Research article

Assessing alternative futures for agriculture in Iowa, U.S.A.

M.V. Santelmann1,*, D. White2,3, K. Freemark4, J.I. Nassauer5, J.M. Eilers6,7, K.B. Vaché8,B.J. Danielson9, R.C. Corry10,11, M.E. Clark12,13, S. Polasky14, R.M. Cruse15, J. Sifneos1,H. Rustigian1, C. Coiner16, J. Wu16 and D. Debinski91Department of Geosciences, Oregon State University, Corvallis, OR 97331; 2Department of Geosciences,Oregon State University; 3Current address: U.S. Environmental Protection Agency, Corvallis, OR 97333, USA;4National Wildlife Research Centre, Canadian Wildlife Service, Ottawa, Ontario, K1A 0H3, USA; 5School ofNatural Resources and the Environment, University of Michigan, Ann Arbor, MI 48109, USA; 6E&SEnvironmental Chemistry, Corvallis OR; 7Current address: MaxDepth Aquatics, Inc., Bend OR 97701, USA;8Department of Bioresource Engineering, Oregon State University, Corvallis, OR 97331, USA; 9Department ofEcology, Evolution, and Organismal Biology, Iowa State University, Ames, IA 50011, USA; 10School ofNatural Resources and the Environment, University of Michigan, Ann Arbor, MI 48109; 11Current address:Department of Environmental Design & Rural Development, University of Guelph, Guelph, Ontario, CanadaN1G 2W1, USA; 12Department of Ecology, Evolution, and Organismal Biology, Iowa State University, Ames,IA 50011; 13Current address: Department of Biological Sciences, North Dakota State University, Fargo, ND58105, USA; 14Departments of Applied Economics and Ecology, Evolution and Behavior, University ofMinnesota, St. Paul-Minneapolis, MN 55108, USA; 15Department of Agronomy, Iowa State University, Ames,IA 50011, USA; 16Department of Agricultural and Resource Economics, Oregon State University, Corvallis,OR 97331, USA; *Author for correspondence (email: santelmm@onid.orst.edu)

Received 17 June 2002; accepted in revised form 15 October 2003

Key words: Agriculture, Biodiversity, Socio-economics, Scenarios, Water quality

Abstract

The contributions of current agricultural practices to environmental degradation and the social problems facingagricultural regions are well known. However, landscape-scale alternatives to current trends have not been fullyexplored nor their potential impacts quantified. To address this research need, our interdisciplinary team designedthree alternative future scenarios for two watersheds in Iowa, USA, and used spatially-explicit models to evalu-ate the potential consequences of changes in farmland management. This paper summarizes and integrates theresults of this interdisciplinary research project into an assessment of the designed alternatives intended to im-prove our understanding of landscape ecology in agricultural ecosystems and to inform agricultural policy. Sce-nario futures were digitized into a Geographic Information System �GIS�, visualized with maps and simulatedimages, and evaluated for multiple endpoints to assess impacts of land use change on water quality, social andeconomic goals, and native flora and fauna. The Biodiversity scenario, targeting restoration of indigenous biodi-versity, ranked higher than the current landscape for all endpoints �biodiversity, water quality, farmer preference,and profitability�. The Biodiversity scenario ranked higher than the Production scenario �which focused on prof-itable agricultural production� in all endpoints but profitability, for which the two scenarios scored similarly, andalso ranked higher than the Water Quality scenario in all endpoints except water quality. The Water Quality sce-nario, which targeted improvement in water quality, ranked highest of all landscapes in potential water qualityand higher than the current landscape and the Production scenario in all but profitability. Our results indicate thatinnovative agricultural practices targeting environmental improvements may be acceptable to farmers and couldsubstantially reduce the environmental impacts of agriculture in this region.

Landscape Ecology 19: 357–374, 2004.© 2004 Kluwer Academic Publishers. Printed in the Netherlands.

357

Introduction

Humans now dominate many of earth’s ecosystems,often with devastating impacts on global biodiversityand biogeochemical cycles �Puckett 1994; Alexanderet al. 1996; Vitousek et al. 1997a; Vitousek et al.1997b; Sala et al. 2000�. Of particular concern is ag-riculture, whose impact on the earth is unparalleledby any other land use in its combination of spatialextent and intensity of influence �Matson et al. 1997�.Whereas some envision “A Geography of Hope” forprivate agricultural lands �USDA-NRCS 1996�, oth-ers see the potential for “a second Silent Spring”�Krebs et al. 1999�.

Agriculture has had an enormous environmentalimpact on the U.S. Corn Belt region. The U.S. Officeof Technology Assessment has designated it as thenation’s first priority problem region with respect tosurface water quality �US OTA 1995�. Three states inthis region �Indiana, Illinois, and Iowa� rank 48, 49and 50 in the amount of natural vegetation remainingout of the 50 states in the U.S. �Klopatek et al. 1979�.Over 70% of the total land base of Iowa is devoted torow crops such as corn �maize� and soybeans, thehighest percentage of any state in the U.S. �US Cen-sus Bureau 2001�. Increasing numbers of livestockare being raised in confined animal feeding opera-tions, reducing pasture land �Freemark and Smith1995; US GAO 1999�. Native flora and fauna havebeen in decline for decades �Farrar 1981�. Equallydramatic are the socioeconomic impacts of currenttrends in agriculture. If trends of the past 25 yearscontinue, in the next 25 years, many Iowa farmerswho now own their land will lose or sell it �USDA-NASS 1999; Freemark 1995�. Farm size will increaseas will the land area in large corporate farms oper-ated by employees. Rural landscapes will be depopu-lated and rural communities will see major changesin demography.

Here, we summarize results of a project to envisionfutures for agricultural landscapes that offer alterna-tives to current conditions and trends �Santelmann etal. 2001; Nassauer et al. 2002�. We maintained mul-tiple goals in the design and evaluation of these in-novative future landscapes. We also introduceddramatic changes in agricultural land use andmanagement, in order to improve water quality andrestore native biodiversity, maintain agriculturalenterprises and economic return to farmers. Weworked to design landscapes that would also be ac-ceptable to people who live and visit there. We asked

not only, “What might these landscapes be like in 25years with continued priority given to corn and soy-bean production?” �the Production scenario�, but also,“what should and could they be like?” if among ourgoals for agricultural watersheds are streams withclean water, natural areas with diverse vegetation,abundant wildlife, and attractive rural landscapespopulated by farmers working their own land �e.g.,the Water Quality and Biodiversity Scenarios�. Ourhypothesized social drivers were varying levels ofpriority given to agricultural production, water qual-ity, and maintenance and restoration of native biodi-versity in public values, support and policy. Nassaueret al. �2002� describe the alternative futures in detail.Here, we focus on the interdisciplinary evaluation andassessment of these alternative futures. Our objectiveis to compare the potential for these alternatives toachieve multiple goals by summarizing and integrat-ing the quantitative impacts of landscape change em-bodied in the scenario futures for multiple endpoints,ranging from economic and social metrics to modeledestimates of native biodiversity and water quality. Wealso compare evaluative approaches used in thisproject, employing a framework based on Levins�1966� paper that could assist other interdisciplinaryprojects in thinking about the advantages and limita-tions of various evaluative methods.

These scenarios were not intended to be prescrip-tions of what to do on specific parcels of land. Rather,they were intended to inform and inspire decision-makers to look beyond the existing landscape and en-vision greater possibilities for agriculture �Nassauerand Corry 2004; Peterson et al. 2003�. The scenarioswere designed to allow the exploration of multipleimpacts of recommended strategies for addressingenvironmental problems associated with agricultureusing spatially-specific models and computer-simu-lated images of the resulting landscapes. This inter-disciplinary assessment of the alternative futuresintegrates the results from disciplinary teams thatevaluated the social, economic, and environmentalimpacts that may result from making changes thathave been recommended to address agricultural andenvironmental problems. Methods used by the disci-plinary teams are outlined briefly here. More detaileddescription of methods used in modeling and evalua-tion of the futures may be found in other publicationsfrom the project �Coiner et al. 2001; Vaché et al.2002; Rustigian et al. 2003�.

358

Methods

Study area

The two watersheds chosen for this study, WalnutCreek �5130 ha� and Buck Creek �8820 ha�, are lo-cated in Iowa, in the heart of the U.S. Corn Belt �Fig-

ure 1�. These watersheds were selected to illustratethe way each scenario might be implemented in dif-ferent physiographic regions, with different sets ofagricultural enterprises. Walnut Creek is located inBoone and Story Counties southwest of Ames, Iowain the Des Moines Lobe physiographic region of Iowa�Prior 1991�. The Des Moines Lobe region is charac-terized by relatively low topographic relief and richprairie soils. Once a mosaic of prairie and prairie pot-hole wetlands, now more than 80% of the land areain the region has been converted to row crops, prima-rily corn and soybeans. Buck Creek, in PoweshiekCounty, is located on the Southern Iowa Drift Plain,an older, more dissected landscape with much moretopographic relief, highly erodible soils, greateramounts of land currently in pasture, forest, and theConservation Reserve Program, and only 45% of theland area in row crops �Freemark and Smith 1995�.Land use and land cover for both watersheds weredigitized into a GIS at 3 m resolution under the USEPA Midwest Agrichemical Surface and SubsurfaceTransport and Effects Research �MASTER� programfrom aerial photographs taken in 1990 �1:20,000� andextensively ground-truthed in 1993 and 1994.

Scenario futures design

Landscape futures for the three scenarios �Nassauerand Corry 1999; Nassauer et al. 2002� were designedby a team of landscape architects using an iterativeprocess �Nassauer and Corry 2004�. The process en-gaged disciplinary experts in agronomy, plant andanimal ecology, wetlands ecology, water quality, hy-drology, agricultural policy, agricultural extension,and GIS, including regional experts as well as projectcollaborators. The futures are not simply digital mapsof land cover in two Iowa watersheds; they representthe plausible outcomes on the landscape of very dif-ferent human priorities for agricultural lands. Theyare not predictions of the future, but are rather an ex-ample of what could happen if normative principleswere used to achieve societal goals of improvementsin water quality, restoration of biodiversity, and eco-nomically healthy rural communities. In their 2001article, Mitsch et al. make specific recommendationsfor practices needed to reduce nutrient loading in theUpper Mississippi River Basin, including:– wetland restoration– riparian buffers– changes in cropping and fertilization

Figure 1. Location of Iowa in the U.S.A. and of the study water-sheds in Iowa.

359

– manure management

Many of these practices �as well as those recom-mended by USDA-NRCS 1996� are elements of ourfuture scenarios, but here they are made spatiallyspecific in the context of existing watersheds, allow-ing the use of spatial models for landscape-levelquantification of multiple impacts, including compari-son of economic, social and environmental outcomes.We chose to use watersheds of this size because landmanagement decisions are often made at the spatialscale of these watersheds, and because research anddecision-making at both smaller and larger scalescould be informed by investigation of responses toland management at this scale.

In the Production scenario �Figure 2, Figure 3�,profitable agricultural production is assumed to be thedominant objective of landscape management. Thisscenario therefore assumes that policy encouragescultivation of all highly productive land with the useof fossil fuels, chemicals, and technology to a degreesimilar to the present. It assumes public support forlarge-scale, high-input agriculture, and trust in thesafety and quality of food produced. Corn and soy-bean row crops are extended to all areas of highlyproductive soils. Best management practices �BMPs�targeting protection of water quality and biodiversitytypical of the region under the 1996 Federal Agricul-tural Improvement and Reform Act �Public Law 104-127 �H.R. 2854� April 04, 1996� are assumed to becomprehensively adopted in 2025. The landscape isdepopulated by 50% compared with 1994. Manyfarmsteads and woodlots are removed, average farmsize doubles, and field size increases up to 130 ha.Livestock are raised almost exclusively in confine-ment operations in a few counties of the state, but notin the study watersheds.

The Water Quality scenario �Figure 2, Figure 3�assumes that agricultural enterprises change inresponse to a �hypothetical� new federal policyenforcing clear, measurable water quality perform-ance standards for surface and groundwater, and sup-porting agricultural enterprises that efficiently reducesoil erosion, reduce sediment delivery to streams,minimize transport of excess nutrients to streams, re-duce the rapid hydrologic response to storm events,and improve aquatic habitat. Forage crops and rota-tional grazing �carefully managed to minimize im-pacts on riparian systems� are widely adopted asprofitable enterprises supported by federal policy tohelp meet water quality performance standards on

erodible land. Conventional BMPs �minimum tillage,rotations, strip cropping, continuous cover, and ani-mal agriculture� are employed. Wider stream buffersof native vegetation and new, innovative BMPs de-tain and clean stormwater. Woodlands are retained forcarefully-managed grazing. Both urban and rural citi-zens appreciate the pastoral appearance of agricul-tural landscapes. Farm vacations and countrysidesecond homes bring urban people into rural areas. Tomanage livestock operations and respond to rural rec-reation demand, 50% more farmers live in these ag-ricultural landscapes in 2025 than under the Produc-tion scenario.

The Biodiversity scenario �Figure 2, Figure 3� as-sumes that technology and agricultural practices re-spond to a �hypothetical� new federal policy toincrease the abundance and diversity of native plantsand animals in the context of agriculture. Public in-vestment creates a comprehensive system of indige-nous species bioreserves of at least 260 ha �640acres�, connected by a network of wide habitat corri-dors that also buffer streams. Federal support encour-ages adoption of innovative, biodiversity best man-agement practices �e.g., perennial strip intercroppingand agro-forestry� in a biodiversity target zone to fur-ther connect and buffer the new bioreserves. Beyondthis target zone, corn and soybeans are grown on soilsthat are highly suitable for cultivation. Livestock en-terprises continue to trend toward confinement opera-tions, constructed to meet rigorous standards forsewage treatment, in a few counties of the state, butnot in the study watersheds. Public investment in re-serves and corridors invites public enjoyment of therural landscape. Trail systems connect the corridorsystem and reserves. While farm size increases andthe number of farms decreases as in the Productionscenario, nearly all farmsteads present in 1994 remaininhabited. Many non-farmers who enjoy rural land-scapes live on farmsteads formerly occupied byfarmers.

The resulting alternative futures are landscape mo-saics characterized by changes in field size, croppingand tillage practices, perennial cover within fields�e.g., grassed waterways, filter strips, field skips� andconversion of cropland to rotational pasture or hay�Figure 2, Figure 3�. The futures also differ in the oc-currence and spatial extent of non-crop habitats inter-spersed among fields �e.g., fencerows, riparian buffersof differing widths, woodland, prairie, and wetlandreserves�. The alternative futures and the decisionmaking rules used to generate them are described in

360

Figure 2. Present landscape �top right� and designed alternative future scenarios for Walnut Creek watershed. Note the increase in land areain row crops at the expense of perennial cover for the Production scenario; the increased amount of land in perennial cover �pasture andforage crops� as well as wider riparian buffers in the Water Quality scenario; and the strip intercropping, wide riparian buffers and extensiveprairie, forest and wetland restorations in the Biodiversity scenario.

361

Figure 3. Present landscape �top left� and designed alternative future scenarios for Buck Creek watershed. Note scenario features similar tothose for Walnut Creek but applied to a different landscape �e.g., the physiography of the watersheds led to the design of forest, savanna, andupland prairie reserves in Buck Creek rather than the riparian forest and prairie/ prairie pothole wetlands which comprised the reserves inWalnut Creek watershed�.

362

more detail in Nassauer and Corry �1999�. Vaché etal. �2002� also provide a table that lists the area �inhectares� of all major land cover types in each sce-nario for both watersheds, and shows the percentchange in area relative to current landscape.

Scenario futures evaluation

The alternative futures were compared to one anotherand to the current condition of these watersheds us-ing GIS-based models and digital simulations of thealternative landscapes. Changes in land use and man-agement in the alternative futures were evaluated fortheir impacts on water quality �discharge, export oftotal suspended sediment �TSS� and of nitrate�, esti-mated economic return, farmer preference, andimpacts on native plants and animals �habitat-areabased estimates of plant, butterfly, and vertebratespecies diversity; spatially-explicit population model-ing for mammal and amphibian species�. For our pur-poses, native plant species were defined as thosedesignated as native to Iowa in the USDA PLANTSdatabase �USDA-NRCS 1997�, and native animalspecies were defined as those designated by The Na-ture Conservancy’s Natural Heritage Program as na-tive to Iowa. More detailed methods are given foreach evaluative approach in the project publicationscited below.

The water quality model SWAT �Soil and WaterAssessment Tool �Arnold et al. 1997�� was used toevaluate scenarios for water quality response �Vachéet al. 2002�. Water quality data used to calibrateSWAT have been collected on Walnut Creek for 11years as part of the USDA Management SystemsEvaluation and Assessment program �Hatfield et al.1999�. Water quality and hydrologic monitoring sta-tions were established on Buck Creek and data werecollected during 1997 and 1998 for model calibrationin that watershed �Shoup 1999�.

For evaluation of the economic impacts of the al-ternative futures, the EPIC model �Erosion Productiv-ity Index Calculator �Williams et al. 1988�� was usedto calculate crop yields. Crop enterprise budgets fromThe Iowa State University Extension Service wereused to calculate production costs. Price data from1987 to 1997 from the National Agricultural Statis-tics Service, indexed to 1998, were used to calculatean average price for conventional crops �e.g., maize,soybeans, oats, alfalfa�. Information on crop yieldsand prices were combined to estimate revenue, fromwhich we subtracted cost to generate estimates of

farmer profit or “return-to-land” �Coiner et al. 2001�.The EPIC model also generates estimates of external-ities such as soil erosion and nitrate losses to runoffand leaching from the rooting zone.

We developed a spatially explicit, content-specificmethod for determining a landscape preference ratingfor alternative future scenarios �Nassauer and Corry1999�. Thirty-two farmers sorted 20 images of land-scape features drawn from the three scenarios intopreference classes, identifying landscapes that were“best for the people of Iowa in 25 years.” A meanpreference score was calculated for each image, withhigher scores indicating higher preference. Eachlandcover class was assigned a preference ratingbased on the mean ratings of the images of that land-cover. A landscape preference rating was calculatedfor each scenario using the product of the mean pref-erence scores of land cover types depicted in repre-sentative images for the scenario and the area in thatland cover. A structured interview format was used toelicit in-depth information about the perceptions ofeach farmer rather than to measure perceptions of arepresentative sample of a larger population �Nas-sauer and Corry unpublished data�. Structured inter-views included a Q-sort �1-5 scale normally distrib-uted� forced-answer component in which farmersrated landscapes on their desirability for the future ofthe people of Iowa. To include farmers who workedin both rolling and flat landscapes, with different cropand enterprise mixes, we sought to interview similarnumbers of farmers in Poweshiek County �n�15�,who were more likely to grow hay, and in StoryCounty �n�17�, who were more likely to grow cornand soybeans. In each county, we interviewed farm-ers of fewer than 80 acres, farmers of more than 640acres, and several farmers of between 80 and 640acres. Mean farm size is 360 acres for each county;median farm size is 210 acres for Poweshiek Countyand 180 acres for Story County. We also targetedsome farmers who had demonstrated their ability toinnovate with landcover; about half of interview par-ticipants had land currently or previously enrolled inthe Conservation Reserve Program.

For evaluating risk to biodiversity, we used twotypes of methods. First, we used a statistical estimateof change in habitat area, weighted by habitat quality�White et al. 1997�, for all butterfly and non-fish ver-tebrate species that occur in central Iowa, or by esti-mated abundance in that habitat �Eilers and Roosa1994� for all plant species that occur in central Iowa,in order to have an index of the response to land use

363

change for all species in a given taxon across a broadrange of taxa �White et al. 1999�. This method is bothfeasible for large numbers of species, and straightfor-ward in its assumptions that habitat area and habitatquality are important influences on species diversity.These assumptions are consistent with the findings ofrecent articles and reviews that indicate the impor-tance of both habitat area and habitat quality in influ-encing species diversity �Rosenzweig 1997; Dupréand Ehrlen 2002� and population density �Bender etal. 1998�. Second, we used spatially explicit popula-tion models �SEPMs� to assess the impact of changesin land use and management on species of interest�Dunning et al. 1995�, and to compare the results ofthese spatially-explicit, process-based modeling ap-proaches to the habitat-area based method. TheSEPMs were used to estimate and compare relativedensities for all mammal species occurring in centralIowa in the alternative futures after 100-year modelruns �Clark and Danielson unpublished data� andabundance of breeding females and species persis-tence through 100-year model runs for four selectedamphibian species �Rustigian 1999; Rustigian et al.2003�.

Integration of results

Each disciplinary research team on the project hasprepared and published articles that present in greaterdetail the methods used in their evaluation of the fu-tures �Coiner et al. 2001; Nassauer et al. 2002; Vachéet al. 2002; Rustigian et al. 2003� and discuss theirresults from a disciplinary perspective. Here, we fo-cus on describing the methods used to summarize andcompare the results of all the evaluative models foreach scenario across disciplines.

One challenge in summarizing the results of an in-terdisciplinary project is deciding how to compare“apples and oranges”; e.g., changes in profitability orfarmer preference with changes in water quality orbiodiversity. In some cases, the choice might be toconvert all impacts into a common currency, e.g.,dollars, and produce a single number or rank for agiven scenario. We chose not to do so here, for mul-tiple reasons, among them the desire to retain as muchinformation as possible for stakeholders to use inmaking their own decisions.

In order to present results of multiple endpoints ona common scale for comparison, we expressed eachendpoint in terms of percent change relative to cur-rent conditions �difference between future and present

value of the endpoint divided by the present value�such that positive change reflected improvement inconditions relative to goals of the scenarios. We con-sidered the following changes as improvements:greater profit, greater area in land use selected byfarmers as “better for the state of Iowa”, increase inland cover with relatively high native plant density orhigh suitability for native animal species, increase inestimated potential abundance of native species, de-crease in export of nitrate and total suspended sedi-ment, and a decrease in annual discharge. When morethan one entity was being used to constitute the indi-cator �such as a set of plant or animal species forbiodiversity�, we used the median of the individualpercent changes.

Each scenario in Walnut Creek �Figure 4a� andBuck Creek �Figure 4b� was evaluated for ten indica-tor endpoints �see below�, for which percent changewas calculated as above.

Water quality response, endpoints 1-3:1� annual discharge in m3/year2� annual export of sediment in tonnes/year3� annual export of nitrate-nitrogen in kg/year

All three endpoints above were estimated using theSWAT model �Vaché et al. 2002�.

Economic profitability, endpoint 4:4� annual return to land �RTL� summed for watershed�U.S. dollars�.

This endpoint was estimated using EPIC asdescribed in methods �above� and Coiner et al.�2001�. This endpoint does not include any incomefrom federal programs targeting environmental im-provements. Such payments would be in addition toprofits shown here, which are only from commoditiesproduced.

Farmer perception, endpoint 5:5� cumulative preference rating for each future, basedon farmer rating of images related to each futurelandcover

This endpoint was calculated from the farmer in-terviews as described above and in Nassauer andCorry �1999�.

Biodiversity response, endpoints 6-10:6� index of native plant biodiversity � median percentchange for all native plant species �n � 932� of indexof abundance in each land cover, weighted by area ofthat land cover�

364

7� index of butterfly biodiversity �median percentchange for all butterfly species �n � 117� in habitatarea in each land cover weighted by suitability of thatland cover for the species�8� index of native vertebrate biodiversity �medianpercent change for all bird, mammal, reptile, and am-phibian species �n � 239� in habitat area in each landcover weighted by suitability of that land cover forthe species�9� index of response for mammal species �medianpercent change for all mammal species �n � 50� ofrelative density in simulation year 100 of species per-

sisting in the watershed�.10� index of response for four amphibian species�median percent change in mean number of breedingfemales in simulation year 100�.

The methods used to calculate changes in suitabil-ity-weighted habitat area for plant and vertebrate spe-cies are those described in White et al. �1997�. Themammal diversity endpoint is from Clark et al. �un-published data� and the amphibian endpoint is basedon data presented in Rustigian et al. �2003�.

Figure 4. Results of the ten endpoints modeled for Walnut Creek �a� and Buck Creek �b�, presented as percent change relative to currentlandscape. The Biodiversity scenario ranks consistently above the Production scenario in all endpoints, and the Water Quality scenario ranksabove the Production scenario in all but economic profitability �endpoint 4�. Endpoints of plant and amphibian biodiversity �endpoints 6 and10� exhibit the greatest change and the greatest differences among scenarios, with the prairie/wetland reserve in the Walnut Creek Biodiver-sity scenario yielding the greatest improvement relative to the present �the change in amphibian biodiversity for Walnut Creek was 916%, offthe regular scale of the figure�.

365

Results

The results of modeling with SWAT �Vaché et al.2002� indicated that the changes in land use andmanagement practices envisioned in both the WaterQuality and Biodiversity scenarios could lead to sub-stantial improvement in water quality. For example,reductions in annual stream discharge for the WaterQuality and Biodiversity scenarios were estimated tobe 30% and 50%, respectively, for Buck Creek, andover 50% for both scenarios in Walnut Creek �end-point 1, Figure 4�. We interpret the decrease in streamdischarge as a positive feature of the scenarios,caused largely by a decrease in storm runoff, thus al-lowing more water for infiltration and groundwaterrecharge. However, even the Production scenarioshowed some improvement relative to the currentlandscape �about 20% for Walnut Creek and 10% forBuck Creek�, primarily resulting from the compre-hensive adoption of conservation tillage in this sce-nario.

Export of TSS �endpoint 2, Figure 4� is similar inresponse to discharge, with the Water Quality andBiodiversity scenarios showing reductions of approx-imately 50% in Walnut Creek, and the Productionscenario similar to the current landscape. In BuckCreek, reductions in TSS for the Water Quality andBiodiversity scenarios were approximately 40%, withthe Production scenario similar to the currentlandscape.

Nitrate export �endpoint 3, Figure 4� was estimatedto show reductions of over 50% in the Water Qualityand Biodiversity scenarios in both watersheds. For theProduction scenario, the model estimated a slight in-crease in nitrate export relative to the currentlandscape for both watersheds. For comparison, in thecurrent landscapes of Walnut Creek and Buck Creekwatersheds, annual discharge estimated by SWAT was9.5 � 106 and 14 � 106 m3 yr–1, respectively; annualexport of nitrate was estimated at 63.4 � 103 and 54.4� 103 kg yr –1, respectively, and annual export ofsediment was estimated to be 532 and 1673 tonnesyr–1, respectively.

The Biodiversity scenario scored highly in terms ofeconomic profitability and farmer perception end-points. Return-to-land �RTL�, a measure of total wa-tershed profit generated from agricultural operations,was almost unchanged in Walnut Creek and 30-40%higher in Buck Creek �endpoint 4, Figure 4� relativeto current estimates of total watershed RTL of $1.55and $1.35 million dollars �U.S.� for Walnut and Buck

Creek, respectively. In Buck Creek, the relativelyhigh RTL in the Biodiversity scenario occurred in partas a result of large areas converted from pasture orset-asides to strip intercropping, estimated to be moreprofitable. Although in Walnut Creek, the Biodiver-sity scenario had reduced area in agricultural produc-tion, the enterprises were sufficiently profitable toyield RTL similar to those estimated for the currentlandscape �Coiner et al. 2001�.

Results from calculations of preference ratings foreach scenario were consistent with farmer interviewcomments about images. In each watershed, theBiodiversity scenario was most preferred, followedby the Water Quality scenario and the current land-scape. The Production scenario was least preferred forboth watersheds �endpoint 5, Figure 4�. Several im-ages of the Water Quality scenario also ranked quitehigh in farmer perceptions, and many farmers inter-viewed showed a preference for the Water Qualityscenario, in which more farmers remain in the water-shed farming their own land. However, the extensive,pastured livestock operations and the corn/bean/oat-alfalfa rotations of the Water Quality scenario werethe least profitable agricultural enterprises modeledgiven recent trends in pricing of agricultural com-modities. Thus, of the two scenarios that showedsubstantial water quality improvements, the Biodiver-sity scenario performs better than the Water Qualityscenario for economic profitability and culturalacceptability �Nassauer and Corry 1999�.

As expected, the Biodiversity scenario rankedhighest in most endpoints for estimating impact tonative biodiversity �Figure 4, endpoints 6-9�. Onlybutterfly diversity in Buck Creek watershed rankedbelow current values for the Biodiversity scenario,reflecting both conversion of pastureland to strip in-tercropping in that scenario �which does not result inimproved habitat for butterflies� and because the re-serves designed for Buck Creek watershed did not in-clude wetland restorations. The soils maps for BuckCreek watershed that were used to assist in scenariodesign did not show extensive areas of former wet-lands in the parts of the watershed where reserveswere located, thus, the Buck Creek bioreserves didnot include restoration of habitat for species associ-ated with wetlands. In Walnut Creek, where pastureand grassland make up less than 4% of the currentlandscape, and where reserves contained extensiveprairie, wetland, and forest restorations, native biodi-versity endpoints �butterfly, plant, and vertebrate di-versity and amphibian abundance�, are substantially

366

higher in the Biodiversity scenario than in the currentlandscape or the Production scenario �Figure 4�. Forterrestrial vertebrate species �endpoint 8, Figure 4�biodiversity is almost double in Walnut Creek in theBiodiversity scenario versus a 17% decrease in theProduction scenario. This overall improvement re-sulted not only from establishment of biodiversity re-serves, a key element for amphibians �Rustigian et al.2003�, but also from changes in cropping and tillagepractices. Conversion to alternative crops and the useof innovative agricultural practices modified large ar-eas �e.g., the 83% of Walnut Creek and 45% of BuckCreek currently in row crops� that provide little habi-tat for native plants and animals in the current land-scape.

Discussion

Ranking of future scenarios; implications of modelresults

Although the high ranking of the Biodiversityscenario for biodiversity endpoints was expected, anunexpected finding is that land use and managementpractices of the Biodiversity scenario are estimated tobe nearly as profitable as current practices, rank high-est in farmer acceptability, and provide water qualityimprovements similar in magnitude to those esti-mated for the Water Quality scenario.

Our water quality modeling results add to thegrowing body of literature indicating that substantialimprovement in water quality downstream from agri-cultural regions will require substantial change in ag-ricultural practices �Schilling and Thompson 2000;Becher et al. 2000�. Cropland comprises over 80% ofthe land cover in some watersheds in the Corn Belt�Freemark and Smith 1995�. Currently-employedBMPs that affect only a small percentage of the land-scape are not sufficient to process or remove the nu-trients and sediment exported from cropland underconventional agricultural practices. Model results forthese watersheds indicate that under the Productionscenario, nitrate export increases slightly and sus-pended sediments decrease slightly. However, underthe Water Quality and Biodiversity scenarios, nearly3-fold decreases in nitrate export and 2-fold decreasesin sediment production are estimated. The WaterQuality and Biodiversity scenarios are quite similarin predicted water quality �endpoints 1-3, Figure 4�,and substantially better than either the Production

scenario or the current landscape. Although the Pro-duction scenario specifies universal applicationthroughout the watershed of conventional BMPs�universal use of conservation tillage, 3-6 m bufferstrips on permanent stream channels, filter strips, andprecision farming targeting fertilizer applications tominimize nutrient additions� water quality benefitsestimated to occur under the Production scenario areminimal. Continued export of high levels of nutrientsand sediment would be expected to continue currenttrends of degradation in the aquatic systems of agri-cultural regions and downstream. These findings areconsistent with the expert opinions of farmers in theregion, expressed in our farmer preference studies, inwhich farmers identified the Biodiversity scenario,the scenario with biodiversity as the leading policyobjective and with novel, challenging land manage-ment practices, to be best for the people of Iowa, andan intensification of agricultural production to be leastgood for the people of Iowa. The Production scenariowas least preferred by farmers interviewed and didnot perform well with respect to water quality andnative biodiversity. Of the three alternatives, the Pro-duction scenario scored lowest in terms of farmerpreferences, water quality and biodiversity endpoints.Only in profitability did the Production scenario farewell; not surprising, since this scenario emphasizedagricultural commodity production. In addition, theproportional improvement in profitability in any sce-nario was relatively low compared to improvement inother endpoints, suggesting that current practices arenear maximum profitability but much improvement ispossible for environmental endpoints.

An important caveat to the interpretation of theeconomic endpoints is that not all economic costs andbenefits resulting from changes in land use and man-agement could be incorporated within the scope ofthis project. Return-to-land only captures profits fromthe agricultural operation. No attempt was made toestimate the level of payments from federal programsthat might be made to encourage adoption of thepractices in the scenarios, nor did we attempt to esti-mate the value of environmental protection orecosystem services. For example, this endpoint doesnot include the economic benefit of improved waterquality or enhanced native biodiversity, or benefits tosociety at the national and international scale, such asthe benefit of increased soil organic carbon �Robin-son et al. 1996� to mitigate global climate change, orthe benefit of reduced nitrate export from agriculturalwatersheds on coastal waters. It does not include eco-

367

nomic costs of chronic non-point source pollution anderosion, or costs of environmental disasters resultingfrom the episodic failure of manure treatment facili-ties that might be associated with the large confinedfeeding operations located outside the study water-sheds. Incorporation of these costs would tend to de-crease the economic profitability of the Productionscenario, and enhance the overall profitability of theWater Quality and Biodiversity scenarios. It was be-yond the scope of this project to investigate the re-gional-scale influences and interactions of thechanges in agricultural practices suggested in the sce-narios on profitability of the agricultural enterprises,however, this could be a topic for subsequentresearch. The research and rules used to develop thealternative futures described here and elsewhere�Nassauer and Corry 1999� provide a foundation forthe investigation of various scenarios for implemen-tation of alternative management practices at aregional scale.

Comparison of evaluative approaches

All methods for evaluating impacts to biodiversityyielded similar results. The Biodiversity scenario, de-signed to enhance native biodiversity, accomplishedits goal. The Water Quality scenario, with highestpriority on water quality improvement also yieldedimprovements in native biodiversity relative to thepresent, and the Production scenario was alwaysworse than the existing landscape at maintaining na-tive biodiversity. The absence of wetland restorationin the Buck Creek implementation of the Biodiversityscenario �because the historical records used toinform the reserve designs did not show wetlands inthese areas� led to much lower improvements in highquality habitat area for butterflies and in amphibianabundance in that watershed. In Walnut Creek, wherea prairie and wetland preserve was part of the designfor the Biodiversity scenario, amphibian abundanceand butterfly biodiversity were substantially im-proved in the Biodiversity scenario �improvements of� 900% and � 100%, respectively�.

Although biodiversity endpoints calculated fromresults of spatially-explicit population modeling ofthe alternative futures and the habitat-area basedmethods both yielded the same rankings of the sce-narios, the use of SEPMs allows us to develop somehypotheses and directions for further research inbiodiversity responses to landscape change. The re-sults of dynamic modeling of mammal communities

�Clark et al. unpublished data� demonstrate thepotential for significant effects from species interac-tions, with direct consequences for land management.For example, increasing grassland patch sizes resultedin reduced densities and diversity of grassland rodentsdue to competition, with reductions in predators ofrodents as a further consequence. These results con-trast with other studies �e.g., Blake and Karr 1987� inwhich increased forest patch size is correlated withgreater diversity of bird species, suggesting the needfor research on the dynamics of mammal communi-ties in grasslands of agricultural regions.

Most efforts for developing landscape-level man-agement strategies have focused on conservationplanning and maintenance of diversity through gen-eral theoretical principles on habitat fragmentation,intensive single-species case studies, or static, habi-tat-based analyses �Pressey et al. 1993; Rabb andSullivan 1995; Collinge 1996�. Comparison of resultsfrom different models used in our project as well asfrom other studies indicates that the effects of bothhabitat loss and interspecific interactions in futurelandscapes must be considered. We emphasize, too,that there are advantages for using data on change inhabitat area to estimate first-order biodiversity im-pacts. This method can rapidly indicate potential ef-fects of landscape change on a broad range of species.In addition, such estimates are easy for watershedmanagers and planners to generate if they canestimate the initial area and change in area for eachtype of land cover, and have a means �e.g., USFWS-GAP 2003� to link species occurrence to land cover.

Contrasts between water quality modeling ap-proaches applied at different scales �the field-based,unit-load approach of the nitrogen component ofEPIC and the spatially-distributed, landscape-levelwater quality model SWAT� illustrate the scale-dependency of water-quality results, as well as limi-tations of field-scale modeling approaches for repre-senting watershed-level processes. Both modelsranked the Water Quality scenario as best fordecreasing nitrate export in surface runoff, followedclosely by the Biodiversity scenario, with nitrate inrunoff from the Production scenario more than twicethat for the Water Quality scenario, but still an im-provement over current practices. However, becauseEPIC is designed primarily for estimation of yield atthe field scale, “externalities” such as nitrate export,are also calculated at the field scale. Thus, nitrogentransformation processes associated with routing offlow through riparian buffers or into wetlands are not

368

modeled in EPIC. The SWAT model, which doesroute flow and models deposition and transformationprocesses in the landscape �including those areas notinvolved in agricultural production� is better suitedfor watershed-level modeling of water quality.

Estimates of nitrate export for Walnut Creek wa-tershed from SWAT were consistently lower thansimilar estimates from EPIC �here, nitrate export fromEPIC modeling is calculated as the sum of nitrate-Nrunoff and leaching below the rooting zone, assum-ing that most of the nitrate that leaches below therooting zone will still be transported to the stream viatile drains�. The nitrate-N exports estimated by SWATfrom Walnut Creek were 93 %, 93%, 36%, and 39%of those estimated by EPIC for the current landscape,the Production scenario, the Water Quality scenario,and the Biodiversity scenario, respectively. Themodel results are thus most similar in landscapes suchas the current or the Production scenario landscape,where minimal land area is dedicated to features suchas wetlands, riparian buffers, or preserves thatpromote nitrogen transformation or uptake/retentionin perennial vegetation. Where such features com-prise substantial areas of the watershed �e.g., the Wa-ter Quality and Biodiversity scenarios�, EPIC andSWAT give very different estimates of nitrogenexport. The estimates of water erosion from fieldsproduced by EPIC are not comparable to the exportof sediment in streams. The EPIC estimates of soilerosion are 25 to 50 times greater than the annual ex-port of sediment in Walnut Creek estimated by SWATin all scenarios, again underscoring the difference be-tween a field-based model and a watershed-basedmodel in which flow is routed and deposition of sedi-ment can occur. Understanding the implications oflandscape change on water quality for entire water-sheds requires application of models designed to in-tegrate processes over larger spatial scales rather thanmodels that sum field scale results for the watershed.Regional modeling of the potential consequences ofimplementing these alternative futures in much largerhydrologic units would be a logical next step.

The economic endpoint used to evaluate the sce-narios is very different from the social endpoint in itsfocus on precise estimation of a subset of the compo-nents of the economic profitability of the scenarios�e.g., crop yields� rather than qualitative indicators ofthe acceptability of the landscape as a whole �e.g., vi-sual preference�. The farmer preference studies usedto explore the cultural acceptability of the futurelandscapes to farmers are not intended to be quanti-

tatively precise. Rather, these studies were intendedto provide an indicator of which scenarios and sce-nario components Iowa farmers believe are “good forthe state of Iowa”, and to help identify elements aboutwhich there is relatively more or less agreement ofopinion. As with all estimates of human attitudes andbehavior, the results have a high level of uncertainty.In addition, because attitudes are not necessarily goodpredictors of behavior �Arcury 1990�, these resultscannot be extrapolated to predict the probability ofimplementation of practices in the scenarios. Thefarmer preference studies could even be viewed asanother iteration of the scenario design process, usedto solicit the expert opinion of producers in the re-gion.

As the previous examples suggest, an importantpart of communicating the results of this interdisci-plinary project is discussion of the methods used, thedegree to which they allow us to answer importantquestions about the alternatives, and the most appro-priate scale at which to apply the results to answerspecific questions. Each discipline involved in sce-nario evaluation has a disciplinary legacy that influ-ences the degree to which the evaluative techniquesof that discipline target realism, precision, or gener-ality �Levins 1966�. Every project has resource con-straints that affect choice of methods. There aretradeoffs between generality, realism, transferability,accuracy and precision in evaluative approaches.Table 1 characterizes the evaluative approaches weused in terms of criteria that might be important todecision makers, both in deciding how much weightto give each endpoint in decision-making and withrespect to guiding efforts to improve on or select themost appropriate evaluative approaches for future ap-plications.

For example, because the farmers we interviewedlive in the counties in which our watersheds occur, weexpect that our results realistically portray theperceptions of local farmers �degree of realism ishigh�, but realize that the generality of these resultsto larger regions of the Midwest could be low. Policymakers interested in knowing about cultural accept-ability of practices in these scenarios across the re-gion might wish to invest in more extensiveinterviews and surveys to explore the generality ofthese results across the region. An investigatorexploring the use of various approaches or models forestimating economic return in a watershed undervarious management plans might consider the degreeto which the choice of model constrains the types of

369

Tabl

e1.

Eva

luat

ive

appr

oach

esus

edto

gene

rate

endp

oint

s1-

10in

Figu

re4,

rate

dfo

rge

nera

lity,

real

ism

,ac

cura

cy,

prec

isio

n,ex

tent

tow

hich

data

are

wid

ely

avai

labl

efo

rm

odel

sof

this

type

,sc

ale

onth

ispr

ojec

tat

whi

chev

alua

tive

appr

oach

was

targ

eted

for

appl

icat

ion

�site

orfie

ld,

wat

ersh

ed�1

03-1

04ha

�,or

basi

n�a

ppro

x.10

4-1

03km

2or

mor

e �.

Eva

luat

ive

appr

oach

Gen

-er

al-

ity

Rea

lism

Acc

urac

yPr

ecis

ion

Dat

aav

aila

bilit

ySc

ale

ofA

ppli-

catio

n

SWA

Tm

odel

high

mod

erat

em

oder

ate

mod

erat

em

oder

ate

wat

ersh

ed,

basi

nE

PIC

mod

elfo

res

timat

ion

ofyi

eld

coup

led

with

cost

ofpr

oduc

tion,

econ

omic

retu

rn

low

high

for

ente

rpri

ses

cons

id-

ered

,bu

tdo

esno

tap

ply

toliv

esto

ckop

erat

ions

high

for

ente

rpri

ses

cons

id-

ered

,bu

tdo

esno

tap

ply

toliv

esto

ckop

erat

ions

prec

ise

and

wel

lca

libra

ted

for

conv

entio

nal

crop

s,pr

ecis

ion

unkn

own

for

alte

rnat

ive

crop

s,no

tus

eful

for

lives

tock

oper

a-tio

ns

high

for

conv

entio

nal

crop

s,lo

wfo

rin

nova

tive

crop

s,la

ndco

ver

not

inco

mm

odity

prod

uctio

nno

tm

odel

ed.

site

,�r

esul

tssu

mm

edfo

rw

a-te

rshe

d�

Farm

erpr

efer

ence

inte

rvie

ws

low

high

high

for

farm

ers

inte

rvie

wed

prec

ise

for

land

cove

rsin

the

spec

ific

land

scap

eco

ntex

tof

the

stud

yar

eas

high

basi

n

Stat

istic

alm

odel

for

plan

t,bu

tterfl

y,ve

rteb

rate

biod

iver

-si

ty

high

low

mod

erat

em

oder

ate

mod

erat

ew

ater

shed

,ba

sin

Spat

ially

expl

icit

popu

latio

nm

odel

for

mam

mal

spec

ies

low

mod

erat

eun

know

nw

ithou

tva

lidat

ion

pote

ntia

llyhi

ghif

mod

elca

nbe

calib

rate

dlo

ww

ater

shed

Spat

ially

expl

icit

popu

latio

nm

odel

for

4am

phib

ian

spp.

low

mod

erat

eun

know

nw

ithou

tva

lidat

ion

pote

ntia

llyhi

ghif

mod

elca

nbe

calib

rate

dlo

ww

ater

shed

370

future enterprises that can be accurately evaluated. Inour case, the choice of EPIC was based on its utilityfor estimating corn and soybean yields, which arehighly important in the present landscape andProduction future, and still important but less so insome of the futures. Thus, the same modelingapproach can be more realistic, accurate and precisefor some future landscapes than for others. The na-ture of economic enterprises comprising all futuresshould be known and considered before the choice ofmodeling approach is decided.

If improved accuracy in modeling impacts to na-tive biodiversity is important to decision makers, theneed to invest in watershed-scale research to be usedfor calibration and validation of models used to esti-mate the response of native plants and animals tolandscape change should be apparent from Table 1.Allocation of research funds to monitoring andassessment and choice of modeling approach to usemay depend on whether it is important to investiga-tors to estimate impacts to all species in the water-shed with moderate accuracy and precision, orimpacts on one or a few species with a high degreeof realism and accuracy.

Beyond ranking of future scenarios for Iowawatersheds

In addition to producing quantitative comparisons ofspatially-specific future scenarios to aid in currentdecision-making for agricultural watersheds, thisproject is valuable as an example of successful inter-disciplinary research. Presentation and publication ofour work has informed and improved subsequent wa-tershed-level conservation and land-use planning ef-forts and scenario studies �Lamy et al. 2002; Baker etal. in press� and augments the work of others in thisdeveloping field �e.g., Peterson et al. 2003; Steinitz etal. 2003; Tress and Tress 2003�. The scenarios we de-signed have explored innovative practices and dra-matically different policy alternatives. We haveapplied fundamental precepts of landscape ecology innovel ways to investigate possible outcomes when theentire agricultural matrix is modified to achieve eco-logical functions. A major strength of this project hasbeen extensive interdisciplinary collaboration that hasfostered design and evaluation of alternatives acrossa broad spectrum of goals and evaluative techniquesin three areas of scenario assessment: cultural, eco-nomic, and ecological.

The future scenarios developed here also providespecific components for watershed-level experimen-tal research that is needed for model validation/evaluation. The lack of long term ecological researchsites in agricultural watersheds, in which suchexperiments could be carried out, is a severe handi-cap to agro-ecosystem research. Agricultural experi-ment stations have a different, equally importantmission which should not be pre-empted for ecosys-tem research. To explore the ecological impacts of al-ternative land use and management strategies at thewatershed scale, long term ecological research in ag-ricultural watersheds similar to what has been donein forested ecosystems �e.g., Likens et al. 1977� isneeded.

Conclusions

The design and evaluation of alternative futures canfacilitate discussion and collaboration among stake-holders and policy makers, allowing evaluation ofeconomic, social, and ecological costs of alternativepriorities for agricultural regions, and visualization oflandscapes that could result from those priorities. Theapproach presented here provides a means to gener-ate and evaluate a broad range of innovative alterna-tives, identify those that are desirable for multipleendpoints, and to focus both policy and research onpromising aspects of those options.

The results presented here indicate that:1. if current trends toward intensification of agricul-

tural production continue, further degradation inwater quality and loss of native biodiversity in ag-ricultural regions will be a likely consequence,

2. substantial improvements in water quality and na-tive biodiversity could be achieved by implement-ing substantial changes in land use and manage-ment practices, if public support and agriculturalpolicy place high priority on clean water and res-toration of habitat for native species,

3. these changes can be designed so that they arecomparable to current practices in profitability andyield much greater environmental improvementsthan practices designed solely with economic profitin mind,

4. these changes may be culturally acceptable tofarmers and preferred to current trends as long asagricultural enterprises remain profitable,

371

5. there is room for substantial improvement relativeto current practices in all endpoints used to evalu-ate landscape response without sacrificing profit-ability, and

6. there is a need for long term ecological research inagricultural systems to study the watershed-levelimpacts of changes in land use and management,to validate the models results presented here, andto allow exploration of environmental responses toagricultural practices across scales.

This study is the first in the U.S. to integrate interdis-ciplinary research on economic profitability, culturalacceptance, water quality, and native biodiversity us-ing a scenario design and evaluation approach for ag-ricultural ecosystems. Using the agricultural land-scape itself as the means to integrate the variousdisciplinary responses to changes in land use andmanagement, we have begun to quantify the socio-economic and ecological benefits and costs ofcontinuing current practices and to compare these tothe benefits and costs of potential change. Althoughmodels can only approximate the watershed-level re-sponse of various endpoints to changes in land use,they allow us to rank alternatives with respect to theireffectiveness in achieving the desired objectives�Starfield 1997�. They provide a starting point forfurther analyses of the scenario elements �such aswetlands, riparian buffers, woodlands, alternativecropping and management practices� that appear todrive changes in the ecological and socioeconomicendpoints. More importantly, they provide a potentialframework for watershed-level field experiments formodel validation and investigation of the effects ofspatial scale in model application, and for quantifica-tion of effects of changes in agricultural land use andmanagement �Ahern 1999�. Once we can quantify theimpacts of such changes, it becomes easier to evalu-ate the outcomes of different policy choices, and totranslate human priorities for healthy agriculturalecosystems into “A Geography of Hope” �USDA-NRCS 1996�.

Acknowledgments

We thank the US-EPA STAR grants program �Waterand Watersheds, grant #R-825335-01� for funding,those involved with the MASTER program for shar-ing with us data essential to this project, and all par-ticipants in the initial project workshop that laid the

groundwork for the scenario designs. We thank thereviewers who helped improve this paper. We extendspecial thanks to the Iowa farmers who participatedin our interviews. This document has been subjectedto the Environmental Protection Agency’s peer andadministrative review and approved for publication.

References

Ahern J. 1999. Spatial Concepts, Planning Strategies and FutureScenarios: A Framework Method for Integrating LandscapeEcology and Landscape Planning. In: Klopatek J.M. and Gard-ner R.H. �eds�, Landscape Ecological Analysis: Issues and Ap-plications. Springer-Verlag, New York, New York, USA, pp.175-201.

Alexander R.B., Smith R. and Schwartz G. 1996. The RegionalTransport of Point and Nonpoint-Source Nitrogen to the Gulf ofMexico. Proceedings of the Gulf of Mexico Hypoxia Manage-ment Conference, December 5-6, 1995, Kenner, Louisiana, U.S.Geological Survey, Reston, Virginia, March 5, 1996.

Arcury T.A. 1990. Environmental Attitude and EnvironmentalKnowledge. Human Organization. 49�4�: 300–304.

Arnold J.G., Williams J. and Maidment D.R. 1997. Continuoustime water and sediment routing model for large basins. Journalof Hydrologic Engineering 121: 171–183.

Baker J.P., Hulse D.W., Gregory S.V., White D., Van Sickle J.,Berger P.A., Dole D. and Schumaker N.H. 2004. Alternative fu-tures for the Willamette River Basin, Oregon. Ecological Appli-cations. In press.

Becher K.D., Schnoebelen D.J. and Ackers K.B. 2000. Nutrientsdischarged to the Mississippi River from eastern Iowa water-sheds. Journal of the American Water Resources Association 36:161–173.

Bender D. J., Contreras T.A. and Fahrig L. 1998. Habitat loss andpopulation decline: a meta-analysis of the patch-size effect.Ecology 79: 517–533.

Blake J.G. and Karr J.R. 1987. Breeding birds of isolated wood-lots: area and habitat relationships. Ecology 68: 1724–1734.

Coiner C., Wu J. and Polasky S. 2001. Economic and environmen-tal implications of alternative landscape designs in the WalnutCreek watershed of Iowa. Ecological Economics 38: 119–139.

Collinge S.K. 1996. Ecological consequences of habitat fragmen-tation: implications for landscape architecture and planning.Landscape and Urban Planning 36: 59–77.

Dunning J.B., Stewart D.J., Danielson B.J., Noon B.R., Root T.L.,Lamberson R.H. and Stevens E.F. 1995. Spatially-explicit popu-lation models: current forms and future uses. Ecological Appli-cations 5: 3–11.

Dupré C. and Ehrlen J. 2002. Habitat configuration, species traitsand plant distributions. Journal of Ecology 90: 796–805.

Eilers L.J. and Roosa D.M. 1994. The Vascular Plants of Iowa.University of Iowa Press, Iowa City, Iowa, USA.

Farrar D. 1981. Perspectives on Iowa’s declining flora and fauna-A symposium. Proceedings of the Iowa Academy of Science 88:1–6.

Freemark K. 1995. Assessing effects of agriculture on terrestrialwildlife: developing a hierarchical approach for the U.S. EPA.Landscape and Urban Planning 31: 99–115.

372

Freemark K. and Smith J. 1995. A landscape retrospective for Wal-nut Creek, Story County. Technical report to the U.S. EPA. Un-published contract report.

Hatfield J.L., Jaynes D.B., Burkart M.R., Cambardella C.A.,Mooreman T.B., Prueger J.H., Smith M.A. 1999. Water Qualityin Walnut Creek Watershed: Setting and Farming Practices.Journal of Environmental Quality 28: 11–24.

Klopatek J.M., Olson R.J., Emerson C.J. and Joness J.L. 1979.Land-use conflicts with natural vegetation in the United States.Environmental Conservation 6: 191–199.

Krebs J.R., Wilson J.D., Bradbury R.B. and Siriwardena G.M.1999. A second Silent Spring? Nature 400: 611–612.

Lamy F., Bolte J., Santelmann M. and Smith C. 2002. Develop-ment and evaluation of multiple-objective decision-makingmethods for watershed management planning. Journal of theAmerican Water Resources Association 38: 517–529.

Levins R. 1966. The strategy of model building in population bi-ology. American Scientist 54: 421–431.

Likens G.E., Bormann F.H., Pierce R.S., Eaton J.S. and JohnsonN.M. 1977. Biogeochemistry of a forested ecosystem. Springer-Verlag, New York, New York, USA.

Matson P.A., Parton W.J., Power A.G. and Swift M.J. 1997. Agri-cultural intensification and ecosystem properties. Science 277:504–508.

Mitsch W.J., Day J.W. Jr., Gilliam J.W., Groffman P.M., Hey D.L.,Randall G.W. and Wang N. 2001. Reducing nitrogen loading tothe Gulf of Mexico from the Mississippi River Basin: Strategiesto counter a persistent ecological problem. Bioscience 51: 373–388.

Nassauer J.I. and Corry R.C.. 1999. Rural Watersheds and Policy.http//:www.snre.umich.edu/nassauer/lab_index.html. Accessed06/25/2003.

Nassauer J.I., Corry R.C. and Cruse R. 2002. The landscape in2025:Alternative landscape future scenarios as a means to con-sider agricultural policy. Journal of Soil and Water Conservation57�2�: 44A–53A.

Nassauer J. I. and Corry R.C. 2004. Using normative scenarios inlandscape ecology. Landscape Ecology 19: this issue.

Peterson G., Cumming A., Carpenter S. 2003. Scenario planning: atool for conservation in an uncertain world. Conservation Biol-ogy 17�2�: 358–366.

Pressey R.L., Humphries C.J., Margules C.R., Vane-Right R.I. andWilliams P.H. 1993. Beyond opportunism: key principles forsystematic reserve selection. Trends in Ecology and Evolution8: 124–128.

Prior J. 1991. Landforms of Iowa. University of Iowa Press, IowaCity, Iowa, USA.

Puckett L.J. 1994. Nonpoint and point sources of nitrogen in majorwatersheds of the United States. U.S. Geological Survey, WaterResources Investigations Report 94-4001, USGS WashingtonDC, USA.

Rabb G.B. and Sullivan T.B. 1995. Coordinating conservation:global networking for species survival. Conservation and Biodi-versity 4: 536–543.

Robinson C.A., Cruse R.M. and Ghaffarzadeh M. 1996. Croppingsystem and nitrogen effects on mollison organic carbon. SoilScience Society of America Journal 60: 264–269.

Rosenzweig M.L. 1997. Heeding the warning in biodiversity’s ba-sic law. Science 284: 276–277.

Rustigian H. 1999. Assessing the potential impacts of alternativelandscape designs on amphibian population dynamics. Master of

Science Research Paper, Department of Geosciences, GeographyProgram, Oregon State University, Corvallis, Oregon, USA.

Rustigian H., Santelmann M. and Schumaker N. 2003. Assessingthe potential impacts of alternative landscape designs onamphibian population dynamics. Landscape Ecology 18: 65–81.

Sala O.E., Stuart Chapin III F., Armesto J.J., Berlow E., Bloom-field J., Dirzo R., Huber-Sanwald E., Huenneke L., Jackson R.B.,Kinzig A., Leemans R., Dodge D.M., Mooney H.A., OesterheldM., Poff N.L., Sykes M.T., Walker B.H., Walker M. and WallD.H. 2000. Global biodiversity scenarios for the year 2100. Sci-ence 287: 1770–1774.

Santelmann M., Freemark K., White D., Nassauer J., Clark M.,Danielson B., Eilers J., Cruse R., Galatowitsch S., Polasky S.,Vaché K., Wu J. 2001. Applying Ecological Principles to Land-Use Decision Making in Agricultural Watersheds. In: Dale V.and Haeuber R. �eds�, Applying Ecological Principles to LandManagement. Springer-Verlag, New York, New York, USA.

Schilling K.E. and Thompson C.A. 2000. Walnut Creek monitor-ing project, Iowa: Monitoring water quality in response to prai-rie restoration. Journal of the American Water ResourcesAssociation 36: 1101–1114.

Schoup M. 1999. Agricultural Runoff of Nutrients and IndicatorOrganisms in Buck Creek, Iowa. Masters of Science Thesis, De-partment of Civil and Environmental Engineering, University ofIowa, Iowa City, Iowa, USA.

Starfield A. 1997. A pragmatic approach to modeling for wildlifemanagement. Journal of Wildlife Management 61: 261–270.

Steinitz C., Arias H., Bassett S., Flaxman M., Goode T., MaddockIII T., Mouat D., Peiser R. and Shearer A.. 2003. Alternative fu-tures for changing landscapes: The Upper San Pedro River Ba-sin in Arizona and Sonora. Island Press, Washington DC, USA.

Tress B. and Tress G.. 2003. Scenario visualization for participa-tory landscape planning- a study from Denmark. Landscape andUrban Planning 64: 161–178.

US Census Bureau. 2001. Statistical Abstract of the United States:2001. U.S. Department of Commerce, Washington, DC, USA.

US GAO. 1999. Report on Animal Agriculture: Waste ManagementPractices. U.S. General Accounting Office Report GAO/RCED-99-205.

US OTA. 1995. Targeting environmental priorities in agriculture:Reforming program strategies. Congress of the United States,Office of Technology Assessment. OTA-ENV-640. GovernmentPrinting Office, Washington DC, USA.

USFWS-GAP. 2003. http://www.gap.uidaho.edu/ Accessed Sep-tember 23, 2003.

USDA-NASS. 1999. Farms and Land in Farms 1993-1997, U. S.Department of Agriculture Statistical Bulletin 955.

USDA-NRCS. 1996. A Geography of Hope, USDA-NaturalResources Conservation Service. 80 pp.

USDA-NRCS. 1997. The PLANTS Database, Version 3.5�http://plants.usda.gov�. National Plant Data Center, BatonRouge, LA 70874-4490 USA. Accessed September 12, 1997.

Vaché K., Eilers J. and Santelmann M.V. 2002. Water QualityModeling for Evaluation of Alternative Agricultural Practices inTwo Iowa Watersheds. Journal of the American Water ResourcesAssociation 38: 773–787.

Vitousek P.M., Mooney H.A., Lubchenco J., Melillo J.M.. 1997a.Human domination of earth’s ecosystems. Science 277: 494–499.

Vitousek P.M., Aber J.D., Howarth R.W., Likens G.E., MatsonP.A., Schindler D.W., Schlesinger W.H., Tilman D.G. 1997b.

373

Human alteration of the global nitrogen cycle: sources and con-sequences. Ecological Applications 7: 737–750.

White D., Minotti P.G., Barczak M.J., Sifneos J.C., Freemark K.E.,Santelmann M., Steinitz C.F., Keister A.R. and Preston E.M.1997. Assessing risks to biodiversity from future landscapechange. Conservation Biology 11: 349–360.

White D., Preston E.M., Freemark K.E. and Kiester A.R.. 1999. Ahierarchical framework for conserving biodiversity. In: Klopatek

J.M. and Gardner R.H. �eds�, Landscape Ecological Analysis:Issues and Applications. Springer, New York, New York, USA,pp. 127–153.

Williams J.R., Jones C.A. and Dyke P.T. 1988. EPIC, the ErosionProductivity Index Calculator, Model Documentation Vol. 1USDA ARS, Temple, Texas, USA.

374