Effects of post-WWII forced displacements on long-term landscape dynamics in the Polish Carpathians

Landscape and Urban Planning 214 (2021) 104164

Available online 17 June 2021

Research Paper

Effects of post-WWII forced displacements on long-term landscape

dynamics in the Polish Carpathians

Andrzej N. Affek

, Jacek Wolski

, Maria Zachwatowicz

, Krzysztof Ostan

, Volker

C. Radeloff

Institute of Geography and Spatial Organization, Polish Academy of Sciences, Twarda 51/55, 00-818 Warsaw, Poland

SILVIS Lab, Department of Forest and Wildlife Ecology, University of Wisconsin-Madison, 1630 Linden Drive, Madison, WI 53706, USA

Faculty of Geography and Regional Studies, University of Warsaw, Krakowskie Przedmie

scie 30, 00-927 Warsaw, Poland

Institute of Geography and Spatial Management, Jagiellonian University, Gronostajowa 7, 30-387 Cracow, Poland

HIGHLIGHTS

• Post-WWII forced displacement caused major forest increase in Polish Carpathians.

• 115k of 181k (63%) of forest increase until 1970 due to displacement.

• Land-use regime switched from agricultural to forest-dominated stable state.

• Displacement caused more forest increase than post-socialist abandonment.

• Displacement areas now one of the largest wilderness areas in Central Europe.

ARTICLE INFO

Keywords:

Resettlements

Land abandonment

Land-use change

Regime shift

Forest area

World War II

ABSTRACT

Armed conicts and major political changes can result in the forced displacement of thousands of people and

may have substantial effects on the environment. However, it is difcult to predict and mitigate long-term

consequences of such displacements, especially when they trigger abrupt land-use changes that result in a

regime shift of the land-use system. Our main goal was to determine the effects of post-WWII forced displace-

ments on long-term landscape dynamics in the Polish Carpathians. After World War II, 630,000 Ukrainians were

forcibly displaced from southeastern Poland, leading to permanent depopulation of mountain borderlands. We

conducted a village-level analysis of forest area change across the Polish Carpathians (1685 villages/cadastral

communities), and a detailed analyses of landscape change and land-cover trajectories in two highly depopulated

test sites. Our source data were pre-war (1850s–1860s and 1930s) and post-war (1970s and 2010s) census data

and topographic maps. We found a substantial forest area increase after displacements, far outpacing the widely

reported forest increase due to the collapse of socialism in early 1990s, and a striking landscape simplication.

Astonishingly, almost two thirds of the post-war (1930s–1970s) forest area increase in the entire Polish Car-

pathians (115,000 ha out of 181,000 ha) was due to the forced displacements. The land-use regimes shifted from

being agriculturally-dominated to being forest-dominated, and approached a stable alternative state. As a result,

a once densely populated rural region has become one of the largest ‘wilderness’ areas in Central Europe, with

vast areas void of human settlements and resurgent wildlife populations. This highlights that forced displace-

ments, which are common during and after armed conicts, can have substantial and long-lasting effects on land

use.

* Corresponding author at: Institute of Geography and Spatial Organization, Polish Academy of Sciences, Twarda 51/55, 00-818 Warsaw, Poland.

E-mail addresses: [email protected] (A.N. Affek), [email protected] (J. Wolski), [email protected] (M. Zachwatowicz), krzysztof.ostan@uj.

edu.pl (K. Ostan), [email protected] (V.C. Radeloff).

Contents lists available at ScienceDirect

Landscape and Urban Planning

journal homepage: www.elsevier.com/locate/landurbplan

https://doi.org/10.1016/j.landurbplan.2021.104164

Received 27 July 2020; Received in revised form 1 June 2021; Accepted 2 June 2021

Landscape and Urban Planning 214 (2021) 104164

1. Introduction

About 41.3 million people worldwide were displaced as of 2019 due

to conicts and violence, with 10.8 million new displacements in 2018

only (IDMC, 2019). Displacement substantially affects the well-being

and socio-cultural identity of people and their relation with land (Pis-

korski, 2011). Furthermore, displacements can alter land use patterns

(Woube, 2005). There is strong evidence that environmental changes

can cause human migrations (see e.g., Abel, Brottrager, Crespo Cuar-

esma, & Muttarak, 2019). Conversely, the permanent displacement of

large populations can also have strong effects on the environment

(Ghimire, 1994; Kim, 1997; Ssekandi, Mburu, Wasonga, Macopiyo, &

Charles, 2017), both in the area where people settled, that is, the target

region, and in the area they left (Gorsevski, Kasischke, Dempewolf,

Loboda, & Grossmann, 2012).

Typically, displacements reduce pressures on the land left behind (e.

g., Baumann, Radeloff, Avedian, & Kuemmerle, 2015; Witmer, 2008;

Yin et al., 2019), but increase pressure in the area where people move.

For example, displacement can cause deforestation for agricultural and

settlement purposes (Ghimire, 1994; Hugo, 1996). The most common

land-use effects in the place of origin are agricultural land abandonment

(Baumann et al., 2015; Witmer, 2008; Yin et al., 2019), bush and forest

regrowth (Landholm, Pradhan, & Kropp, 2019), and sometimes the

complete cessation of any kind of human activity (Kim, 1997). There are

cases though of land use intensication following displacements,

particularly when armed groups use agriculture as an income source

(Eklund, Degerald, Brandt, Prishchepov, & Pilesj

o, 2017; Landholm

et al., 2019) or civilians expand agriculture due to food insecurity (Alix-

Garcia, Bartlett, & Saah, 2013). The exact post-displacement land-cover

trajectories depend on both the magnitude of displacement and on

subsequent population movements, both of which are determined by

political and economic conditions (Baumann et al., 2015; Witmer, 2008;

Yin et al., 2019). However, most of the studies that examine the effects

of armed conicts on land use do not explicitly address the effects of

displacements but rather the overall effects of the conicts, and most are

focused on immediate effects. Less is known about displacements’ long-

lasting consequences, especially their environmental legacies and if the

resulting land-use changes are reversible or not.

Displacements can be considered through the lens of systems theory

as a trigger causing an abrupt change of the social-ecological system

(Ramankutty & Coomes, 2016; Walker et al., 2006). Such an abrupt

change may be reversible or not, depending on the strength of the self-

reinforcing processes. If the abrupt change is irreversible, it results in a

shift of the land-use regime and the establishment of an alternative

stable state (Müller et al., 2014). However, achieving fully stable states

in case of complex social-ecological systems is rare due to intrinsic

stochasticity and the multimodal, open character of these systems

(Ramankutty & Coomes, 2016; Walker et al., 2006). Irrespective of how

stable the alternative states are, there is a need to understand both the

potential risks and the consequences of regime shifts (Biggs, Peterson, &

Rocha, 2018). Insight into system dynamics before and after forced

displacements could help identify whether such displacement can

trigger abrupt changes and shifts toward alternative stable states. More

broadly, it could help recognize how vulnerable systems are to external

triggers and link land-use regime shifts with land cover changes (Ram-

ankutty & Coomes, 2016). Historical land-use regime shifts related to

displacement may thus complement the existing Regime Shifts Database

(Biggs et al., 2018) with abrupt systemic changes that are not yet rep-

resented there, thereby increasing the strength of meta-analyses of

current and future land-use regime shifts.

Studying the effects of post-war displacements, and capturing past

human-environment interactions, may shed new light on the possible

land-use effects of present day forced displacement, and on the effects of

other causes of migrations such as poverty, natural disasters, deserti-

cation, and climate change (MacDonald et al., 2000; Ssekandi et al.,

2017). A powerful example of such a legacy is World War II because of

its great magnitude, strong effects on land system, and widespread post-

war displacements (Machlis & Hanson, 2008).

In Central and Eastern Europe millions of people were forcibly dis-

placed after World War II (Rieber, 2000). These displacements were

mostly caused by post-war border and political shifts, policies aimed to

create ethnically homogenous nation-states, and – in the case of German

communities – by the notion of their collective responsibility for Nazi

war crimes (Eberhardt, 2011; Rieber, 2000). In many European regions,

particularly in the rural and marginalized mountainous borderlands,

forced displacements resulted in permanent depopulation, which greatly

changed land use (Affek, Zachwatowicz, & Solon, 2020; Bi

cík &

anek, 1994). The main change was forest expansion, both through

natural succession and active planting (e.g., Kozak, Estreguil, & Troll,

2007; Wolski, 2007). However, forest area increases related to politi-

cally driven forced displacements overlapped spatially and temporally

with the broader process of forest transition, i.e., the economically-

driven reversal from decreasing to expanding forest areas (Mather,

1992), which in the Carpathian region began in the Interwar period

(Kozak et al., 2007; Kozak, 2010; Munteanu et al., 2014). Furthermore,

the long-term effects of post-WWII displacements are potentially

confounded by widespread forest area increases due to land abandon-

ment after the collapse of socialism in the early 1990s (Grifths, Müller,

Kuemmerle, & Hostert, 2013; Kuemmerle et al., 2008). Therefore, it is

difcult to distinguish the effects of conict-related forced displace-

ments from other factors inuencing land use (Baumann & Kuemmerle,

2016; Baumann et al., 2015), and necessary to assess not only the

magnitude of reforestation, but also to identify potential changes in

pathways and the speed of reforestation, in order to differentiate abrupt

land-use changes from gradual changes (Ratajczak et al., 2018).

The main goal of our study was to determine the effects of forced

displacement on land cover. Specically, we examined the widespread

post-WWII forced and permanent displacement of the Ukrainian popu-

lation that had inhabited the Carpathians within the post-war Polish

borders. Our main research questions were: what were the short- and

long-term effects of displacements on (1) forest area, and (2) landscape

composition. To answer these questions we formulated the following

specic objectives:

1a. To determine village-level forest area change in the Polish Car-

pathians over two 40-year periods (1930s-1970s and 1970s-2010s);

1b. To quantify the independent effect of displacements on forest

area change and determine the share and area of new forests caused

by the displacements;

2. To determine landscape changes and trajectories of land-cover

change in depopulated, post-Ukrainian villages.

2. Methods

2.1. Study area

Our study area was the Carpathians within the current borders of

Poland, which we further limited for modeling purposes (i.e., to ensure

data availability and consistency) to the borders of the former Austrian

province of Galicia, and to whole villages according to their pre-WWII

extent (in total 1685 villages covering 17,000 km

, which corresponds

to 85% of the Polish Carpathians) (Fig. 1). Displacements were

concentrated in the south-eastern third of the study area, so that the

north-western part of the Polish Carpathians, where no displacement

took place, served as a control (Fig. 2A). The topography in the eastern

part ranges from 240 to 1346 m above sea level, and in the western from

210 to 2499 m a.s.l. The potential natural vegetation is temperate forest,

with spruce dominating above 1150 m a.s.l. (only in the western part),

beech (Fagus sylvatica) and r (Abies alba) from 600 to 1275 m a.s.l., and

oak (Quercus robur) and hornbeam (Carpinus betulus) dominating in the

lower parts. Only 0.6% of the study area is above timberline (Matusz-

kiewicz, 2008).

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

Before World War II, the south-eastern part of today’s Polish Car-

pathians was a Polish-Ukrainian ethnic borderland. Especially at the

turn of the 19th and 20th century, there was massive outmigration to the

USA and South America (Pollack, 2010), despite of which, the popula-

tion density in our study area continued to rise though due to the very

high birth rate until the late 1930s (ranging from < 30/km

at higher

elevations to > 120/km

in the foothills) (Soja, 2008). After the devel-

opment of railways at the end of the 19th century, the extensive Car-

pathian forests started to be heavily exploited, but due to complex

topography, limited access, and logging being selective and largely

focused on the most valuable trees, overall forest area remained rela-

tively stable up to WWII (Affek et al., 2020; Kozak, 2010; Munteanu

et al., 2014; Wolski, 2007). Concomitantly, a gradual shift away from

pastoralism, which was widespread both in the form of transhumance in

the upper parts and silvopastoralism in the lower parts, and toward crop-

based agriculture began, which resulted in the gradual abandonment of

high-elevation pastures (Kubijowicz, 1926). However, many local rural

communities were culturally conservative and did not adopt new tech-

nologies and cultural patterns coming from the Western world (Wolski,

2016).

In the 1940s the Polish Carpathians and their foothills experienced

one of the greatest forced displacements in the modern history of

Europe. First, from 1944 to 1946 a total of 480,000 Ukrainians were

displaced from the provinces of Rzesz

ow, Krak

ow, and Lublin to the

Fig. 1. A – location of the study area in Europe, B – study area with pre-war administrative village-level division (1685 villages) used for modeling (grey borders) and

test sites representing post-displacement Carpathian foothills (“Przemy

sl Foothills”) and middle mountains (“Bieszczady Mountains”) (orange borders). Background

layer: ArcGIS Basemap (World Hillshade). (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

Fig. 2. A – % of Ukrainian population in 1939 (displaced in 1940s); B – real forest area change in 1930s-1970s; C – real forest area change in 1970s-2010s; D –

modeled counterfactual forest area change for 1930s-1970s (as if there had been no displacements); C – modeled forest area change for 1930s-1970s; F – modeled

forest area change for 1970s-2010s. All data at the village level (N = 1685). Forest area changes in % of village area.

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

Ukrainian Soviet Socialist Republic (Eberhardt, 2011). This was part of a

larger population exchange between the post-war Poland and the Soviet

Union, aimed at ethnic consolidation after the delineation of the new

border along the Curzon Line (Eberhardt, 2011). The second wave of

displacements took place in 1947 (the so-called Operation Vistula),

when the remaining 140,000 Ukrainians including Poles from mixed

families, were displaced within the Polish post-war territory. The action

was carried out by the Soviet-installed Polish communist authorities

with the aim of removing material support and assistance to the

Ukrainian Insurgent Army (UPA) and bring an end to its guerilla activ-

ities (Motyka, 2011). However, the dispersion of Ukrainians throughout

the entire area of the “Recovered Territories” of northern and western

Poland, and later efforts to polonize them suggest that the creation of an

ethnically homogeneous nation-state was an additional goal (Snyder,

1999). The implementation and acceptance of such ethnic policy in

broader society was fostered by strong Polish-Ukrainian antagonism,

which developed already in the interwar period (Wolski, 2016), and

peaked after the Ukrainian Insurgent Army massacres of Poles in Vol-

hynia and Eastern Galicia in 1943 (Motyka, 2011; Snyder, 1999). As a

sidenote, the displaced Carpatho-Ruthenian ethnic groups belonging to

the East Slavs, that is Lemkos and Boykos peoples, are often regarded as

a subgroup of Ukrainian people. They were either Eastern Orthodox or

Greek-Catholic, while Poles were Roman-Catholic, thus religion was the

main determinant of national identity in the ethnic borderland up to

World War II, ahead of language and culture (Eberhardt, 2011).

Acknowledging the ethnic diversity of the East Slavs living in the Car-

pathian region, we retained the terminology used in the demographic

sources (Eberhardt, 2011; Kubìjovi

c, 1983) and call the entire displaced

population ‘Ukrainians’.

As a result of displacements, 126 Carpathian villages were

completely depopulated, and in an additional 250 villages only a few

Polish families remained (Appendix A, Fig. A.1). Due to the policy of no-

return, destruction of settlements and often adverse natural conditions

these villages have for the most part not been resettled (Pudło, 1992),

and especially in remote mountain areas the depopulation was

permanent.

In addition of our analyses of the entire Polish Carpathians, we

selected two test sites where before WWII>90% of population were

Ukrainians to analyze the detailed changes in landscape composition

and the trajectories of land-cover change after displacement. One test

site (“Przemy

sl Foothills”) represents the Carpathian foothills, and the

other (“Bieszczady Mountains”) the Carpathian middle mountains.

“Przemy

sl Foothills” covers 6096 ha and comprises 7 pre-war villages,

while “Bieszczady Mountains” covers 6170 ha and comprises 3 pre-war

villages. Population dropped from 4830 and 1590 people (79 and 26/

) in 1939 to 58 and 77 (0.9 and 0.4/km

) people in 2011, in

“Przemy

sl Foothills” and “Bieszczady Mountains” respectively

(Kubìjovi

c, 1983; https://stat.gov.pl). We selected those 10 villages,

because they are representative of post-displacement villages in the

foothills and middle mountains, where the depopulation resulting from

displacements was permanent.

2.2. Effects of displacement on forest area across the Polish Carpathians

To estimate the effect of the forced displacements on forest area, we

conducted a village-level analysis across the Polish Carpathians (in total

1685 villages/cadastral communities, “Katastralgemeinden” in German).

For this, we reconstructed the mid-19th century high-resolution village-

level administrative borders of the former province of Galicia (part of

Austro-Hungarian Empire) for the Carpathians within the current Polish

borders. Our map sources were the Administrativ Karte von den

onigreichen Galizien und Lodomerien of Kummersberg at a scale of ca.

1:115,000, published in 1855, current data from National Register of

Boundaries, and the second military survey of the Habsburg Empire

1:28,800 of 1860s. We calculated village-level forest area in the Car-

pathians in the 1930s and 1970s based on topographic maps

(Bednarczyk, Kaim, & Ostan, 2016; Ostan et al., 2017). Forest area in

the 2010s was derived from the national Database of Topographic Ob-

jects 1:10,000 (BDOT10k). By combining the data from the three time

points, we obtained two maps of village-level forest area change (in

percentages of village area) for the 1930s-1970s and the 1970s-2010s.

For each village, we also obtained past land use and ownership infor-

mation from the 1868 census (Skorowidz, 1868), and ethnic structure

from the 1939 census (Kubìjovi

c, 1983). Topography-related variables

were derived from the Digital Terrain Elevation Data for Poland (DTED

Level 2 with 30-m resolution).

In order to quantify the extent to which land-cover changes were

caused by the displacement of the Ukrainian population, we parame-

terized regression models of village-level forest area change from the

1930s to the 1970s, and from 1970s to 2010s. In these models, the share

of the population that was Ukrainian in 1939 served as a proxy measure

of displacement intensity, because the entire Ukrainian population was

displaced and forcibly re-settled elsewhere (Fig. 2A). We also included

other potential predictors related to environmental and socioeconomic

conditions, such as initial land use and ownership pattern, population

density, accessibility (distance to markets), soils and topography

(Table 1). We preselected those variables that in univariate models were

at least weakly correlated (Spearman rho < − 0.15 or > 0.15) with forest

area change (Appendix B, Tables B.1 and B5). We identied seven vil-

lages as strong outliers (out of 1685), six of which were in the control

group, and removed them because they would have been sources of

potential bias. As a robustness check, we parameterized models with all

data points, and their results were similar (results not shown). To

quantify the relationship between each of our candidate predictors and

forest area change, we tted a series of linear models containing all

possible combinations of our covariates, using the ‘dredge()’ function in

R package “MuMIn”. We examined the scatterplots to see if there were

non-linear relationships, but found none that were obviously so, and

when we tested the squared terms of variables (e.g., we observed steeper

forest area increase in villages with > 90% Ukrainians), they explained

less variance than when they entered the model without the squared

term (results not shown). To identify the best model, we ranked them all

according to their Bayesian Information Criterion (BIC), which penalizes

over-parameterized models. We assessed the total explanatory power of

models by calculating their adjusted R

, and checked for multi-

collinearity by calculating the Variance Ination Factor (VIF). Hair,

Black, Babin, and Anderson (2010) suggest that a VIF of 10 represents a

high degree of collinearity and recommend lower VIF thresholds. We

therefore employed a conservative VIF of 4 in our models, which was

also a natural break among our variables.

To assess the relative importance of the predictors included in the

best model, we applied hierarchical partitioning analysis to calculate the

independent, joint, and total contributions of each predictor to overall

explained variance (Chevan & Sutherland, 1991). We used the ‘hier.part

()’ function to apply hierarchical partitioning in R package “hier.part”.

We also assessed potential biases arising from spatial autocorrelation by

calculating and mapping Moran’s I of the model residuals (see Appendix

C).

To assess the overall change in forest area due to the displacements,

we applied our regression model to calculate the counterfactual increase

in forest area that would be expected if a village had no Ukrainians, and

subtracted that from the prediction based on the actual ethnic compo-

sition, using the following formula:

OFAC =

∑

i=1

(FAC − FACr)/100 (1)

where OFAC – overall forest area change only due to displacements (in

ha), N – number of villages, A

– area (in ha) of ith village, FAC – modeled

forest area change (in % of village area), FACr – modeled counterfactual

forest area change (in % of village area) if there had been no displace-

ments. We calculated and mapped FAC and FACr for each village, con-

verted the values from percentages of village area to hectares, and

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

subtracted the latter from the former, which provided us with an esti-

mate of the portion of the forest area increase in the Polish Carpathians

due to the displacements (OFAC). To account for uncertainty of

modeling, we calculated the 95% condence interval (CI) of the esti-

mated overall forest area change. We calculated our CI by summing up

the intervals for predictions of the mean response of each village’s

predicted difference in forest change (FAC-FACr). Lastly, to directly

compare the effect of displacement on forest area change between the

two periods, we selected a subset of relevant variables, that is variables

that were important in at least one period, and modeled again both time

periods, this time with the same set of variables.

2.3. Effects on the landscape in two test sites

To show the spatially-explicit landscape effects of forced displace-

ments we mapped land cover in detail in our two selected test sites

(“Przemy

sl Foothills” and “Bieszczady Mountains”), where vast majority

of inhabitants were forcibly displaced (Affek et al., 2020; Wolski, 2007).

We compared the most detailed pre-war land-cover map (the Austrian

1:2880 cadastral map developed in the mid-19th century) with post-war

1:10,000 topographic map from the 1970s, and with BDOT10k for the

2010s. We assumed the 1970s and 2010s were appropriate time points

to capture immediate and long-term landscape effects of displacements

in the 1940s respectively. The pre-war map of 1930s that we used to

model forest area has a resolution that is too low (1:100,000) and land

cover classes that are too simplied to combine them with the detailed

post-war land-cover maps. That is why we compared the detailed post-

war land cover maps for our 10 selected villages with similarly

detailed maps for the mid-19th century instead. In prior in-depth

studies, we found that landscape pattern between the mid-19th cen-

tury and WWII was relatively stable in both test sites (<5% net change)

(Affek et al., 2020; Wolski, 2007), thus we assumed the older cadastral

maps, which are widely used in historical landscape research (Dolej

s &

Forejt, 2019), reect the pre-war state adequately. Furthermore, no

other high-resolution data sources (such as aerial photographs) were

available for that period. The map sources and how we analyzed them

are described in detail in Appendix D.

Resulting from our land-cover analyses were three high resolution

land-cover maps (1850s, 1970s, and 2010s) for “Przemy

sl Foothills” and

“Bieszczady Mountains” which were consistent in the level of detail and

land-cover classes. We mapped six classes for each of the three dates:

built-up area, orchard, arable eld, grassland, transitional woodland-

shrub, and forest. Via land-cover transition matrices, we calculated

the trajectories of change from the 1850s to the 1970s, and then to the

2010s. We performed the spatial analyses of the vector data layers in

ArcMap 10.2.2 (ESRI ArcGIS Advanced).

3. Results

3.1. Effects of forced displacement on forest area

Between the 1930s and the 1970s, forest area increased from

617,600 to 798,700 ha (i.e., by 181,100 ha, 29.3% growth, 9.1 per-

centage points increase) across the Polish Carpathians, and from

534,700 ha to 702,900 ha (168,200 ha, 31.4%, 9.9 percentage points) in

the 1685 villages, which we included in our modeling. Our village-level

analysis showed substantial spatial variation in forest area change

though, with much larger forest area increases (>20 percentage points

per village) in the south-eastern part of the study area (Fig. 2C). The

pattern was strongly correlated to the distribution of pre-war Ukrainians

(Pearson’s r = 0.802, p < 0.0001), (Table 2).

To determine the independent effect of displacements on forest area

change in the 1930s–1970s we applied our highest-ranked model, which

included 5 covariates: Ukrainian population in 1939, slope, peasantry

pastures in 1860s, forest area in 1930s, and distance to towns with >

20,000 citizens (Appendix B, Table B.4). The selected model explained

70% of the observed variance (adjusted R

= 0.70) with a standard error

of the estimate of 7.25 percentage points and a maximum VIF of 2.9.

Hierarchical partitioning of the explained variance showed that the

Ukrainian population in 1939, i.e., our proxy for forced displacement,

was responsible for 75% of all independent effects (53% of the total

variance), far ahead of the second most important factor, which was

distance to markets (Fig. 3, Appendix B, Table B.4). Villages that had a

Table 1

Source data of variables considered in our modeling.

Variables Source

type

Source Model

for

1930s-

1970s

Model

for

1970s-

2010s

1 Forest area change

1930s-1970s [% of

village area]

maps (Bednarczyk

et al., 2016;

Ostan et al.,

2017)

R P

2 Forest area change

1970s-2010s [% of

village area]

spatial

data

(BDOT10k;

Ostan et al.,

2017)

– R

3 Forest area in

1860s [% of village

area]

census (Skorowidz,

1868)

P P

4 Forest area in

1930s [% of village

area] “F”

map (Bednarczyk

et al., 2016)

P P

5 Forest area in

1970s [% of village

area]

map (Ostan et al.,

2017)

P P

6 Ukrainian

population in 1939

[% of village

population] “U”

census (Kubìjovi

1983)

P P

7 Population density

in 1857 [persons/

]

census (Skorowidz,

1868)

P P

8 Elevation [mean

for a village]

DEM DTEDlevel2 P P

9 Slope [mean for a

village] “S”

DEM DTEDlevel2 P P

10 Distance to town

with population

above 20,000 [km]

“D”

map,

census

various sources P P

11 Soil agricultural

suitability [mean

for a village]

map (Skiba &

Drewnik, 2003)

P P

12 Land owned by

landlords in 1860s

[% of village area]

census (Skorowidz,

1868)

P P

13 Landlord forests in

1860s [% of village

area]

census (Skorowidz,

1868)

P P

14 Landlord arable

elds in 1860s [%

of village area]

census (Skorowidz,

1868)

P P

15 Landlord pastures

in 1860s [% of

village area]

census (Skorowidz,

1868)

P P

16 Landlord gardens

in 1860s [% of

village area]

census (Skorowidz,

1868)

P P

17 Peasantry forests

in 1860s [% of

village area]

census (Skorowidz,

1868)

P P

18 Peasantry arable

elds in 1860s [%

of village area]

census (Skorowidz,

1868)

P P

19 Peasantry pastures

in 1860s [% of

village area] “P”

census (Skorowidz,

1868)

P P

20 Peasantry gardens

in 1860s [% of

village area]

census (Skorowidz,

1868)

P P

R – response variable; P – predictor variable

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

higher share of Ukrainians prior to WWII, and those located further

away from towns, with steeper slopes, and higher share of initial forest

area and peasantry pastures had the highest post-war increases in forest

area.

The unstandardized full regression equation used to calculate the

modeled forest area change was:

FAC = − 3.434 + 0.255 U + 0.224 D + 0.721 S − 0.181F + 0.119P, (2)

where FAC – modeled forest area change 1930s–1970s, U – Ukrainian

population in 1939, D – distance to town, S – slope, F – forest area in

1930s, P – peasantry pastures in 1860s (see also Table 1). The map of

modeled forest area change for 1930s–1970s closely resembles the

actual forest area change (Fig. 2C and E), while the map of counter-

factual forest area change indicates that some areas (less elevated and

closer to towns) would have had a decrease in forest area, instead of an

increase, if it had not been for the displacements (Fig. 2B and E). When

we applied Eq. (1) to calculate the forest area increase due to the forced

displacements (OFAC), we obtained an estimate of 114,700 ha (±4670

ha with 95% condence). In other words 63% of the overall forest area

increase (114,700 out of 181,000 ha) between 1930s and 1970s in the

entire Polish Carpathians was due to the forced displacement. Further-

more, in villages inhabited by ≥ 90% Ukrainian population in 1939 (260

villages), the displacement caused 73,100 ha of the 88,400 ha of total

forest area increase (83%).

In the 1970s–2010s, the forest area in the Polish Carpathians

continued to expand, but generally at a slower pace (Fig. 2D). Inter-

estingly, the largest increases occurred then in the densely populated

foothills, where no post-war forced displacements took place, while

areas where earlier displacements were common had only moderate

forest area increase. Indeed, we found a weak but signicant negative

correlation between forest area change in the 1970s-2010s and pre-war

share of Ukrainian population (r = − 0.137, p < 0.0001). The map of

modeled forest area change for that period shows the above-mentioned

regularities even more clearly (Fig. 2F). However, our explanatory

Table 2

Village-level forest area change (1930s-1970s, and 1970s-2010s) in percentage points in villages of different ethnic structure.

1930s-1970s 1970s-2010s

Share of Ukrainian population N of villages Min Max Mean SD Min Max Mean SD

<5% 1,180 − 21.21 23.47 2.46 3.97 − 13.04 28.23 7.54 4.92

5–90% 238 − 0.94 63.26 17.36 12.84 − 1.81 25.10 7.21 4.70

>90% 260 − 3.24 67.39 30.48 13.76 − 3.28 36.36 5.58 4.36

All differences signicant at p < 0.001

Fig. 3. Hierarchical partitioning results showing the independent effects of each variable in the models of forest area change in 1930s-1970s (top) and 1970s-2010s

(middle). In the bottom panel, only villages with Ukrainian population in 1939 > 0% were considered.

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

variables only weekly predicted forest area change (adjusted R

= 0.28).

The independent effect of displacement constituted only 3% of all in-

dependent effects, and was far behind the independent effect for soil

agricultural suitability (30%), elevation (26%), slope (24%) and forest

area in 1970s (11%). The best model for that period included two

interaction effects (Ukrainian population × elevation, and Ukrainian

population × slope), which we did not include when we assessed the

independent effects of predictors. In villages with a higher share of

Ukrainians in 1939 (>90%) forest area increase was negatively corre-

lated with slope, but that correlation was positive for villages with a

lower share of Ukrainians (<5%). However, when we parametrized the

model only for the post-displacement villages (N = 523), the overall

model performance increased (adjusted R

= 0.32), as did the inde-

pendent effect of ethnicity (14%) and initial forest area (37%). Inter-

estingly, the forest area increase in 1970s–2010s was signicantly

negatively correlated with elevation, slope, share of Ukrainians and the

initial forest area, and positively with soil agricultural suitability (see

Appendix B, Tables B.5 to B.8 for details). The comparison of the two

periods based on the models with the identical subset of seven variables

showed that the independent effect of displacement was 63 times larger

for the 1930s–1970s than for the 1970s–2010s (0.508 versus 0.008),

which is equivalent of 72.5% and 3.5% of all independent effects in each

model, respectively (see Appendix B, Table B.9). The relationship be-

tween the displacement and forest area change was positive in the rst

period, and negative in the second period.

3.2. Effects on the landscape

In “Przemy

sl Foothills” and “Bieszczady Mountains”, where we

conducted our detailed land-cover change analyses, each village had ≥

90% Ukrainian population prior to WWII, and land cover changed

dramatically after the post-WWII displacements (Fig. 4). The direction of

change was similar in both areas (reforestation), but the magnitude was

much larger in the foothills, where some of the reforestation occurred

already before displacements. As a result, while the land cover was

substantially different between the “Przemy

sl Foothills” and “Bieszc-

zady Mountains” in the pre-war period, they were already quite similar

in the 1970s, and almost identical in the 2010s (Fig. 4). From the 1850s

to the 1930s, forest area was essentially stable and changed only from

56% in the 1850s to 55% in 1930s in the “Bieszczady Mountains” and

from 33% in the 1850s to 37% in 1930s in the “Przemy

sl Foothills”. By

the 1970s, the forest area had increased substantially, to 75% and 77%

respectively, and continued to grow thereafter, reaching 84% in the

2010s in both areas. Built-up areas and arable elds disappeared almost

completely after WWII. This was particularly striking in the foothills,

where it resulted in the shift of the dominant land cover from arable

elds to forest.

The land-cover trajectories from the 1850s to the 2010s were similar

in both areas, with permanent forest being the most common sequence,

followed by post-war transitions from open farmland (either grassland

or arable land) to forest (Table 3). Whereas forests and grassland were

Fig. 4. Landscape pattern and composition in 3 decades in two test sites representing Carpathian foothills and middle mountains with population displacement

in 1940s.

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

the main land-cover types in the mountains before displacements, and

the foothills used to be dominated by farmland, both areas were largely

void of arable land and built-up areas by the 2010s, and completely

covered by forest, bushes, and grassland. As a result of displacements,

markedly different pre-war landscapes of Carpathian middle mountains

and foothills became very similar by the end of our study period.

4. Discussion

We found that forced displacements in the Polish Carpathians after

WWII explained 63% of all the forest area increase from the 1930s to the

1970s, highlighting that displacements can trigger widespread and

permanent land use changes. In general, the trend of forest area increase

and agricultural land abandonment since WWII is common throughout

the Carpathians and many mountain regions of the world (Lasanta et al.,

2017; MacDonald et al., 2000). These processes are usually conse-

quences of a multitude of factors, which often operate at broad scales

and are interrelated, including long-ranged demographic changes,

technological advances, socio-economic and institutional reforms

launching new rules of land management, or external trends in land use

patterns (see also Baumann et al., 2015). However, the magnitude of

changes that took place in the eastern part of Polish Carpathians and

high correspondence with the distribution of displaced Ukrainian pop-

ulation clearly indicate that the displacement factor was the main cause

for agricultural abandonment and subsequent increases in forest area.

Similar landscape changes occurred after WWII in other post-

displacement borderland regions of Poland (Latocha, 2012) and other

European countries [e.g., in Czechia (Bi

cík &

anek, 1994) and

Slovenia (Hladnik, 2005)], and in other continents where armed con-

icts force people to leave their land (Baumann & Kuemmerle, 2016).

Forced displacements are widespread, and may have strong effects

on land cover (Hugo, 1996; Woube, 2005), but it is not clear how long

those effects persist. We found that the post-WWII forest area change in

the Polish Carpathians that resulted from the forced displacements,

persisted for 70 years, and were much more important than other factors

such as access to markets, topography, and pre-war land use and

ownership patterns. Indeed, we estimated that almost two-thirds of post-

war forest area increase was due to the forced displacements (see also

Kozak, 2010; Kozak et al., 2007; Munteanu et al., 2014).

Forest area increase continued after the 1970s, and especially the

collapse of socialism in early 1990s, which lead to agricultural aban-

donment, also fostered forest area increases (Kolecka et al., 2017; Kozak,

2010; Kuemmerle et al., 2008). In our study area, the only exception to

this general trend was a substantial forest area decrease in the Silesian

Beskids (Western Carpathians) due to massive bark beetle outbreak in

the 2000s (Grodzki, 2007), and that decrease is likely temporary.

However, much to our surprise, the post-socialist forest area increase

was considerably smaller than the increase after WWII. Furthermore, the

post-socialist forest area increase was lower in post-displacement areas

compared to the rest of the Polish Carpathians. The relative importance

of the different factors explaining forest area change 1970s-2010s in the

displaced areas and the direction of relationships indicated that the

legacy of the displacement continued, but the effect was much weaker

and opposite to that from the previous period. In 1970s-2010s, increases

in forest area in displaced areas were more common in villages at lower

elevations, and with better soils and less steep slopes, which is not the

common pattern of forest expansion. We suggest that this unusual

pattern of afforestation is due to a saturation effect (Schneeberger,

Bürgi, & Kienast, 2007), in that forest area increased so much in the

displacement areas after WWII that only small plots were still available

for the new forests to grow. The negative correlation (Spearman’s rho =

− 0.398, p < 0.0001) between initial, 1970s, forest area and 1970s-

2010s forest area increase further supports this interpretation.

The major driving forces and patterns of change in the post-1990

period are common for the entire Carpathians, because all Carpathian

countries were under socialist governance between 1945 and 1991 and

all countries except Ukraine are now members of the EU (see e.g.,

Grifths et al., 2013; Munteanu et al., 2014). Across the Carpathians,

forest area increases was rapid after the 1990 because that is when

agricultural land was abandoned (Grifths et al., 2013; Kuemmerle

et al., 2008; Munteanu et al., 2014). However, the parts of Poland where

forced displacement took place followed different path. Here, the

magnitude and pace of land use change in the post-war period was much

higher than in the post-socialist era, and they were already abandoned

and underwent afforestation and succession 40 years earlier (Affek et al.,

2020; Wolski, 2007).

In addition to the increases in forest area, we found major declines in

landscape diversity after displacements. The pre-war small elds,

meadows and pastures were either afforested and taken over by State

Forests, underwent natural forest succession, or were converted to large-

scale agricultural monocultures managed by agricultural cooperatives

or State Farms. The majority of buildings were burned or demolished

and the cultural continuity and centuries-old landscape characteristics

were largely lost (Affek et al., 2020; Kozak et al., 2007; Soja, 2008).

Nonetheless, some traces of past cultural landscape are still visible in the

microtopography (e.g., hollow ways, agricultural terraces, and stone

mounds) and vegetation (e.g., post-grazing beeches, old fruit trees)

(Affek, 2016; Wolski, 2007). After the transition from the centrally-

planned to a market-driven economy in the 1990s, most of the collec-

tive farms collapsed and their agricultural land was either privatized or

afforested and added to the State Forests (Affek et al., 2020; Kuemmerle

et al., 2008). However, even the creation of strong individual property

rights was not sufcient to maintain agriculture, because limiting

availability of agricultural inputs combined with limited access to

markets prevented landowners from reaping the gains of specialization

and increasing agricultural productivity (Rozelle & Swinnen, 2004).

Agricultural land abandonment is a common immediate effect of

forced displacements during wars. For example, it occurred also in

Bosnia and Herzegovina (Witmer, 2008) and the Caucasus region

(Baumann et al., 2015; Yin et al., 2019). Our results highlight how long-

lasting forest area increase can be, because we analyzed land cover

changes up to 70 years after the displacement. Most prior analysis of the

effects of wars on land use examined on a decade or so afterwards (e.g.,

Witmer, 2008; Yin et al., 2019), except of some studies of post-WWII

displacements in Poland and Czechia, which also reported long-lasting

effects (e.g., Bi

cík &

anek, 1994; Latocha, 2012).

Table 3

Land-cover trajectories (1850s → 1970s → 2010s) for the two test sites

(“Przemy

sl Foothills” and “Bieszczady Mountains”).

“Przemy

sl Foothills” “Bieszczady Mountains”

% of the

test site*

Trajectory % of the

test site*

Trajectory

31.94 permanent forest 56.19 permanent forest

31.86 arable eld → forest →

forest

8.65 grassland → forest →

forest

8.64 grassland → forest →

forest

6.66 arable eld → forest →

forest

7.46 arable eld → grassland

→ grassland

5.99 permanent grassland

3.09 arable eld → grassland

→ forest

5.90 arable eld → grassland

→ grassland

2.31 trans → forest → forest 3.43 grassland → trans →

forest

1.89 arable eld → trans →

forest

2.60 trans → forest → forest

1.31 permanent grassland 1.73 arable eld → trans →

forest

11.51 other 1.60 grassland → grassland

→ forest

1.26 arable eld → grassland

→ forest

5.98 other

* Only trajectories above 1% of the test site area are shown.

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

A common problem, limiting the scope of land change assessments in

conict zones, is the restricted access to the areas of interest. Therefore,

models of land change in conict zones in Eastern Africa (Gorsevski

et al., 2012) and Caucasus (Baumann et al., 2015; Yin et al., 2019) were

based on satellite images and did not include local demographic data,

which means that they did not account for displacement explicitly. Our

approach made use of spatio-temporal land cover patterns derived from

historical and contemporary maps combined with a detailed village-

level demographic census. Accounting for land use and land owner-

ship legacies reaching back to the feudal system, while controlling for

other potential drivers (e.g., topography) enabled us to extract ethnicity

as the factor for displacement and quantify, for the rst time, forest area

change arising from post-war displacements, thereby substantiating

prior circumstantial evidence that displacements were the main cause

for the widespread forest cover increase in the Polish Carpathians (Affek

et al., 2020; Kozak et al., 2007).

The landscape change after Carpathian displacements can by inter-

preted in the theoretical framework proposed by Ramankutty and

Coomes (2016) as an abrupt change of a middle-scale land-use regime

observable at the regional level. The agriculturally-dominated pre-

displacement land-use regime, which had only minor forest area uc-

tuations related to legal, economic and ownership changes, was main-

tained by reinforcing forces including (1) sociocultural conservatism, (2)

very high birth rate balancing rural outmigration, (3) poor transport

infrastructure and (4) mountainous borderland location. Preconditions

(early-warning signals) that the pre-war land-use regime was potentially

susceptible to a future shift include tensions between neighboring ethnic

groups stemming from the emergence of nationalist movements and

ideologies in the 1920s and 1930s, World War II and its resulting border

changes, and the introduction of the new communist political system in

Poland.

We found that the land-cover changes due to the forced displacement

were irreversible. This means that the displacement represents a shock

event (Baumann et al., 2015) or a catastrophic driver (Kozak et al.,

2007). Displacement belongs to the type-3 triggers of land-use regime

shifts (rapid demographic change) (Ramankutty & Coomes, 2016). The

distinguishing feature of ethnic displacements is the disappearance of a

culturally entrenched mode of land use in addition to demographic

change. We found that over time a new, forest-dominated land-use

regime was established, with forest cover complemented by a substan-

tial share of fallow land and a small number of collective animal farms.

Several self-reinforcing processes prevented a return to the old regime,

including the destruction of pre-war settlements, ownership and agri-

cultural reforms (i.e., nationalization and collectivization of farmland),

laws preventing the return of the displaced people, borderland location

combined with a long-held belief in the temporary nature of post-war

order, harsh living conditions in the mountains, poor infrastructure,

low protability of agricultural production and a centrally planned

economy. That is why only about 6000 (3%) of the displaced families

returned (Pudło, 1992), and a state campaign of settling Poles from

other regions during the late 1950s and 1960s had very limited success

(Wolski, 2016). We argue that what limited the resilience of the pre-war

land-use regime to the displacements were both the characteristics of the

pre-war land system and the strength of the post-displacement rein-

forcing processes. That is why land abandonment did not revert and

instead became permanent, even where environmental conditions were

favorable, and despite economic incentives for agriculture (Pudło,

1992). Irreversible shifts in land cover due to forced displacement may

be likely elsewhere too, especially if unregulated boundaries, uncertain

land tenure, landmines, and legal restrictions prevent displaced people

from returning or new settlers to replace them (Baumann et al., 2015;

Witmer, 2008; Yin et al., 2019).

Permanent shifts of the land system may trigger rewilding (Baumann

et al., 2015; Kim, 1997). In the Carpathians, the land-use change

following displacements entailed large-scale natural succession and

forest regeneration processes. Once densely populated rural region

became one of the largest wilderness areas in Central Europe, with vast

areas free from human settlements, and with resurgent wildlife. It is

currently the area with the lowest light pollution in Poland (

Scię

zor,

Kubala, & Kaszowski, 2012) and the largest mountain population of

purebred European bison in the world (Perzanowski & Olech, 2014).

Shock events such as displacements have the potential to shift the

land systems into an alternative stable state (Müller et al., 2014). The

forest saturation effect observed in the post-displacement areas from the

1970s to the 2010s suggests that the land system began to approach a

new stable state 30 years after the perturbation occurred. Nonetheless,

we share the view of Ramankutty and Coomes (2016) and Walker et al.

(2006) that land systems are unlikely to be fully stable and in the case of

post-displacement Carpathian areas the path towards stable state ended

with the end of socialist period and the emergence of new drivers.

5. Conclusions

We found that permanent and widespread land-cover changes took

place over both 30 and 70 years after forced displacement following

WWII. A new land-use regime was established, effectively maintained by

a diverse set of self-reinforcing processes, which means that displace-

ments triggered a land-use regime shift, from agriculturally-dominated

to forest-dominated. The observed forest saturation effect in the post-

displacement areas, when the collapse of socialism in the early 1990s

triggered widespread abandonment elsewhere, indicate that land-

system approached a new stable state and the new forces were further

strengthening the post-war land-use regime. Our result show that pol-

icies related to displacements, especially prohibitions to return to the

area, played a key role in shaping the post-displacement landscape. Our

ndings have important socioeconomic and environmental implications

because they highlight the strong land use legacies of the post-WWII

displacements on Carpathian landscapes for seventy years, which may

be relevant elsewhere when predicting the landscape consequences of

forced displacements.

CRediT authorship contribution statement

Andrzej N. Affek: Conceptualization, Data curation, Formal anal-

ysis, Funding acquisition, Investigation, Methodology, Project admin-

istration, Resources, Software, Validation, Visualization, Writing -

original draft, Writing - review & editing. Jacek Wolski: Investigation,

Resources, Writing - original draft, Writing - review & editing. Maria

Zachwatowicz: Writing - original draft, Writing - review & editing.

Krzysztof Ostan: Resources, Writing - review & editing. Volker C.

Radeloff: Conceptualization, Funding acquisition, Methodology, Re-

sources, Supervision, Writing - review & editing.

Acknowledgements

This work was supported by the Polish National Agency for Aca-

demic Exchange [Grant No. PPN/BEK/2018/1/00501], and NASA’s

Land Cover and Land Use Change Program. We thank S. Skiba for

sharing the soil map of the Polish Carpathians and L. Farwell for sta-

tistical advice. We also thank C. Munteanu and two anonymous re-

viewers for their valuable suggestions and comments that helped greatly

to improve the manuscript.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.

org/10.1016/j.landurbplan.2021.104164.

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

References

Abel, G. J., Brottrager, M., Crespo Cuaresma, J., & Muttarak, R. (2019). Climate, conict

and forced migration. Global Environmental Change, 54, 239–249. https://doi.org/

10.1016/j.gloenvcha.2018.12.003.

Affek, A. N. (2016). Past Carpathian landscape recorded in the microtopography.

Geographia Polonica, 89(3), 415–424. https://doi.org/10.7163/GPol.0062.

Affek, A. N., Zachwatowicz, M., & Solon, J. (2020). Long-term landscape dynamics in the

depopulated Carpathian Foothills: A Wiar River basin case study. Geographia

Polonica, 93(1), 5–23. https://doi.org/10.7163/Gpol10.7163/GPol.2020.110.7163/

GPol.0160.

Alix-Garcia, J., Bartlett, A., & Saah, D. (2013). The landscape of conict: IDPs, aid and

land-use change in Darfur. Journal of Economic Geography, 13(4), 589–617. https://

doi.org/10.1093/jeg/lbs044.

Baumann, M., & Kuemmerle, T. (2016). The impacts of warfare and armed conict on

land systems. Journal of Land Use Science, 11(6), 672–688. https://doi.org/10.1080/

1747423X.2016.1241317.

Baumann, M., Radeloff, V. C., Avedian, V., & Kuemmerle, T. (2015). Land-use change in

the Caucasus during and after the Nagorno-Karabakh conict. Regional Environmental

Change, 15(8), 1703–1716. https://doi.org/10.1007/s10113-014-0728-3.

Bednarczyk, B., Kaim, D., & Ostan, K. (2016). Forest cover change or misinterpretation?

On dependent and independent vectorisation approaches. Prace Geograczne, 146,

19–30. https://doi.org/10.4467/20833113PG.16.015.5545.

cík, I., &

anek, V. (1994). Post-war changes of the land-use structure in Bohemia

and Moravia: Case study Sudetenland. GeoJournal, 32(3), 253–259. https://doi.org/

10.1007/BF01122117.

Biggs, R., Peterson, G. D., & Rocha, J. C. (2018). The regime shifts database: A framework

for analyzing regime shifts in social-ecological systems. Ecology and Society, 23(3),

Art. 9. https://doi.org/10.5751/ES-10264-230309.

Chevan, A., & Sutherland, M. (1991). Hierarchical partitioning. American Statistician, 45

(2), 90–96. https://doi.org/10.1080/00031305.1991.10475776.

Dolej

s, M., & Forejt, M. (2019). Franziscean cadastre in landscape structure research: A

systematic review. Quaestiones Geographicae, 38(1), 131–144. https://doi.org/

10.2478/quageo-2019-0013.

Eberhardt, P. (2011). Political migrations on Polish territories (1939-1950). Monograe

12. Warszawa: Instytut Geograi i Przestrzennego Zagospodarowania PAN.

Eklund, L., Degerald, M., Brandt, M., Prishchepov, A. V., & Pilesj

o, P. (2017). How

conict affects land use: Agricultural activity in areas seized by the Islamic State.

Environmental Research Letters, 12(5), 054004. https://doi.org/10.1088/1748-9326/

aa673a.

Ghimire, K. (1994). Refugees and deforestation. Internal Migration Quarterly Review, 32

(4), 561–568.

Gorsevski, V., Kasischke, E., Dempewolf, J., Loboda, T., & Grossmann, F. (2012). Analysis

of the Impacts of armed conict on the Eastern Afromontane forest region on the

South Sudan - Uganda border using multitemporal Landsat imagery. Remote Sensing

of Environment, 118, 10–20. https://doi.org/10.1016/j.rse.2011.10.023.

Grifths, P., Müller, D., Kuemmerle, T., & Hostert, P. (2013). Agricultural land change in

the Carpathian ecoregion after the breakdown of socialism and expansion of the

European Union. Environmental Research Letters, 8(4), 045024. https://doi.org/

10.1088/1748-9326/8/4/045024.

Grodzki, W. (2007). Spatio-temporal patterns of the Norway spruce decline in the Beskid

Slaski and

Zywiecki (Western Carpathians) in southern Poland. Journal of Forest

Science, 53, 38–44. https://doi.org/10.17221/2155-jfs.

Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate Data

Analysis (7th ed.). Upper saddle River, New Jersey: Pearson Education International.

https://doi.org/10.1016/j.ijpharm.2011.02.019.

Hladnik, D. (2005). Spatial structure of disturbed landscapes in Slovenia. Ecological

Engineering, 24(1–2), 17–27. https://10.1016/j.ecoleng.2004.12.004.

Hugo, G. (1996). Environmental Concerns and International Migration. International

Migration Review, 30(1), 105–131.

IDMC. (2019). Global Report on Internal Displacement 2019. Geneva. Retrieved from

https://www.internal-displacement.org/sites/default/les/publications/documents

/2019-IDMC-GRID.pdf.

Kim, K. C. (1997). Preserving biodiversity in Korea’s demilitarized zone. Science, 278

(5336), 242–243. https://doi.org/10.1126/science.278.5336.242.

Kolecka, N., Kozak, J., Kaim, D., Dobosz, M., Ostan, K., Ostapowicz, K., … Price, B.

(2017). Understanding farmland abandonment in the Polish Carpathians. Applied

Geography, 88, 62–72. https://doi.org/10.1016/j.apgeog.2017.09.002.

Kozak, J. (2010). Forest cover changes and their drivers in the Polish Carpathian

Mountains since 1800. In H. Nagendra, & J. Southworth (Eds.), Reforesting

landscapes: linking pattern and process. Landscape Series 10 (pp. 253–273).

Netherlands: Springer. https://doi.org/10.1007/978-1-4020-9656-3.

Kozak, J., Estreguil, C., & Troll, M. (2007). Forest cover changes in the northern

Carpathians in the 20th century: A slow transition. Journal of Land Use Science, 2(2),

127–146. https://doi.org/10.1080/17474230701218244.

Kubìjovi

c, V. (1983). Ethnic groups of the South-Western Ukraine (Haly

cyna – Galicia) 1.

1. 1939. Wiesbaden: National statistics of Haly

cyna – Galicia. Otto Harrasowitz.

Kubijowicz, W. (1926).

Zycie pasterskie w Beskidach Wschodnich (Shepherd life in the

Eastern Beskids). Prace Instytutu Geogracznego Uniwersytetu Jagiello

nskiego 5.

Krak

ow: Uniwersytet Jagiello

nski.

Kuemmerle, T., Hostert, P., Radeloff, V. C., van der Linden, S., Perzanowski, K., &

Kruhlov, I. (2008). Cross-border comparison of post-socialist farmland abandonment

in the Carpathians. Ecosystems, 11(4), 614–628. https://doi.org/10.1007/s10021-

008-9146-z.

Landholm, D. M., Pradhan, P., & Kropp, J. P. (2019). Diverging forest land use dynamics

induced by armed conict across the tropics. Global Environmental Change, 56,

86–94. https://doi.org/10.1016/j.gloenvcha.2019.03.006.

Lasanta, T., Arn

aez, J., Pascual, N., Ruiz-Fla

no, P., Errea, M. P., & Lana-Renault, N.

(2017). Space–time process and drivers of land abandonment in Europe. Catena, 149,

810–823. https://doi.org/10.1016/j.catena.2016.02.024.

Latocha, A. (2012). Changes in the rural landscape of the Polish Sudety Mountains in the

post-war period. Geographia Polonica, 85(4), 13–21. https://doi.org/10.7163/

GPol.10.7163/GPol.2012.4.21.

MacDonald, D., Crabtree, J. R., Wiesinger, G., Dax, T., Stamou, N., Fleury, P., …

Gibon, A. (2000). Agricultural abandonment in mountain areas of Europe:

Environmental consequences and policy response. Journal of Environmental

Management, 59(1), 47–69. https://doi.org/10.1006/jema.1999.0335.

Machlis, G. E., & Hanson, T. (2008). Warfare ecology. BioScience, 58(8), 729–736.

https://doi.org/10.1641/B580809.

Mather, A. S. (1992). The forest transition. Area, 24(4), 367–379.

Matuszkiewicz, J. M. (2008). Potential natural vegetation of Poland. Retrieved May 18,

2020, from https://www.igipz.pan.pl/potential-vegetation-zgik.html.

Motyka, G. (2011). Od rzezi woły

nskiej do akcji “Wisła”. Konikt polsko-ukrai

nski 1943-

1947 (From the Volhynia massacre to the “Vistula” operation. Polish-Ukrainian

conict 1943-1947). Krak

ow: Wydawnictwo Literackie.

Müller, D., Sun, Z., Vongvisouk, T., Pugmacher, D., Xu, J., & Mertz, O. (2014). Regime

shifts limit the predictability of land-system change. Global Environmental Change, 28

(1), 75–83. https://doi.org/10.1016/j.gloenvcha.2014.06.003.

Munteanu, C., Kuemmerle, T., Boltiziar, M., Butsic, V., Gimmi, U., Lúbo

s Halada, …

Radeloff, V. C. (2014). Forest and agricultural land change in the Carpathian region -

A meta-analysis of long-term patterns and drivers of change. Land Use Policy, 38,

685–697. https://doi.org/10.1016/j.landusepol.2014.01.012.

Ostan, K., Iwanowski, M., Kozak, J., Cacko, A., Gimmi, U., Kaim, D., … Ostapowicz, K.

(2017). Forest cover mask from historical topographic maps based on image

processing. Geoscience Data Journal, 4(1), 29–39. https://doi.org/10.1002/gdj3.46.

Perzanowski, K., & Olech, W. (2014). The case study - restitution of the wisent (Bison

bonasus) to the Carpathians. In M. Meletti, & J. Burton (Eds.), Ecology, Evolution and

Behaviour of Wild Cattle: Implications for Conservation. Cambridge: Cambridge

University Press. https://doi.org/10.1017/CBO9781139568098.024.

Piskorski, J. M. (2011). Wygna

ncy. Przesiedlenia i uchod

zcy w dwudziestowiecznej Europie

(Exiles. Displacements and refugees in 20th-century Europe). Warszawa: Pa

nstwowy

Instytut Wydawniczy.

Pollack, M. (2010). Kaiser von Amerika. Die große Flucht aus Galizien (Emperor of America.

The great escape from Galicia). Wien: Zsolnay.

Pudło, K. (1992). Dzieje Łemk

ow po drugiej wojnie

swiatowej (Zarys problematyki) (The

history of Lemkos after the Second World War (Outline of problems)). In J.

Czajkowski (Ed.), Łemkowie w historii i kulturze Karpat (Lemkos in the history and

culture of the Carpathians), 1 (pp. 351–379). Rzesz

ow: Muzeum Budownictwa

Ludowego w Sanoku, Editions Spotkania.

Ramankutty, N., & Coomes, O. T. (2016). Land-use regime shifts: An analytical

framework and agenda for future landuse research. Ecology and Society, 21(2).

https://doi.org/10.5751/ES-08370-210201.

Ratajczak, Z., Carpenter, S. R., Ives, A. R., Kucharik, C. J., Ramiadantsoa, T.,

Stegner, M. A., … Turner, M. G. (2018). Abrupt change in ecological systems:

Inference and diagnosis. Trends in Ecology and Evolution, 33(7), 513–526. https://doi.

org/10.1016/j.tree.2018.04.013.

Rieber, A. J. (Ed.). (2000). Forced Migration in Central and Eastern Europe, 1939–1950.

London-Portland: Frank Cass.

Rozelle, S., & Swinnen, J. F. M. (2004). Success and failure of reform: Insights from the

transition of agriculture. Journal of Economic Literature, 42(2), 404–456. https://doi.

org/10.1257/0022051041409048.

Schneeberger, N., Bürgi, M., & Kienast, P. D. F. (2007). Rates of landscape change at the

northern fringe of the Swiss Alps: Historical and recent tendencies. Landscape and

Urban Planning, 80(1–2), 127–136. https://doi.org/10.1016/j.

landurbplan.2006.06.006.

Scię

zor, T., Kubala, M., & Kaszowski, W. (2012). Light pollution of the mountain areas in

Poland. Archives of Environmental Protection, 38(4), 59–69. https://doi.org/10.2478/

v10265-012-0042-4.

Skiba, S., & Drewnik, M. (2003). Mapa gleb obszaru Karpat w granicach Polski (Soil map

of the Carpathians within the borders of Poland). Roczniki Bieszczadzkie, 11, 15–20.

Skorowidz. (1868). Skorowidz wszystkich miejscowo

sci poło

zonych w Kr

olestwie Galicyi

i Lodomeryi wraz z Wielkiem Księstwem Krakowskiem (Index of all localities in the

Kingdom of Galicia and Lodomerya, including the Grand Duchy of Krak

ow). Lw

ow:

c.k. galicyjska drukarnia rządowa.

Snyder, T. (1999). “To Resolve the Ukrainian Problem Once and for All”: The Ethnic

Cleansing of Ukrainians in Poland, 1943–1947. Journal of Cold War Studies, 1(2),

86–120. https://doi.org/10.1162/15203979952559531.

Soja, M. (2008). Cykle rozwoju ludno

sci Karpat Polskich w XIX i XX wieku (Population

growth cycles in the Polish Carpathian Mountains during the 19th and 20th centuries).

Krak

ow: Instytut Geograi i Gospodarki Przestrzennej Uniwersytetu Jagiello

nskiego.

Ssekandi, J., Mburu, J., Wasonga, O., Macopiyo, L., & Charles, F. (2017). Effects of post

eviction resettlement on land-use and cover change in Uganda’s oil exploration

areas. Journal of Environmental Protection, 08(10), 1144–1157. https://doi.org/

10.4236/jep.2017.810072.

Walker, B., Gunderson, L., Kinzig, A., Folke, C., Carpenter, S., & Schultz, L. (2006).

A handful of heuristics and some propositions for understanding resilience in social-

ecological systems. Ecology and Society, 11(1), Art. 13. https://doi.org/10.5751/ES-

01530-110113.

Witmer, F. D. W. (2008). Detecting war-induced abandoned agricultural land in

northeast Bosnia using multispectral, multitemporal Landsat TM imagery.

A.N. Affek et al.

Landscape and Urban Planning 214 (2021) 104164

International Journal of Remote Sensing, 29(13), 3805–3831. https://doi.org/

10.1080/01431160801891879.

Wolski, J. (Ed.). (2016). Bojkowszczyzna Zachodnia - wczoraj, dzi

s i jutro (The Western

Boyko Region - yesterday, today and tomorrow). Monograe 17 (Vol. 1 and 2).

Warszawa: Instytut Geograi i Przestrzennego Zagospodarowania PAN.

Wolski, J. (2007). Przekształcenia krajobrazu wiejskiego Bieszczad

ow Wysokich w ciągu

ostatnich 150 lat (Transformations of the High Bieszczady Mountains rural

landscape during the last 150 years) (Vol. 214). Warszawa: Instytut Geograi i

Przestrzennego Zagospodarowania PAN.

Woube, M. (2005). Effects of Resettlement Schemes on the Biophysical and Human

Environments. The Case of the Gambela Region. Ethiopia. Boca Raton: Universal

Publishers.

Yin, H.e., Butsic, V., Buchner, J., Kuemmerle, T., Prishchepov, A. V., Baumann, M., …

Radeloff, V. C. (2019). Agricultural abandonment and re-cultivation during and after

the Chechen Wars in the northern Caucasus. Global Environmental Change, 55,

149–159. https://doi.org/10.1016/j.gloenvcha.2019.01.005.

A.N. Affek et al.