One Century of Pasture Dynamics in a Hilly Area of Eastern Europe, as Revealed by the Land-Use Change Approach

Abstract

Land use is paramount to sustainable development, and in the past, important changes happened under the influence of various factors. Revealing these changes in a meaningful manner, not just as total statistics but also as fluxes and at a spatial level, allows us to detect and associate them with the factors involved. We show a study case in Iași County, Romania, using a raster approach to change detection for a land-use-type database that extends to the 1920s. The database was created from topographic, remote sensing, and field data collected between 1920 and 2006, with five intervals between 1960, 1980, 1990, and 2000, starting from CORINE Land Cover data. These periods mark the socio-political and natural changes in the study area. The change detection results are well matched with the identified drivers and their spatial distribution. The fluctuations between land-use types provide a good way to create drivers’ associations. Our analysis can be easily applied to any other concerned areas and could be used as base references for any legislative intention to determine land-use-type changes because it can be learned from past conversions with regard to failures or examples of good practice.

1. Introduction

The current paradigm towards which modern society tends is that of sustainable development. Though its implementation is often criticized, sustainability remains an ongoing issue considering the new climate change [1] threats. The concept of sustainable development begins with the rational use of resources. Land is the most valuable resource, as it represents the support on which all human activities rely and is the source of materials which are necessary for the evolution of society. In this context, the land is seen in its physical context as the entire complexity of the physical features of the Earth’s surface, above and beyond the ground level, including vegetation and soil, but not wildlife [2]. Human use of land resources gives rise to the ‘land-use’ concept [3], which consists of all the relations between humans and the physical land seen as a resource in terms of the manipulations required for the land’s usage [4]. Depending on the purpose for which the land is used (food production, shelter provision, recreation, extraction and processing of materials, etc. [3]), the degree of anthropic modification is variable, and the land cover is more or less natural, up to the point at which we speak about human landscapes [2]. Hence, land use is inherently linked to land cover, which is considered the biophysical attribute of the Earth’s surface [5].

Land use and land cover are shaped under the influence of two broad sets of forces—human needs and environmental characteristics and processes [3]. Land-use science is born through this complex reality of land use and land-cover changes in space and time and the connections and feedback with social, economic, cultural, political, environmental, and ecological aspects [6]. The dynamics between the environment and society and their interactions are broadly expressed by land-use changes at different spatial levels and over different periods. The scientific interest in land-use change [7] has demonstrated both the central role of the changes and their complexity [8]; most of the time, the changes are unidirectional and permanent, causing unwanted consequences on the environment and the population. The causes, effects, and control of land-use change have become topics of crucial importance to contemporary society. The primary concern today is the expansion of urban and peri-urban areas, but also non-urban space [9]. Rural issues include intensification of agricultural production, animal husbandry, agricultural land abandonment, deforestation and agricultural land degradation, loss of wetlands for agricultural or other purposes, forest fragmentation by roads, and multiple appropriate management of public forests and area lands.

The historical approach to land-use dynamics is both broad and complex. The importance of the subject is evident due to the long list of real and perceived controversies that can take part in land-use change decisions, the most prominent being land reforms. The extent of land-use change also varies depending on the period examined. The size and pace of human activities on the land surface have accelerated over the past 300 years [3,8]. The majority of the land has a high degree of anthropization after 1960, and already transformed lands have been managed more intensively to increase the yield of agricultural and forest products [10].

When analyzing land-use and land-cover change, it is first necessary to conceptualize the meaning of “change” to identify real-world situations. At a fundamental level, land use and land-cover use (quantitative) refer to change according to the extent of area (increases or decreases) of a particular type of land use or land cover [3]. It is important to note that even at this level, the detection and measurement of change depend on the spatial scale; the greater the level of spatial detail, the greater the changes in the extent of land use and land cover. The latter can be detected and recorded. In the case of land-cover change, the specialized literature distinguishes between two types of changes: transformation and modification [4,11].

Land-use changes comprise two different elements: changes in land cover, i.e., a shift from one land-cover type to another, such as the conversion of forest lands to croplands, and changes in land-use intensity, also called land-cover change; that is, a change in the power with which a particular type of coverage is used [5]. Land-cover changes result from natural processes, such as climate variations, volcanic eruptions, changes in river channels or sea levels, etc. However, most current and recent land-cover changes are due to human actions [12]. Ref. [13] suggests that land use (both deliberate and accidental) alters the land cover in three ways: by transforming or changing it to a qualitatively different state; by creating an alteration or quantitative change in condition, without complete transformation; maintaining its condition against natural change agents. Similarly, land-use change can involve either (a) conversion from one type of use to another; that is, changes in the mix and pattern of land use in an area, or (b) change in a particular type of land use [3]. Thus land-use changes may be more qualitative at lower levels of spatial and temporal detail but quantitative over a broader time.

The benefit of analyzing long periods concerns land-use patterns as an effect of the long-term interaction between human society and the natural environment. This research provides information on the changes produced by the human–nature relationship, which is on the border between natural and social sciences. Like a mirror, interaction reflects the human intervention in the environment. It is an important insight, although it only provides a few clues about the nature of the driving forces. One must examine these driving forces behind land-use changes and their changing nature with equal interest.

Land-use research is not just a source of data in this endeavor. The most important contribution is the analysis and explanation of the spatial model, factors, and relationships regarding the practical use of the study area. This analysis helps reveal trends that change over time and the role of different decision levels, allowing the formulation of realistic future land-use forecasts. Moreover, the historical knowledge of land use can be the premise for realizing a conceptual framework that can offer possibilities to exploit the data obtained for optimal land use.

The paper’s main objective is to analyze the land use in Iași county, starting with the 20th century, based on cartographic materials, satellite and aerial images, and other available databases, focusing on pastureland. Throughout history, grasslands, especially pastures, have been used to produce food for livestock, fiber, and fuel [14]. The conversion of forests to livestock pastures or any other natural ecosystem into agricultural land has been frequently identified during the development of human civilization [15,16] as a primary driver of progress since the agricultural revolution that followed the Neolithic period [17,18,19]. Simultaneously, this conversion increased erosion and generated land degradation [20,21] in some areas, especially dryer regions such as the Moldavian Plateau in eastern Romania. The dryness of this region is not so high as in Dobrogea, for example [22]. Still, climate change modeling pinpoints northeastern Romania as an area with groundwater level decline [23] due to a temperature rise, despite an increase in rainfall. However, there is a lack of studies about the differences between land degradation and productivity on cropland and pastures, and on their conversions in-between. Provisional studies in Dobrogea show that cropland productivity is highly dependent on water availability [24,25], which will be a problem in Northeastern Romania in the future [23].

2. Study Area

2.1. Physical Geography

The territory of Iași County (5477 km²) is located in northeastern Romania in the central part of the Moldavian Plateau between the Moldova river valley in the west and the Prut river in the east. Geomorphologically, hills of various geomorphometry dominate the landforms of the Moldavian Plateau’s central part, a defining feature of this area [26] which is typical for northeastern Romania. Regarding the landform subunits, the area overlays the Jijia Hills in the northeast, Prut Corridor in the east, and Central Moldavian Plateau in the south. The west side limits include the Ruginoasa-Strunga Saddle, Mare-Hârlău Hill, Siret Corridor, Fălticeni Plateau, and Moldova Corridor (Figure 1).

The Jijia Hills subunit [27] is an area of low hills up to 240 m in height separated by the Suceava Plateau to the west and the Central Moldavian Plateau to the south by a steep scarp with altitude differences of about 200 m. The Suceava Plateau includes the western landform subunits overlaying the northwestern part of the county and is defined by massive relief with elevations higher than 400 m. The Central Moldavian Plateau in the southern part of the area consists of medium-height landforms from 300 m to 450 m. Due to the sandstone–limestone geology, the massive hills present plateau-like hilltops.

Climatically, Iași county is a dry area with average multiannual temperatures of 9.4 °C. However, there is an altitude-induced variability of mean temperature between the lowland Prut and its tributaries floodplains (30 to 80 m asl), the Siret floodplain (200 m asl), and the low hills (100–200 m asl) or high hills (200–600 m asl), of around 1 °C [28]. Similarly, rainfall values vary between 560 mm (Iași) and 790 mm (Bârnova), and these variations are highly dependent on the elevation and the geographical position [29,30,31].

2.2. Human and Political Context

The administrative-territorial organization of Iași County has been relatively stable over the last 300 years (for details, see [32]). The settlement network was fully established during the reign of Alexandru Ioan Cuza after 1860 when the number of settlements increased from approx. 300 to 400. The population evolution data based on the census indicate a continuous increase of inhabitants, from 300,000 in 1912 to 976,586 in 2021, doubling in the second half of the twentieth century due to the demographic policy approach of the communist regime. Most of the population settled in rural areas, but in the 19th century, the movement to urban areas started, and after 1950, this exodus produced fast urbanization, while today a balance exists.

In the inter-war period, the transition was made from a subsistence economy to a capitalist economy. Agricultural area and agricultural production have increased, with corn and cereals being predominant [33] (in 1938, accounted for 82% of the cropland), but failing to reach their full potential, especially due to the lack of technology. There was a predominance of small properties, which fragmented from year to year (in 1930, 30% of the properties were under 5 ha; in fact, many parcels were under 1 ha, and they were disposed as long, and narrow strips along the hillslopes, mainly due to particle inheritance—[33,34]—in 1941 58% of the parcels were under 3 hectares which, a surface which is considered to be in the lower part of self-sustaining cultivation). It was the industry that evolved the most, being favored by politics at the expense of agriculture. The investment increased three times between 1924 and 1939 [33].

During the communist period (1946–1989), Romanian economy was nationalized (http://www.cdep.ro/pdfs/strategie.pdf, accessed on 21 December 2022), and the private sector’s contribution was reduced (12.8%). Planning and management were hyper-centralized and very rigid. The production being often concentrated in large units, the economy’s rigidity was pronounced. After the 1980s, the standard of living dropped quite significantly, and Romania entered a stage of economic underdevelopment. After the fall of the communist regimes in Eastern Europe, Romania, together with the other former communist states, entered the capitalist economic system, which presupposed integration into the global economy. One of the major changes was the restitution to the previous owners of agricultural lands that were collectivized. The first period, from 1990 to 2000, was characterized [35] by a general economic decline due to the deterioration of economic structures and the loss of supply and sales markets, with rising inflation and unemployment and worsening living conditions in general. This transition has generated several dysfunctions, some of which are still felt today. This situation caused relative stability in land use, in general, the dynamics being marked by phenomena associated with the lack of investments and resources to maintain the situation during the communist period.

During the period after 2000, there has been an improvement in the economic situation, with an increase in all economic branches and an increase in investment. This situation directly generated the expansion of built-up land to the detriment of cropland to an unprecedented level. During this period, many mistakes were made in converting land use, despite there being more and more talk of sustainable development and land-use planning. This is also most evident in the loss of forest lands, vineyards, orchards, and croplands.

3. Materials and Methods

3.1. Materials

In the Moldavia region of Romania, for widespread analysis of the evolution of land use, maps can be used only after 1890–1920, as the information for previous periods is scarce and is represented by archaeological discoveries or historical documents that cannot be georeferenced. An important criterion in choosing these sources was their eligibility for digitizing land use. The oldest cartographic representations from the 18th century are the generalized interpretation of the Earth’s surface with little accuracy and partial coverage. For this reason, they do not meet the mandatory requirements and have not been included in the analysis.

The used data sources (

Table 1. The cartographic and remote sensing sources used for the land-use digitization.

Data Source	Type	Scale	Resolution	Field Year
Moldavian Topographic Atlas, Cassini projection	Topographic map	1: 50,000	-	1893–1894
Army plans, Lambert-Cholesky projection	Topographic map	1:20,000	-	1916–1959
Topographic maps	Topographic map	1:25,000	-	1954–1959
	Topographic map	1:25,000	-	1972–1981
	Topographic map	1:5000	-	1972–1989
	Topographic map	1:2000–10,000	-	1980–1989
Aerial topographic surveys	Aerial imagery	-	0.25–0.5 m	1964, 1978, 1982
Corona KH-4B satellite imagery	Satellite imagery	-	1–25 m	(1960–1974)
Landsat 5–7 satellite imagery	Satellite imagery	-	15–30 m	(1970–2006)
Orthophoto-imagery	Aerial imagery	-	0.5 m	2005, 2008
World Imagery	Satellite and aerial imagery	-	one meter or better	update every 3–5 years
CORINE Land Cover	Land-cover	1:100,000	25 ha, 100 m	1990, 2000, 2006

) have full coverage for the study area. To simplify the presentation, the six datasets used in this study were labeled 1920, 1960, 1980, 1990, 2000, and 2006. In reality, for the 1920 to 1980 datasets, the field data have variable timing around the framed years: (i) for 1920, the contours and the river network of the Moldavian Topographic Atlas (1:50,000—Cassini projection) and the Army plans (1:20,000—Lambert-Cholesky projection) are similar, while the forest cover and the settlement network are different; since the pastures are represented by conventional signs and the cropland is not represented at all on the Army plans, the aerial images from the 1950 dataset and the presence of agricultural roads on the Army plans were used to infer the pasture and cropland for the 1920 dataset; (ii) for 1960, since the pastures are represented by conventional signs and the cropland is not represented at all on the 1960–1962 topographic maps, the aerial images from 1950 and the presence of agricultural roads were used to infer the pasture and cropland for the 1960 dataset; (iii) for 1980, there is a long period of aerial imagery used in the 1984 printed topographic maps: 1972–1981; additionally the Corona high-resolution imagery is usable for the cropland vs. pasture differentiation.

The georeferencing process ensured the alignment of all the sources with a single defined coordination system, namely the Stereo 70 projection, S-42 Romania datum (EPSG:3844). Corner coordinates were used for the topographic maps, while georeferencing ground checkpoints were used for the non-topographic sources, such as the Corona images. These elements of the map recognizable on different datasets (for example, on the topographic maps) are assumed not to have changed position for a long time [36], e.g., intersections, roads, bridges, places of worship, etc. Despite the possible errors due to the georeferencing process, the digitization was performed in such a way that the errors were minimized.

The methodology of land-cover extraction was the following: (i) the 2000 CLC dataset was used as a reference regarding the level of generalization, and (ii) for 1990, 2000, and 2006 datasets, the CLC data were checked against satellite and aerial imagery and field checks, some errors being resolved; (iii) for 1920, 1960, and 1980, the presented data sources were used to obtain CLC-style land cover; the newer temporal dataset was used to assess if there was a change in land-cover, and if there was a change, this was implemented; otherwise, there was no change to the polygon; so, starting from the 1990 CLC dataset, any changes found in the 1980 datasets were applied to the 1990 polygons and so on, down to the 1920 dataset.

After the land-cover extraction, the land use was attributed according to

Table 2. The equivalence between the CLC classes and the land-use types used in this study.

CLC Level 1	CLC Level 2	CLC Level 3	Land-Use
1 Artificial surfaces	11 Urban fabric	111 Continuous urban fabric	1 Urban, built-up and related land
		112 Discontinuous urban fabric
	12 Industrial, Commercial, and transport units	121 Industrial or commercial units
		122 Road and rail networks and associated land
		123 Port areas
		124 Airports
	13 Mine, dump, and construction sites	131 Mineral extraction sites
		132 Dump sites
		133 Construction sites
	14 Artificial, non-agricultural vegetated areas	141 Green urban areas
		142 Sport and leisure facilities
2 Agricultural areas	21 Arable land	211 Non-irrigated arable land	2 Arable agricultural land (cropland)
		212 Permanently irrigated land	-
		213 Rice fields	-
	22 Permanent crops	221 Vineyards	3 Vineyard land
		222 Fruit trees and berry plantations	4 Orchard land
		223 Olive groves	-
	23 Pastures	231 Pastures	5 Pastureland
	24 Heterogeneous agricultural areas	241 Annual crops associated with permanent crops	2 Arable agricultural land (cropland)
		242 Complex cultivation patterns
		243 Land principally occupied by agriculture, with significant areas of natural vegetation
		244 Agro-forestry areas	-
3 Forest and semi-natural areas	31 Forests	311 Broad-leaved forest	6 Forest and other wooded lands
		312 Coniferous forest
		313 Mixed forest
	32 Scrub and/or herbaceous vegetation associations	321 Natural grasslands	7 Natural grassland (meadows)
		322 Moors and heathland	-
		323 Sclerophyllous vegetation
		324 Transitional woodland-shrub	6 Forest and other wooded lands
	33 Open spaces with little or no vegetation	331 Beaches, dunes, sands	8 Open land without, or with insignificant, vegetation cover (barren land)
		332 Bare rocks
		333 Sparsely vegetated areas
		334 Burnt areas	-
		335 Glaciers and perpetual snow
4 Wetlands	41 Inland wetlands	411 Inland marshes	9 Wet open land (wetland)
		412 Peat bogs
	42 Maritime wetlands	421 Salt marshes	-
		422 Salines
		423 Intertidal flats
5 Water bodies	51 Inland waters	511 Water courses	10 Waters
		512 Water bodies
	52 Marine waters	521 Coastal lagoons	-
		522 Estuaries
		523 Sea and ocean

.

3.2. Methods

Change detection in the context of remote sensing [37] is defined as the process of identifying differences in the state of an object or phenomenon by observing them at different times. This method involves the ability to quantify temporal effects using multitemporal datasets. Detection of changes in the GIS environment is a process that helps to measure the intensity of changes between two or more time intervals. Change detection often involves comparing two satellite images or two aerial photographs with different acquisition data covering the same area of interest [38].

The process of detecting changes is a standard method used to analyze land-use dynamics. It involves using two or more spatial layers, by which a comparison determines if there are changes in the categories and classes represented by them. In general, the detection of changes is carried out on raster layers. There are several change detection methodologies depending on the type of aspect targeted.

The most straightforward methodology involves the raw local difference of pixels [39] between two rasters (representing two different time intervals t₁ and t₂), using integers as codes for land-use categories and their combinations. Creating a contingency matrix for 1, 2, or more categories allows the computation of descriptive statistics and the overall fluxes between classes (Figure 2). The spatial results can be shown by styling the resulting raster with combination codes. This approach is implemented in SAGA GIS [40] as the Confusion Matrix (Two Grids) module from Imagery/Classification tool library. We used the R stat [41] coding environment coupled with the raster [42] and RSAGA [43] packages to filter the spatial results for the generalized classes of no change, positive and negative, and to derive Sankey plots [44]. The usage of the raster approach with its problems related to the pixel size constraints and the variations of the boundaries due to manual digitization and/or georeferencing errors induced edge errors, where small positive or negative errors appear along the borders of the land-use polygons that are rasterized for the application of raster algebra. In order to remove these artifacts, we used the Local Moran’s I measure [45] to identify the one-pixel edges in the difference raster, where these pixels will be surrounded by other class pixels. In this case, the Local Moran’s I measure close to zero shows spatial randomness and a lack of correlation with the neighborhood, which is precisely the case for edge errors. These edges were generalized to the majority class of the neighborhood. The 5 m resolution was chosen as a good compromise between accuracy, computation time, and result size.

The presented change-detection approach allows us to assess the gains and losses between land-use/land-cover types in a region over several periods and to provide the following information [38]: (i) change of area and type of change, (ii) spatial distribution of modified classes, (iii) the trajectory of the changes of the land-use classes, and (iv) evaluation of the accuracy of change-detection results.

4. Results

The results of the land use extraction for the six periods and five intervals are shown in Figure 3 and in the supplementary Figures S1–S5. The correction of the CLC datasets (Figures S3–S5) against remote sensing and field data showed up to 5% positive changes in the pastureland-cover class (from the total surface of the county), which are mainly attributed to the incorrect consideration of the pastures as wet areas or croplands. This situation is characteristic of floodplains. Here, depending on the climatic and hydrologic regime, there might be periods when water is present at the surface. The area will be classified as a swamp by the interpreter. The majority of the time, the area is covered by grass and used as a pasture, but this will be found if multiple sources are used, and fieldwork is performed. At the same time, these pastures areas, in some years, might be used as croplands, so deciding upon the land use requires investigation of multitemporal images to see if the arable use was present not more than three years to maintain the pasture use class [46].

The total sums of land-use surfaces are shown in Figure 4, and the dynamics of balance (losses and gains) in Figure 5, as an overall of the land-use changes. The Sankey plots that present the changes from and to pasturelands are shown in Figure 6 and Figure 7.

In the 1920 to 1960 period, the pasture surfaces increased considerably in the study area (Figure 3, Figure 4 and Figure 5), with spatial gain resulting from the abandonment of some agricultural lands, the deforestation of some forest areas that failed to regenerate, and the drying of the lands with excess humidity. In this interval, the conversion of pastures to cropland was the most extensive (Figure 4, Figure 5, Figure 6 and Figure 7). One can see the close connection between the land-use classes of cropland and pastures (Figure 6 and Figure 7). Cropland is also increasing, especially in wetland conversion. Spatially (Figure 8), the wetland to cropland conversion is the most intensive in the major river floodplains of Prut catchment: Prut, Jijia, Bahlui, and Jijioara (Figure 8a–c). The conversion of forest and cropland to pastures (Figure 8) is widespread in all the administrative units, as is the conversion of pastures to built-up and cropland (Figure S9).

The 1960–1980 period (Figure 4, Figure 5, Figures S1, S2, S6, and S9) is mainly characterized by the decrease of cropland and pastureland class, with the increase of urban and built-up, orchards and vineyards at balance level. Regarding the fluxes between land uses, we can remark (Figure 6 and Figure 7) that the cropland conversion to pastures and vice-versa characterize the main changes. Spatially cropland is converted to pastures at all administrative levels (Figure 6), while pastures are converted to cropland along the major floodplains of Siret and Bahlui river and in built-up urban areas (Figure 9).

The 1980–1990 period (Figure 4, Figure 5, Figures S3 and S7) is mainly characterized by the continuation of the decrease in pastureland class, all the other classes having low variations in balance. Regarding the fluxes between land uses (Figure 6 and Figure 7) we find the same changes as those in the 1960–1980 period, but with a lower intensity. Spatially, cropland is converted to pastures mainly along the floodplains (Figure S7), while pastures converted to forest land on hillslopes are affected by erosion and gullying (Figure S10).

The 1990–2006 period (Figure 4, Figure 5, Figures S4, S8 and S11) is mainly characterized by the increase in pasturelands and urban and built-up areas, while there is a decrease in orchards and vineyards land uses at balance level. The fluxes of land-use changes are still mainly characterized by the conversions from cropland to pasture and vice versa. Spatially, conversion to pastures is widespread (Figure S8), while clusters of big spatial change from pastures are located on the floodplains and in the proximity of the urban areas (Figure S11).

According to the data obtained (Figure 4 and Figure 5), pasturelands’ surface increased considerably after the 1920s, up to 1960 and between 1990 and 2000, with a decrease between 1960 and 1990. A low increase appeared between 2000 and 2006. As fluxes (Figure 6 and Figure 7), the pasture to cropland and vice versa occurs the most frequently at all temporal levels, with the exception of 1920 to 1960, when the conversion of wetland to pastures is higher than those of cropland to pastures. Spatially (Figure 8, Figure 9, Figures S6 and S11) changes are medium to low in size and widespread, with the exception of the conversions along the major floodplains and around big urban areas.

5. Discussion

From a physical–geographical point of view, taking into account that the foundations of land use in the studied area were laid in the medieval and modern periods, the general land use was influenced by physical and geographical factors, both in terms of favorability and restrictions:

(i)

The major landforms, characterized by the presence of a highland area in the central-eastern part, imposed an extension of the inhabited areas from the contact area to the external depression areas from the west and east and inside the highland areas;

(ii)

The minor landforms, represented by the ridges and structural interfluves and structural valleys with very wide major floodplains on one side and the Siret and Prut rivers valleys with major floodplains and extended terraces, imposed the location predominantly on river terraces or slopes;

(iii)

The general direction of winds, excess humidity, and flooding of floodplains [47] has, until recently, restricted the use of structural plateaus, hillslopes, and major floodplains as inhabited areas;

(iv)

The uneven loess layer distributed at the level of the cuesta dip slopes or plateaus interfluvial ridges [48] generated the presence of the aquifer at great depths, and the chernozem soils determined a favorability for arable agriculture so that these areas were dedicated exclusively to croplands; the slopes of the steep cuesta hillslopes, through the geomorphological processes controlled by the slope and lithology, favored the development of pastures;

(v)

On the moderately sloping slopes, but very frequently affected by landslides of the translational type, relict and stabilized, the village built-up areas were favored, especially by the climatic shelter and the presence of groundwater from the landslides at shallow depths [26]; the reactivation of landslides was counteracted by the use of traditional materials for the construction of houses, which could be easily rebuilt; however, there are frequent cases in which the village has been moved due to the exacerbation of these phenomena [49].

In the inter-war period, deviations from the optimal model imposed by the physical-geographical factors appeared through:

(i)

The extension of cropland on sloping areas to the detriment of pastureland, even if this measure has led to soil erosion;

(ii)

Against the background of some dry years, a series of floodplains naturally drained due to the decrease in the phreatic level and were used as cropland or pastureland;

(iii)

On the background of the extension of the inhabited areas, there is the choice of some inhabited sites that will prove to be unsuitable, demonstrating the effects of landslides and bank erosion [36].

During the communist period, the optimal physical and geographical use of the land was supplemented by a series of measures adopted to minimize the various types of restrictions imposed by nature:

(i)

The drainage of floodplain areas, where the cropland and construction of reservoirs were introduced, which diminished flood risk, allowing the extension of the built-up lands;

(ii)

The stabilization of hillslope areas affected by geomorphological processes, either by plantations or by specific arrangements (terracing, drainage), areas that could be converted into vineyards, orchards, cropland, or even built-up land;

(iii)

Additionally, during this period, through measures of conversion of the land use, a series of “mistakes” made during the inter-war period were eliminated, especially at the level of the use of hillslopes as croplands.

The persistence of land use with loss or gain around the same spatial core is noted, with certain exceptions regarding specific categories or temporal intervals, where transitions from one land-use category to another happened.

The temporal intervals for which spatial use data exist, and which were used in this study, capture a series of stages with well-defined socio-economic and political characteristics that influenced land-use dynamics: (i) the inter-war period, with milestones represented by the 1920 and 1960 datasets; (ii) the first part of the communist period: the 1960 and 1980 datasets; (iii) the last part of the communist period, the 1980 and 1990 datasets; (iv) the post-communist period, 1990, 2000, and 2006 datasets.

Land for pastoral use most often overlaps with low-fertility soils (soils with an A ochric horizon suitable only for occasional crops every few years), and erosion processes do not favor agricultural production that makes field crops sustainable, or drained meadows, where the restrictions are also of a pedological (alluvial soils, predominantly clayey or with a skeleton) and hydrological nature (shallow or fluctuating groundwater table will influence the development of gleyed horizons).

Regarding land-use change, the pasture class is often a transition class. If we analyze a sequence of changes in land use, this class can be considered an intermediate class between deforested land and its introduction in time as agricultural land. This legitimacy is used as a matrix in land-use prediction models [50]. In the studied area, the pasturelands had the same role as the intermediate class in relation to the croplands, in the sense that when the situation demanded it (extensive agriculture from the inter-war period), the pastures were converted into croplands, and vice versa (during the communist period when spatial planning has been applied).

The local organization of the communities also conditions a large number of pasture areas, so almost every community has at least one communal pasture area (called “islaz”), necessary for grazing, located either on floodplains or on steep hillslopes.

In the current approach, we did not use 2012 and 2018 datasets because of their increased resolution (derived from satellite data interpretation) and methodologic changes that can generate confusion [51] at the regional or country scale. In that sense, a representative example is provided by built-up areas, which were extracted with a “house-size” resolution, a much more detailed insight compared with previous datasets. For example, if a pasture on a floodplain was located inside a built-up area, in the pre-2012 datasets, this was included in the built-up area due to the size restrictions of the CLC methodology, while in the post-2006 datasets, this pasture was extracted as a separated polygon. Because the work for the pre-CLC datasets was carried out between 2014 and 2018, the 2012 and 2018 datasets were not available for correction, and a correction to match their increased resolution will require the reconsideration of all the datasets (and this might not be feasible, since the improved resolution might not be available from 1920s sources).

The 1920 land-use dataset presents the situation related to the Kingdom of Romania, before the Great Union (in 1918), with its effects at the level of administrative reorganization and agrarian reforms. These effects are cumulatively observed in the dataset from 1960. This dataset also partially includes a series of effects of the measures taken by the communist regime, but it is much smaller because the 1960 set presents the situation on the ground from 1954–1959. Additionally, the agrarian reforms of the communist regime were finalized in late 1960. The main vectors of land dynamics during this period were agrarian reform, population growth, modernization of society, and economic growth.

Regarding the landforms’ influence on land use, a series of considerations can be made by relating the slope and the landforms (Figure 10) to each type of land use. The major floodplains were separated [52,53] from the gently inclined and the steeply inclined hillslopes: the slope value of 8.1° was chosen as a threshold to separate the hillslopes, corresponding to the value of 18%, which is the value above which agrotechnical measures were applied [54].

Cropland and pastureland slope statistics (Figure 10a) show a decrease in slope for cropland and an increase for pastureland, but this situation is clear when the statistics are computed without considering the floodplains. When floodplains are included in the hillslope computation, the pasturelands slopes decreased between 1920 and 1960, a situation explained by the conversion of the wetlands from floodplains to pastures between 1920 and 1960. The importance of considering landforms and land use, besides slope, becomes clear considering the situation presented above, so an analysis of land-use type by landforms is shown in Figure 10b. Cropland has been extending continuously to floodplains and decreasing from steep hillslopes since 1920, with a light increase between 2000 and 2006. Pasture proportion on floodplains increases abruptly from 1920 to 1960, decreasing continuously but not reaching 1920s levels. The proportion of steep slopes covered by pastures decreases from 1920 to 1960, then slowly increases to levels slightly higher than in 1920. These statistics show that cropland extended between 1920 and 1960 on steep hillslopes while pastures decreased. After 1960, the croplands from steep hillslopes were converted to pastures.

From 1945 to 1960, there was a dry period [31,55] during which many reservoirs emptied, a situation recognizable on the 1960 topographic maps. This situation probably favored the conversion of the wetlands to pastures, but at this point, we cannot estimate which was the main driver: the climate or the human intervention through drainage work.

Within the territory of Iași county, after the inter-war period, during the communist regime, several important changes took place at the level of several types of uses. These were imposed by a series of measures that mainly aimed to protect the urban areas from the effects of floods, which were sometimes catastrophic [47]. Thus, in the first few years of the communist regime, the Iași Branch of the Romanian Popular Republic Academy of Sciences (1952) drew up the Technical–Economic Report on the comprehensive hydro-technical development of the Iași geographical region, material that was the basis for arguing the need for intervention. The intervention involved the complex hydro-technical arrangement, provided by flow control through the construction of reservoirs and levees, followed by the management of excess moisture through drainage systems in the floodplains. Later, soil-erosion control measures were imposed both to minimize the effect of erosion on reservoirs (clogging them with the sedimentary material produced by erosion) and to improve the quality of the land through land-improvement works [54].

Massive investigations were made with regard to reclamation works for soil-erosion control measures, for the creation of an irrigation network that would use water from the Prut river and a series of reservoirs to counteract the effect of droughts on crops (which had modest productions, even in the context of some qualitative soils). These works took place until after 1980 in a timeline that can be summarized in certain specific stages: (i) rectification of the channels of the Bahlui, Jijia, and some tributaries (Bahluieț)—1940–1960 with levee construction; (ii) construction of the large reservoirs, stage 1960–1965; (iii) completing the network of reservoirs: 1975 stage; (iv) development of the Jijia River works for flood control (after 1980); (v) drainage works; (vi) arrangement of irrigation systems; (vii) soil-erosion control measures works; (viii) drainage, irrigation, and soil-erosion control works in Siretului floodplain between Pașcani and Roman (after 1988); (ix) completion of the flood protection network through the construction of permanent, non-permanent accumulations and polders, after 1980, which continues today.

The 1980 and 1990 datasets cumulatively highlight the effects of the communist period, with the 1990 dataset being a benchmark for the changes that characterized this period. The 1980 dataset is characterized by the main effects induced by the communist regime in the period up to 1990, maintaining the same trends, with lower values, but also providing a series of variations introduced by the characteristic of this period, when the communist regime restricted a series of investments.

The main vectors of land-use dynamics during the communist period were the nationalization of land, the generalized implementation of territorial planning, and population growth.

Considering the obtained results on pasture dynamics, we can summarize that the situation of cropland and pastureland in the county’s floodplains is supported by spatial data, remote-sensing images, and scientific and historical sources. During the inter-war period, important areas of the floodplains, which were occupied by marshes, pastures, or meadows, were cultivated due to the need for agricultural resources. We cannot make accurate statements regarding the conversion method. We can only support the hypothesis that, depending on the hydrological regime specific to each sector of floodplains or the intensity of droughts [31,55], there were years when these surfaces became free of excess moisture, naturally becoming meadows that were used as pastures or even as cropland. This type of land conversion was certainly not permanent (as the maps show), depending on the hydrological regime, in a direct relationship with the fertility of the alluvial soils. After 1960, in some areas (Figures S6 and S7), due to the hydro-technical and drainage work on the floodplains, these pastures and some cropland replaced the wetlands.

The steep hillslopes there were used as cropland during the inter-war period, especially due to the over-fragmentation of the parcels visible on the aerial imagery from 1950 (Figure 11). After 1960, many of these cropland parcels were unified and converted to pasture after the collectivization of the agricultural lands (Figure 12), or even forested as soil-conservation works (Figure 13 and Figure 14). After 1990, the land was retroceded to the previous owners in the same location and some croplands still present over-fragmentation (Figure 15), although some parcels are located on plateaus (Figure 13).

The restitution of agricultural lands was made based on Law 18 of 1991, which provided restitution on the old site in the hilly area, possibly on the old site in the plain area and not necessarily on the old site, in an area of a minimum of 0.5 ha per person and a maximum of 10 ha per family. The unclaimed lands remained in the common property of the respective territorial-administrative unit. The regions occupied by orchards, vineyards, greenhouses, ponds, fish farms, nurseries, administrative and agro-zootechnical constructions, as well as those necessary for the fodder base related to the existing zootechnical production capacities in the agricultural production cooperatives, were favored by law to form some association. People who were not owners before collectivization also became owners. The existing urban property on 1 January 1990 was preserved regardless of the old owners, who were compensated on request. Former CAPs (“Cooperativa agricolă de producţie”—Collective Agricultural Production) were able to gain legal status by becoming companies, and former employees were able to become shareholders. The communal settlements remained in the administration of the respective territorial-administrative unit.

This law also provided a series of rules about changing the category of land use and state funding of works to prevent and combat soil degradation and pollution processes, but which were not implemented often because their implementation involved many costs that were not covered by the state of the new owners. These costs, which during the communist period were borne by the state, could not be paid by individuals, so a number of conversions came about naturally. Thus, many lands with vineyard and orchard use were either allowed to degrade and transformed into pastures or were destroyed and plowed and thus transformed into cropland. Conversions often violated the law, as Law 18 clearly imposed for the conversion of agricultural land taxes paid on such conversions to a fund from which should subsequently finance the land reclamation works. The law was violated this way with the tacit consent of the local, county, and national authorities. Law 18 also established the legal framework for future grouping and amalgamation of land, the possible rectification of irrationally located plots, and the rational use of agricultural holdings based on studies and projects.

6. Conclusions

From an applied perspective, the aerial photo interpretation of the remote sensing databases, map analysis and the change-detection methods used allowed us to identify the spatial distribution of the land use and their dynamics for the last century in Iași County, Northeastern Romania. The spatial distribution and the main paths of changes that were identified allowed us also to associate, in a meaningful manner, the processes that drove the land-use change: urbanization, agricultural conversions, management of flood-prone and of wetlands areas, and management of forest areas, respectively, with other changes that appeared due to various processes.

Agriculture experienced two stages of development: the extensification of the inter-war period and the intensification of agriculture during the communist period. These phenomena could be quantified with the help of the methodology we used. The extension of the cropland for the period between 1920 and 1960 on the steep hillslopes, which were previously covered by pastures, is a process that mainly characterizes the study area and which is clearly shown by the analyzed datasets. This was an organic evolution due to undeveloped agrotechnology, social and historical changes, and anthropic pressure. The wetlands naturally changed to pastureland or cropland during dry years. After 1960, hydro-technical works allowed the preservation of cropland, but especially pastures on the floodplains, since flood protection measures removed the excess water and the wetlands.

The communist regime established after 1945 planned and implemented measures for soil conservation that reestablished pastures on steep hillslopes; also, in the context of a drought period in 1950 and the need for flood protection, floodplain wetlands were converted to pasturelands and croplands along all the floodplains in the study area.

After 1990, the changes in pasturelands were not so intense, but there was an extension of pasturelands on unmaintained croplands (especially in the 1990s).

In summary, we identified that the pasturelands were dominantly converted to cropland during the big inter-war crisis and later, during the communist period, from cropland to pasturelands as a consequence of centralized land-use planning. After 1990, the surfaces of pasturelands continues to grow as a consequence of agricultural land abandonment [32]. The knowledge of the past changes in land use related to pastures should be the basis of future sustainable politics regarding this land-use type. Our results need to be considered when policies target conversions because considering that we identified the drivers and related the changes to them, both statistically, but also spatially and as fluxes, much can be learned from the previous conversions.