saskatchewan, and eastern Alberta, just south of the boreal fores t. This general region is known for high production of waterfowl during thc 20th cen tury (Reynolds et al. 2006).

The lowest CC I scores (0 to 0.22; yellow in figure 8) were from the western and eastern extremes of the PP R. To the west, northern ~'I ontana , south\\,estcTll Saskatchewan, and southeastern Alberta form a subregion where wetland and waterfowl productivity would have been strongly limited in most, but not all, rea rs by insufficicnt moistu re and a very Ion!) covcr-cycle return time over much of the 20th cen tury. In contrast, the low CC I scores in the east, includ­ing Iowa and most of southwestern Minnesota, identify a subregion where productivit), would have been limi ted by a slow cover cycle, prolonged lake-marsh conditions, and 100 much water. Wedged between these categories was a small, light-green area ( 15.4%; table 2) of moderate productivity potential.

The warmer climate scenarios produced an eastward shift of all CCI classes (figurc 8). The area of re llow wit h the least favorable climate increased from 37.6% of the map during the historic period to 39% and 62.5% of the map under the 2°C and 4°C scenarios, respectively (table 2). The da rk green area of high potcntial productivity nearly disappeared fro m the map, sh ift ing to the east and droppi ng sharply from 47% in the historic period to 30.8% and 12.5%, respectively, under the ZOC and 4°C scenarios (table 2). Under the 4°C scenario. ncariy all of eastern Nort h Dakota was covered by the driest index scores, while eastern South Dakota ret:lined a more favorilble cover-cycle category (figure 8). Overall, the part of the PPR with the highest CCI scores historically was largely transformed under the more extreme climate sce­na rio into a region with low to moderate CCI scores.

www./iioscicI1CCll/ag.org

,

\ Cover cycle index

0 0.0 - 0.22

0.23 - 0.43 0.44 - 1.0

, , ,____ '

: Historic /

Temperature

, ,

: I

...--__ I , -----.-

,

\ -

-, -.

~'.' ~-U_'_-J' " , , I , r-----' , , ,

I , c.....-,------1,

L

Articles

Figure 8. Mllps of tile cover-cycle ;mlex for II,e prairie pOl/lOle regia" based Oil IliS/o ric went/ler data and II/ree dimate scenarios (2 degrees e eIS;'I! IOCj warming; 4°C warmillg; 4°C 11'("",illg (//1(110% iI/crease ill precipitatioll).

Eastern portions of the PPR that wefe historically too wet to produce the highest eel scores could see improvements under a slightly wanner climate. Under the 2°e scenario, the ext reme southeastern section of the PPR (Minnesota and Iowa) shifted inlo the high-productivity category, suggesting thaI the drier climate could create more desirable water-to­cover interspersion ratios and morc dynamic wetlands. Simi­larly, the only part of the PPR with high CCI scores under the 4°C scenario was in southwestern Minnesota and Iowa. The \·VETSIM model produced a similar finding for s{'miperma­nent wetlands (Johnson et al. 2005). Increasing precipitation by 10% virtually canceled out a 2°C higher air temperature, a result also found by Voldseth and colleagues (2007, 2009). Although greenhouse gases may produce a more favorable wetland climate in this region in the future, it currently has the f{'w{'st undrained wetlands (Dahl 1990) and th{' least amount of wat{'rfowln{'sting habitat within the PPR.

Seasonal wetlands in the West. The hindcas! CCI analysis by WLS identifi{'d a 150- to 200-kilometer north-south run­ning SW3th just east of the Missouri River (forming the

lVII'\\'. b iose if" w' /11l1g. org

western PPR boundary in the United States), through both Dakotas and northward into Canada, producing the most productive and dynamic semipermanent wetlands du ring the 20th century (dark-green polygon in figure 8). The drier, western portion of this swath is occupied by the Missouri Coteau, a dead-icc moraine with the highest wetla nd densi ­ties in the PPR (e.g., Edmunds County, South Dakota, has approximately 45,000 natural wetlands; Johnson and Hig­gins 1997). Seasonal wetlands comprise 72% of the wetland area in this county (Johnson and Higgins 1997), providing the foundation for extremely high waterfowl production during wet weather extremes.

The capacity of a landscape to support waterfowl is largely determined by the abundance of seasonal wetlands, because breeding pairs isolate themselves from other individuals of the same species when establishing territory in the spring. More important, seasonal wetlands in places like Edmunds Coullly, South Dakota, arc embedded with in abunda nt grass];lIld landscapes that reduce predation and increase nest success, which is a determinant of population growth in mallard ducks (Hockman et a1. 2002).

Fcbfllllry 20 10 I \'01. 60 No.2 ' BioScil.'nce 135

•

Articles

Table 2. Percelltage of till? prairie potllole regioll iI, lI,e th ree cover-cycle j"dex Ctl legories IIl1fier I,islor;c (///(/ scenario climales. Til e climatic regioll prodllcing tire most dyrwmic wet/al/ds \\fas dark gree /lj moderately {iy,wmie, [igla gree,,; ami least dYllamic, yellow (see figure 8).

Simulation Dark green Light green Yellow Total

HistoriC 47.0 15.4

Plus 2'C 30.8 30.2

Plus 4' C 12.5 25.0

Plus 4"e + 10% 32.5 35.0

- 40 0 Historic 0 ~ Plus 2°C -0 30 Plus 4°C -V> ~

'" " 20 >--0 ~

10 " .c E ~

z 0 60 80 100 120 140 160 Hydroperiod frequency (days)

Fig/Ire 9. AIIIllIai Ilydroperio(/ frequ ellcy m/Cllitiled usillg WETLA NDSCAPE{or seasollalwet/alld 51 (HId a 100-yetlr weatl,er dtlta set and two climate scella rios for tile Millot, Nortll Dakota, weather statiOIl. A 100-dtly Itydroperiod (vertiml bar) tlpproximates tile avertlge mi"imllm Ilydro­periodllecess(lry for mallY waterfowl (/tI(1 ampltibitltlS (e.g., leopard frog) to complete tlleir life cycles.

Simulations with WLS indica te that substantial reductions in the hydropcriod of seasonal wetlands would accompany climate warming on or near the :-'·1issou ri Cote'llI. Si mula­tions for the Minot, North Dakota, weather station using seasonal wetbnd S I (median of three model seasonal wetlands) produced a cu rve of hydroperiod frequency duro ing the 20th cen tury (figure 9). Annual hydroperiods of at least 60 days would have occurred in nearly 40% of the years, whereas lengths of 150 days would have occurred in only about 5% of the yea rs. A hydroperiod of 100 days, a duration that generally corresponds to the median value for many vertebrates, including waterfowl (lkl1rose 1980) and amphibians (Wagner 1997), 10 complete min imum life-history requirements, would have occurred in 22 of 100 years. Some dabbling ducks such as pint ails ca n complete their breeding cycle (pairing, incubation. brood rearing, fledging, molting) in as few as 70 days (Austin and Miller (995), while some diving ducks, such as canvasbacks, require a minimum of 130 days (Be)lrose 1980).

A 2°C I .. armer climate cut the lOO-day an nual hyd rope­riod frequency by two-t hirds, from 22 years to 7 years (figure 9). A 4°C warmer climate nearly eliminates a hyd roperiod equal to or longer than 100 days. Thus, in only I of 100 years

t 36 BioScience ' Fe/millry 2010 I \101. 60 No. 2

37.6 100.0

39.0 100.0

62.5 100.0

32.5 100.0

1.4 SP4 (Historic)

1.2 -- $P4 (Plus 4°C)

- $1 (Historic) E 1.0 -- $1 (Plus 4°C) -~ T1 (Historic) a. 0.8 T1 (Plus 4°q --" "0

0.6 ~

" -~ 0.4

0.2 "- .... 0.0 -

0 20 40 60 80 100 Duration: percentage of time steps

Figll re 10. A comparison of wetlalld stage-dllmtiotl wrves from tile simlllatioll model WETLA NDSCAPEfor I,istoric climate alld a 4 degrees Celsills climate challge scetUirio for tile Actillemy, SOIlt/, Dakota, weat/ler statiol/.

under the warmest scenario would this model wetland have an annual hydroperiod of sufficient length to enable many wetland vertebrate species reliant on seasonal wetlands to complete their life q'CIes. The shallow depth, high ratio of surface area to volume, presence of surface water in sum­mer when evaporative demand is the highest, and weak grou ndwater support make season;ll wetlands perhaps the most vulnerable of the wetland complex to climate warm­ing. Although semipermanent wetlands with a naturally longer hydroperiod would provide some alternative habitat for seasonal wetlands in marginally dry yea rs, thcir small areal extent in the western PPR would not make up the dif­ference.

Water-depth frequency aRaIysis. The effect of a warmer climate on the hydrology of prairie wetlands can be visualized by the construction of wetland stagc (depth ) duration curves (Price 1994). For example, during the historic period at Academy, South Dakota (the station that produced the most dynamic wetlands) , semipermanent wetland 51' had a depth exceeding 50 cent imeters (cm) nearly 50% of the time (figu re 10). Un ­der a seasonally uniform warming of 4°C, a depth exceeding 50 cm occurred less than 20% of the time. For seasonal wet­land 51, water depths exceeding 25 cm declined in frequency under the 4°C scenario from approximately 30% to 15%; for

www.bioscimulI!ag.org

temporary wetland T I, frequency at tllat depth dropped from

approximately 10% to 7% (figure 10). O'"crall, the warmer scenario reduced the time spent by ;t ll wetland permanence types at the grealer water depths. but the magnitude o f the effect was directly related to wetland hydropcriod, with the

semipermanent and seasonal ,,,,('tlands l>cing affcrled the most, and the temporary wetland the )C;lSt.

Wetlands of the complex also shifted permanence type in response to a warmer climate. Under the 4°C scenario. the semipermanent Ivctland exhibited depth frequencies nearly identical to the se3sona! wetland in the historic dimate, and

the .seasonal wetland under the same scenario closely matched the duration curve of the temporary '\INland. Thus, a 4°C

warming was sufficient to change the water regime of a future semipermanent wetland into that of a historic seasonal, and a future seasonal wetland into that of a historic temporary.

Resilience of the prairie wetland complex The respective responses of members of the wetland com­plex to climate change were not as predicted o r expected.

Scveralmajor factors determined the responses. First, water

regime characteristics track each other across permanence types because I,'etlands in the complex experience nearly identical weather, which leads to a muderately coordinated

response among members of the complex to weather vari­

ability. Second, other factors produce individuality among wetlands and Ivetland types in their water regimes ,md re­sponses to climate cha nge. These <Ire mostly physiographic

factors such <IS catch ment area, slope, soils, and p<ltterns of snow accumulation that affect the do\vnslope delivery of

precipilation 10 the ",elland. Groundwater relations are a third factor as described by Winter (2000). Vegetation and

land use cont ribule to a fourth factor that is correlated with physiographic faclors that mainly afft'Ct eVOIpotranspira­tion and runoff (van der Karnp et al. 2003. VoldsC'lh el al.

2007). Hence, this multiplicity of fac tors, some of which arc relatively fixed in space (the enl'ironmentaltemplate or physiographic setting), and others that are quite variable in both space and in time, complicates the response of the

indi\' iduOlI wetlands of the complex to climate change, but in combi nation, these factors create the incredibly di\'erse and rich heterogenei ty in pattern and process across prai rie

wetland landscapes. Evidence from modeling, however, re\'ealed identifiable

patterns of response by the three wetland types to climate

ch"nge. First WOlS the unexpectedly gre"ter resil iency of tem ­porary wetlands, the most labi le of the classes. The model

tempor"ries recharged with water in the spring under all scenarios at nearly historic levels under our assumption

of uniform temperature increases across scOlsons. Once

recharged, temporaries lost wateT through evapotranspira­tion and groundwater recharge. Greater evapotranspi ra­

tion ra tes in a warmer climate produced shorter hydrope­

riods than under simu lated historic conditions; however, the shortening was less than expected-a given increment of ai r temperature ilffected evapotranspi ration rates less

\\1V1V. b j mcit, /lCt' 1I111g. 0 r8

Artic/es

in spring, when temporaries hold water, than in summer,

when they do not. To illustrate. WLS calculated that a 4De increase in air temperature at the Minot weather station in 1982, a representative year, increOlsed the ra te of aver­age (\;Iily su rfOlce wilter evapotrOlnspiration from model

wetlOlnd SP4 less in May ( 1.6 111m per day) than in Ju ly

(3.7 111m per day). ThIlS, by naturally drying out by Iilte spring in most years, temporary wetlands escape the more

se\'ere consequences of ,I warmer climate in slimmer. It was coun terintuitive thOlt small , shallow recharge wetlands

would be more resilient to climate warming than deeper discharge wetlands. This apparent resilience of tempo­rary wetlands. however, could be lessened with different

assumptions about the future seasonali ty of temperature

OInd precipitation. \"' inters that warm more than summers, as <Ire projected for many regions (lpee 2007), would increase sublimOl tioll and reduce snowpack, thus lowering

the spring rise and shortening hydroperiods. Conversely, semipermanent wetlands that usually have

standing water in the hottest part of the growing season were

rel'llil·ely more affected by the disproportionately higher evapotranspiration rates than were temporary wetlands. Greater evapotranspiration losses occurred in both the wet­

land itself and in the catchment. The WLS simulation adjusts runoff according to soil moisture levels. Percent reductions

in runoff, however, were gre,lteS! for semipermanent wet­];\Ilds because of the much drier soils in a greenhouse cl i­

mate in the summer compared with spring. The disproportionale effects o f warming on semiper­

malu'nt wetlands have major consequences for vertebrates by greatl}' shortening the length of the hydropcriod for the whole complex. This model result contrasts with Winter's

(2000) assessment that semipermanent wetlands will be the most resilient permanence type to climate change because

they have more groundwater support , especially those in glacial deposits with high permeability. Clearly, a wetland is semipermanent becallse of groundwater support (Winter and Rosenberry 1995, van der K,unp and Hayashi 2(09); however, it appears from our modeling that the effect of this "bonus"

water, if locally derivl"d, is depleted quite quickly in long-term simulations under chronically warmer climates tilat cause a pcrrnanence type shift to the seasonal class. van der Kamp and Hayashi ( 1998) reported that extended periods of drought led

to a lowering of the local water table around the wetland. The influence of groundwater on wetland permanence

de!,ends on the permeability of glacial deposits associOlted with the wetland basin. Winter OInd Woo ( 1990) estimated

thOlt grollndwOlter supplied 5% to 25% of the water input

to semipermanent wetlands. Carroll and colleagues (2005)

estimated thai groundwater contributed 19% of the inflow to thei r model wetland. Modeled gro undwater inflows for

the semipermanent wetlands at Orchid Meadows ranged

from 4% to 15% percent of the water budget. This range suggests that our WLS semipermanent wetlands occupied glacial deposits from low to moderate permeability. Had some of ou r model wetlands been associOlied with highly

Febru(lry 20/0 I Vol. 60 No.2 · BioScience 137

Arlicles

permeable glacial deposits contributing 15% to 25% percent of the waler budget, Ihe modeled effects of climate warming on sem ipermanent wetlands may ha\'c been less sc\'crc.

Conclusions and recommendations WL$ simulations (figure 10) showed that all three perma­nence types of w('tlands lost sign ificant hydropcriod under both l OC and 4°C ,,'arming scenarios, unless accompanied by a minimum increase in precipitation of 5% to 7% per degree of warming. These results strongly ind icate that thc prairie wetland complex is, as a unit. highly vulnerable to climate warming of the magnitude projected by global circu­lalion models (lpec 2007). None of thc permanence types of the complex escapes serious waler regime consequences under these dimate scena rios.

Our model experiments suggest that the vulnerability of the members of prairie wetl:md complexes to climate warming lind drying, as defined by our scemlrios, generally increilscs in this order: temporary wetlands. semipermanent \\'etlands, seasonal wetlands. The modeling results for SC'dsonal wetlllnds were the most alarming, particularly in the western PPR, \\'here they are often the wetlands with the longest hydropcriod in the complexes. and as such, on which vertebrates depend to complete their rclath'ely long life cycles. The highly evaporative summertime greenhouse climate \\'iII Mpush back" hard against the extended hydropcriods of seasonal wetlands made in wet springs. All but the \'ery wellest of the historic "boom" yeilrs for waterfowl production in the more :Irid regions of the PPR may be Mbust" reilrs in a 4°C \\'armer climate.

This pattern of accelerated drying raises additional con · cems for the future conservat ion of waterfowl populations. Waterfo\,,rl have already ahef(~d timing of their migration to arrive earlier on the breeding grounds (lI.·lurphy-Klassen et al. 2005), but the key to maintaining populations will be in their abil ity to adapt to earlier drying ;Ifter their ;lTrival. Early drying may be an ecological trap (Schlaepfer et al. 2002, Batt in 2004), whereby migrating ducks are attr:leted to wet b:lsins in early spring but cannot fl edge their young when wet lands dry up too quickly in the more evapor:ltive greenhouse climate. Survival of mallard ducklings in North Dakota was 7.6 times lower when fewer seasonal wetlands were :lvailable during drought than when water was abun­dant in theS(' s.1me \\'etlands in subsequent years (Krapu et al. 2006). Indeed, the function:11 loss of the many wetlands thaI attract ducks 10 breeding grounds in spri ng makes it difficult to imagine how 10 maintain walerfowl popubtions at today's levels under an :lltered climate. Sorenson ;md col­leagues (1998) estimated that a doubling of ca rbon dioxide could cut the US mid-continent breeding duck population in half, from an average of 5 million to between 2.1 million and 2.7 million birds.

Climate change also poses a conservation challenge f:l rther east, along the fringe of the PPR in Iowa and Minnesota, I"here virtually all wetlilnds hllve 1)t."'Cn drained and grasslands have been plowed for agricultural production. Ilefore tillage, this region provided favo rable breeding opportunities for

138 BioScience ' February 2010 / Vol. 60 No. 2

walerfOI,,1 escaping the more frequent droughts in the western PPR. Ironically, the effects of climate change 3re projected by our research to greatly increase the frequency and scvcrity of drought in the western PPH, where the highest wetland dl'nsi­tics and most grassland nesting habitats are currently found. Mitigation in Ihe eastern PPR is problem:ltic because of the high cost of wetland ;md grassland restoration where land prices :lnd crop production arc the highest in the PPR.

The WLS simulations provide wetland scien tists with their first look at prairie \\'etland complexes in future green­house climates. Improvements in modeling and climate predictability are needed and ongoing; howel'er, the body of research conducted thus far on prllirie wetlllnds and climate I'ariability and climate cha nge by both us (e.g., Poiani and Johnson 1991, Poiani et a!. 1996, Johnson ct al. 2005, Millett ct a!. 2009, Voldset h et a!. 2009) and by others (L1rson 1995, Covich et a!. 1997, Clair 1998, Sorenson et al. 1998, Winter 2000 ) has reached sim ilar conclusions on thc seriousness of the problem, despite different analyticlli lIpproaches. The main points are: (a) prairie wetlands in general are highly sensitive to climate warming; (b) I\'etlands in the drier, west ­ern PPR are most vulnerable to climatl' warming; (c) melll ' bers of the wNland com pll'x will rcspond differently to cli­mate change, and longer-hydroperiod wetlands are perhaps the most sensi tive; (d) shortened \vetland hydropC'Tiods \\'ill severely affect vertebrates because of their longer life-cycle r('quiremenls; {el in 3 grccnhollse climate, more of the PPH will be too dry or without functional wetlands and nesting habitat to support historic I('\'cis of \":lIerfowl breeding; and (0 adaptation of farming pr;letices in wetland watersheds may buffer the effects of climatc change on wetlands.

These findings appea r solid enough to sen'e as the foun­dation from which to develop management plans to prepare for and adapt to climate change in the PPR as recommended by wildlife conservation groups (Anderson and Sorenson 200 1, Galley 2004). Adaptive managl'1llcnt would greatly benefit from a larger network of long-term wetland moni ­toring si tes in the PPR that could ])etter detect ea rly signs of warm ing on waler levels and hydropcriod, and to serve as a lest for model projections (Conley and van der Kamp 2001 ). Only three long-term wetland monitoring field si tes are cur­rently operational to assist in detecting future trends (va n der Kamp and Hayashi 2009). The climate and wetlands of the PPR should be watched more closely in the ruture to look for signs of higher evaporative demand and reduced hydroperiod to prepare for a PPR wilh less-prod uctive wet ­lands and felver waterfowl (M illett ct al. 2009).

Our findings on clim:lle change and prairie wetlands can be applied to other wetland ecosystems beyond North America only in very general terms, because of the stunning heterogeneity of wetland ecosystems globally (Mitsch and Gosselink 2007, Keddy et al. 20(9). However, ou r analytical approach, in which hydrologically driven ecological simu­lation models are developed, para meterized, and tested at long-term research and monitoring sitl's and thell scaled up to landscapes and regions, can be applied to diverse wetl:U1d

11'11'11'. II iosrier Ice" UI g. {Irs

ecosystems. Once h)'drolog)' is modeled successfully, the next challenge is to link hyd rology directly to species populations, guilds, and commu nities of organisms-the ul timate goal of ecologiC:11 rese'Hch. We cont inu(' on this path illld antici­pate integrating our find ings with those of other gro ups 10

complete a broilder geographical synthesis of the effects of climate change on wetland ecosystems worldwide.

Acknowledgments This work was funded by the US Gcological Survey, Biological Rcsourc('S Division Climate Changt; Program; and by the US Environmental Protect ion Agency (EPA), Science 10 Achien' JU.'Suits (STAR) program. manag(od by the EPA's Office of Re­scarch and Development, National Center for Environmental Research. STA R research supports the agency's mission 10

5:.tfeguard human health and the ellvironmenl.lom Tornow, of the US Fish and Wildlife Service in Madison, South Dakota. has made Ihe Orchid Meadows field si te (Severson Waterfowl Pro­duction Area) available to our project on along-term basis. Su­san I~ttcher assisted in field d'lta collection at Orchid Mead­ows for many years. We acknowledge Ihe foundat ional research in prair ie wetland ecology by K.1Ten Poiani, George Swanson, Arnold van der Valk, Milton Weller. ilnd Thomas Winter.

References cited ,\nderwn MG. Smcn:;(lIl t.G. 2001. Glob~1 cl im.Lle ~h,mge J nd "'a1erfo",l:

I\daptution in the face of un(er lainl y. Trans .... t ions of th,' North Ameri ·

can Wildlifl' and Nal urnl R~'WUrcl'S ConferelKe 66: 307-319. AUlOtin IE. Milkr M. 1995. Northern I'liltail (Am,s tlWIII ). 1 n I'uol,· A. cd. The

lIirds of North Am,·ric .. Online. Corncll l.abor~ tory of Orni thology. (3 Ik<:ember 2009; 1m p:llbn ... bj .,Is. Comrlt. rli ulb"ll )

II;J.nin J. 2004. When good ~nimals lo\'e bad habi tats: I'.eologieal traps and

Ihe conser-.llion of Jnimal po])UIJtions. Con)(,I'\-Jtion lIoology 18: 1482- 1491.

lIcllrO)(' Fe. 19110. Ducks. Gf:'t'SC. ~nd s..·~ns of Nonh ,'merka. 3rd cd. Stackpole.

llrowil j-, 1. Dinsmore IJ. 1986. Implic.lIions of lll.ITsh ,ize ;md isolation

for marsh bi rd manJgemelll. IOUTUJI of Wi ldlife Managemcn t 50: 392-397.

!lurk,'u VR. cI al. 2005. Nonlincar d)"T.ami .;s in \-.:osystem respon)(' to c1imal ic change: Cue studi,--s and policy iml'lical ion~. Ecological Com· pkxuy 2: 357- 394.

Carroll R. l>ohll G, Tracy J. Willler TC. Snuth R. 2005. Simubl iOIl of a $<'l11i· permanent weiland buin in lhe Couon .. ·ood 1.:11<..' AreJ. ea,l·cenl rJI

North DakOlJ. Journal of 11)'drologic Engin' ..... ring 10: iQ-34.

Clair T,'. 1993. Cllladian frcsh"OIlcr ....... tIJnd$ and djm~te change. Clinutic Change 40: 163-16S.

Conll'y I'M. \'a n del' KlUllp G. 200 I. ,\ loniloring t h~ hyd rology of Canadian prairie wetlands 10 dcte<:l lhe effects of c1imale chang<' alld kind 11)('

cha IIg'·s. EIl" iron mental j-, loniloring and Assc~smcnt 67; 19>-215.

Co\'ich AI'. I' ritl Sc, I.amb PI. " Iarmtf nD, /1.-1.111 hew~ WI. I'oiani IV', Prel'as FE. Richman Mil. Winter TC, 1997. l'Olential effects of climale chnge

on aqUJI ie e<:05ystn1lS of the Great I'lains uf North America. H)·drologi· eall'rocCSlooeS II: 993-1021.

Dahl TE. 1990. Wcllands lossa in the United St3!CS 1 180'~ 10 1980'5. US I:hh

and Wildlife: Service N~lional Wetlands hl"e:mory Project. I:"irhairn SE.l>insmort H. 2001. loc:;11 ami bud SC',1pc·lnd influenet'S on

wetl~nd bird communities of the prJirie pothole region of low:!, USA. Wetbnds 21: 41-47.

Galley KI~M. ed. 200~ . Global Climat,· Chanlle and Wildlife in North Am~Tic<,I.. The Wildlife Suciet y. ·1~-.:hnic., 1 Kevicw 04·2.

IV"~". /J ioS( ic, ICC" 111 g. 0 rg

Articles

Gleason KA. I..aubhan MK. Euli~ Nil. 2008. Ecosy~tem Servict'S IXri\'Cd from Wetland Con)('r\';ltion I't-dclicn in thl' Unilcd Siales I'r.lirie Pot·

hole Kl,;ion with 3n Emphasis on lhc US IXl'artmcnl of Agriculture Con5t·rv.ation Kcscrl't" and Wetlands Rl-sel'\'e: I'rOllrdms. US Gcological Sur\'ey.l'roft'SSionJII~Jpcr 1745.

H()eklllJn ST. Mill~ IS, Howerter DW. Devrics )11. Ii:1Il II. 2002. Sensitivity

anJI~i5 of I he life q'de of mid'(Olliinelli lll.dIJTds. Journal of Wildlife Managelllnll 66: 883-900.

(II'CC Ilnll·rgo\,.:rn men tal Pan t! Oil Climale Chang~. 2007. Climate Change 2001; The Physica l Science lIasi$. Cambridge University Prcss.

10hl1:;(>11 nit Hi!;!;in' KF. 1997. Weiland Kesourcc:s of Ent" TIl Soulh I}akota. South Dakola Stale Ulli\'Crsi ty.

10hnsol1 WC, lIotucher SE. Poiani KA, GUnlenspcrgen GR. 1{J(W. Influl'nce of wC~lher e~tremt'S on the W;lter lewis of gbdatcd prair~ wetlands. Wcll~lWS 24: 385-398.

lohnson WC. Milieu SV. Gilmanov T. Voldseth KA, Guntenspcrgcn Gil., Naugle DE. 2005. Vulnernbitit), of northern prairie welbnds to climate change. lIioS<:ience 55: %3-872.

Keddy PI\. Fraser 1.11. Suloml-shch AI. Junk WI. Campbell DK,Armyo MTK . Alho CIK. 2009. \\'\·t ~nd ",o nderful: Thl' world'~ largest ",el lands arc

con$\'r\'~lion priorities. Sioxience 59: J9-51. Kra]>u GI~ I'iclz I'J.Brandt DA, CoK RK Jr. 2006. Mallard brood mOI'cnlCnts,

wet!Jlld usc. J 'W duckling sur\'ilOiI during and following a pr.liric:

drought. Journal of Wildlife Managcnlell110: 1436-1444. Kunduwia. ZW. Mata lJ.Arnell N. 0011 p, Kaha l I'. jimmC'l B, Milkr K.Oki

T, Sen Z. ShikJomano,<1. 2001. Frcsh-"OIter rcsourct'S and their manag<" m~nt. Clinlale Change 2007; Imp;1cts, Ad3plalion ,lIW Vulncrdbili ly.

umbridge Uniwrsi tr Pr~ss. Larson D. 1995. Effects of climate on numbers of nonhern prai r ie wl1lands.

(limati( ChJIIge 30: 169-1110.

,\IJgnlison II. el al. 2000. Hi.loricallrends in bko: an(1 riwr ice cover in the Nort hern l lemisphere. S<:ience 289: 1143- 1746.

!M!'.A I Milknniulll Ecosystem ASS('~Ill\·"t. 2005. F.cosystems and Human Wdl. bdng: Wetlands ~lId Walcr Synthesis. World I\csour(c:s InSli tUle.

Millen IIV. john""n WC, GU111clbl'crgen GR. 2009. Cl imate trends of the North American pr.li rie pothole region 1906-2000. Climatic Change 93: 243-267.

Miuch WI. Gossdink IG. 2001. Wetlands. 4th cd. Wiley. Murkin II. \';In der "alk AG. Cbrk WR. 2000. I'rnirie WeIland Ecology:

The Contribution of the Marsh Ecology Research Progr.lm. Iowa Stale Uniwrsi ly Pr.:-ss.

Murphy· KIJSM'n 11,\1. Underwood TI, Scaly SG. C~yrnyj AA. 2005. Long· terlll t rends in spring arrival datl-S of millrallt birds al Delta Marsh.

Mani toba. in rclation to cl imate challge. The ,'uk 122: 1130- 11 48. NauS'" DE. lI i!;!;ins KF. Nusser SM. Joh nson Wc, 1999. Sca"'·dl·I)t·ndl·n1

habi tal 11)(' in lhree species of prJi r ie "'etland bi rds. l.andscapc EcolOllY 14; 261- 276.

N~ugle DE. Juhnson RR. Esley ME. J'liggins Kf. 200 I. A lands("Jpc 3Jl'PfOO(h to con)('rving wetland bird habi tat in the Prai rie lOot hole Region of cali! · ern Sou[h O~kot a. Wetl~nds 21: 1-11.

Ojima 1:>S.I.ackell 1M. 2002. Pr,-p;aring ror a Changing Climate: The Poten· tial Conscquenct'S of Climate Va riability and Change--Ccnlral Greal P!Jins. Cent ral Great I'la ins Siccrillg COmmi .. l ..... and Assn-sment Teanl. Colorado Sial,· Uni,·ersity.

l'uiani KA, 10hnson WC. 199!.Global wMmh'1I 311d pra irie ,,'t"llallds: Con.

S<'<luelKes for ""Jtcrfo",1 habi tat. lIioxience 41 ; 61 I ~ 18. ---. 1993. A SIMliJ I simulalioll model of hyd rology and vcse tation

dynamics in 5t'mi'l""rmanenl prairie wetlalld~. F.coJogkJI AppliC"a t ions 3; 279-293.

l'oiani KA. lohnson We. Winter TC, Swanson G. 1996. Climate change and norlhl'TIl pr~irie ",etlands: Simulationi of long. term d)'Tlamics. l.imnol·

ogy and Ocl'JllOgrnphy 41 : 871-&11. Pricc IS. 1994. Wa ler I"wl regimn in prJirie sloughs. Canadian Water Re·

sourcn Journal 19: 9>-106. Kl·ynolds R I~ ShafferTI.., Loesch CR. Cox RR Jr. 2006. Thl' farm bill and

duck production in th" prJiri" pothole region: Increasing the bencfi lS. Wild life Society B"lIetin J4: %3- 914.

february 20/0 / Vol. 6Q No. Z • BioScience 139

Articles

So;hlaepfcr MA, C( 011. 2002. F.cologial and c\'OJulionuy lrap§. Trends 111 IOcology and E"olulion 17: 474--480.

Sorenson LG. Goldberg R. Root TL. Anderson MG. ]998.I'oI<:nl;01I effects

of global warming on ",all'rfo",1 populations bTl't.-dillS in the northnn Greal "[Olins. Clim,uic Ch:mgc 40: 343-369.

5Il"'"Jri RE, Kanlrud HA. 1978. Classification of NJIUr~1 I'ond~ and Lakes in ,h ... Glaciated Prairie RC&ion. BurNu of Sport Fi.!hNles and Wildlife, US Fish and Wildlife SC'r\"iCo.'. Resource Public<llion 92.

5"-anson GA. ~k)'cr /111, Adonlailis VA. 1985. Foods consumed by br\-cdiJlIj mallards on wt'liands of wUlh·ccntral North lJakola. Journal of \\'ild­lif .. ~t3nas{"mcnl 49; 197-203.

Swanson GA, Euliss Nil,! lallson n. /llushcl OM. 2003. [)'narnics of a prai­

rie pothole ""'Iland complex: Implications for wetbnd management. PJgn :i5-94 in Winter Te. ed. H)·drological. Chcmiul, ~nd Kiologic~1 CharJct .. ristiC$ of a I'rairie PothoJ.. Wetland Complex undcr Highly Variable aimate Conditions-ttw, Collonwood Lale Area, East·Cenlral Nonh Dakota. US Geological Surwy. Professional I'~ptr 167:'.

\'an der Kamp G, Haya~hi M. 1998. Thc groundwater recharge funclion of small wNlands in 1 he .sfmi-arid nonhl'rn prairiC$. Greal Plains Rcs~.'arch 8: 39-56.

---. 2009. GrOlllld"'Jter-w~tIJnd l·wsY.lcm intCfaction in Ihe scmiarid glaciated plains of North America. Hydrogeology lournal 11: 203-214.

v~n Ikr Kamp G. HarJshi :\1 , Gallen D. 2003. Comparing the h)'drology of gra~ and cuitinlcd calChml'nts in ttw, §emi·arid Can;adian prairies..

IIrdrological Processes 17: SS9-:'75. van der Valk AG. 2005. Wdler·lc\"tilluct uations in Nort h Ameri"," prairie

wetlands. Hydrobiologia 539: 171-188. van der Valk AG. D;J."iJi CB. 1978. The TOll' of !<.'1:d bankt in Ihe .... getalion

dynamics of prairie glaciJI marshes. Ecology S9: 322-33S. Voldscth RA. JohnloOn we. Gilmallo\' T. GUlltenspergen GR. Milieu IIV.

2007. Model estirnatl'S of land-use effects on water I .. "cls of northern

prairie "l:tlands. Ecologic;11 AppliCltions 17: 527-5-10. Vold5C1h RA.lohnloOn WC,Guntenspergt"n GR.Gilmano\· T, Millcn B. 2009.

Adaptation of farming practices could buffer cffect~ of climale change 011 northern prairi .. wetlands. WetlJnds 29: 63:.-647.

W.lgner G. 1997. Studies of the north .. rn li.'Opard frog (Rmrn pipiell$) ill Alberta. Alberta En\'ironmental Prolection, Wildlif,' Management I)i\';'

sion. Wildlife SUtus Rcpor! no. 9.

I . ( . ' @ ,..

0 0 • -•

140 BioScience ' f.'ebfllllry 10/0 I \'01. 60 No.2

•

• "

"

Welkr MW. 1988. I§sun and 31'Proachcs in aSSo.'SSlng cumulatin' impJClS on walerbird habit~t in ..... d~nd$. En"ironm .. nlJl I-ol3nagemenl 12:

695-701. ---. 1994. hl'shwat<'r MJrshes. 3rd ,-d. Unil'CT~ity of Minncwta Press. ---. 1999. \\'ctland Birds. I labital Resources and Cous.-rvJtion Implica·

tions.. Cambridge Uni \"CTsit), Press.

Weller MW, Fredricluon Ul. 1974. A";an t'Cologr of a managed gl..Jdal I1Imh. ].ivinS Bird 12: 269-29 1.

Weller MW. Spalcher CEo 196:'. Role of Ibbiw in the Distribution and Abundance of ,\Ia~h Birds. Iowa Sl~tc Uniwrsity Agricultural and

Ilome Economics F"'IN.'rimenlal Station. Sp«ial Repor! no. 43. Winter TC. 2000. The "ulnerability of wetlands 10 climate ChJIlSl': A h)"dro-

10Sk bndscapc per' IN.'Ctiw. Journal of the Amerk;U1 \Vater ReloOurees A§SOCiation 36: 305-311-

Winler Te. R~nlxrrr DO. 1995, The imeuclion of groundw.l1er with prairit> polholC$ in Itw, Cononwood Lake arn, east·c~mral North J)~­

kOIJ, 1979-1990.lourn~1 ofH)-drolog)' IS: 193-221. Wimcr TC. Woo MK. 1990. I-I)'drology oflak .. s and ... etlands. Pases 159-

187 in Wolman MG, Riggs He, l'ds. Surface Water H)'d rolosy: The

G .. olog), of North America. \'01. I. Geological Society of ,\merica. Wolfram S. 1999. Th .. Mathcmatica Hook. 4th ed. Cambridge Uni\'enit ),

Press.

W umfT jalw$lm (cM/ef.joluUDI1@sds/,u".rdlO ). Rlehrmi A. \'a/dst'tJr. mid

Crllig O/"lI'sky IIfr with Ihe lJ.opnrlmell/ "f IIDrtiCllilU,e. FOft'Slry, Lmlllst'llpr, ","ll'drks. m Smul, lJdkotll Slilte Ulli,"f'(si/), ill /lrooliugs. Ilmt \ I'rm.., is Willi 1/'" I"parlmr'll of EII"il"O"'''(lIIlIl Slii/lirs (l/ St. Oillf Col/.og,·. Norlhfield. MIII­IlNOIiI. Glnm R. Gmllfllsprrgeu is willi tltc US Grologiwl SII,."('y Biologim/ Rr$DI'rcrs Dlyisi01I III PaIUJcrll/ Wildlift RtSt'llrr/r CrlllC'T ilr Urrrl?l, M"ry/Ilml. B,.,er Millett IS WIth lilt Dtp.lflmffll of Grogmplly tl/ Sorrth Ckrko/Il Sltl/l

U,mYrsi ,y in Brookings. [Joil\'id E. Nnllg/( is ",ill, tllr Wildlife Biology Progmm as tl" U'''' 'I''rsi,}, of Mmw",,,. MimI/I". MiT,"," T"I/",rr is ,..,11, 'hr GIS Olltrr of Excel/flier <II 50Mir /)"kolll SIIIIr U"i"t'rsily itl Rrrxl~i"gs. Ro>cltmry \ V. N. Cm,ol/ is ,..illr fkst'fl Rt·St'llreh l"stilUlr. Utli""rsil}, of Nr.mla. ill RrrrtJ. /01111 Tr<l()' is wilh tlrr M.llrll \\'1lur RrwurcN RrSt'lirc/r Critter. III lire UII"' ... rsily of 11/11/'0 III BoiSt'.

I • •

• •

• .,.. <D' I-flRP/N

I>'IV"'. b iose iellct'111 II g. org