The spatial distribution and expansion of Eucalyptus in its hotspots: Implications on agricultural landscapes

Abstract

Fast coppicing plantations like Eucalyptus are becoming an ever increasingly important land use system globally, including the Eucalyptus hotspot highlands of Northwestern Ethiopia. However, comprehensive information regarding species composition is essential for proper planning and policy decisions. The current study mapped the spatial distribution of Eucalyptus globulus (hereafter referred to as Eucalyptus) and identified the key push factors for its expansion. The study used a mapping procedure that uses Landsat imagery together with ground truth data based on supervised training of a pixel-by-pixel classification algorithm within image regions to distinguish areas of Eucalyptus plantations from other classes. High-resolution multispectral and multi-temporal remote-sensing images were combined with ground truth data to produce robust features of Eucalyptus plantation distribution maps. Heckman's Two-Stage econometric model was also employed for determining the major driving factors of Eucalyptus expansion. The results of the mapping algorithm were Eucalyptus plantation distribution maps of 30 × 30 m resolution that showed temporal changes from 1999 to 2021. The findings revealed that Eucalyptus coverage increased by 55% during the period from 1999 to 2010 and the change expressively increased to 69% in 2021 with respect to the reference period. The study also found that a number of push factors influenced the size of land planted with Eucalyptus. The developed maps showing the spatial distribution and expansion of Eucalyptus will help policymakers properly manage the ecosystems and agricultural landscapes of Eucalyptus growing areas.

1. Introduction

Ecological changes due to land management dynamics are thought to be attributed to the conversion of different land covers to fast coppicing plantations like Eucalyptus. Originating from Australia, several species of the genus Eucalyptus were introduced to different parts of the world in the late 18th century [1]. Later, large-scale plantations of Eucalyptus were intensified due to the increasing need for wood processing industries and the construction sector [2]. In Ethiopia, Eucalyptus globulus (also named as the Tasmanian blue gum), was introduced in 1895 by Emperor Menelik II [3]. For its fast-growing tendency and high economic contribution, the expansion of Eucalyptus was supported by different countries [1]. Given the industrial contributions of Eucalyptus, many studies have shown the apparent undesirable effects of the tree on the environment and ecosystems [4,5].

Today, extensive areas are being converted to Eucalyptus plantations and ultimately resulting in the alteration of the cultural landscape and the natural ecosystem. As a quick-growing and invasive species, Eucalyptus can disrupt the natural composition of habitats and biodiversity, and thus monitoring and managing such species should be given much emphasis [6,7]. Mapping enables to design management plans, especially for plant species of invasive behavior that are serious threats to biodiversity. In order to map and monitor forest areas at wider spatial and temporal scales, remote sensing is a commonly applied technique. These days, the uses of Landsat imagery with different techniques are considered the fundamental approaches for producing wider-scale land cover change maps and environmental management.

Following the advancement of observatory sciences, many regional or global land cover maps have been produced [8,9]. Mapping vegetation at the species level can help monitor their size and spatial distributions and design specific modeling for the inhabiting tree species [10]. In order to map species of certain vegetation, remote-sensing data have been commonly used for their wide availability and palatability for analysis [11]. Tracking such land cover changes at a regional scale is of critical importance to analyze the transformation of the cultural landscape of an area and the impacts of such changes on environmental policies. Land cover maps are usually produced from the classification of remote sensing data [12]. There are different types of classification algorithms that include supervised decision trees, unsupervised classification followed by post-classification refinement, and unsupervised and clustered supervised classifications [13]. However, maps developed using such classification algorithms have coarse spatial resolution and have limited importance for monitoring specific vegetation maps.

The intriguing question of how specific tree species are spatially distributed has inspired many scholars within plant ecology [14]. Mapping vegetation at the species level is however quite difficult because of spectral mimic nature of bands and within-class variation arising from species diversity, environmental conditions, and other factors like farming activities [15,16]. Because of this, supervised classification algorithm was employed for mapping Eucalyptus plantations in the study area. Various studies employed the method for mapping individual species in different parts of the world. Recent advances in remote sensing enable the mapping of actual species distributions, yet previous studies remain at a local scale [17,18]. A study by Ref. [19] observed inter-seasonal spectral variation that allowed the accurate discernment of species of trees using automated classification methods. Following the launch of many satellites like Landsat, Sentinel, and Moderate Resolution Imaging Spectroradiometer having different temporal and spatial resolutions, numerous studies were carried out. Images obtained from such satellites have different resolutions ranging from less than 1 m to greater than 100 m 20]. The resolution feature of the images determines the type of map to be developed. Today, images of lower resolutions are required for mapping vegetations while plots lower than 0.0256 km² are difficult to map as they appear as mixed pixels of different species [21].

Satellite imagery was widely used in crop-type mapping with 30 m spatial resolution [22]. The launch of a 10 m Sentinel-2 satellites with high temporal resolution resulted in further improvements in crop-type mapping and marking of regional or field borders [23]. A study by Ref. [24] showed the conveniences of mapping different crop types in different landscapes using Sentinel-2 imagery and machine learning algorithms. Another study by Ref. [25] mapped individual tree species in the temperate forest using Sentinel −2 images.

Supervised image classification methods utilize the spatial information of pixels for reducing the effect of stippled noise in classified maps. However, many automated classification techniques fail to map specific vegetation and are characterized by a high tendency of misclassification. The rapid expansion of Eucalyptus in the study area has drawn ecological and social attentions from two perspectives. On one hand, fertile lands that are suitable for crop production are considerably diminishing and on the other hand, the conversion of cultivable lands to Eucalyptus plantations is changing the ecology. This calls for proper management and monitoring of the current land use system and predictions for the future. Eucalyptus with such an expansion trend can change the ecosystem, especially the species composition of the habitat, there should be a mechanism for monitoring such changing species. Employing remote sensing approaches and field surveys can help monitor such alarmingly expanding species [26]. Though there are previous studies carried out in the study area including [6,27] their main focus is on land use land cover changes for the whole set of classes. Studies that specifically focus on mapping Eucalyptus vegetation at the species level are lacking in the study area.

The main goal of the current study is therefore to map the spatial distribution of Eucalyptus in the study area and examine the major driving factors of its expansion. The study explicitly shows the temporal and spatial trends at the expense of other land classes which can help design future management policy options. Temporally, Eucalyptus expansion maps covered from 1999 to 2021 and showed how remotely sensed high-resolution, multi-temporal information of plantation covers can help policymakers and environmentalists and ecologists make proper ecological planning. Unique to this study is that it is the first to address species-level mapping for showing the trend of Eucalyptus plantation expansion in the study regions.

2. Materials and methods

The flow of the current study is as follows: selection of study regions, choice of data sets, acquisition of satellite and socioeconomic data, selection of appropriate algorithms for mapping and spatiotemporal analysis, analysis of socioeconomic data, and presentation.

2.1. Study site

The current study was undertaken in the Eucalyptus hotspot areas of Northwestern Ethiopia with a geographic coverage of 10°21′23.59″ to 10°37′28.05″N and 37°53′09.02″ to 37°40′25.95″E. This region covers the major Eucalyptus growing districts namely, Senan, Gozamin and Machakel covering about 241,129 ha. This location featured by varying topography and altitude (2500 masl- 3900 masl). plantations. Rainfall in the area is unimodal with average annual of 1300 mm usually occurring in the months of May to September with a yearly average temperature of 18 °C.

In terms of agroecological setup, the study region is characterized by diverse landscape features with slopes ranging from nearly plain to very steep (greater than 45%) slopes. The majority of the land is used for the cultivation of food crops while Eucalyptus plantations take the next highest share. Commonly grown food crops in the area include maize, teff (Eragrostis tef), wheat, barley, potato, and beans. Eucalyptus is the most vastly planted exotic tree plantation in the study area [28]. Map of the study area is located in Fig. 1.

Fig. 1.

Map of the study sites.

2.2. Satellite data

2.2.1. Data acquisition and preparation

Multi-temporal high-resolution data from Landsat4-5 (C2L2) and Landsat8 C2L2 satellite images of the years 1999, 2009, and 2021 were downloaded from USGS (2022) official site. Images were then processed and made ready for the interactive supervised classification method. Using ground truth data, Eucalyptus maps developed were validated across the study regions. The base year was considered as land re-allocation which initiated dramatic expansion of Eucalyptus plantations was done in 1997. However, due to the absence of quality images in this year, the study made use of Landsat images of 1999 as a reference period. After about 11 years, there have been social conflicts of interest associated with afforested lands including shading and allelopathic problems to the neighboring plots. In 2021, additional areas were succeeded with the Eucalyptus plantation. In order to have a clear sky season, cloud-free images (0 ground cloud cover) were acquired. Surface features were identified with the interpretation of true and false color composites. With the integration of existing topographic maps and administrative data (secondary), land cover changes were classified and mapped with the help of Arc GIS 10.8.

2.2.2. Land use classification and change detection

In order to categorize each of the Landsat images digitally, a hybrid classification technique was employed as this technique better enables identifying the spectral variability of individual cover types [29]. Adopting a hybrid classification method is preferred over unsupervised or supervised classification and it has been suggested by various researchers [[30], [31], [32]].

Assessing land use patterns for scientific utilization of the given area and existing natural resources has become predominantly important. Recently, the alarmingly increasing population has led to severe resource completion and a decline in the use value of natural resources. Analysis of land use land cover changes with wider spatial and temporal dimensions can help ensure optimized utilization of the existing natural resources. Land use mapping helps examine the best possible optimal utilization of resources and scientific management of a given area with its natural resources [33]. The application of recent methods of GIS and remote sensing is vital for capturing spatiotemporal land cover changes emphasizing on tracking the conversion of cultivable lands to Eucalyptus plantations.

In order to examine changes in Eucalyptus plantation cover for the last 22 years, the expansion rate and conversion of arable lands with Eucalyptus plantation were estimated through land cover changes using GIS. High-resolution successive Landsat images (with a resolution of 30 × 30 m) of the years from 1999 to 2021 were taken for the analysis as the multispectral character and spatial resolution of Landsat imagery is much preferred for studies focusing on environmental monitoring and agricultural sciences [34]. The land use classes considered in this study are shown in Table 1.

Table 1.

Land use classes of the study area.

S. No	Land Use Class	Description
1.	Eucalyptus	Plantations of Eucalyptus
2.	Other forests	Forest plantations which are both native and indigenous located in various landscapes
3.	Croplands	Cultivable lands that are assigned for the production of food crops
4.	Grazing lands	Communal lands which are left for free grazing
5.	Bushlands	Lands with short vegetation cover
6.	Built ups	Residential and non-residential buildings
7	Water bodies	Rivers, open water sources, and ponds

2.2.3. Image processing

Satellite images (Landsat 4–5 and Landsat8) of respective periods (1999–2021) were obtained from the official site of the [35] USG (United States Geological Survey) agency. Then, images were georeferenced together with the appropriate image interpretation. In recent years, researchers have proposed many vegetation mapping methods. In comparison with various novel classifiers, the conventional interactive supervised classification method has generally been used because of its ease of application, simple operation, and good performance. For this study, supervised classification algorithm was [33]. Supervised classification has been widely used in the literature [36]. After we identified land use classes with visual observation of the Landsat images, identified classes were crosschecked with recent Google Earth images and true ground data.

2.2.4. Accuracy assessment

A classification data to be useful in change detection, accuracy assessment is very essential [37]. Accuracy assessment was performed considering the required number of points, where about 50% of them were taken from Eucalyptus forest-covered areas and the other 50% from non-forest areas. Following the [38] standard procedure, the accuracy of the different maps produced from the classifiers was done based on the computation of the error matrix statistics. Then, the producer's accuracy, user's accuracy, and overall accuracy were calculated and accuracy levels were 81.13%, 78.09%, and 72.20% respectively. The spatial distribution of Eucalyptus was compared to the one obtained by the overlay of the common areas classified as Eucalyptus. In order to assess the accuracy of the classified maps, we created a set of random points from the ground truth data and compared them with the classified images using a confusion matrix.

2.2.5. Sampling for survey data

In order to draw the required sample from the study population, a multi-step sampling procedure was followed. On the basis of their Eucalyptus growing potentials, three growing regions were considered (Table 2).

Table 2.

Sample taken per study region.

Study region	Major farming scheme	Altitude	Number of households	Sample
I. Plain lands	Wheat, teff, and maize-based farming with Eucalyptus	2500–2900	2977	127
II. Lands with slight slopes	Barley, wheat, potato, and beans-dominant farming t with Eucalyptus	2900–3400	3593	151
III. Hilly lands	Wheat, barley, and potato with profuse plantation of Eucalyptus	3400–3800	3004	128
Total			9574	406

From each of the study regions, we drew respective samples following the [39] sample size formula. A total of 406 questionnaires were dispatched for collecting household-level data. With the exclusion of 18 improperly filled questionnaires, a final total sample of 388 (with a 95.6% response rate) was used for analysis. In addition to the formal questionnaire, 3 consecutive focus groups and 2 key informants’ discussions were carried out each involving 8–11 individuals.

2.2.6. Analysis of socioeconomic data

In order to identify the drivers influencing households' decisions in intensive plantation of Eucalyptus and the size of land covered with Eucalyptus in the study area, we employed the Heckman Two-Stages econometric model [40,41] using STATA 16 software. The model is preferred as it helps ensure determine households’ decision to involve in the plantation of Eucalyptus and the size of land planted with Eucalyptus. The model has also useful in terms of its ability to eliminate selectivity bias. In this study, involvement in intensive plantation of Eucalyptus for five years and above was taken as a selection criterion.

The Heckman Two-stage model, where a Probit model (selection equation) and a regression model (with correction for selectivity bias) was specified. The first stage of the model is the equation that catches the involvement in the Eucalyptus plantation (eq. (1)). The second stage of the model (eq. (2)) is the fitted Ordinary Least Squares regression which helps identify the drivers that influence the size of land planted with Eucalyptus [41] which is estimated with the Probit estimate of the Inverse Mill's ratio (λ).

The first stage of the model (Probit equation), can be specified as:

$$Y_i={\beta }_i{x}_i+{\epsilon }_i$$ (1)

where

${\epsilon }_i\sim N\left(\mathrm{0,1}\right)$

i = 1, 2, …n.

Y_i = a dummy variable indicating the decision on intensive Eucalyptus plantation as.

Y_i = 1 if Y_i>; Otherwise Y_i = 0.

β_i = unknown parameters to be estimated in the Probit equation.

${\mathrm{\chi }}_I$ = are explanatory variables that determine the decision in intensive plantation of Eucalyptus.

${\epsilon }_i$ = Random term which is assumed to be bivariate and normally distributed with correlation coefficient, ρ).

Then, with the inclusion of the Inverse Mill's Ratio (λ), the second stage of the model can be expressed as:

$$Y_i={\beta }_i{x}_i+\mathrm{\mu }{\mathrm{\lambda }}_i+{\epsilon }_i$$ (2)

where

${\epsilon }_i\sim N\left(0,{\delta }^2\right)$

i = 1, 2, …, n.

Y_i = Size of land planted with Eucalyptus (ha) (which is only observed if a household's decision to involve in intensive plantation of Eucalyptus is yes.

β_i = Unknown parameters.

${\mathrm{\chi }}_i$ = Explanatory variables influencing the size of land planted with Eucalyptus

μ = Parameter indicating the impact of households’ plantation decision on the size of land planted with Eucalyptus.

$\mathrm{\lambda }$ = Inverse Mills ratio (correction factor for selection bias).

${\epsilon }_i$ = Random term (assumed to be bivariate and normally distributed with correlation coefficient, δ²).

Different variables were hypothesized to have a significant push effect on the expansion of Eucalyputs in the study area (Table 3).

Table 3.

Variables and their hypothesized effect on the size of land planted with Eucalyptus.

Variables	Description	Hypothesized Effect
Number of oxen	Continuous	–
Management cost of Eucalyptus	Continuous	–
Experience (years)	Continuous	+
Number of income sources	Continuous (1,2,3, etc.)	–
Slope	1 = plain; 2 = gentle slope; 3 = steep slope	+
Fertility of plots (%)	Continuous	–
Neighborhood effect	0 = Eucalyptus has an effect) on adjoining plots; 1 = Eucalyptus has not effect)	+
Tenure security	0 = Plantation doesn't ensure tenure security	+
	1 = Plantation ensures tenure security
Total size of land (ha)	Continues	+
Number of other trees owned	Continuous	–
Age	Continuous	+
Sex of household head	Dummy (0 = female; 1 = male)	+

3. Results

3.1. Spatial distribution and expansion of Eucalyptus

The total area of land in the study regions is estimated to be about 241,129 ha and the majority of the land is suited for the plantation of Eucalyptus. Whilst the dominant farming system in the study area is cultivation of food crops, Eucalyptus is succeeding and replacing crop and other land covers at an alarming rate (Fig. 2, Fig. 3). Based on the classification, the estimated area of land covered by Eucalyptus plantation in 1999 was about 832 ha (3.71% of the area land potentially suitable for the plantation of Eucalyptus). The figure quickly grew to 1837 ha (8.20%) in 2010 (Table 4) (see Fig. 4).

Fig. 2.

The distribution of Eucalyptus from 1999, 2010, and 2021 (left to right).

Fig. 3.

The distribution of Eucalyptus and other land cover classes in 2021.

Table 4.

The estimated proportion of area with Eucalyptus plantation.

Year	1999	2010	2021
Total estimated area covered by Eucalyptus (ha)	832	1837	2692
Proportion of area covered by Eucalyptus (%)	3.71	8.20	12.01
Change in area of plantation from the reference period (%)	__	54.71	69.10

Fig. 4.

Google earth partial view of the area in Senan district (Image trimmed on April 16, 2022).

Varying levels of spectral separability of bands were observed between the study periods. The quality of spectral separability increased in the order 1999, 2010, and 2021 for the different land cover categories. There was a spectral confusion between the categories Eucalyptus and other forests covers for the 1999 Landsat 4–5 imagery and it was challenging to separate the seven classes visibly. From 2010 to 2021, a tremendous area of land has been covered with Eucalyptus figuring roughly about 2692 ha with an increase in the area of Eucalyptus coverage from the reference period by 69.01%. These changes are visibly observed in Fig. 3.

It was noticed that the distribution of Eucalyptus in 1999 was confined mainly near settlement areas (especially towns), river banks, and water bodies, and the majority of arable lands were devoid of Eucalyptus plantations (Fig. 2). Since 1999, however, a significant proportion of arable lands were planted with Eucalyptus. Growers confirmed that they began to convert their arable lands to Eucalyptus plantations following the land redistribution program which was taking place in the Amhara region. People assume that plots with plantation plantations will secure land ownership and this has initiated the majority to focus on using their plots to Eucalyptus plantations. The increasing need for construction and fuel wood and its major contribution as a source of cash to smallholder growers pushed growers to extensively plant Eucalyptus.

Prior to 1999, indigenous alternative tree species were not totally eliminated and people had the option to use those indigenous species to fulfill their wood need for construction and fuel. During that time, Eucalyptus has been planted mainly as homestead plantations and at the borders of plots. Later, the roles of all indigenous tree species were totally substituted by Eucalyptus and growers began plantation plantations on cultivable plots. In this regard, the period was taken as a benchmark for the expansion of Eucalyptus in the study area. Growers begin converting significant portions of their croplands for many socioeconomic and policy-related factors which will be discussed in Section 2.3.2.

In 2010, growers observed the returns and became initiated to expand Eucalyptus more as a trickle-down effect. During this time, urban areas previously planted with Eucalyptus were partly converted to built-up areas but intensive land cover classes were converted to Eucalyptus plantations. From 2010 to 2021, the changes were so rapid due to the rising interest of people in planting fast-growing wood products. We noticed that the expansion of Eucalyptus in the higher altitude areas of steep slopes and terrain features is mainly driven by continued degradation and decline in productivity of cultivable lands. Such lands are characterized by a high level of acidity and low fertility which however are still suitable for the plantation of Eucalyptus.

Different land use classes (Eucalyptus, other forests, croplands, grazing lands, bushlands, built-up areas, and water bodies) (Fig. 3). It was observed that Eucalyptus is extensively succeeding arable lands mainly in the mid and highland altitude areas. The intensity of land use change (from other classes to Eucalyptus) was much higher for food crop lands followed by grazing and bushlands.

3.1.1. The trend of land cover changes and push factors

Because of high population density, rugged topography (up to 45% slope), and intensive agricultural practices, the study area is characterized by serious degradation and a decline in productivity. Growers confirmed that they prefer replacing their croplands with Eucalyptus for many reasons. In the first instance, they argue that Eucalyptus has higher profit and quick returns as compared to returns from crops and secondly, Eucalyptus requires lesser cultural management than crops.

It was observed that because of the growing interest of people for the tree, cultivable lands are decreasing each year plantationsand extensive areas are being converted to Eucalyptus plantations (Figrue 4). Eucalyptus is also being planted on grazing and bush lands pushing borders of communal areas. This has caused a displacement of grazing lands to new areas which consequently led to the degradation of lands causing serious impact on crops and livestock in the study area.

The Heckman Two-Stage model depicts the major push factors of Eucalyptus expansion in the study area (Table 5). We found that grower households in the study area are focusing on the plantation of Eucalyptus for a number of socioeconomic reasons. plantations.

Table 5.

Result of the Heckman Two-stage model.

Involvement in plantation for >5 years			Number of obs = 388
			Censored obs = 29
			Uncensored obs = 359
	Coefficient	Standard Error	Z	P > |Z|
Number of oxen	0.0074	0.0043	1.70	0.089
Cost of management	−0.0323	0.0275	−1.18	0.240
Experience	0.0077	0.0023	3.32	0.001
Sources of income	0.1190	0.0312	3.81	0.000
Slope	0.0980	0.0258	3.79	0.000
Fertility of plots	−0.0581	0.0243	−2.39	0.017
Neighborhood effect	0.0248	0.0164	1.51	0.132
Tenure security	0.0886	0.0534	1.66	0.097
Total size of land	0.0721	0.0251	2.87	0.004
Number of other trees owned	0.0743	0.0302	2.46	0.014
Sex	0.0424	0.0308	1.38	0.169
Age of households	0.2104	0.1336	1.58	0.115
Constant	−0.4041	0.2291	−1.76	0.078
Land planted with Eucalyptus (ha)
Number of oxen	0.0197	0.0404	0.49	0.627
Cost of management	−0.0559	0.2024	−0.28	0.781
Experience	0.0289	0.0225	1.28	0.199
Sources of income	0.0057	0.2788	0.02	0.984
Slope	0.4429	0.2044	2.17	0.030
Fertility of plots	0.0739	0.1807	0.41	0.682
Neighborhood effect	0.5241	0.1801	2.91	0.002
Tenure security	0.7556	0.2702	2.80	0.005
Total size of land	0.2308	0.2115	1.09	0.275
Number of other trees owned	0.0156	0.0226	0.69	0.491
Age of households	2.0192	0.3380	5.97	0.000
Constant	−2.8271	0.9270	−3.05	0.002
Lambda (λ)	0.2828	0.1770	1.60	0.110
rho = 1.0000		Sigma = 0.2828

Findings indicate that a number of push factors were found to significantly influence both growers' tendency of involved in Eucalyptus plantation and the size of land planted with Eucalyptus. Accordingly, the major driving factors that push the conversion of different land uses to Eucalyptus plantations were identified. The number of oxen owned influenced (positively and significantly at P < 0.1) households' involvement in Eucalyptus plantations. Experience of households, the number of income sources, slope, and total size of land holding were potential factors that significantly and positively (P < 0.01) influenced households’ decision to involve in Eucalyptus plantation. On the other hand, the amount of land planted with Eucalyptus was significantly and positively influenced by the slope of plots (P < 0.05). This is because sloppy areas are characterized by a high degree of marginality and lower fertility. Such plots are preferred for plantations than to cultivation of crops. Another prominent reason for the expansion of Eucalyptus in the study area was the spillover effect of the plantation of Eucalyptus (i.e., the effect on the neighborhood farms). The neighborhood effect was significant at a 1% probability. Newly planted Eucalyptus plantations will apparently have light hindering and allelopathic effects on adjoining plots which in turn obliges other growers to convert their arable lands to Eucalyptus plantations.

Findings of this study confirmed that households engage in intensive plantation of Eucalyptus as they believe that plots planted with Eucalyptus will be more secure than arable lands. Tenure security significantly (P < 0.01) and positively influenced the size of land planted with Eucalyptus. Age of households significantly and positively (P < 0.01) determined the size of land planted with Eucalyptus. This is because aged households experience a shortage of labor in the family and want to focus on plantations which require lesser management costs as compared to the cultivation of food crops.

3.2. Discussion

The findings presented in this study show the conversion of different land classes such as grazing lands, other forest lands, bushlands, and water bodies to Eucalyptus plantations. This can pose a risk on the ecology and natural biodiversity and thus the detection of such alterations of the natural landscape is essential for sustainable management of existing land resources. During the period from 1999 to 2021, Eucalyptus has been expanding rapidly and is changing the natural ecosystem and biodiversity with different implications.

3.2.1. Implication on cultivation of food crops

As an invasive and fast coppicing tree species, Eucalyptus highly impedes the growth of neighboring plants by hindering light, aggressively competing for nutrients, and causing allelopathic effects on neighboring fields [[42], [43], [44]]. Cultivable lands are getting diminished and being converted to Eucalyptus plantations from year to year. Because of the rising market need for construction and fuel wood and for other socioeconomic reasons, there is a shift in growers’ attention from the cultivation of food crops to plantation of Eucalyptus which may lead to decline in cultivated crops and ultimately become a threat to food security. A study [45] indicated the invasive behavior of large-scale plantations of Eucalyptus and its negative impact on ecological well-being.

Eucalyptus can pose a burden on the ecosystem and negatively affect biodiversity. Studies by Refs. [27,46] also showed low herbaceous species richness in Eucalyptus plantations than in native forests. A comparable finding was concluded by Ref. [47] who observed a change in native Californian ecosystem processes and reduction in biological diversity and indigenous plant species and displacement of wild habitats following the invasion by Eucalyptus.

Several factors significantly influenced the preferences of farmers to focus on Eucalyptus plantations. The cumulative effect of this has resulted in a shift in landuse fromcultivable lands to plantation of Eucalyptus (a change from 54.71% in 2010 to 69.10% in 2021). This apparently has a serious impact on the cultivation of food crops.

Information obtained from key informants' discussion revealed that the growing wood market and absence of landuse policy have increased growers’ tendency to convert their cultivable plots to Eucalyptus plantations. A study by Ref. [48] noted that the increasing need for wood products initiated farmers to focus on tree plantations. Informal land transactions and permission of land rents up to 25 years remained other driving factors for the expansion of Eucalyptus. Group discussions held with growers also indicated that households who have no sufficient family labor prefer to rent out their plots for Eucalyptus plantations and live in towns. This has contributed to the conversion of extensive area of land to Eucalyptus plantations.

3.2.2. Implication on the agricultural landscape

The conversion of different land covers to plantations of Eucalyptus can bring a general alteration of the natural landscape and scene of an area. During the period from 1999 to 2021, the proportion of land covered with Eucalyptus quadrupled (changed from 3.71% to 12.01%). This trend has ultimately changed the agricultural land of the area (mainly the conversion of croplands to the plantation of Eucalyptus). Eucalyptus is generally thought to cause changes in the landscape and scenery of the area and can be viewed from positive and negative perspectives [49].

Afforestation of other land classes with Eucalyptus can result in soil physicochemical properties and organic matter dynamics. This change in turn affects the growth potential of the area because of change in. Different scholars including [3,50,51], and [52] presented that Eucalyptus is vastly planted by smallholder growers for its positive attributes such as good yield, fast-coppicing, and quick growing tendency in a variety of environments. Such rapid shift in land use system posses a threat to future cultivation of food crops and food security in the area. The rapid expansion of Eucalyptus in the study area is causing a dramatic transformation of the agricultural and other land classes. The existing farmlands are steadily decreasing following the yearly increasing plantation of Eucalyptus which is consequently leading to the elimination of indigenous crop varieties.

3.2.3. Implications on ecology and biodiversity

The conversion of cultivable lands to Eucalyptus is apparently altering the ecological setup of the area by changing the scene and population dynamics of flora and fauna. As a non-native species, the expansion of Eucalyptus can result in a transformation of the cultural landscape and a change in the natural ecosystem. Studies by Refs. [20,53] assessed the potential spread and invasive nature of Eucalyptus in the southeastern US which could impact ecosystem properties and functions. It was observed that short-rotation woody cropping systems have the potential to influence biological diversity. A similar finding by Ref. [54] noted that Eucalyptus has a high tolerance to harsh environmental conditions, pests, and diseases. By its very nature, Eucalyptus exhibits a vigorous growing habit that suppresses the growth of understory and decreases species richness. This consequently leads to increased biodiversity erosion in its niches of plantations [3]. The condition has become a big threat to indigenous crop varieties, tree, and animal species that are native to the study area.

Birds and browsing animals may find Eucalyptus less palatable [55]. Because of their faster growing habit, plantations of exotic species are usually taller than other plants of equal age and their shade may affect nearby crops by reducing the sunlight needed for growth. A study by Ref. [56] found that the diversity and abundance of bird species tended to be lower in such forest species than in reference forests while individual bird species responses were highly variable. Birds and browsing animals may find Eucalyptus unsuitable for shelter and food [55] and because of its morphological nature, Eucalyptus is not convenient for cavity-nesting bird species due to its limited availability of large stems. Thus, Eucalyptus has a moderate capacity to harbor species of fauna.

Studies including [57,58] reported lower plant and animal diversity in Eucalyptus plantations relative to native forests. The study by Ref. [58] in Brazil found fewer species of lizards and selected invertebrates in Eucalyptus plantations than in primary Atlantic forests. A study by Ref. [59] in Tasmania found that native land snails and millipedes were less diverse in Eucalyptus plantations than in native forests.

3.2.4. Socioeconomic concerns

There are a number of social issues related to the expansion of Eucalyptus. It was found that the experience of growers in Eucalyptus production, the number of income sources of households, and the slope of plots were the major driving factors that positively and significantly influenced the involvement of growers in Eucalyptus plantations for more than five years. Similarly, the slope of plots, neighborhood effect, tenure security, and age of households affected the size of land planted with Eucalyptus. The neighborhood effect of the plantation of Eucalyptus was one of the driving factors for the expansion of Eucalyptus in the study area. Social disagreements and conflicts as a result of different effects of Eucalyptus on neighboring plots like blocking from light, litterfall, blockage of the scene, and allelopathic effects of the tree in the root zones of neighboring plots are among the main social effects of the plant.

Different studies including [42,43,45] reported the negative impacts of the tree especially the allelopathic effects around the root zone and its shading effect under the tree canopy. The study stated that large-scale plantations of Eucalyptus have negative impacts and invasive behavior on the environment. All these effects of the tree can apparently trigger social tensions and conflicts.

As plantations need a significant number of manpower, especially during establishment (nursery operations, silvicultural activities, plantation, and overall management of the tree plantations), it is possible to consider the contributions of the tree in creating employment opportunities for the community. A study by Ref. [60] indicated the role of the sector in employment in rural areas. It was found that its neighborhood/trickledown effect was one of the push factors for the expansion of Eucalyptus plantations. Eucalyptus also has a significant contribution to the income of smallholder farmers Eucalyptus contributed significantly to forestry in terms of economic income. Eucalyptus has an immense contribution to infrastructure (poles, posts, and construction), the making of farm tools, and fuel.

3.3. Conclusions

The current study stresses on mapping the spatial distribution of Eucalyptus using a supervised classification algorithm in the main Eucalyptus growing highlands of Northwestern Ethiopia over a period of 21 years and assessed the major driving factors and land covered with Eucalyptus employing the Heckman Two-stage econometric model. The study suggested appropriate policy measures for proper management of land, especially in areas where the dominant population supports their livelihood with the cultivation of food crops.

An exception to this study is that it is the first of its kind in the study region to map the spatial and temporal distribution of Eucalyptus at the species level. The study also identified the main push factor for the conversion of other land classes to Eucalyptus plantations. Though the current study has overall strengths of the study in classifying the land classes successfully and proper identification of the spatial distribution of Eucalyptus in the study regions, it is not without limitations. Due to the inseparability of classes for the spectral mimicry of bands of different land covers, it was challenging to identify the specific Eucalyptus species vegetation for the Landsat images taken in 1999. This was due to the unavailability of satellite images with a better spatial resolution during this period.

Based on the findings of this study, it is concluded that land cover transformations were driven by a combination of many push factors. The results showed that many areas are being rapidly converted to Eucalyptus plantations and call for proper land use and management. Whilst Eucalyptus has socioeconomic contributions, plantations are at the expense of arable lands leading to dramatic land conversions.

The study provided valuable intuitions to policymakers and researchers with useful information for foreseeing and monitoring the socioeconomic and ecological impacts arising from the expansion of Eucalyptus at the expense of other land classes. Predicting and monitoring environmental changes as a result of human-induced factors and exactly identifying conditions that allow such ecological transformations over time are vitally important for future land use planning. Based on the findings, studies emphasizing on the socio-economic and ecological impacts of Eucalyptus are suggested especially in countries with extensive Eucalyptus plantations in the perspective of sustainability.

Author contribution statement

Amare Tesfaw: Conceived and designed the study; Analyzed and interpreted the data; Wrote the paper.

Feyera Senbeta & Dawit Alemu: Analyzed and interpreted the data; Wrote the paper.

Ermias Teferi: Contributed analysis tools or data; Wrote the paper.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement

Data will be made available on request.

Declaration of interest's statement

The authors declared that they have no competing interests.

Appendix A. Supplementary data

The following is the supplementary data related to this article: