1 Introduction

Climate and land use changes have widely been recognized as the two most important drivers of vegetation dynamics worldwide and thus remain critical global phenomenon that attracts the attention of policy makers and researchers. The frequency with which these elements affect vegetation, however, may vary depending on geographic location and the level of human involvement [1,2,3,4]. The climate-induced impacts are particularly pronounced in the fragile tropical dryland systems given their high sensitivity to climatic anomalies (such as the frequent occurrences of extreme heat, increasing aridity, and erratic rainfall patterns) and due to the lack of capacity to establish resilient systems. The impacts are even more evident when the climate anomalies are combined with anthropogenic activities [2].

Studies have already revealed that tropical dry vegetation resources, despite their enormous socio-economic and ecological significance, are under greater risk attributed mainly to threats emanating from the interplay of climatic and anthropogenic factors [5]. They are among the most exploited ecosystems and are being transformed into agricultural lands at an alarming rate [6]. Although the impacts of anthropogenic disturbances on these ecosystems seem to outweigh that of the climate-induced impacts, climate change is also a huge concern in the drylands; it is expected to worsen degradation caused by human activities. The dryland vegetation in particular is highly sensitive to the changes in rainfall regimes across the dry tropical regions [7,8,9]. Overall, the climate-related changes and their interplay with anthropogenic factors will continue adding further stress to these fragile ecosystems, with significant implications for the lives of billions of people.

Under the ideal scenario of constant climatic conditions, vegetation cover will likely increase significantly in areas where clear and efficient environmental protection measures are implemented. However, it is still unclear whether vegetation changes are more sensitive to changes in climate variables or environmental policies, despite the wealth of literature and scientific research on factors affecting vegetation cover [3, 4, 10,11,12,13,14,15]. It is possible to pinpoint the primary factors influencing vegetation cover in any given geographic location by examining the spatial and temporal characteristics of vegetation changes at the local scale and linking them to the underlying changes in anthropogenic and climatic factors. From there, it will be possible to identify the impacting aspects and create focused intervention strategies.

Several studies conducted hitherto have already analysed the inter-annual variations of NDVI-based vegetation greening/browning and their relationships to precipitation and temperature [16, 17]. Nevertheless, studies that integrate the effects of human activities and climate variability are still lacking in these ecosystems. Therefore, understanding the relationships between vegetation growth and climate and anthropogenic factors is vital for global change studies and developing management strategies to address the challenges posed by climate change and human activities. Therefore, this study was initiated to investigate the effects of climate variability and environmental policy changes on the spatial distribution and pattern of vegetation cover using high-resolution MODIS (Moderate Resolution Imaging Spectroradiometer) NDVI, and AVHRR (Advanced Very High Resolution Radiometer) image datasets over the past four decades.

2 Materials and methods

2.1 Study area

The study area, Tigray region (Fig. 1), is one of the national regional states of Ethiopia located in the northmost part of the country between 12015′N to 14057′N latitude and 36027′E to 39059′E longitude. Tigray region is bordered by Eritrea to the north, Sudan to the west, and with Ethiopian regions of Amhara and Afar to the south and east, respectively [18]. The region has an estimated total area of 53,638 square kilometres [19].

Fig. 1
figure 1

Map of the study area

Full size image

According to Sebastian [20] Tigray is divided into two agro-ecological zones as semi-arid tropical cold and semi-arid tropical warm. These agro-ecological zones are characterized by mountain ranges, flat areas and other hilly landscapes. Most precipitation is received during the rainy season, which lasts three months (July, August and September) in most parts of Tigray. The land cover is referred to as degraded land primarily due to centuries of overexploitation and overpopulation. The highland areas are characterized by higher population pressure than the lowland areas.

2.2 Data sources and processing

2.2.1 NDVI dataset

The Normalized Difference Vegetation Index (NDVI), a satellite–derived vegetation index, is an important tool for analyzing vegetation dynamics in response to climate variability [21, 22]. NDVI is the most widely used index for monitoring spatiotemporal vegetation dynamics for different vegetation types [23,24,25,26,27,28,29]. The index, which ranges between − 1 (for water bodies) and + 1 (for dense vegetation) [30], is computed as follows:

$$NDVI=\frac{NIR-RED}{NIR+RED}$$
(1)

where: NIR (Near Infrared) and RED denote spectral reflectance in the near-infrared (800–1000 nm) and red (620–750 nm), respectively.

The NDVI data derived from the NOAA–AVHRR sensors have been widely used to quantify vegetation activity across various regions of the world [16, 31,32,33]. Recently, the NDVI dataset derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) improved the spatial resolution of the previous products and enabled a thorough examination of vegetation dynamics and trends at various scales [32, 34]. This study used a combination of NDVI data derived from AVHRR and MODIS sensors, which were obtained from the United States Geological Survey (USGS) database (downloaded from http://earthexplorer.usgs.gov/). The MODIS (MOD13Q1) NDVI data, with a 16-day interval and 232-m resolution, was used for the period 2000 to 2022; whereas the low-resolution vegetation cover AVHRR data was used for the period 1982 to 1999. The MODIS NDVI values were corrected by a scale factor (i.e., 0.0001). The land imaging component of the MODIS sensor combines characteristics of AVHRR (Advanced Very High Resolution Radiometer) and Landsat sensors to provide improved monitoring of the earth’s surface at global scales [35].

2.2.2 Data on climate variables

Precipitation is determinant for vegetation growth and productivity in the dryland ecosystems, and it is widely used in the analysis of vegetation dynamics. To examine trends in the spatiotemporal variations of precipitation and its relations with vegetation dynamics in the study region, daily precipitation data was obtained from the product of Climate Hazards group Infrared Precipitation with Stations (CHIRPS) time series data spanning from 1982 to 2022. The CHIRPS precipitation data had a resolution of 0.05 by 0.05 degrees and was accessed from the United States Geological Survey (USGS). To illustrate the long-term trend of global temperature change, smoothed global surface temperature anomaly data was obtained from https://climate.nasa.gov/vital-signs/global-temperature/ for the period 1880 to 2020.

2.2.3 Data on environmental policy

An intensive literature search of all available published and unpublished scientific reports was made to obtain data related to environmental policies formulated in the region and the country between 1982 and 2022. The scientific internet searcher engines (e.g., using Web of Science journals) were used to search for the various scientific articles on environmental policy and their effect on environmental rehabilitation with particular reference to the study region. During the online literature search, focus was given to themes such as land, environment, and forest policies formulated between 1982 and 2022. Then, the impact of such policy implementations on environmental rehabilitation and restoration endeavours was carefully examined in relation to the climate variability and changes in vegetation cover in the study region.

2.3 Data analysis

2.3.1 Data preparation and processing

Before using them in further analyses, the lower resolution AVHRR vegetation cover data was sampled to a higher resolution that matches the resolution of the MODIS NDVI data (i.e., 232 m × 232 m). The CHIRPS precipitation data was also resampled to the same pixel size as the MODIS NDVI data. The data were further processed and analyzed using spatial statistics tools in R programming language and ArcMap.

2.3.2 Spatial pattern analysis

Spatial analysis techniques can provide useful insights into the spatial relationships between different variables [36]. In this study, Moran's Index (Eq. 2), which measures the degree of spatial clustering or dispersion of a variable in a given area, was used to examine the spatial autocorrelation between values of variables at a particular location. The calculation of Moran's I is similar to that of the correlation coefficient, and its expected values vary between − 1 and + 1. A positive spatial autocorrelation corresponds to a value of the I statistic that is greater than zero, while a negative autocorrelation is represented by a value less than zero [37].

$$I=\frac{n}{{S}_{o}}\frac{{\sum }_{i=1}^{n}{\sum }_{j=1}^{n}{\omega }_{i,j{z}_{i}{z}_{j}}}{{\sum }_{i=1}^{n}{z}_{i}^{2}}$$
(2)
$${S}_{o}={\sum }_{i=1}^{n}{\sum }_{j=1}^{n}{\omega }_{i,j{z}_{i}{z}_{j}}$$

where, I is the Moran’s I statistic \({z}_{i}\) is the deviation of a variable for feature i from its mean (\({x}_{i}\)−\(\widehat{X}\)), \({\omega }_{i,j}\) is the spatial weight between features i and j. n is equal to the total number of features, and So is the aggregate of all the spatial weights.

The zI-score for the statistic is computed as:

$${z}_{I}=\frac{I-E\left[I\right]}{\sqrt{V\left[I\right]}}$$
(3)

where, I is the observed Moran’s I statistic, E[I] is the expectation of Moran’s I statistic and \(\sqrt{V\left[I\right]}\) is the square root of the variance of Moran’s I statistic.

To effectively measure any association in the spatially distributed NDVI and precipitation variables and detect the local pockets of dependence that may not show up using Moran’s I spatial autocorrelation test, a hotspot analysis using Getis-Ord Gi* statistic was used. The hotspot analysis was primarily used to identify the statistically significant hot spots and cold spots in the NDVI and precipitation variables in the study area. The G statistics was selected because it makes it possible to evaluate the spatial association of a variable within a specified distance of a single point [36].

2.3.3 Temporal trend analysis

The Mann–Kendall’s trend test was used to analyze the temporal trend of change in NDVI and precipitation between 1982 and 2022. The Mann–Kendall’s trend test is a non-parametric trend test first proposed by Mann [38]. It was then further studied by Kendall [39] and improved by Hirsch, Slack and Smith [40], and Hirsch and Slack [41]. The Mann–Kendall test is based on the calculation of the Kendall-Tau measure of association between two samples. The null hypothesis for the Mann–Kendall’s test is that there is no trend in the series being evaluated. When calculating the trend test, increasing ranks are automatically generated to represent an increasing time series, which simplifies the calculation. The Sens’s slope used for the test and its variance were computed using Eqs. 4 and 5.

$$S={\sum }_{i=1}^{n-1}{\sum }_{j=i+1}^{n}Sgn\left({x}_{j}-{x}_{i}\right)$$
(4)
$$Var\left(S\right)=\frac{n\left(n-1\right)\left(2n+5\right)}{18}$$
(5)

where, n is the number of observations and \({x}_{i}\left(i=1\dots .n\right)\) are the independent observations.

The test was used to examine the presence of a statistical upward or downward trend characterized by its magnitude. The magnitude in the Mann–Kendall’s test is described by the Sen’s slope.

2.3.4 Linear relationship analysis

Pearson's Correlation Coefficient (PPC), also known as the correlation coefficient, was used to check if there is a linear relationship between NDVI and precipitation over the study period. The PPC is useful to check the linear relationship between two continuous sets of data. It is used to determine the direction of the linear relationship between the tested variables, with its values usually ranging from + 1 to − 1.

2.3.5 Functional (causal) relationship analysis


Regression analysis (Eq. 6) was used to verify the availability of functional relationships. A similar study by Wang, Price and Rich [42] used correlation analysis and regression analysis to examine the potential relationship between NDVI, precipitation and temperature. This method was, therefore chosen in particular to identify possible causes and relationships between the changes in NDVI, and policy implementation and changes in climatic variables during the selected study period.

$$\widehat{y}=a+bx$$
(6)
$$b=\frac{\sum xy}{\sum {x}^{2}}; a=\overline{y}-b\overline{x}$$

Where, \(\widehat{y}\) is predicted y, b is the slope, and a is the y-intercept.

2.3.6 Residual analysis

The difference between the observed y-values and the predicted y-values is called the residual. The size of the residuals from a regression equation is a measure of model fit. Large residuals indicate poor model fit. In this study, the residual analysis approach was used to separate NDVI changes caused by policy formulation and implementation from those resulting from changes in precipitation and temperature [43]. The NDVI residuals for the entire study area were calculated as differences between the predicted and observed NDVI values (Eq. 7).

$$Residual=Observed-Predicted$$
(7)

3 Results and discussion

3.1 Spatial pattern of NDVI and precipitation

When the values of variables are grouped in close proximity, it indicates a strong correlation between these two factors [37, 44]. Based on the concept of Moran's I for spatial autocorrelation [37] and hot spot analysis, the Getis-Ord Gi* statistic [36] was used to examine the spatial cluster in NDVI and precipitation. The spatial autocorrelation test was performed for four different periods: 1982–1999, 2000–2019, and 1982–2022, and four seasons. For the periodic analysis, the year 1999 was used as a benchmark because abrupt changes in NDVI were observed in the same year [45], and 2019 was used to compare vegetation changes before and after the Tigray War. Figure 2 shows the spatial distribution of vegetation cover over the study period.

Fig. 2
figure 2

Illustration of the average yearly NDVI values calculated over a 41-year period (1982 to 2022), using both AVHRR and MOD13Q1 yearly NDVI data as inputs. The long-term trends depicted in the figure provide valuable insights into the dynamics of vegetation cover in the Tigray

Full size image

To fully understand the spatial relationship between precipitation and NDVI, we conducted a spatial autocorrelation analysis of NDVI and precipitation for each of the four seasons: autumn (September, October, November), winter (December, January, February), spring (March, April, May), and summer (June, July, August). The purpose of this analysis was to identify any potential seasonal overlap between NDVI and precipitation during the study periods. Our results showed the presence of clustered spatial patterns in both NDVI and precipitation across all periods with a p-value of less than 0.01 (Tables 1 and 2). Moreover, with a range of + 2.58 to − 2.58 at a 99% confidence level [46], the z-scores for all periods were greater than 63.78, indicating that the probability of a clustered pattern forming by chance alone was less than 1%.

Table 1 Global Moran’s I summary table showing spatial autocorrelation of long-term periodic average NDVI and CHIRPS average data over the past four decades
Full size table
Table 2 Spatial autocorrelation test result for NDVI and precipitation for the period 1982 to 2022
Full size table

To investigate and map the spatial autocorrelation in detail, hotspot analysis, with a search radius of 10 km, was used to group similar pixel values into different groups. The hotspot analysis resulted in three primary groups—namely, hotspot, cold-spot and non-significant group. While the hotspot represented clusters of high-value pixels, the cold spot represented clusters of low-value pixels. The pixels with values that cannot be attributed to any of the groups were therefore regarded as not significant. According to the hotspot analysis (Fig. 3), statistically significant (p < 0.05) spatial clustered high precipitation values were generally observed for all study periods and seasons in the south-western part of Tigray. For the periodic NDVI, the statistically significant high values were distributed in south-west as well as in some parts of southern Tigray. Conversely, most of the eastern, north-eastern and central parts were characterized by lower significant (p < 0.05) values and are therefore referred to as cold spots. As shown in Fig. 3, the high precipitation values were highly localized to the south-western parts of Tigray while the coldspots are highly localized to the eastern and southern parts. In comparison, the periodic NDVI hotspots are highly distributed in south-west, south-central and southern parts. The periodic NDVI coldspots, however, were distributed in north-eastern, eastern and south-eastern parts of the study area.

Fig. 3
figure 3

The hot and cold spots of periodic NDVI and Precipitation calculated using Getis-Ord Gi* statistic based hotspot analysis. Hot spots (in red) are areas with significantly higher values of NDVI and precipitation than surrounding areas, while cold spots (in blue) are areas with significantly lower values

Full size image

In contrast to the periodic spatial pattern, the hotspot analysis results calculated from long-term seasonal averages over a 41-year period (Fig. 4) exhibited even greater discrepancies than the periodic averages. Irregularities were observed in the spatial distribution of hotspots and cold spots in both seasonal NDVI and precipitation (Fig. 4) across all seasons. During the summer season, higher NDVI values were dominant in the precipitation cold spots of the eastern and southeastern parts of the study area, illustrating a non-overlapping nature of hotspots and cold spots in NDVI and precipitation at the seasonal level. Furthermore, the winter season in the semi-arid tropical region is characterized by limited moisture, resulting in minimal vegetation growth across most parts of the study area. However, localized areas with high groundwater and perennial streams may exhibit vigorous growth. In contrast, during the summer season, vegetation growth is more widespread throughout the study area due to the availability of moisture from rainfall. Consequently, the winter season exhibits a higher Moran's I value (i.e., 1.03), indicating a more clustered pattern of vegetation greenness compared to the relatively distributed but still clustered vegetation greenness observed in the summer season (Moran’s I = 0.91). Generally, the results of the hotspot analysis using the G stat indicated an inconsistent relationship in the spatial distribution of NDVI and precipitation hotspots and cold spots, suggesting small to non-overlapping spatial patterns between them.

Fig. 4
figure 4

The hot and cold spots of long-term seasonal averages of NDVI and Precipitation calculated using Getis-Ord Gi* statistic based hotspot analysis. Hot spots (in red) represent areas with significantly higher values of NDVI and precipitation than surrounding areas, while cold spots (in blue) represent areas with significantly lower values

Full size image

Previous studies have reported a direct and strong relationship between the spatial pattern of NDVI and seasonal rainfall [42, 47,48,49,50]. This relationship highlights the significant influence of rainfall on the distribution and variability of NDVI across different regions and periods. However, a recent study conducted in the Kafta Sheraro National Park in Western Tigray [51] identified human activities as a major driving force for cover type changes over extended periods. This finding suggests that factors beyond rainfall, such as anthropogenic impacts, play a crucial role in shaping the vegetation dynamics in the area over longer periods. Interestingly, another study conducted in the upper Heihe River of Northwest China [52] found no clear relationship between NDVI and precipitation, despite being in different climatic zones. This finding is consistent with a study on the Chinese Loess Plateau [53], which observed a significant increase in vegetation cover following the implementation of the Grain for Green project. These findings indicate that local factors and interventions can override the traditional relationship between NDVI and precipitation.

In addition to climate and human activities, variations in topography may also contribute to the spatial variation in vegetation cover [54]. Therefore, considering the analysis results and the reviewed literature, it raises the question of whether other factors, besides precipitation, also influence the spatial variation in NDVI within the study area.

3.2 Temporal pattern

In this study, we used data from four study periods (1982–2022, 1982–1999, 2000–2018, and 2019–2022) and four seasons (autumn, winter, spring, and summer) to better understand the temporal patterns and relationships of NDVI and precipitation. The year 1999 was used as a breakpoint for analysis because NDVI values changed abruptly between 1998 and 2000 (Fig. 5). The years 2019–2022 were chosen to examine vegetation cover changes during the Tigray War. The study period 1982–2022 was chosen to examine the overall long-term temporal patterns over four decades.

Fig. 5
figure 5

Temporal distribution of mean annual NDVI and annual precipitation. The years for which the environmental policies are formulated and implemented are indicated as yellow circular dots

Full size image

Mann–Kendall’s trend test was used to evaluate the trend of change over time, and Pearson's correlation coefficient was employed to investigate any linear association between NDVI and precipitation. Results of the Mann–Kendall trend test showed a significant increase in NDVI between 1982 and 2022 (SS = 0.007, r = 0.79, p < 0.01). Similarly, an upward trend was observed between 1982 and 1999 at p < 0.001 (Table 3). However, a statistically significant decrease in NDVI was observed between 2000 and 2018 (SS = − 0.003, r = − 0.77, p < 0.01). Nevertheless, no discernible NDVI trend was observed between 2018 and 2022. In addition, no statistically significant upward or downward trend in precipitation was observed at all study periods.

Table 3 Mann–Kendall trend test result calculated from periodic means
Full size table

Seasonally, both the NDVI and precipitation showed an increasing linear trend (SSndvi = 0.004, SSrf = 0.36, p < 0.01) for autumn. For the summer season, however, only NDVI showed a significantly (at p < 0.01) increasing trend (SSndvi = 0.003, rndvi = 0.7) (Table 4). No visible significant NDVI and precipitation trend was observed for either the winter or spring seasons.

Table 4 Mann–Kendall trend test result calculated from long-term seasonal means
Full size table

A time series plot (Fig. 6) was generated to explore the temporal patterns of NDVI and precipitation over a long period (over four decades). In comparison, the rate of change of the NDVI was higher (r2 = 0.62) than the rate at which precipitation changed over time (r2 = 0.1).

Fig. 6
figure 6

Time series of annual a average NDVI and b precipitation (CHIRPS)

Full size image

The scatter plot in Fig. 7 shows the linear relationships between NDVI and precipitation both periodically and seasonally. The results show moderate correlation between the two variables in autumn (r2 = 0.65) and summer (r2 = 0.60). However, weak linear relationship was observed for winter and spring. Similarly, the long-term annual trend in 1982–1999 (r2 = 0.55) and 1982–2022 (r2 = 0.54) indicated a moderate correlation. This indicates a generally inconsistent linear relationship between NDVI and precipitation in the study area.

Fig. 7
figure 7

Seasonal and periodic scatterplots depicting the relationship between NDVI and precipitation across the four-decade study period (1982–2022). The deep brown areas represent regions of high cluster density, indicating strongly correlated NDVI and precipitation values. Conversely, the blue-shaded areas represent regions with weakly correlated NDVI and precipitation values

Full size image

3.3 Functional relationships between NDVI and precipitation

A simple linear regression analysis was used to examine the response of NDVI to the changes in precipitation over the study period. Accordingly, the analysis results showed no significant causal relationship between NDVI and precipitation (Fig. 8). The fact that there is no significant correlation, and the small value of the regression coefficient (r2 = 0.19) shows the lack of measurable causality between the increase in NDVI and the precipitation pattern between 1982 and 2022.

Fig. 8
figure 8

Linear regression analysis graph showing NDVI predictability by precipitation between 1982 and 2022. The graph indicates low predictability of vegetation dynamics by precipitation

Full size image

The findings of this study show that there is no strong, linear relationship between vegetation cover change and precipitation in Tigray. This is similar to the finding of a study in the Yangtze River watershed area [55], which found no significant correlation between NDVI and precipitation. The findings suggest that precipitation is not the only factor that limits vegetation growth. Additionally, the relationship between vegetation and precipitation may be more complex than a simple linear relationship resulting in different levels of response to precipitation [56].

3.4 Residual analysis

Residual analysis is a commonly used approach to identify areas experiencing human degradation by removing the effect of precipitation from the NDVI time series [43]. Accordingly, the approach was employed to perform multiple linear regression analyses to obtain the residual values between 1982 and 2022. The NDVI residuals were obtained by subtracting the predicted NDVI from the observed NDVI values and were then examined for trends over time. Following the concept of the residual analysis, the researchers assumed that statistically insignificant trend changes in NDVI could be explained by the climate trend, while statistically significant trend changes in NDVI residuals could be attributed to human activities. The results of the analysis showed a statistically significant decreasing trend (r = − 0.55, z-stat = − 11.58, p < 0.01) between 1982 and 2022, indicating that changes in NDVI values cannot be attributed solely to precipitation. This suggests that other factors, such as biotic and anthropogenic factors [57], may have contributed to the rapid vegetation changes observed in Tigray over the past 41 years.

3.5 The trend in temperature

The Earth's temperature has been steadily rising for decades, and scientific pieces of evidence confirm that this trend is continuing [58]. The global average temperature has increased by about 0.15 °C per decade since the early 1980s, indicating a long-term trend of increasing atmospheric greenhouse gas concentrations primarily due to human activities such as burning fossil fuels and deforestation [59, 60]. In recent years, temperatures have reached record highs, with 2016 and 2020 being the warmest years on record [58]. Figure 9 shows the annual temperature anomalies both for the global and local scales. Although the magnitude by which the local temperature changed (r2 = 0.37) is much smaller than the global temperature change (r2 = 0.85), it is true that the local temperature in the study area also changed significantly over the years. To combat land degradation in Tigray, various government and non-government actors have launched a massive reforestation campaign to rehabilitate degraded areas. However, despite the efforts, there was an overall 52% seedling survival rate [61]. While other factors contributed to the low survival rate, the increasing temperature trend has emerged as one of the dominant factors by increasing water stress through increased evapotranspiration.

Fig. 9
figure 9

The graph illustrates the trend in global temperature anomalies from 1880 to 2020, smoothed using a 30-year period (1951 to 1980) as a baseline (NASA, 2023). The yellow-shaded area in the graph highlights a significant and rapid increase in global temperature during the analyzed period (1982 to 2022)

Full size image

This, therefore, may have affected the seedling survival rate in Tigray region. The increase in temperature is especially visible from around 1998. This implies that the stagnant vegetation covers might have been caused by the increase in temperature over the study area through its negative influence on the seedling survival by causing water stress.

In tropical regions like Tigray, where exposure time is usually higher than the global average, rising temperatures can negatively impact vegetation growth due to increased water stress [62]. Furthermore, high temperatures can lead to more frequent and severe drought events, causing widespread vegetation die-offs [63]. Therefore, while increasing temperatures may lead to increased vegetation growth in some instances, this is not always the case and depends on factors such as water availability and altitude. In semi-arid regions like Tigray, increased temperatures may lead to decreased vegetation growth due to increased water stress. The steadily increasing temperature, therefore, could be taken as one of the major factors causing the insignificant vegetation change in Tigray since 2000. In general, the increasing temperature trend poses a significant challenge to vegetation growth in Tigray. Despite the massive reforestation campaign, the low seedling survival rate highlights the need for more effective measures to combat the effects of climate change on vegetation growth.

3.6 Environmental laws, policy frameworks, strategies and programs

Ethiopia, like many other countries in sub-Saharan Africa, is severely affected by widespread land degradation and desertification, mainly due to improper land use and poor land management practices, overgrazing and deforestation driven by rapid population growth. Recognizing this fact, the Ethiopian government has taken important and encouraging steps to address environmental issues and support sustainable development. In recent decades, several policies, strategies and laws have been introduced to support the country's goals of sustainable development and environmental protection [64]. Table 5 summarises a list of the most important policies, strategies and legal instruments relevant to the realization of sustainable management and environmental protection.

Table 5 List of key environmental laws, policy frameworks, policies and programs
Full size table

The 1995 Constitution of the Federal Democratic Republic of Ethiopia (FDRE) (Proc. 1/1995) [65], which enshrines the concept of sustainable development and environmental rights, provided the legal framework for all environmental policies and laws [66]. Accordingly, the Environmental Protection Agency (EPA) was established in 1995 by Proclamation No. 9/1995, followed by the Environment Policy of Ethiopia (EPE) in 1997 [67]. The 1997 Environmental Policy helped formulate the various subsequent environmental proclamations; and laid the foundation for environmental management in Ethiopia. It envisaged the integration of environment and development at the policy, planning and managerial levels in order to improve decision-making [68]. In addition to formulating and implementing various environmental laws, policies, strategies and programs, Ethiopia has also demonstrated its commitment to environmental management and sustainable development through the ratification of key environmental conventions.

The regional state of Tigray in northern Ethiopia faces food security problems due to severe land degradation and climate-related recurring droughts. In particular, the sequence of dry years between the late 1970s and late 1980s was the driest decade of the last century in the northern Ethiopian highlands. In response to these challenges, the government and people of Tigray, together with the Ethiopian federal government and NGOs, have implemented various internationally recognized land restoration and ecosystem-based adaptation programs since the early 1990s and written success stories [69]. Some of these interventions include area enclosures, and physical and biological soil and water conservation measures. Area enclosure—the closing of highly degraded areas to livestock, cultivation and indiscriminate tree felling, remained a key intervention and contributed a lot to the restoration of degraded areas in the region. Increased biodiversity, vegetation cover and density, and improvement in soil fertility have been reported in protected areas [70,71,72]. Many studies confirmed that the revegetation practices implemented in the Tigray region, particularly in relation to area enclosures, afforestation and reforestation, were successful [73]. Some of the land reclamation work was also carried out through the Food for Work program, which was later expanded to all food insecure areas of the country through the Safety Net program during PASDEP and the country's GTP. In 2017, the Tigray region was recognized internationally as the Gold Medal Winner in the World Future Council Award for the World's Best Policies to Combat Desertification and Land Degradation (https://www.worldfuturecouncil.org/p/champions/). In general, the Tigray region of Ethiopia shows that the restoration of degraded land can become a reality.

Therefore, it is believed that these large-scale land rehabilitation operations, facilitated by changes in ownership and natural resource management policies and legal frameworks, influenced vegetation cover, explaining the increasing vegetation cover from 2000 to 2022 compared to the low NDVI values between 1982 and 1999 (Fig. 5). According to Nyssen, Haile [74], the Tigray region of Ethiopia is now greener than at any time in the last 140 years. This is not merely due to an increase in precipitation, but largely due to human investment to restore productivity to degraded land. Several recent studies have also shown an increase in vegetation greening over large areas of the Sahel, invalidating claims of widespread irreversible desertification in the African Sahel [74,75,76].

Although precipitation turns out to be the dominant causative factor for the region's revegetation, a human-caused factor superimposed on the climate trend is also cited as a causative factor for such changes [75]. Other similar studies [77,78,79,80,81,82] elsewhere revealed the sensitivity of vegetation dynamics to human-induced factors as compared to the climatic variability while acknowledging the complexity of the effects from the interplay of these factors across different vegetation and land use types. Ecological restoration programs and policies are widely reported to be effective in reversing environmental degradation and promoting vegetation restoration, particularly in ecologically vulnerable areas [78, 83]. Nevertheless, the complex relationship between climate change and ill-informed human-induced activities may negatively affect the positive effects of ecological restoration policies, entailing the need to take great caution and further rigorous work in this regard. Therefore, in agreement with these studies, our study confirms the assertion that people have had a measurable impact on vegetation dynamics in the study region and thus it is crucial to put effective restoration programs and policies in place to facilitate sustainable land management and address the challenges posed by climate change.

4 Conclusion

This study used satellite-based vegetation index (AVHRR and MODIS NDVI) and gridded high-resolution CHIRPS precipitation time series data to investigate the spatiotemporal characteristics and relationships between vegetation change trends and climate variability in the semi-arid Tigray region of northern Ethiopia. Further analysis, using the residual trend analysis, distinguished between the effects of climatic variability and those of the environmental policies formulated in the region between 1982 and 2022. Results of the study revealed that changes in climate variables alone cannot fully account for the significant change in NDVI values observed in the study area from 1999 onwards. The limited linear correlation between NDVI values and climate variables suggests that other factors, such as human activities, may have contributed to the observed changes. Specifically, the residual analysis indicates that rehabilitation and land restoration programs implemented following the introduction of environmental policies played a more substantial role in altering vegetation cover than the climate variables studied.

To gain a deeper understanding of this relationship, future research should examine potential variations considering different vegetation formations and climate variability in the study area. Researchers should also pay close attention to how different vegetation formations respond to these factors, as some may be more sensitive to one or the other. Additionally, policymakers should focus on developing and implementing effective environmental policies that account for all climatic and human-induced factors across different land cover types to promote effective vegetation growth. By doing so, it may be possible to modify the stagnant temporal pattern of vegetation since 1999 and achieve an increasing pattern again. It is also important to monitor the effectiveness of these policies over time and adjust them as necessary to ensure continued success in promoting vegetation growth.

Overall, the results of this study provided valuable scientific evidence on the spatiotemporal changes in vegetation growth. It explored the impact of climatic variables and environmental restoration policies on vegetation growth over the past four decades in the semi-arid Tigray region of northern Ethiopia. The findings shed light on the intricate relationship between climate variables, human-induced factors, and vegetation growth. This suggests that decision-makers and practitioners should prioritize formulating and implementing environmental policies to restore degraded ecosystems and mitigate the effects of human activities and climate change. By adopting a comprehensive approach that considers all these factors, it becomes feasible to foster sustainable vegetation growth and mitigate climate change effects, facilitating the sustainable development of fragile dryland ecosystems.