lost in transition: determinants of post-socialist ...muelleda/download/mueller...

Lost in transition: determinants of post-socialist croplandabandonment in Romania

D. Mullera,b*, T. Kuemmerleb,c, M. Rusud and P. Griffithsb

aLeibniz Institute of Agricultural Development in Central and Eastern Europe (IAMO), Halle (Saale),Germany; bGeomatics Lab, Geography Department, Humboldt-Universitat zu Berlin, Berlin,

Germany; cDepartment of Forest and Wildlife Ecology, University of Wisconsin-Madison, Madison,WI, USA; dInstitute of Agricultural Economics, Romanian Academy, Bucharest, Romania

The transition from command to market-oriented economies drastically affected landownership and land management in Eastern Europe and resulted in widespread croplandabandonment. To examine these phenomena, we analysed the causes of post-socialistcropland abandonment in Arges County in southern Romania between 1990 and 2005.Based on Landsat-derived maps of cropland use and a suite of environmental andsocioeconomic variables hypothesized to drive cropland abandonment, we estimatedspatially explicit logistic regression models for two periods (1990–1995 and 1995–2005)and three elevation groups. Our results showed that isolated cropland patches were morelikely to become abandoned than more homogenous cropland areas. Unfavorabletopography was an important determinant of abandonment in the plain and, to a lesserextent, hilly areas, but not in the mountains where locations with adverse market accessand higher farm fragmentation exhibited higher likelihoods of cropland abandonment.

Keywords: cropland abandonment; logistic regression; spatial model; land-use change;Romania; Eastern Europe

1. Introduction

Land-use practices are the main reason for the conversion of natural ecosystems into human-dominated landscapes (Foley et al. 2005; Kareiva, Watts, McDonald, and Boucher 2007),particularly in the tropics where croplands continue to replace forests at high rates (Leperset al. 2005; Hansen et al. 2008). However, in other areas of the world, land-cover trajectorieshave reversed when countries reach an industrialized stage, as well as when population sizestabilizes (Mather and Needle 1998; Rudel et al. 2005). This often results in a twofoldpattern of intensified cultivation in areas favourable for farming, and the abandonment ofcultivation where farming conditions are marginal (Baldock, Beaufoy, Brouwer, andGodeschalk 1996; MacDonald et al. 2000; Strijker 2005).

Cropland abandonment has widespread effects on ecosystem services, for example,increased carbon sequestration when cropland reverts back to forests (Silver, Ostertag, andLugo 2000), decreased soil erosion (Tasser, Mader, and Tappeiner 2003), or increased waterquality (Hunsaker and Levine 1995). On the other hand, traditional cultural landscapes areincreasingly lost where cropland is abandoned (Government Service for Land and WaterManagement 2005; Palang et al. 2006), which often results in the decline of biodiversity

Journal of Land Use ScienceVol. 4, Nos. 1–2, March 2009, 109–129

*Corresponding author. Email: [email protected]

ISSN 1747-423X print/ISSN 1747-4248 online# 2009 Taylor & FrancisDOI: 10.1080/17474230802645881

http://www.informaworld.com

Downloaded By: [Müller, Daniel] At: 16:30 21 February 2009

(Brouwer, Baldock, and La Chapelle 2001; Cremene et al. 2005). Yet despite these effects,the rates and spatial patterns, as well as the underlying causes of cropland abandonment, arelargely unknown in many parts of the world.

Cropland abandonment is often triggered by broad-scale political or socioeconomicshocks (Yeloff and Van Geel 2007). Central and Eastern European countries have experi-enced drastic and rapid changes in their political and socioeconomic structures since thecollapse of socialist regimes. The departure from command economies was accompanied bya restructuring of the respective farming sectors, as well as land reforms that triggeredmassive ownership transfers of both natural resources and capital assets (Mathijs andSwinnen 1998; Lerman, Csaki, and Feder 2004; Rozelle and Swinnen 2004). Rural popula-tion compositions have changed dramatically due to large internal and external migrationmovements, especially of younger population segments (Angelstam, Boresjo-Bronge,Mikusinski, Sporrong, and Wastfelt 2003; Palang et al. 2006; Muller and Munroe 2008).As a result, cropland abandonment became widespread in the post-socialist era in EasternEurope and European Russia (Bicik, Jelecek, and Stepanek 2001; Ioffe, Nefedova, andZaslavsky 2004).

Several studies have mapped the rates and spatial patterns of post-socialist croplandabandonment in Eastern Europe, for example, in Albania (Muller and Sikor 2006), Estonia(Peterson and Aunap 1998), Latvia (Nikodemus, Bell, Grine, and Liepins 2005), and theCarpathians (Kozak, Estreguil, and Troll 2007; Kuemmerle et al. 2008a). These studiesshowed marked variability in abandonment rates throughout the region. Unfortunately, moststudies did not employ quantitative and spatially explicit modelling techniques to assess theunderlying determinants of cropland abandonment (for an exception, see Muller andMunroe 2008, in Albania). Yet such methods are powerful tools for linking observed land-use changes to a suite of environmental and socioeconomic driving factors.

Overall, the determinants of post-socialist cropland abandonment remain largely unex-plored. This is unfortunate, because post-socialist cropland abandonment in Eastern Europeoffers unique opportunities to study the role of changing institutions and socioeconomics forland-use change. Indeed, cropland abandonment may continue to be the primary land-usechange in many industrialized countries (Verburg, Schulp, Witte, and Veldkamp 2006).Understanding the determinants of post-socialist cropland abandonment may provide valu-able management insights, inform policy-makers, and help to improve future land-usescenarios. Moreover, statistical modelling of land-use change facilitates the testing oftheoretical assumptions and hypotheses (Munroe and Muller 2007).

Studying the determinants of cropland abandonment in Romania is interesting fromseveral perspectives. For example, Romania is the second largest new member state of theEuropean Union with the largest share of farmland (65%) among the new accessioncountries (FAO 2006). Romania followed a unique land reform strategy that combinedrestitution to former owners, or their heirs, with a redistribution of ownership rights to formercooperative workers who had contributed land to the cooperative. But land markets inRomania remained inactive until the late 1990s, with few reported sales and leases(Swinnen, Vranken, and Stanley 2006). This resulted in a farming structure characterizedby a large share of small farms with little commercial activities (Dawidson 2005). Inaddition, the land reform also created a large class of absentee landowners who did notreside in rural areas, as well as many old landowners (Chirca and Tesliuc 1999; Mathijs andNoev 2004). Overall, the effect of the changing land property situation on the rate, extent,and spatial patterns of cropland abandonment is unclear.

In a previous study, we assessed post-socialist land-use change in Arges County, southernRomania, based on Landsat Thematic Mapper and Enhanced Thematic Mapper Plus images

110 D. Muller et al.


from 1990 to 2005 (Kuemmerle, Muller, Griffiths, and Rusu 2009). Building upon this work,the focus here is on assessing the underlying causes of cropland abandonment, the largest post-socialist land use in Arges County. To accomplish this, we linked cropland abandonment fromtwo time periods (1990–1995 and 1995–2005) to a suite of bio- and geophysical, socio-economic, and policy-related variables using spatially explicit logistic regressions models.Beforehand, we hypothesized that variables affecting agricultural suitability and farm profits,as well as rural population dynamics, were the prevailing factors that contributed to croplandabandonment in Arges County. Specifically, our objectives were as follows:

(1) To quantify the direction and strength of the causal influences of each variable oncropland abandonment during the two periods.

(2) To compare the causal influence of the variables in the regression model for theentire study area with models estimated for three elevation subgroups.

2. The study area

We studied Arges County in central Romania (Figure 1) because the region encompasses awide range of environmental conditions. The county is part of the historical province

Figure 1. Study area.Source: Authors.

Journal of Land Use Science 111


Muntenia and covers an area of 6826 km2, has a mean annual rainfall of 750 mm and a meanannual temperature of 7�C. Predominant soil types are Argillic soils in the plains, Cambisolsin the hilly areas, and Podzols in the mountainous areas. Biogeophysical variables showclear north–south patterns. For instance, elevation and rainfall are higher in the north,whereas temperature is lower.

The southern region of Arges County consists of a plain (,250 m above sea level),whereas the hilly zone (250–1500 m) contains the county’s capital and major marketcenter Pitesti (174,000 inhabitants in 2003 (NIS 2004), which is linked to the country’scapital Bucharest by highway. Close to Pitesti is the main factory of the carmakerDacia, the largest private employer in the region. The mountainous northern part ofthe study area (above 1500 m), a part of the Carpathian Mountains, is characterized byrugged terrain and includes Romania’s highest mountain, the Moldoveanu Peak(2544 m).

3. Data and variables

3.1. Land-cover maps and dependent variables

Land-cover maps for the years 1990, 1995, and 2005 were available from a previous study(Kuemmerle et al. 2009). These maps were generated by classifying the land-cover typesforest, cropland, and permanent grassland (including shrubland) from Landsat TM andETM + images at a spatial resolution of 30 m · 30 m using a hybrid classification approach(Bauer et al. 1994; Kuemmerle, Radeloff, Perzanowski, and Hostert 2006). Settlements,roads, and water bodies were masked prior to classification based on available topographicmaps. The accuracy of these land-cover maps was assessed based on a random sample of765 ground truth points mapped in the field. Overall accuracy was 91.0% (Kappa = 0.86),92.5% (0.89), and 91.0% (0.86) for 1990, 1995, and 2005, respectively. Land-coverchanges were mapped via post-classification map comparison (Coppin, Jonckheere,Nackaerts, Muys, and Lambin 2004), which yielded two change maps for the periods of1990–1995 and 1995–2005. A detailed description of the change analyses is provided inKuemmerle et al. (2009).

Cropland decreased strongly in Arges County during transition, but within distinctspatial clusters (Figure 2). In total, 21% of the cropland was abandoned in 1990, whichequaled an area of 512 km2 and represented a decrease from 35 to 28% relative to thecounty’s area. Cropland abandonment was more prevalent in the 1990–1995 period (17%,equaling 415 km2) compared to the 1995–2005 period (5%, 97 km2) (Kuemmerle et al.2009).

Patterns of abandonment varied considerably along altitudinal gradients (Kuemmerleet al. 2009). In 2005, cropland covered 70% of the land surface in the plains (1754 km2 of6286 km2), 11% in the hilly areas, and a mere 2% in the mountains (Figure 3). In themountainous areas, cropland decreased from 88 to 40 km2 despite a slight increasebetween 1995 and 2005. In the hilly area, 227 km2 (41% of the 1990 cropland) wereabandoned in the first and 68 km2 (21%) in the second period, respectively. In the plains,170 km2 or 10% were abandoned and almost three-quarters of this decline tookplace between 1990 and 1995. In sum, Arges experienced the highest amountsof cropland abandonment in the hilly zone and the highest rates in the mountainousnorth.



Figure 2. Spatial patterns of cropland abandonment.Note: Elevation subgroups derived from terciles based on mean commune elevation.Source: Authors, data from Kuemmerle et al. (2009).

1754

555

88

1633

328

20

1584

260

40

0

500

1,000

1,500

2,000

2,500

km²

1990 1995 2005

MountainHillyPlain

Figure 3. Cropland changes by elevation subgroup.Source: Authors.



3.2. Cropland density

Land-use change at a particular location often depends on its surroundings. For example, thedecision to abandon a cropland pixel may depend on the presence of nearby cropland. Theinclusion of a spatially lagged average of the dependent variable as an additional regressorcan account for this, but may lead to endogeneity in the lag variable, as well as potentiallybiased coefficients (Anselin 2002). To control for such endogeneity, we constructed avariable containing the number of neighbouring cropland pixels at the beginning of eachchange period within a three-by-three window. This approach is similar to Geoghegan,Waininger, and Bockstael (1997) and Geoghegan et al. (2001), who included patterns ofsurrounding land uses in their estimations to capture the diversity of land use. Gellrich, Baur,Koch, and Zimmermann (2007) used a binary variable if surrounding observations wereequal to their dependent variable. Our approach has the additional advantage that it partlycontrols for potential endogeneity between the dependent and the independent variables byincluding a temporal lag. Besides, the density variable is an ordinal measure that captures thehomogeneity of cropland in the immediate surroundings.

3.3. Time-invariant location factors

We used the Shuttle Radar Topography Mission digital elevation model (DEM) with aspatial resolution of 90 m (Slater et al. 2006). The DEM was used in the classificationprocess (Kuemmerle et al. 2009) and therefore not employed as a covariate. Instead, wecalculated terrain roughness (the slope curvature), which is not correlated with elevation.Roughness values were multiplied by a factor of 10 to match the data range of the othervariables.

The road network consisted of four different road categories: the highway linkingBucharest with Pitesti (category 1), European roads (2), national roads (3), and countyroads (4). To assess the influence of road presence on cropland abandonment, threeEuclidean distance measures were calculated cumulatively. First, using road categories1 and 2; second, using road categories 1, 2, and 3; and, third, using all four road categories.

We also include a variable that captured the distance of each location to the northern limitof the study region (subsequently labeled ‘Y coordinate’) to control for spatial dependence inthe model residuals (Tasser, Walde, Tappeiner, Teutsch, and Noggler 2007) and to accountfor the existing north–south gradient. No systematic patterns were observed in the east–westdirection and we therefore did not include an X coordinate.We excluded temperature and soildata due to collinearity with the Y indicator variable.

3.4. Rural commune census

A census of all rural communes was carried out in Arges during the summer and autumn of2005. Covering socioeconomic and demographic developments, input use, farm fragmenta-tion, and productivity parameters, the census captured exogenous influences, as well asendogenous changes within communes, that were hypothesized to be relevant for croplandabandonment. Municipalities and towns were excluded from the primary data collectionbecause we focused on land use in rural areas. We used a structured questionnaire for groupinterviews with selected key informants such as commune mayors, agricultural and forestryagents, and cadastral officers. All 16 interviewers were trained in survey and interviewingtechniques and carried out the interviews in groups of two. The interviewers were instructedto facilitate a discussion among the respondents to verify and check responses with all



participants. Temporal coverage was achieved through recall techniques (Groves 1989) oflandmark events that the respondents could easily remember and that approximatelymatched the date of the land-cover maps. The first year (1989) depicts the situation duringsocialism and 1996 was chosen as an intermediate point in time when the Romanianpresidential election was held.

We calculated livestock units from the census data to assess the effect of livestockbreeding on cropland use. We used the livestock unit coefficients for countries of theOrganisation for Economic Co-operation and Development, OECD (Chilonda and Otte2006). The cattle, sheep, and goat numbers were multiplied by 0.9, 0.1, and 0.1, respectively,and divided by the commune area to yield livestock unit densities as covariates for 1989 and1996.We further used the number of tractors per commune to proxy agricultural input levels.Several other potentially interesting measures of input intensity (e.g., irrigated area, use ofcertified seeds, or the use of fertilizer) could not be included due to missing values or lack ofvariation across the communes. Land reform effects were captured by two variables: First,we used the average number of plots owned by respondents of a commune after land reformimplementation. More plots are equivalent to a higher physical fragmentation of the farm,because plot locations are typically noncontiguous. Second, the share of cropland owned byindividuals within each commune was calculated to obtain a measure for the type ofagricultural production organization. Both variables were only included in the second periodbecause they did not vary significantly in 1989.

3.5. Population and migration proxies

Weused interpolation techniques to approximate a spatially continuous population distributionbased on the village population data gathered in the rural commune census. This populationsurface served as a proxy for the demand for cropland at the pixel level. Villages with missingrecords for 1989 or 1996 (about 10% of the cases) were replaced with official census data fromthe National Institute of Statistics (NIS 2004). We disaggregated the commune-level popula-tion data from the NIS census data to village level by distributing the official data proportion-ally among the missing villages within a commune. We then calculated the centroids of thevillage boundaries, linked the population data to these points, and used a third-order inversedistance weighted interpolation to derive continuous population surfaces. Based on visualassessments of the density maps and field experience, we decided to include up to five nearestneighbours and a maximum interpolation distance of 5 km (representing the maximum sphereof spatial influence of inhabitants from a village). This yielded two population density maps,one each from 1989 and 1996. In addition, we included the household density per squarekilometer in 1989 and 1996 as independent variables in the models.

Emigration was proxied by the number of individuals who left the commune perma-nently. The migration variable was the only variable that spanned the entire estimationperiod, because it was collected for the periods 1989–1996 and 1996–2004. To avoidendogeneity bias (see Section 5.2), we tested the robustness of the emigration variable byestimating the models with and without migration. Coefficients, signs of other variables, andmodel fit did not change, and we therefore decided to include this potentially importantdriver in our models. Lastly, temporary emigration in 1996 was included as a variable tocapture the economic effect of remittance returns from seasonal migration on croplandabandonment. At the beginning of the 1990s, migration returns were close to zero in allcommunes and therefore temporary migration was not considered as a covariate in the firstestimation period from 1990 to 1995.



3.6. Hypothesized influence of the covariates

In total, we considered 14 covariates that we a priori expected to influence croplandabandonment (Table 1). We assumed that undulating terrain (higher roughness) wouldincrease cropland abandonment, as agricultural production may concentrate on locationsthat are better suited for market-oriented production (MacDonald et al. 2000; Ioffe et al.2004). Similarly, we hypothesized higher abandonment at more marginal locations for thevariables that proxy road access.

Higher rural population density was assumed to have a negative bearing on land abandon-ment via the increased demand for cultivated land (Angelstam et al. 2003; Ioffe et al. 2004).Wefurther expected a lower efficiency of resource consumption per capita in smaller households, asa result of the varying resource use intensity among households and individuals (Liu, Daily,Ehrlich, and Luck 2003). In Arges County, household size decreased, whereas the number ofhouseholds increased (own data). Thus, higher per capita consumption in smaller householdsmay have stimulated abandonment, whereas higher household density may have increasedcropland demand. Temporary and permanent emigration were unevenly distributed across agegroups, with the young and economically active typically being the first to leave. Therefore,emigration possibly induced labour shortages, thereby increasing abandonment (Nikodemuset al. 2005; Muller and Munroe 2008).

More tractors per commune (a proxy for agricultural input intensity) potentiallyincreased agricultural production, reflected a higher market orientation and was labour-saving, possibly reducing overall abandonment. On the other hand, labour scarcity due toselective migration could have induced farmers to abandon marginal plots and concentratetheir capital and labour inputs on the best plots. Moreover, marginal plots may not be suitablefor mechanization. Thus, the ultimate effect of tractor numbers on the amount of croplandused was a priori unclear. The effect of livestock density on cropland abandonment may alsohave been twofold: Intensive livestock rearing may stimulate fodder crop cultivation,thereby keeping abandonment rates low. On the other hand, extensive livestock productionmay amplify cropland abandonment due to converting cropland to pastures. Overall, theeffect of livestock density on abandonment was unclear a priori, because we have no spatialdata on the intensity of livestock production.

Table 1. Independent variables with hypothesized influence on cropland abandonment.

Variable Level Unit Influence

Y coordinate Pixel Kilometers from northern edge +Roughness Pixel Mean in 9 · 9 window +Distance to European road Pixel 100 m from category 1 and 2 +Distance to main road Pixel 100 m from category 1, 2, and 3 +Distance to county road pixel 100 m from category 1, 2, 3, and 4 +Cropland in neighborhood Pixel Surrounding number in 3 · 3 window -Population Pixel Interpolated -Household density Commune Households per km2 ?Emigration (permanent) Commune Individuals +Temporary emigration Commune Individuals +Tractors Commune Number ?Livestock density Commune Units according to OECD per km2 ?Cropland owned individually Commune % of cropland -Parcels Commune Average number per farm +

Source: Authors.



More plots owned by a household (the farm fragmentation proxy) presumably resulted inproduction inefficiencies via increased labour requirements and reduced the effectiveness ofand the incentive for mechanization (van Dijk 2003), thereby fostering abandonment.Similarly, land owned by the state or by legal and family associations may have been morelikely to be kept in cultivation due to higher input use and mechanization levels, as well asbetter access to financial capital.

4. Statistical approach

4.1. Data integration and sample selection

We combine the variables derived from the commune census with the spatially explicit databased on the commune boundaries by assigning the same value for all pixels within acommune. All calculations were conducted using a spatial resolution of 28.5 m to match theland-cover maps, resulting in 8,395,490 observations within Arges County. All grids wereconverted to integer values to reduce computing time and the distance measures wererescaled to 100 m steps to facilitate the interpretation of the statistical results.

The final data set was the result of five sampling steps. First, an inside buffer of 1 kmwaserased from the boundary of the study area to avoid influences from neighbouring counties(Muller and Zeller 2002). This buffer size was based on visual inspection, but sensitivitytests with larger buffer sizes did not alter the results. Second, spatial coding was carried out toselect non-adjacent pixels at a specified distance from the nearest selected neighbour toreduce spatial autocorrelation in the response variable (Besag 1974; Nelson and Hellerstein1997). To choose an appropriate distance between selected observations, we tested differentcoding schemes ranging from distances of 6–15 pixels between selected neighbours, anddecided to use a distance of 9 pixels (228 m) as a compromise between the correction ofspatial dependency among observations and the rapidly decreasing number of observations.Third, only observations that were covered by cropland at the beginning of the respectivechange period were selected. In that way we accounted for the temporal dependence ofcropland abandonment on the state of land cover at the start of the respective period (Tasseret al. 2007; Muller and Munroe 2008). This also ensured that abandoned pixels in the firstperiod did not affect coefficients in the second period. Fourth, disproportionate (or balanced)sampling was applied to the binary-dependent variable, because the number of observationswith stable cropland (= 0) was much larger than the number of abandoned pixels (= 1) – as isoften the case in land-use/land-cover studies (Serneels and Lambin 2001; Gellrich et al.2007). We randomly selected 1000 observations from both the presence and absence group.Such disproportionate sampling in logistic regression analysis does not affect the coeffi-cients, but can affect the constant term and, therefore, the predicted probabilities (Maddala2001). The constant has to be decreased by ln(p1) - ln(p0) to obtain correct predictions,where p1 and p0 are the proportions of the sampled observations from the presence andabsence group, respectively, and ln is the natural logarithm.

The models were estimated for the entire study area (hereafter called the global model).Although such results can be highly relevant for policy-making, global models average outthe spatial heterogeneity (Osborne and Suarez-Seoane 2002). This may be especiallyproblematic for an ecologically diverse region such as Arges County. We, therefore, calcu-lated elevation terciles for the mean elevation within a commune to obtain three elevationsubgroups that displayed high elevation homogeneity within a group, but strongly differedfrom each other. Three additional models (local models) were fitted for the elevationsubgroups that are subsequently referred to as the plain model, hilly model, and mountain



model for the low, the medium, and the high elevation terciles, respectively. Every subgroupcontained 31 communes, but they differed in their spatial extent (and the number of pixels)due to varying commune sizes. We employed identical sampling strategies for the subgroupsand randomly selected 500 abandoned and 500 non-abandoned observations.

4.2. Estimation strategy

To understand the causal determinants of cropland abandonment in Arges County at thepixel and at the commune level, we fitted maximum likelihood-based binary logisticregressions. Two logistic regression models of cropland abandonment were estimated, onefor the 1990–1995 period and one for the 1995–2005 period. Estimating two change periodsallowed us to describe the temporal changes in the influences of the independent variablesthat affected the use of cropland. All time-variant variables were included using values fromthe beginning of the respective change period. We consider these temporally lagged vari-ables as exogenous covariates of cropland abandonment, because the lags avoid the endo-geneity bias which arises from a possible simultaneity of cropland abandonment and thecovariates (Muller and Zeller 2002; Perz and Skole 2003). This assumes that later abandon-ment did not influence the state of a variable at the beginning of the respective period. Thisregression setup allows us to go beyond mere associations between variables and permits, inconnection with field-based evidence, causal interpretations of the effects of the independentvariables on cropland abandonment.

The multilevel data structure required corrections to ensure the independence of obser-vations. Pixels within communes may exhibit dependence, whereas pixels across communesare more likely to be independent (Muller and Munroe 2005; Overmars and Verburg 2006;Gellrich et al. 2007). To control for these within-group effects, robust estimation techniqueswere employed that account for potential correlations within communes. We used a variantof the Huber and White estimator that affects the variance–covariance matrix by estimatingrobust standard errors, but leaves the estimated coefficients unaffected (Froot 1989;Gutierrez and Drukker 2005). Within-group adjustment also has the additional advantageof controlling for potential spatial autocorrelation in the residuals by accounting for correla-tions of observations within communes (Gellrich et al. 2007).

In total, eight models were estimated: one global and three local models for each of thetwo periods. The variance inflation factor (VIF) was used in all models to test for multi-collinearity in the covariates. VIF assesses whether the variance of one explanatory variableis independent from all other explanatory variables (Chatterjee, Hadi, and Price 2000). Theonly variable that had to be deleted from the subgroup estimations (but not the global model)was the Y coordinate, which was positively and significantly correlated with terrain rough-ness. No serious multicollinearity was detected for all other variables with a mean VIF below1.5 and the largest VIF below 10. Several ad hoc methods to correct for spatial effects wereemployed: the Y coordinate (Section 4.3) in the global regressions, the coding scheme(Section 5.1), and the roughness variable, which was included as a spatial average in anine-by-nine rectangular window (Nelson, Harris, and Stone 2001).

The degree to which the statistical estimations explain the occurrence of land abandon-ment was assessed using five goodness-of-fit statistics: First, McFadden’s adjusted R2 (adj.R2) was used to assess the regression fit. Adj. R2 is a commonmeasure in logistic regressionsthat accounts for the number of parameters included (Ben-Akiva and Lerman 1985). Second,the percentage of correctly predicted observations (PC) was calculated from the predictedprobabilities by assuming the highest predicted probability as the predicted value. Third, thearea under the curve (AUC) of the receiver operating characteristics (ROC) was estimated



using a nonparametric ROC curve (Metz 1978). The AUC measures the share of thecorrectly predicted positive (abandoned) from all predicted positive values against theshare of the correctly predicted negative (stable cropland) against all negative values. TheAUC is derived by varying the probability threshold, which results in an AUC between 0.5for a random map and 1.0 for a perfect prediction (Pontius and Schneider 2001). Fourth, theKappa statistic quantifies the locational accuracy between predicted and observed datacompared to a random map (Cohen 1960). A perfect prediction yields Kappa = 1 andKappa approaches 0 if correct predictions are equal to those of a random map. Last, wecalculated the Bayesian Information Criterion (BIC), which compares subgroup models byjudging the distance between the fitted model and the observed data based on the log-likelihood function (Schwarz 1978). The more negative the BIC, the better the model fit.

5. Results

The regression results suggested that the determinants of cropland abandonment variedconsiderably for the elevation subgroups in significance and strengths of the influences.Globally, the density of cropland and topography were important determinants while marketaccess and farm fragmentation emerged as driving factors of cropland abandonment in thehilly and mountainous areas.

5.1. Descriptive statistics of covariates

The concentration of cropland in neighboring locations at the beginning of the respectiveperiod showed the strongest relationship with cropland abandonment (Figures 4 and 5),suggesting that abandonment was more likely to occur where cropland was less homoge-nous. The extent of abandonment was higher further north (lower Y coordinate) and atlocations with rougher terrain in the first and second period. In the second period, abandon-ment was associated with proximity to roads, which was indicated by lower medians andlower ranges of the road distance variables. Fewer tractors per commune in both periods anda larger share of individually owned cropland in the second period appeared to be connectedwith the occurrence of cropland abandonment. No clear relationship was found for the otherindependent variables.

5.2. Determinants of cropland abandonment

Goodness-of-fit statistics for the eight logistic regression models are summarized in Table 2.The two global models displayed excellent fit with AUC of 0.84 (81% correct predictions)and 0.89 (79%) for the periods of 1990–1995 and 1995–2005, respectively. Kappa was goodwith 0.48 and 0.43. The model fit was somewhat lower for the three elevation subgroups,especially for the first period in the plain and mountainous areas. The models performedreasonably well for hilly areas in both periods and considerably better for the plains in thesecond period. Models for the mountainous areas showed lower goodness-of-fit in bothperiods with AUC of 0.7.

To facilitate the readability of the results (in total, more than 100 coefficients plusassociated statistics), we omitted the tables with coefficients, standard errors, and p-values.1

Instead, we chose a graphical description of the estimation results. Figure 6 visualizes theodds ratios, their 95% confidence intervals, and p-values in a compact form, with one graphfor each independent variable. Each plot in Figure 6 contains the global and subgroup



0

5

No Yes

Y coordinate

0

20

40

No Yes

Roughness

0

200

400

No Yes

Distance to European road

0

100

200

No Yes

Distance to main road

0

20

40

No Yes

Distance to county road

0

5

10

No Yes

Cropland in neighbourhood, 1990

0

10

20

No Yes

Population (interpolated), 1989

0

50

No Yes

Household density, 1989

0

50

100

No Yes

Emigration, 1989–1996

0

50

100

No Yes

Livestock density, 1989

0

50

100

150

No Yes

Tractors, 1989

Figure 4. Box plots of independent variables by cropland abandonment, 1990–1995.Note: Yes = abandoned; no = stable cropland. The box plots omit outliers, that is, observations below (above) 1.5times the interquartile range of the first (third) quartile. See Table 1 for the units of the variables. For the sake ofbrevity, the box plots are not shown for the elevation subgroups, but can be obtained from the authors upon request.Source: Authors.



0

5

No Yes

Y coordinate

0102030

No Yes

Roughness

0

200

400

No Yes

Distance to European road

0

100

200

No Yes

Distance to main road

0

20

40

No Yes

Distance to county road

0

5

10

No Yes

Cropland in neighbourhood, 1995

0

10

20

No Yes

Population (interpolated), 1996

0204060

No Yes

Household density, 1996

0

50

100

No Yes

Emigration, 1996–2004

0204060

No Yes

Temporary emigration, 1996

0

20

40

No Yes

Livestock density, 1996

0

50

100

No Yes

Tractors, 1996

406080

100

No Yes

Cropland used individually, 1996

2468

No Yes

Parcels, 1996

Figure 5. Box plots of independent variables by cropland abandonment, 1995–2005.Source: Authors.



regressions for both periods to facilitate comparability of models. Coefficients with sig-nificance levels below 20% were omitted.

The concentration of cropland in neighbouring locations had the largest quantitativeimpact on the occurrence of abandonment. In the global model, one additional pixel ofcropland in the immediate neighbourhood decreased the odds of abandonment by 19%(equivalent to an odds ratio of 0.81) in the first period and by 28% in the second period.Spatially homogenous cropland patterns were particularly important in the plains, where anadditional cropland pixel in the neighbourhood decreased the likelihood of abandonment by34% in both periods. In the hilly area, the decrease was about 17% in both periods, whereasin mountains the decrease in the odds ratio was considerably lower.

Terrain roughness had a significant influence on cropland abandonment in most modelswith the expected positive sign. While the effect was small in the mountainous subgroupduring the first period and vanished in the second, terrain undulation was associated withincreased abandonment in areas of low andmedium elevation. The global model masked thiselevation-specific effect.

The effects of distance to the European and main roads were insignificant or small inboth periods. But pixels located farther away from county roads were more likely to becomeabandoned in the hilly and mountainous subgroup in both periods. This effect was largestin the mountains and between 1995 and 2005, when the probability of abandonmentincreased by 7% every 100m further away from a local road. Road access did not substantiallyaffect abandonment in the plains. At the county level, local variations of road influenceaveraged out.

Our models did not reveal strong quantitative evidence for a relationship of croplandabandonment with population and household density or migration. Higher populationdensity counteracted cropland abandonment for the entire study area in the second period,but was insignificant for the elevation subgroups. Household density made abandonmentless probable in the first period and more probable in the second. Permanent and temporaryemigration had a small negative influence in the mountainous subgroup during the secondperiod.

Globally, a higher density of livestock decreased the likelihood of abandonment in thesecond period. Again, the global results masked subgroup variations. For example, in thehilly areas a higher livestock density was associated with higher abandonment in both

Table 2. Goodness-of-fit.

PeriodElevationsubgroup Adj. R2 PC AUC Kappa BIC

1990–1995 Plain 0.047 0.519 0.667 0.038 -5,5331990–1995 Hilly 0.124 0.677 0.746 0.354 -5,6391990–1995 Mountainous 0.079 0.529 0.711 0.058 -5,576

1995–2005 Plain 0.250 0.620 0.828 0.240 -5,8001995–2005 Hilly 0.092 0.657 0.720 0.314 -5,5801995–2005 Mountainous 0.064 0.596 0.696 0.185 -5,430

1990–1995 Global 0.294 0.741 0.844 0.481 -13,1771995–2005 Global 0.369 0.714 0.892 0.427 -13,368

Note: The global (subgroup) estimations are based on 2000 (1000) observations with 1000 (500) observations from thepresent and absent group, respectively (Section 5.1).Source: Authors.



Figure 6. Odds ratios, p-values, and confidence intervals for the logistic regressions.Source: Authors.

Journalof

Land

Use

Science123


periods. Finally, higher farm fragmentation within a commune positively influenced thelikelihood of cropland abandonment in the mountainous region, where the regression resultssuggest that the ownership of an additional parcel per farmer increased the odds of abandon-ment by 28%. Farm fragmentation had nomeasurable effect on cropland abandonment in thehilly and plain areas.

6. Discussion

The study region underwent drastic changes in livelihood strategies, which shaped land useand land cover during the post-socialist period. Cropland abandonment was widespread inArges County and most of this abandonment occurred at the onset of transition. Croplandbecame more stable between 1995 and 2005, but abandonment proceeded at high rates insome areas. Cropland abandonment differed notably by topographic conditions. The highestabsolute decrease was observed in hilly areas, whereas the highest abandonment rate wasfound in the mountains. We identified topographic variables, the spatial homogeneity ofcropland use, and accessibility as the most important determinants of this process. However,the influence of these drivers differed considerably among elevation subgroups in signifi-cance and strength of influence.

The logistic regression models for the entire county fitted the data very well. Theexplanatory power of the models for the elevation subgroups was lower, particularly in themountainous areas in both periods and in the plains in the first period. Reasons for the lowerfit were possibly the small variation of commune-level data in the subgroup models. Inaddition, only half as many observations entered the regressions in the subgroup models.Nevertheless, the subgroup results emphasized that great care is required when interpretingand inferring from the results of one regression model for a study region as large and diverseas Arges. Global coefficients masked important subregional variations and only disaggre-gating the study area along a topographic gradient revealed remarkable local variations in thedeterminants of cropland abandonment.

For example, more isolated patches of cropland were more likely to become abandoned,but this effect was three times stronger in the plains compared to the hilly or mountainoussubgroups in both periods. This result may in part reflect the patchiness of the agriculturallandscape in the mountainous areas, where many of the small and isolated plots remained incultivation, albeit at low cultivation intensities. Cropland in the plains had a larger tendencyto become more homogenous over time. The higher commercial orientation of farmers andthe higher share of legal and family associations in the plains may have contributed to thistrend toward more consolidated land-use pattern in the plains.

Cultivation in the plain and hilly areas was significantly more likely on flat terrain,indicating the importance of topographic variation in agricultural production. In the moun-tains, terrain roughness had no discernible effects; rather, adverse road access and higherfarm fragmentation contributed to cropland abandonment. Higher transportation costs mayhave reduced the profitability of farming to the point where farmers gave up cultivation onremote plots. Higher farm fragmentation negatively affected the change in cropland area inthe mountainous areas. We assume that the higher fragmentation may have become a criticalfactor for cropland use for the low-intensity, semi-subsistence farming systems in themountains, because it increased labour and management requirements. Higher input require-ments may have induced farmers to stop cultivating on marginal plots.

Population and migration proxies were weaker predictors of cropland abandonment thanexpected. Several reasons may explain their small or insignificant influence. First, theinsecure status of many emigrants may have prompted people to keep land in cultivation



in order to reduce the risks involved in migration. In addition, land rentals are sometimespursued on a yearly basis for a small return in cash or kind (own data). The effect ofemigration on abandonment in the southern plains may have been further absorbed bylegal associations or state farms that rent in land on a year-by-year basis from labour-scarcefamilies. Second, a great deal of domestic migration was temporary and many emigrantsreturn for peak work periods to help those who have stayed behind. And occasionally, themigrants’ remittances may serve to hire in additional labour. Third, our data do not capturethe age structure of farming households. Many are in retirement age and may soon dis-continue farming, while the younger generations attempt to find work elsewhere and areunlikely to return (Cartwright 2003).

From a macroeconomic perspective, we suggest twomajor trends contributed to the highamount of cropland abandonment in Arges. First, the ratio of output to input prices wasdeteriorating in agriculture during the post-socialist transition (Rozelle and Swinnen 2004).This development is apparent in Romania for example, in the decreasing share of agriculturein the gross domestic product from 24% in 1990 to 10% in 2005 (World Bank 2007).Concurrently, the opportunity costs of agricultural labour were rising across Eastern Europe,which encouraged part of the workforce to move away from agricultural employment towardother economic sectors, similar to processes observed inWestern Europe in the past (Strijker2005). But, contrary to other East European countries, agricultural employment in Romaniadid not decrease substantially after 1989 and there was less emigration from rural areas(World Bank 2007). Yet official statistics may not capture several important demographicprocesses that affect cropland use such as the loss of younger population segments, rur-al–urban commuting, and seasonal migration (Dorondel 2007; Kuemmerle et al. 2009).Commuters and temporary emigrants often continue to be registered in rural areas, thusresulting in biased agricultural workforce estimates.

A limited number of income opportunities began to emerge in rural tourism in thenorthern part of the study area along the foothills of the Carpathians, where the amount oftourist accommodation has increased significantly (NIS 2004; Dorondel 2007). Low-inputfarming may continue in such areas, as it is an essential element of the landscape that attractsvisitors and generates income. Such monetary incentives may have been one factor in theslowdown of abandonment between 1995 and 2005 in the mountainous subgroup. The plainareas in the south possess more suitable natural conditions and better market access, both ofwhich have benefited the development of profitable farming. Farming structures includelarger management units, which possibly reduced abandonment. Farming may be under thehighest threat in the hilly areas where the economically active are relocating to the capitalPitesti. Farming in the hilly areas is further constrained by difficult terrain conditions andfarm structures dominated by small-scale holdings. Abandonment is likely to continue athigh rates in the hilly areas without outside interventions.

We conjecture that Romania may have entered a period of accelerated forest transition.Whereas socialism may have slowed down forest expansion because of government supportfor agricultural production and the largely political decision to provide rural employment,Romania may now start following the economic development pathway of the forest transi-tion (Rudel et al. 2005). Indeed, some of the abandoned cropland has already experiencedsecondary forest succession (Kuemmerle et al. 2009).

Our analysis can be extended in several ways. First, the inclusion of spatially explicitmeasures of cultivation intensities may yield valuable additional insights into more subtleland-use modifications, as well as the share of farmers’ market participation. Second,household-level data may better capture socioeconomic and demographic trends in Arges.For example, household data would provide information on changes in the age structure of



rural households, thereby facilitating the further exploration of emigration’s influence on thelabour force. Third, grazing activities were only indirectly incorporated because areas usedfor fodder production were included in the cropland category. But separating pastures,grassland, and abandoned areas is difficult due to spectral collinearity among these classes.Spatial information on livestock distribution, though challenging to gather, may add to abetter understanding of changes in pasture usage during transition.

AcknowledgmentsWe thank the editors Anette Reenberg and Lars Jorgensen for putting this special issue together. DarlaMunroe, Mario Gellrich, Jim Curtiss, and two anonymous reviewers provided valuable and construc-tive comments. Funding from the Deutsche Forschungsgemeinschaft (DFG) under the EmmyNoether-Programme is gratefully acknowledged.

Note1. The tables with the estimation results can be obtained from the authors upon request.

ReferencesAngelstam, P., Boresjo-Bronge, L., Mikusinski, G., Sporrong, U., and Wastfelt, A. (2003), ‘‘Assessing

Village Authenticity with Satellite Images: A Method to Identify Intact Cultural Landscapes inEurope,’’ Ambio, 32, 594–604.

Anselin, L. (2002), ‘‘Under the Hood: Issues in the Specification and Interpretation of SpatialRegression Models,’’ Agricultural Economics, 27, 247–267.

Baldock, D., Beaufoy, G., Brouwer, F., and Godeschalk, F. (1996), Farming at the Margins:Abandonment or Redeployment of Agricultural Land in Europe, London/The Hague: Institutefor European Environmental Policy.

Bauer, M.E., Burk, T.E., Ek, A.R., Coppin, P.R., Lime, S.D., Walsh, T.A., Walters, D.K., Befort, W.,and Heinzen, D.F. (1994), ‘‘Satellite Inventory of Minnesota Forest Resources,’’ PhotogrammetricEngineering & Remote Sensing, 60, 287–298.

Ben-Akiva, M.E., and Lerman, S.R. (1985), Discrete Choice Analysis: Theory and Application toTravel Demand, Cambridge, Massachusetts: Massachusetts Institute of Technology.

Besag, J.E. (1974), ‘‘Spatial Interaction and the Statistical Analysis of Lattice Systems,’’ Journal of theRoyal Statistical Society B, 36, 192–236.

Bicik, I., Jelecek, L., and Stepanek, V. (2001), ‘‘Land-Use Changes and Their Social Driving Forces inCzechia in the 19th and 20th Centuries,’’ Land Use Policy, 18, 65–73.

Brouwer, F., Baldock, D., and La Chapelle, C. (2001), Agriculture and Nature Conservation in theCandidate Countries: Perspectives in Interaction, The Netherlands: Ministry of Agriculture,Nature Management and Fisheries.

Cartwright, A. (2003), ‘‘Private Farming in Romania: What are Old People Going to Do with TheirLand?,’’ in The Postsocialist Agrarian Question: Property Relations and the Rural Condition, eds.C. Hann and Munster: The ‘‘Property Relations’’ Group, LIT Verlag, pp. 171–188.

Chatterjee, S., Hadi, A.S., and Price, B. (2000),Regression Analysis by Example, NewYork: JohnWiley&Sons.

Chilonda, P., and Otte, J. (2006) , ‘‘Indicators to Monitor Trends in Livestock Production at National,Regional and International Levels,’’ Livestock Research for Rural Development, Vol. 18, Article117. Retrieved 22 November, 2006: http://www.cipav.org.co/lrrd/lrrd18/8/chill8117.htm.

Chirca, C., and Tesliuc, E.D. (1999), From Rural Poverty to Rural Development. Bucharest: WorldBank and National Commission for Statistics.

Cohen, J. (1960), ‘‘A Coefficient of Agreement for Nominal Scales,’’ Educational & PsychologicalMeasurement, 20, 37–46.

Coppin, P., Jonckheere, I., Nackaerts, K., Muys, B., and Lambin E. F. (2004), ‘‘Digital ChangeDetection Methods in Ecosystem Monitoring: A Review,’’ International Journal of RemoteSensing, 25, 1565–1596.



Cremene, C., Groza, G., Rakosy, L., Schileyko, A.A., Baur, A., Erhardt, A., and Baur, B. (2005),‘‘Alterations of Steppe-Like Grasslands in Eastern Europe: A Threat to Regional BiodiversityHotspots,’’ Conservation Biology, 19, 1606–1618.

Dawidson, K. (2005), ‘‘Redistribution of Land in Post-Communist Romania,’’ Eurasian Geographyand Economics, 46, 618–632.

Dorondel, S. (2007), ‘‘Agrarian Transformation, Social Differentiation, and Land Use Change inPostsocialist Romania,’’ Doctoral dissertation. Humboldt-Universitat zu Berlin.

FAO (2006), ‘‘FAOSTAT Data,’’ Food and Agriculture Organization [online]. Available from: http://faostat.fao.org/[Accessed 5 December 2006].

Foley, J.A., Defries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., Chapin, F.S., Coe, M.T.,Daily, G.C., Gibbs, H.K., Helkowski, J.H., Holloway, T., Howard, E.A., Kucharik, C.J., Monfreda,C., Patz, J.A., Prentice, I.C., Ramankutty, N., and Snyder, P.K. (2005), ‘‘Global Consequences ofLand Use,’’ Science, 309, 570–574.

Froot, K.A. (1989), ‘‘Consistent Covariance Matrix Estimation with Cross-Sectional Dependence andHeteroskedasticity in Financial Data,’’ The Journal of Financial and Quantitative Analysis, 24,333–355.

Gellrich, M., Baur, P., Koch, B., and Zimmermann, N.E. (2007), ‘‘Agricultural Land Abandonment andNatural Forest Re-Growth in the Swiss Mountains: A Spatially Explicit Economic Analysis,’’Agriculture, Ecosystems & Environment, 118, 93–108.

Geoghegan, J., Villar, S.C., Klepeis, P., Mendoza, P.M., Ogneva-Himmelberger, Y., Chowdhury, R.R.,Turner, I.B.L., and Vance, C. (2001), ‘‘Modeling Tropical Deforestation in the Southern YucatanPeninsular Region: Comparing Survey and Satellite Data,’’ Agriculture, Ecosystems& Environment,85, 25–46.

Geoghegan, J., Waininger, L., and Bockstael, N.E. (1997), ‘‘Spatial Landscape Indices in a HedonicFramework: An Ecological Economics Analysis Using GIS,’’ Ecological Economics, 23,251–264.

Government Service for Land and Water Management (2005), ‘‘Land Abandonment, Biodiversity andthe CAP in Relation to the 1st and 2nd Pillars of the EU’s Common Agricultural Policy,’’ Outcomeof an International Seminar in Sigulda, , Latvia, 7–8 October, 2004. Utrecht: DLG.

Groves, R.M. (1989), Survey Errors and Survey Costs, New York: Wiley & Sons.Gutierrez, R., and Drukker, D.M. (2005), Reference for Cluster-Correlated Robust Variance

Calculation [online]. StataCorp. Available from: http://www.stata.com/support/faqs/stat/robust_ref.html [Accessed 8 May 2006].

Hansen, M.C., Stehman, S.V., Potapov, P.V., Loveland, T.R., Townshend, J.R.G., Defries, R.S.,Pittman, K.W., Arunarwati, B., Stolle, F., Steininger, M.K., Carroll, M., and Dimiceli, C.(2008), ‘‘Humid Tropical Forest Clearing from 2000 to 2005 Quantified by UsingMultitemporal and Multiresolution Remotely Sensed Data,’’ Proceedings of the NationalAcademy of Sciences, 105, 9439–9444.

Hunsaker, C.T., and Levine, D.A. (1995), ‘‘Hierarchical Approaches to the Study of Water Quality inRivers,’’ Bioscience, 45, 193–203.

Ioffe, G., Nefedova, T., and Zaslavsky, I. (2004), ‘‘From Spatial Continuity to Fragmentation: The Caseof Russian Farming,’’ Annals of the Association of American Geographers, 94, 913–943.

Kareiva, P., Watts, S., McDonald, R., and Boucher, T. (2007), ‘‘Domesticated Nature: ShapingLandscapes and Ecosystems for Human Welfare,’’ Science, 316, 1866–1869.

Kozak, J., Estreguil, C., and Troll, M. (2007), ‘‘Forest Cover Changes in the Northern Carpathians inthe 20th Century: A Slow Transition,’’ Journal of Land Use Science, 2, 127–146.

Kuemmerle, T., Hostert, P., Radeloff, V., Van Der Linden, S., Perzanowski, K., and Kruhlov, I. (2008),‘‘Cross-Border Comparison of Post-Socialist Farmland Abandonment in the Carpathians,’’Ecosystems, 11, 614–628.

Kuemmerle, T., Muller, D., Griffiths, P., and Rusu,M. (2009), ‘‘Land Use Change in Romania After theCollapse of Socialism,’’ Regional Environmental Change, http://dx.doi.org/10.1007/s10113-008-0050-z.

Kuemmerle, T., Radeloff, V.C., Perzanowski, K., and Hostert, P. (2006), ‘‘Cross-Border Comparison ofLand Cover and Landscape Pattern in Eastern Europe Using a Hybrid Classification Technique,’’Remote Sensing of Environment, 103, 449–464.

Lepers, E., Lambin, E.F., Janetos, A.C., Defries, R., Achard, F., Ramankutty, N., and Scholes, R.J.(2005), ‘‘A Synthesis of Information on Rapid Land-Cover Change for the Period 1981–2000,’’BioScience, 55, 115–124.



http://www.stata.com/support/faqs/stat/robust_ref.html

http://www.stata.com/support/faqs/stat/robust_ref.html

Lerman, Z., Csaki, C., and Feder, G. (2004), Agriculture in Transition: Land Policies and EvolvingFarm Structures in Post-Soviet Countries, Lanham, Maryland: Lexington Books.

Liu, J., Daily, G.C., Ehrlich, P.R., and Luck, G.W. (2003), ‘‘Effects of Household Dynamics onResource Consumption and Biodiversity,’’ Nature, 421, 530–533.

MacDonald, D., Crabtree, J.R., Wiesinger, G., Dax, T., Stamou, N., Fleury, P., Gutierrez Lazpita, J.,and Gibon, A. (2000), ‘‘Agricultural Abandonment in Mountain Areas of Europe: EnvironmentalConsequences and Policy Response,’’ Journal of Environmental Management, 59, 47–69.

Maddala, G.S. (2001), Introduction to Econometrics, Chichester: John Wiley & Sons.Mather, A.S., and Needle, C.L. (1998), ‘‘The Forest Transition: A Theoretical Basis,’’ Area, 30,

117–124.Mathijs, E., and Noev, N. (2004), ‘‘Subsistence Farming in Central and Eastern Europe: Empirical

Evidence fromAlbania, Bulgaria, Hungary, andRomania,’’Eastern European Economics, 42, 72–89.Mathijs, E., and Swinnen, J.F.M. (1998), ‘‘The Economics of Agricultural Decollectivization in East

Central Europe and the Former Soviet Union,’’ Economic Development and Cultural Change, 47,1–26.

Metz, C.E. (1978), ‘‘Basic Principles of ROC Analysis,’’ Seminars in Nuclear Medcine, 8, 283–298.Muller, D., and Munroe, D.K. (2005), ‘‘Tradeoffs between Rural Development Policies and Forest

Protection: Spatially Explicit Modeling in the Central Highlands of Vietnam,’’ Land Economics,81, 412–425.

Muller D., and Munroe, D.K. (2008), ‘‘Changing Rural Landscapes in Albania: CroplandAbandonment and Forest Clearing in the Postsocialist Transition,’’ Annals of the Association ofAmerican Geographers, 98, 855–876.

Muller, D., and Sikor, T. (2006), ‘‘Effects of Postsocialist Reforms on Land Cover and Land Use inSouth-Eastern Albania,’’ Applied Geography, 26, 175–191.

Muller, D., and Zeller, M. (2002), ‘‘Land Use Dynamics in the Central Highlands of Vietnam: A SpatialModel Combining Village Survey Data with Satellite Imagery Interpretation,’’ AgriculturalEconomics, 27, 333–354.

Munroe, D.K., and Muller, D. (2007), ‘‘Issues in Spatially Explicit Statistical Land-Use/Cover Change(LUCC)Models: Examples fromWestern Honduras and the Central Highlands of Vietnam,’’ LandUse Policy, 24, 521–530.

Nelson, G.C., Harris, V., and Stone, S.W. (2001), ‘‘Deforestation, Land Use, and Property Rights:Empirical Evidence from Darie�n, Panama,’’ Land Economics, 77, 187–205.

Nelson, G.C., and Hellerstein, D. (1997), ‘‘Do Roads Cause Deforestation? Using Satellite Images inEconometric Analysis of Land Use,’’ American Journal of Agricultural Economics, 79, 80–88.

Nikodemus, O., Bell, S., Grine, I., and Liepins, I. (2005), ‘‘The Impact of Economic, Social andPolitical Factors on the Landscape Structure of the Vidzeme Uplands in Latvia,’’ Landscape andUrban Planning, 70, 57–67.

NIS (2004), Database of Localities, Bucharest: National Institute of Statistics.Osborne, P.E., and Suarez-Seoane, S. (2002), ‘‘Should Data be Partitioned Spatially Before Building

Large-Scale Distribution Models?,’’ Ecological Modelling, 157, 249–259.Overmars, K.P., and Verburg, P.H. (2006), ‘‘Multilevel Modelling of Land Use from Field to Village

Level in the Philippines,’’ Agricultural Systems, 89, 435–456.Palang, H., Printsmann, A., Gyuro, E.K., Urbanc, M., Skowronek, E., and Woloszyn, W. (2006),

‘‘The Forgotten Rural Landscapes of Central and Eastern Europe,’’ Landscape Ecology, 21,347–357.

Perz, S.G., and Skole, D.L. (2003), ‘‘Social Determinants of Secondary Forests in the BrazilianAmazon,’’ Social Science Research, 32, 25–60.

Peterson, U., and Aunap, R. (1998), ‘‘Changes in Agricultural Land Use in Estonia in the 1990sDetected with Multitemporal Landsat MSS Imagery,’’ Landscape and Urban Planning, 41,193–201.

Pontius, R.G.J., and Schneider, L.C. (2001), ‘‘Land-Cover Change Model Validation by an ROCMethod for the Ipswich Watershed, Massachusetts, USA,’’ Agriculture, Ecosystems &Environment, 85, 239–248.

Rozelle, S., and Swinnen, J.F.M. (2004), ‘‘Success and Failure of Reform: Insights from the Transitionof Agriculture,’’ Journal of Economic Literature, 42, 404–456.

Rudel, T.K., Coomes, O.T., Moran, E., Achard, F., Angelsen, A., Xu, J., and Lambin, E. (2005),‘‘Forest Transitions: Towards a Global Understanding of Land Use Change,’’ GlobalEnvironmental Change Part A, 15, 23–31.



Schwarz, G. (1978), ‘‘Estimating the Dimension of a Model,’’ The Annals of Statistics, 6, 461–464.Serneels, S., and Lambin, E.F. (2001), ‘‘Proximate Causes of Land-Use Change in Narok District,

Kenya: A Spatial Statistical Model,’’ Agriculture, Ecosystems & Environment, 85, 65–81.Silver, W.L., Ostertag, R., and Lugo, A.E. (2000), ‘‘The Potential for Carbon Sequestration Through

Reforestation of Abandoned Tropical Agricultural and Pasture Lands,’’ Restoration Ecology, 8,394–407.

Slater, J.A., Garvey, G., Johnston, C., Haase, J., Heady, B., Kroenung, G., and Little, J. (2006), ‘‘TheSRTMData ‘Finishing’ Process and Products,’’ Photogrammetric Engineering & Remote Sensing,72, 237–247.

Strijker, D. (2005), ‘‘Marginal Lands in Europe: Causes of Decline,’’ Basic and Applied Ecology, 6,99–106.

Swinnen, J.F.M., Vranken, L., and Stanley, V. (2006), Emerging Challenges of Land Rental Markets: AReview of the Available Evidence for Central and Eastern Europe and Former Soviet UnionCountries, Washington, D.C.: The World Bank.

Tasser, E., Mader, M., and Tappeiner, U. (2003), ‘‘Effects of Land Use in Alpine Grasslands on theProbability of Landslides,’’ Basic and Applied Ecology, 4, 271–280.

Tasser, E., Walde, J., Tappeiner, U., Teutsch, A., and Noggler, W. (2007), ‘‘Land-Use Changes andNatural Reforestation in the Eastern Central Alps,’’ Agriculture, Ecosystems & Environment, 118,115–129.

Van Dijk, T. (2003), ‘‘Scenarios of Central European Land Fragmentation,’’ Land Use Policy, 20,149–158.

Verburg, P.H., Schulp, C.J.E., Witte, N., and Veldkamp, A. (2006), ‘‘Downscaling of Land Use ChangeScenarios to Assess the Dynamics of European Landscapes,’’ Agriculture, Ecosystems &Environment, 114, 39–56.

World Bank (2007), World Development Indicators 2007, Washington, D.C.: The World Bank.Yeloff, D., and VanGeel, B. (2007), ‘‘Abandonment of Farmland and Vegetation Succession Following

the Eurasian Plague Pandemic of Ad 1347–52,’’ Journal of Biogeography, 34, 575–582.



lost in transition: determinants of post-socialist ...muelleda/download/mueller...

Documents