Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2026 Jun 19;16:28138. doi: 10.1038/s41598-026-58168-9

Projecting the spatio-temporal habitat suitability of Striga hermonthica under climate change scenarios in Ethiopia using ensemble modeling

Mikiyas Abebe 1,2,✉, Zerihun Woldu 1, Zemede Asfaw 1, Bikila Warkineh 1
PMCID: PMC13554200  PMID: 42321392

Abstract

Striga hermonthica is a root-parasitic plant that attacks many important cereal crops, including sorghum, teff, and rice. It is one of the most destructive agricultural weeds, posing a major threat to food security in Ethiopia. This study aimed to map suitable habitats for S. hermonthica in Ethiopia and to evaluate potential changes in its range under projected climate scenarios for 2050 and 2070. One hundred thirty-two spatially filtered occurrence records were used in distribution modeling. Prior to modeling, the environmental variables were tested for multicollinearity using Pearson correlation and Variance Inflation Factor, leading to the selection of 10 uncorrelated predictors. The ensemble model, built using seven algorithms with 10 replicates each, demonstrated strong performance (mean AUC = 0.92; TSS = 0.75). The most influential environmental variables were precipitation during the warmest quarter and temperature seasonality. Under current climate conditions, approximately 234,326.35 km² were found suitable for the weed. Projections under both intermediate and high-emission scenarios indicate a likely expansion of its range. Northern, northwestern, western, and eastern lowlands of Ethiopia are comparatively identified as risk area for this weed. Thus, all stakeholders, including researchers, policymakers, farmers and governmental and non-governmental organizations, should take early expansion-controlling management plan like create awareness for farmers.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-58168-9.

Keywords: Climate scenarios, Ensemble modeling, Environmental variables, Habitat suitability, Species distribution, Striga hermonthica

Subject terms: Climate sciences, Ecology, Ecology, Environmental sciences, Plant sciences

Introduction

Nowadays, climate change is one of the major global issues, due to its impacts on the distribution of biodiversity and human welfare. The inter-governmental panel on climate change1 reported that the temperature of the globe is increasing alarmingly. Moreover, in its previous reports, the IPCC2 highlighted that by 2100, the concentration of CO2 in the atmosphere is expected to double, and the pattern of precipitation will vary from the present situation. This earlier report had further indicated that in 2100, the average global temperature will rise by 1.4 to 5.8Inline graphic. Similarly, IPCC1 projected that the concentrations of CH4, CO2 and N2O increase by 156%, 47%, and 23%, respectively, from their present concentration levels. This change in climate will affect the existence of many plant species causing significant shifts from their present geographical habitat or to lead vulnerable species to local or global extinction, depending on their potential to tolerate the environmental changes. Moreover, climate change will have pronounced impacts on global food security3. Identifying suitable habitats for certain species is essential for developing effective conservation actions and policy guidelines for economically important plants. It is also important to control the expansion of economically insignificant plants that have negative impacts on agriculture, such as Striga hermonthica.

Striga hermonthica (Del.) Benth., commonly referred to as witchweed, is a hairy annual herb featuring tubular corollas and belongs to the family Orobanchaceae (previously classified under the family Scrophulariaceae)4. This semi-parasitic plant hampers the growth of several crops such as sorghum, maize, rice, millet, and other grasses by attaching to the roots of host plants and extracting vital nutrients from them5. Each Striga hermonthica plant can produce hundreds of thousands of minute seeds that remain dormant in the soil for many years. These seeds germinate only when they detect chemical signals released by the roots of nearby host plants6.Moreover, its seed can disperse through various mechanisms such as animals, wind, and water. Such property of Striga hermonthica allow it to colonize various habitats and pose threats to the agricultural economy in numerous countries7.

In Ethiopia, seven known species of Striga have been recorded: Striga hermonthica, Striga asiatica, Striga aspera, Striga latericea, Striga gesnerioides, Striga pinatifida, and Striga forbesii5,6. Currently, these species have significant impacts on various crops, causing a reduction in productivity to more than 30%8. Of these, Striga hermonthica is the most dominant and is widespread, especially in North, Northeastern, and Eastern Ethiopia. Striga hermonthica prefers infertile soil, hot arid and semiarid areas, and temperature ranges between 25 and 35 °C for its germination and survival9.

Previous studies, including David, Obiakara10, have applied species distribution models (SDMs) to assess the potential geographic range of three Striga species across the African continent. However, due to the broad spatial focus of these studies, they provide limited resolution regarding the distribution intensity and infestation levels of each Striga species at the country level.

Gebreslasie, Tessema11 recognized Striga hermonthica as the most prevalent Striga species in Ethiopia. However, their findings suggested that S. asiatica and S. gesnerioides were even more dominant, despite the fact that current climatic conditions in Ethiopia’s highlands are unsuitable for these two species. Their model output showed the highlands as suitable habitats for S. asiatica and S. gesnerioides, likely due to the limited occurrence data, only three records from GBIF and published sources representing Ethiopia, while effective and accurate modeling, field occurrence data is essential Gomes, IJff12.

Another limitation of their study was the absence of socioeconomic factors, such as population dynamics, land use changes, and policy frameworks, as their modeling relied solely on RCP-based climate projections. Overall, the report provided insufficient detail on both the current and future distribution of Striga hermonthica in Ethiopia. Therefore, country-specific species distribution modeling is essential to support effective management and control strategies.

In addition, studying the significant impact of Striga hermonthica on essential crops like sorghum, maize, and teff, along with its broader economic implications for agriculture, is vital for assessing both its current and future potential habitats. This understanding will assist in developing effective management strategies. Species Distribution Modeling (SDM) provides a valuable tool for this purpose. Therefore, this study aims to identify suitable habitats for S. hermonthica under present and projected climate conditions and to quantify potential changes across different climate scenarios. Figure 1.

Fig. 1.

Fig. 1

Open in a new tab

Map of Ethiopia with elevation and occurrence points of Striga hermonthica. This plant prefer infertile soil, arid and semiarid environments with hot environmental condition. The map was prepared using ArcGIS version 10.8 (Esri, https://www.esri.com).

Results

Model performance

The AUC and TSS values in 10 replications and seven algorithms ranged from 0.88 to 0.95 and 0.70 to 0.80, respectively. While the lowest AUC and TSS were recorded in the GLM algorithm, the highest were recorded in the RF algorithm. On the other hand, sensitivity and specificity ranged from 0.76 to 0.88 and from 0.88 to 0.95, respectively. The standard deviation across all evaluation metrics was low and indicated stable model performance. On the other hand, the AUC and TSS values of six algorithms for the host plant Sorghum bicolor range from 0.90 (RF) to 0.94 (BRT) and 0.67 (BRT) to 0.79 (RF), respectively, indicating acceptable modeling performance. Similarly, sensitivity and specificity ranged from 0.86 (MARS and SVM) to 0.91 (GAM) and 0.80 (BRT) to 0.91 (RF) with low standard deviation across evaluation metrics (Table 1).

Table 1.

Model performance evaluation of different algorithms and their mean value, which comprises area under receiver operating characteristics (AUC), true skill statistic (TSS), sensitivity, and specificity. While AUC is threshold-independent, the other listed metrics are threshold-based metrics.

Species Evaluation method Algorithms Ensemble
brt rf glm gam maxent mars svm
S.hermonthica AUC 0.92 0.95 0.88 0.92 0.92 0.92 0.92 0.92 ± 0.03
TSS 0.73 0.80 0.70 0.77 0.74 0.76 0.74 0.75 ± 0.06
Sensitivity 0.80 0.88 0.76 0.84 0.86 0.84 0.82 0.83 ± 0.06
Specificity 0.93 0.92 0.95 0.92 0.88 0.91 0.93 0.92 ± 0.05
S. bicolor AUC 0.94 0.90 0.92 0.92 0.91 0.91 0.92 ± 0.02
TSS 0.67 0.79 0.73 0.74 0.72 0.71 0.73 ± 0.02
Sensitivity 0.87 0.88 0.91 0.90 0.86 0.86 0.88 ± 0.05
Specificity 0.80 0.91 0.82 0.84 0.87 0.85 0.85 ± 0.06
Open in a new tab

The higher performance of RF in predicting the accuracy of Striga hermonthica can be attributed to its ability to capture complex and nonlinear relationships between species occurrence and environmental variables. In contrast, GLMs tend to perform less effectively due to their dependence on linear assumptions. The comparatively high specificity values suggest lower commission errors, while moderate to high sensitivity values reflect acceptable detection of species presence across the algorithms.

The observed variation in performance among different algorithms in both the host and parasitic plants indicates that there are differences among the algorithms in capturing species-environment relationships and balancing omission and commission errors. The reliance on a single algorithm can introduce uncertainty and bias in habitat suitability predictions. Therefore, employing an ensemble approach is crucial for enhancing prediction robustness and minimizing uncertainties that are specific to individual algorithms.

Relative contribution of environmental variables

Not all environmental predictors contribute equally to the distribution of species; some predictors have a significant impact, while others have a lesser effect. The ensemble results of the model indicated that the two primary environmental predictors significantly influencing the distribution of Striga hermonthica were precipitation during the warmest quarter (Bio18) and temperature seasonality (Bio4). Aspect and total nitrogen content at a depth of 0–20 cm have a relatively low impact on the distribution of Striga hermonthica (Fig. 2a). Though the identified environmental variables have impacts on the distribution of Striga hermonthica, its infestation is also influenced by the availability of host plants, as well as the environmental factors and soil.

Fig. 2.

Fig. 2

Open in a new tab

Relative contribution of environmental variables in the distribution of Striga hermonthica (a) and Sorghum bicolor (b). The values represents the mean value of all algorithms.

Temperature seasonality (Bio4), elevation, and warmest quarter (Bio18) were the three primary environmental variables influencing the distribution of Sorghum bicolor. On the other hand, aspect, slope, and nitrogen content at a depth of 0–20 cm have comparatively low influence on the distribution out of the ten predictors (Fig. 2b).

The response curve indicates that the spread of Striga hermonthica increases with precipitation during the warmest quarter, beginning at 100 mm and extending to 500 mm, which is optimal for the effective distribution of Striga hermonthica. It also prefers a maximum Bio4 value of 200, which corresponds with a monthly mean temperature standard deviation of 2 °C (Bio4 = standard deviation × 100) (Fig. 3a). Similarly, Sorghum prefer precipitation in the warmest quarter, ranging from 100 to 500 mm, for effective distribution. Additionally, they thrive at elevations between 1000 and 2000 m. The Bio4 value of 200 corresponds to a monthly mean temperature standard deviation of 2 °C, which is ideal for sorghum growth (Fig. 3b).

Fig. 3.

Fig. 3

Open in a new tab

Response curve for Striga hermonthica (a) and Sorghum bicolor (b). X axis represents the range of values of the environmental variables, Y axis represents probability of occurrence as range of values.

Potential distribution of Striga hermonthica and Sorghum bicolor under the current climate condition

Under the current climatic conditions, about 234,326.35 km² (20.69%) of Ethiopia’s total geographical area is suitable habitat for Striga hermonthica. Of this, 219,137.61 km² (93.52%) is classified as less suitable, while 10221.04 km² (4.36%) and 4967.70 km² (2.12%) are categorized as moderately and highly suitable habitats, respectively (Table 2). In this climatic conditions, Striga hermonthica has spread across eastern Amhara, including areas such as Kalu, Kobo, Bati, and Kewet. It has also infested parts of western Amhara, notably Quara, Metema, and Achefer. The weed expanded into northern and eastern Tigray, woredas such as Adiyabo, Wukro, and Ahferom. Central Ethiopia, Particularly Dulecha, Sululta, and Mieso, is also experiencing its spread. Furthermore, eastern Ethiopia includes Dire Dawa, Jijiga, and Tefiri Bere been infested as well. It has also spread to the areas of Teltele, Burji, Yabelo, and Bore. The areas identified as highly and moderately suitable for this weed are mainly located in the eastern Amhara region, specifically between Kobo and Kewet, and northern Tigray, such as Esimbela, and Tahtay Adiyabo (Fig. 4a).

Table 2.

Predict change of gain and loss of suitable habitat in the year 2050 and 2070 of different scenarios for S. hermonthica and sorghum. It shows the current and future suitable habitat range size, loss and gain as result of climate change.

Scenarios Remain suitable habitat Gain Loss Net change Habitat predicted to be suitable (km2)
km2 % Km2 % Km2 % Km2 %
S. hermonthica Current 234326.35
SSP 4.5 (2050) 168770.49 77.16 180911.50 82.71 49955.76 22.84 130955.74 59.87 349681.99
SSP 8.5 (2050) 137011.46 65.89 221420.62 106.48 70931.98 34.11 150488.65 72.37 358432.08
SSP 4.5 (2070) 139112.50 62.76 198252.73 89.44 82537.90 37.24 115714.82 52.21 337365.23
SSP 8.5 (2070) 106177.96 50.27 260893.41 123.53 105,025 49.73 155868.42 73.80 367071.37
Sorghum Current 340836.74
SSP 4.5 (2050) 115,188.33 33.80 209561.29 61.49 225630.47 66.20 −16069.18 −4.71 324749.62
SSP 8.5 (2050) 110587.76 32.45 203012.74 59.56 230251.67 67.55 −27238.93 −7.99 313600.5
SSP 4.5 (2070) 108696.33 31.89 177921.1 52.2 232140.41 68.11 −54219.31 −15.91 286617.43
SSP 8.5 (2070) 103014.13 3022 210711.95 61.82 237821.24 69.78 −27109.28 −7.96 313726.09
Open in a new tab

Fig. 4.

Fig. 4

Fig. 4

Open in a new tab

Shows potentially suitable habitats for Striga hermonthica (a) and Sorghum bicolor (b) under current and future climate conditions across different scenarios. The first row of fig. illustrates the localities of unsuitable, less suitable, moderate, and highly suitable habitats under current climate conditions. The second and third rows of figures similarly show the locations of unsuitable, less suitable, moderate, and highly suitable habitats projected under climate change scenarios, specifically the middle pathway and worst emission scenarios of SSP4.5 and SSP8.5 for the years 2050 and the 2070s.

Under current climatic conditions, Sorghum bicolor is cultivated across approximately 340,836.74 km² in Ethiopia. Of this area, 274,737.35 km2 is classified as less suitable, while 50,900.68 km² and 15,235.37 km² are considered moderately and highly suitable areas, respectively. Like the parasitic weed, Striga hermonthica, the highly suitable and moderately suitable areas for Sorghum bicolor extend to lower elevations, particularly from Waja to Berhet. This region includes woredas such as Raya Azebo, Kobo, Werebabo, Bati, Kalu, Dawa Chefa, Shewarobit, and Berhet. Additionally, these suitable areas also reach into eastern Ethiopia, encompassing Merti, Anchar, Mieso, Doba, Kersa, and Jarso. Furthermore, suitable regions extend to Gorje, including Wara Jareso, Baso Liben, and Gindeberet (Fig. 4b). The spatial overlap between the highly suitable habitats of Striga hermonthica and Sorghum bicolor increases the risk of infestation and potential vulnerability of sorghum production in those areas.

Future predicted distribution of Striga hermonthica and Sorghum bicolor

The model results indicated that by 2050, under both the middle pathway (SSP4.5) and the worst emission (SSP8.5) scenarios, the suitable habitats for the parasitic weed Striga hermonthica are projected to increase by 130,955.74 km² and 150,488.65 km², respectively, compared to the current distribution pattern. Thus, the suitable habitat for the year 2050 will be predicted to be 349,681.99 km² and 358,432.08 km² in the SSP4.5 and SSP8.5 scenarios, respectively (Table 2). In contrast to the expansion of suitable habitats for the parasitic weed, the suitable habitats for Sorghum bicolor declined in both the middle pathway (SSP 4.5) and the worst emission scenarios (SSP 8.5) by the years 2050 and 2070.

Under the SSP 4.5 scenario of the year 2050, the areas of highly suitable habitats for this weed decline by a net 2754.90 km2 compared to its current distribution pattern. In contrast, moderately and less suitable habitats are projected to expand by 25,438.68 km2 and 98931.93 km2, respectively. Bati, Dawa Chefa, Kalu, Kobo, and Sanja are among the areas that are expected to be highly and moderately suitable. Moreover, about 165,359.10 km² of newly suitable habitat is expected to emerge in the western, northwestern, southeastern, and other regions of Ethiopia. In contrast, about 43,743.38 km² of currently suitable areas are projected to become unsuitable, particularly in the Tigray region, southern Ethiopia, east-central Ethiopia, and eastern Ethiopia. On the other hand, under SSP 4.5 in 2050, the total suitable habitat for Sorghum bicolor is projected to be 324,749.62 km². This represents a decrease of 16,069.18 km² compared to the current total suitable habitat. It is also anticipated that 12,698.11 km² of the previously highly suitable area will lose its suitability. Additionally, 22.46% of the previously moderately suitable habitat is expected to experience a reduction in its suitability level. Woredas such as Zikuala, Sekota, Belesa, Ewa, Chifra, Illu, Gimbicho, Akaki, several woredas in Gurage, Gololcha and Hareshen, are predicted to become unsuitable for habitation. In contrast, northwestern and western woredas like Guba, Metema, and Tsegede are anticipated to become newly suitable habitats. Additionally, the suitability level for Tahetay Adyabo and Kafta Hummera is expected to increase compared to its previous status.

Under SSP 8.5 for the year 2050, the total suitable area for this weed is projected to increase by 150,488.65 km² compared to the current suitable habitat range. We predict a net decline of 36.79% in highly suitable areas under the worst emission scenario (SSP 8.5) for the year 2050. In contrast, habitats classified as moderately and lowly suitable are expected to expand significantly, compared to the current geographical range size (Table 3). Eastern Amhara (including Kobo, Bati, Kalu, Kewet), Arkayl, Sanja, Metema, Agarfa, Amibara, and around Jijiga will become moderately and highly suitable habitats (Fig. 4a). In this year and scenario, about 195,088.75 km² of previously unsuitable areas are expected to become newly suitable habitats, particularly in the northwestern Ethiopia (Kafta Humera, and Sanja), the Gambela, and the Somali regions (Kelafo, Gode and Debewayin). Meanwhile, approximately 63,469 km² of previously suitable habitat, including Zikula, Sekota, and Hareshen will become unsuitable for this weed. On the other hand, under SSP 8.5 for the year 2050, the suitable habitat for Sorghum bicolor, is projected to decline by 27,238.93 km². Many woredas in the eastern Amhara and Tigray regions, as well as central and eastern Ethiopia, are experiencing a decline in their previous suitability levels for sorghum. In contrast, the suitability levels of sorghum habitats are increasing in the central, northern, and western woredas of the Tigray region. Additionally, western Ethiopia and the Kefa zone are expected to become new areas suitable for sorghum. Highly suitable habitats for sorghum are anticipated to be found in Kalu, Dawa Chefa, and Tehuledere, all located within the eastern Amhara region under this scenario (Fig. 4b). The results indicated that both the host and the parasitic plants share similar moderately and highly suitable habitats this year and scenario, such as Kalu, Bati, Dawa Chefa, and Kobo. Additionally, in the same year, both species experienced a loss of suitable habitat from similar geographical areas; for instance, Hareshen, Zikula, Sekota, and Dehana demonstrate the weed’s dependence on the host plant.

Table 3.

Habitat gain and loss across suitability levels under SSP 4.5 and SSP 8.5 for 2050 and 2070, compared to current suitability level. It show the magnitude loss and gain of less, moderate and highly suitable as result of projected climate change.

Scenario Level of suitability Remain suitable Gain Loss Net change Habitat predicted to be suitable (km2)
km2 % km2 % km2 % km2 %
SSP 4.5 2050 L. suitable 160061.91 73.06 157962.04 72.10 59030.11 26.94 98931.93 45.27 318023.95
M. suitable 6762.09 66.17 28895.52 282.76 3456.84 33.83 25438.68 248.93 35657.61
H. suitable 1946.49 39.18 266.32 5.36 3021.22 60.82 −2754.90 −55.46 2212.81
SSP 8.5 2050 L. suitable 131420.57 59.99 182963.44 83.51 87668.11 40.01 95295.31 43.50 314384.01
M. suitable 4582.00 44.84 43788.92 428.51 5636.93 55.16 38152.00 373.35 48370.92
H. suitable 1008.89 20.31 2131.24 42.90 3958.81 79.69 −1827.57 −36.79 3140.13
SSP 4.5 2070 L. suitable 131615.60 60.07 182427.53 83.26 87479.02 39.93 94948.51 43.34 314043.13
M. suitable 5971.97 58.44 22887.07 223.97 4246.95 41.56 18640.12 182.41 28859.05
H. suitable 1524.92 30.70 199.47 4.02 3442.78 69.30 −3243.31 −65.29 1724.40
SSP 8.5 2070 L. suitable 102546.18 46.80 221306.09 101.01 116552.76 53.20 104753.33 47.81 323852.26
M. suitable 3528.51 34.53 46719.38 457.18 6690.42 65.47 40028.96 391.71 50247.88
H. suitable 103.28 2.08 1724.68 34.72 4864.43 97.92 −3139.74 −63.23 1827.96
Open in a new tab

In 2070, under SSP 4.5, the suitable habitats increase by a net gain of 198,252.73 km² compared to the current geographical suitability pattern. In contrast to this the suitable habitat for the host plant, Sorghum bicolor, will declined by 54,219.31 km2. While the highly suitable geographical habitats of Striga hermonthica decline by 3,243.31 km² from the current range size, moderately and lowly suitable habitats will expand by a net gain of 18,640.12 km² (182.41%) and 94,948.51 km² (43.34%), respectively. On the other hand, the highly, moderately, and low suitable habitats of sorghum are expected to decrease by 5,359.17 km² (35.17%), 19,823.99 km² (38.95%), and 29,036.15 km² (10.57%), respectively. The highly suitable habitats of Striga hermonthica are expected in the Bati, Dawa Chefa, Ofla and Kalu woreda, while the moderately suitable areas are likely to be found in Kafta Humera, Tahtay Adiyabo, Kobo, and Bati. Raya Azebo, Kobo, Guba Lafto, Kalu, Bati, Tahtay Koraro, and Midda Weremo are among the woredas that are expected to be highly and moderately suitable habitats for sorghum. In this year and scenario, both the host and plants also share similar moderately and highly suitable habitats like Kalu, Bati, and Dawa Chefa, and those areas are at risk for sorghum cultivation (Fig. 4).

By 2070, under the SSP 8.5 scenario, the total area of suitable habitats striga is projected to be 367,071.37 km², representing a net gain of 155,868.42 km² (73.80%) compared to the current distribution range. While the areas classified as less and moderately suitable are expected to expand by 104,753.33 km2 and 40,028.96 km2, respectively, the areas designated as highly suitable are predicted to shrink by 3139.74 km2, diverging from the current expansion trend (Table 3). Moderate suitability is expected in northern and northwestern Ethiopia, including woredas such as Kafta Humera, Tahtay Adiyabo, Laelay Adiyabo, Sanja, Welkait and Metma. Moreover, eastern Amhara is the habitat where moderately and highly suitable areas will occur (Fig. 5).

Fig. 5.

Fig. 5

Fig. 5

Open in a new tab

Map depicted change in suitable habitat in the future different scenarios comparatively with the current Striga hermonthica (a) and Sorghum bicolor (b). The map shows where suitable habitat change occurred from the current ranges as result climate change. The map were created using ArcGIS version 10.8.

On the other hand, the total suitable habitat of Sorghum bicolor decreased by 27,109.28 km² from the current suitable habitat range and is projected to reach 313,726.09 km². While the highly and low-suitable habitats are expected to decline by 10,071.64 km² (66.10%) and 20,994.54 km² (7.65%), respectively, the moderately suitable habitat is projected to expand by 3,954.89 km² (7.7%) compared with the current suitable range. Raya Azebo, Gulina, Kalu, Dawa Chefa, Bati, and Teferi Ber woredas are identified as moderately and highly suitable areas for sorghum cultivation in 2070 under the worst emission scenario. Additionally, new areas of suitability are expected to emerge in northwestern, western, and southwestern Ethiopia, specifically within the Keffa Zone and Jimma. Raya Azebo, Tehuledere, Bati, Kobo, and Guba Lafto woredas are among the moderately and highly suitable habitats for both the host and weed plants. Thus, those woredas are among the identified risk areas for producing sorghum (Fig. 6b). Those mentioned moderate and highly suitable woredas represent priority hotspots for targeted monitoring and integrated management interventions under future climate change.

Fig. 6.

Fig. 6

Open in a new tab

Area of suitable habitats predicted under current and future climate scenarios (A), area of suitable area based on suitable level (B).

Suitable habitat overlaps between the host and the parasitic weed

Under the current climatic conditions, 180,616.15 km² of habitats are suitable for both the host plant and the parasitic weed (Table 4). Of the total suitable habitat for sorghum, 52.99% is also suitable for the weeds. Additionally, 77.08% of the habitat that is suitable for the weed is similarly suitable for the host plant, sorghum. During this time, the host and weed plants overlap in various woredas of the Great Rift Valley, extending into northern Ethiopia. This range also includes eastern Ethiopia, with areas such as Misrak Gashamo and Degehabur (Fig. 7).

Table 4.

Area and percentage of suitable habitat overlap between the host and weed plants under current and future climatic conditions.

Period Suitable areas for host plant Suitable habitat for weed Area overlap Host plant (%) Weed (%)
Current 340,836.74 234,326.35 180,616.15 52.99 77.08
2050 SSP 4.5 324,749.62 349,681.99 198,770.76 61.21 56.84
2050 SSP 8.5 313,600.5 358,432.08 188,425.69 60.08 52.57
2070 SSP 4.5 286,617.43 337,365.23 158,815.84 55.41 47.08
2070 SSP 8.5 313,726.90 367,071.37 169,967.41 54.17 46.30
Open in a new tab

Fig. 7.

Fig. 7

Open in a new tab

Suitable habitat overlaps between the host and the weed plants across time series.

In both the middle pathway scenario (SSP 4.5) and the worst emission scenario (SSP 8.5) for the year 2050, the projected overlap area is expected to increase compared to current levels. Under the SSP 4.5 scenario for 2050, the projected total overlap area is 198,770.76 km². Among the total suitable habitats for sorghum, 61.21% are also suitable for the growth of the parasitic weed, Striga hermonthica. Similarly, 56.84% of the habitats suitable for Striga hermonthica are similarly suitable for sorghum. In the same scenario and year, Striga hermonthica has expanded to both higher and lower elevation ranges, such as Meket, Bugena, Jama, Gimbicho, and Chifera, which are areas unsuitable for sorghum, implies utilizes other host plants. Moreover, new suitable habitats are expected for both the host and the weed in northwestern Ethiopia, like Metema, Quara, Sanja, and Guba. Under the SSP8.5 scenario for 2050, the two plant species are projected to share a total of 188,425.69 km² of suitable habitat. Of the total suitable habitats for sorghum, it is anticipated that 60.08% will also be suitable for the weed plants.

Under both SSP 4.5 and SSP 8.5 for the year 2070, the total overlap area of Sorghum bicolor and Striga hermonthica is projected to decrease compared to the current year and the year 2050. In SSP 4.5, approximately 55.51% of the suitable habitat for the host plant is similarly suitable for the weed, while in SSP 8.5, 54.17% of sorghum’s suitable habitats are also suitable for the weed. In both scenarios for this year, the overlapping area in eastern Ethiopia, including the Afedern, Miesso, and Lome woredas, decreases. Conversely, northwestern Ethiopia, which includes Guba, Quara, Metema, Sanja, Welkiet, Tegede, and Kafta Humera, remains among the suitable habitats for both the host plant and the weed.

Discussion

Climate change, driven by both anthropogenic and natural factors, affects the extent and suitability of species habitats, influencing them either positively or negatively13. This climate change is important for its better distribution the parasitic weed, Striga hermontica, which poses a significant threat to cereal production in sub-Saharan Africa, including Ethiopia, by diminishing agricultural productivity, particularly the cultivation of sorghum, maize, rice, and teff. This parasitic weed is well adapted to environments characterized by drought and low soil fertility. Consequently, the anticipated changes in climate may significantly contribute to the increased spread and infestation of this weed. Thus, identifying the suitable habitats by considering the main environmental variables and bioclimatic factors is crucial for both conserving economically significant species and controlling the spread of species that have negatively impacted human economies.

The ensemble model results of our study show that precipitation during the warmest quarter (Bio18) and temperature seasonality (Bio4) strongly influence the distribution and infestation of Striga hermonthica. This is because this parasitic weed requires adequate moisture in the warmest quarter to break dormancy, attach to its host crop, and carry out essential physiological functions. Similar to our finding, Zewudie, Ding14 showed that precipitation in the warmest quarter has a significant impact on the distribution of Eragrostis tef. In most parts of Ethiopia, precipitation in the warmest quarter occurs between mid-June and September15. However, some parts of Ethiopia, including eastern, southeastern, and southern regions, experience this conditions between March and May16. During this season, host crops such as sorghum and maize are planted and are available17.

The other environmental variable, seasonal temperature, significantly influences the expansion of Striga hermonthica. This is due to the fact that this semi-parasitic weed requires low seasonality (stable temperature) for its germination and to respond to the host root exudates18. In contrast, highly seasonal temperatures hinder germination by disrupting the synchronization between Striga hermonthica and its host plants. Additionally, as Striga hermonthica is a partial parasite, seasonal temperature variations influence its distribution by affecting the physiological activities of host plants2.

As Striga hermonthica is a semi-parasitic weed, and its occurrence and distribution also depend on the availability and distribution of host crops such as Sorghum bicolor, teff, rice, sugarcane, and finger millet, as well as climate conditions. As a result, regions that support these host plants also influence the weed’s distribution. This finding implies that areas favorable for host plants may have a higher potential for infestation. For example, soil nitrogen content influences the distribution of sorghum, which in turn indirectly affects the distribution of Striga hermonthica. Moreover, continuous cultivation of cereal crops and degraded agricultural systems may further promote the persistence and spread of the weed across agro-ecosystems. Therefore, the distribution of Striga hermonthica is regulated not only by climatic suitability but also by the interaction among host crop availability, soil fertility conditions, and agricultural land-use practices. However, due to its ability to infest multiple host plants, Striga hermonthica may spread to various regions of Ethiopia in the future, depending on the availability of these host plants.

The distribution of a species is largely constrained by the availability of suitable habitats, which are shaped by both natural and human-induced factors such as soil characteristics, moisture availability, land use and land cover changes, and various climatic conditions19. Therefore, understanding the current potential range of a species is essential for projecting its future distribution. Under present climatic conditions, Striga hermonthica is predominantly found in the lowland regions of northern and eastern Ethiopia. This pattern is partly due to the shallow and poorly developed soils that characterize much of Ethiopia’s lowlands20. Additionally, these areas typically receive less than 700 mm of annual rainfall and experience relatively low climatic stability21. S. hermonthica thrives under such conditions, favoring poor soils with low organic matter content and annual moisture levels between 400 and 700 mm22,23. The widespread cultivation of host crops such as sorghum, rice, and maize in these lowland regions further contributes to the weed’s successful spread and establishment. Additionally, the dominance of cereal-based agriculture, along with the continuous cultivation of host plants, may contribute to the persistence of Striga hermontica seed banks in agricultural fields. Furthermore, poor soil fertility management increases the risk of infestation by this parasitic weed. The current ongoing land use changes may also create more favorable conditions for this parasitic weed due to the expansion of cultivated land driven by increasing demand.

While the extent of impacts varies across different scenarios, future climate change is expected to significantly affect the suitable habitats of species and their distribution, either positively or negatively. Our projections indicate that by 2050 and 2070, under SSP 4.5 and SSP 8.5 scenarios, climatically suitable habitats for Striga hermonthica may expand beyond the current suitable range. Similar to our findings, Gufi, Manaye24 reported that Tamarindus indica may experience expansion in climatically suitable habitats beyond its current distribution pattern, which has been documented in the Tigray region.

The findings suggest that the climatically suitable habitat of Striga hermonthica mayexpand under the high-emission (worst-case) scenario than under the intermediate emission pathway for both 2050 and 2070. This species thrives in warm conditions, particularly within a temperature range of 25–35 °C, and requires 400–700 mm of moisture to break seed dormancy and establish on host plants. According to IPCC1, worst-case emission scenarios are expected to result in higher temperatures compared to moderate pathways, thereby creating more favorable conditions for the potential establishment and persistence of S. hermonthica. Additionally, climate projections indicate that both temperature and rainfall will increase in northern and northwestern Ethiopia, while central and lowland areas will experience reduced and more erratic rainfall patterns. These changes could create new suitable habitats for S. hermonthica, particularly in the northern and northwestern regions. In these areas, irregular and intense rainfall may accelerate soil erosion and deplete organic matter25, further enhancing conditions that favor the growth of this parasitic weed. However, these projections represent potential climatic suitability rather than actual field-level infestation or confirmed spread, and therefore should be interpreted with caution.

As a facultative (partially) parasitic plant, the spread of Striga hermonthica is closely linked to the distribution and climate resilience of its host species. Its expansion is not solely driven by environmental conditions favorable to the parasite itself but also by the ability of host crops to survive and persist under changing climates. Major host plants such as sorghum and maize belong to the C₄ photosynthetic pathway, which allows them to perform efficiently under high temperatures, intense sunlight, and low moisture conditions26. These traits enable them to withstand the increasingly harsh environmental conditions projected under future climate scenarios. Moreover, S. hermonthica has different host plants, including Sorghum bicolor, Eragrostis tef, Zea mays, and rice. This presence of multiple host plants could contribute to its expansion in future projected climatic conditions.

Under the current climatic conditions, the overlap area is 180,616.15 km². For the year 2050, the potential overlapping area is projected to increase under both SSP 4.5 and SSP 8.5 scenarios, along with a rise in the share of suitable habitat for the host plant. An increase in the overlap of suitable habitats for both the host and weed plants indicates the potential expansion of the weed, which may significantly affect the agroecosystem and reduce cereal productivity, including sorghum production. Although the weed shares a substantial suitable habitat with sorghum cultivation areas, its suitable habitat is also expanding into areas unsuitable for sorghum cultivation. This indicates that Striga hermonthica has a strong potential to spread into new regions at both higher and lower elevations, even in the absence of sorghum production, by utilizing alternative host plants such as rice, finger millet, teff, maize, and other cereal crops. The overlap of two plant species is projected to decline from the current overlap potential by the year 2070 for both SSP 4.5 and SSP 8.5, potentially reaching 158,815.84 km² and 169,967 km², respectively. This decline is linked to the anticipated loss of previously suitable habitats for sorghum, particularly in the Sekota, Sahala, and Great Rift Valley woredas of Ethiopia, despite the weed persisting in those areas. This suggests that the weed has the potential to survive in harsh environments if it can obtain alternative host plants.

In both the SSP2-4.5 (intermediate) and SSP5-8.5 (high-emission) scenarios for 2050 and 2070, these resilient C₄ crops are expected to continue growing across large parts of Ethiopia, particularly in lowland and semi-arid regions, and under large scale irrigated farms. This continued presence of suitable hosts supports the projected expansion of S. hermonthica, as the parasite depends on host root exudates to trigger germination and complete its life cycle. Thus, the climate adaptability of host crops inadvertently facilitates the future spread of this parasitic weed under both moderate and severe climate change trajectories.

Conclusion

Precipitation of the warmest quarter (Bio18) and temperature seasonality (Bio4) are the two significant environmental predictors influencing the distribution of Striga hermonthica. Under both SSP 4.5 and SSP 8.5 scenarios for the years 2050 and 2070, the results depicted an expansion of suitable habitat compared to its current distribution. In comparison, the worst emission is better for expansion of Striga hermonthica than the middle emission scenario. The results indicated that in all scenarios and years, highly suitable habitats are expected to decline compared to the current total suitable areas. In contrast, low and moderately suitable habitats are projected to expand beyond their current geographical ranges. The northern, northwestern, and southeastern regions are projected to have newly suitable habitats.

Striga hermonthica significantly reduces the yield of important staple crops such as sorghum, maize, and teff, posing a major threat to food security in affected regions. The areas identified as highly and moderately suitable for weed occurrence include Kobo, Tehuledere, Bati, and Guba Lafto. Additionally, potential increases in suitability have been noted in regions such as Mieso, Kafta Humera, and Tahetay Adiyabo. Furthermore, newly suitable habitats are projected in northwestern and western Ethiopia, including Bensangul Gumaz, Gambela, Metema, and Snaja. These areas require priority attention for monitoring the spread of the weed. To effectively control and limit the spread of this parasitic weed, coordinated action is needed from all relevant stakeholders, including policymakers, agricultural extension agents, researchers, and local farming communities. One practical and sustainable approach is the promotion of organic soil amendments, such as compost. As highlighted by Woldemariam, Damot27, the application of nitrogen-containing fertilizers improves soil fertility and structure, which in turn creates unfavorable conditions for the germination and establishment of Striga hermonthica. In addition to enhancing soil fertility, crop rotation, such as using crops from the Fabaceae family, limiting agricultural expansion into natural ecosystems may help control the spread of the weed by disrupting its persistent seed banks. Agricultural experts should actively engage with farmers through training and awareness programs to encourage the adoption of such practices, which not only suppress the weed but also enhance overall soil health and crop productivity. Integrated management strategies that combine improved agronomic practices with ecological knowledge are essential to reduce the long-term impact of this weed on smallholder farming systems.

Limitation of the study

Although the ensemble modeling framework reduced dependence on a single algorithm, some uncertainty may still exist in the future habitat projections due to the use of a single GCM. Different climate models may produce varying future climate predictions and habitat suitability patterns. In addition, while the present study incorporated host-crop distribution through overlap analysis with sorghum, Striga hermonthica can also parasitize several other host crops, including maize, teff, rice, and finger millet, which were not explicitly included in the modeling framework. Therefore, future studies could incorporate multiple GCMs, broader host-crop datasets, and additional land-use dynamics to improve the robustness of future projections.

Methods

Study area

Habitat suitability identification for Striga hermonthica was conducted across the entire geographical extent of Ethiopia. Ethiopia, located in the Horn of Africa, spans from 3° to 15° N latitude and from 33° to 48° longitude. The total area of the country is about 1.13 × 106 km², with altitudinal variations ranging from 125 m below sea level at the Danakil Depression to 4,533 m above sea level at Ras Dashen Mountain28. This significant altitudinal variation contributes to the country’s rich biodiversity. Striga hermonthica has been observed in the farmlands of Ethiopia’s lowlands and mid-elevation areas, where host plants like sorghum, maize, and tef are cultivated.

Geographic occurrence records

The occurrence records for this parasitic weed were obtained from four sources: field collection with Kobo Toolbox (https://kobotoolbox.org) using a smartphone (n = 141), collections deposited at the National Herbarium of Ethiopia (Addis Ababa University; n = 18), different published literatures (n = 10), and the Global Biodiversity Information Facility (www.gbif.org) database (n = 3). Similarly, the occurrence data for Sorghum bicolor was obtained from field collection (n = 178), Notational Herbarium of Ethiopia (n = 35), Global Biodiversity Information Facility (n = 32), and various published literature (n = 13). The voucher specimen of Striga hermonthica, with collection number MA001, was collected, pressed, and transported to the National Herbarium of Ethiopia (ETH) for identification. Finally, the specimen was compared with the authenticated specimen in the herbarium and confirmed by experts. On the other hand, for the secondary species occurrence data, we have used the geo-referenced occurrence data collected from 1970 to 2024. To maintain temporal consistency with the bioclimatic predictor variables, occurrence data prior to 1970 was excluded.

The 172 and 258 recorded occurrence data points for Striga hermonthica and Sorgum bicolor, respectively were entered into a Microsoft Excel spreadsheet and saved in comma-delimited format (CSV) (Supplementary Table 1). Finally, the CSV data was converted into a shapefile and rarified using SDM Toolbox version 2.6 in ArcGIS version 10.8, with a grid resolution of 1 × 1 km to get rid of auto-spatial duplication, as outlined by Hernández-Lambraño, Carbonell29. Consequently, 132 and 248 occurrence records were retained for use in the SDM modeling for Striga hermonthica and Sorgum bicolor, respectively. Moreover, to check the accuracy of the occurrence point we used Google earth pro. Figure 8.

Fig. 8.

Fig. 8

Open in a new tab

Image of Striga hermonthica (purple color) taken from sorghum farmlands around Woledi town.

Ecological predictor variables

Various ecological predictors, such as bioclimatic variables, topographic features, and soil data, were used for modeling habitat suitability. The bioclimatic variables were obtained from the freely available WorldClim database version 2.1 (https://www.worldclim.org/data/cmip6/cmip6climate.html) at 30 arc seconds (~ 1 km2) resolution, based on interpolated climate data from 1970 to 200030. We also utilized the topographic environmental predictors (aspect, elevation, and slope), which were acquired from the digital elevation shuttle radar topography mission (SRTM, http://srtm.csi.cgiar.org)31. On the other hand, the soil data were obtained from the International Soil Reference and Information Centre (ISRIC, https://www.isric.org)32. Prior to running species distribution models (SDMs), all predictor variable data were re-projected, cropped and resampled using R software version 4.4.1 to match the bioclimatic variable resolution of 1 km², ensuring consistency across pixel dimensions. The current bioclimatic coordinate reference system (CRS) was used as a reference for the project and resampled.

The impact of climate change on the distribution of Striga hermonthica was predicted based on the Global Climate Model (GCM) of HadGEM3-GC31-LL (the UK’s physical climate model) under the Coupled Model Inter-comparison Project (CMIP6). The model was selected due to its better performance in SDM modeling for Ethiopia33. The years 2050 and 2070, with Shared Socioeconomic Pathways (SSP 4.5, the medium pathway of the future greenhouse gas emissions, and SSP 8.5, the worst emission scenario), were selected for SDM prediction.

Environmental predictor selection

Using all the downloaded environmental variables for SDM gives redundant information and results in model overfitting34. Moreover, not all the environmental predictors contribute equally to determining the suitable habitat for species35. Thus, selecting environmental predictors that significantly contribute to the distribution of the desired species is important for ensuring the robustness of the model. To do this, first, Pearson correlation analysis was done to group the environmental predictors that have an r < 0.7 correlation coefficient with the methodology outlined by36. Then, to remove multicollinearity effects between the environmental variables, variance inflation factors (VIF < 5) was conducted using usdm package in R environment Olivoto, de Souza37(Appendix 1). As a result, out of the initial 34 predictor variables, 10 were selected for modeling the prediction of Striga hermonthica (Table 5).

Table 5.

Environmental variables used in the modeling with their description and unit.

Environmental variable Description Unit
Bio2 Mean Diurnal Range (Mean of monthly (max temp - min temp)) °C
Bio4 Temperature Seasonality (standard deviation ×100) °C
Bio5 Max Temperature of Warmest Month °C
Bio14 Precipitation of Driest Month mm
Bio15 Precipitation Seasonality (Coefficient of Variation) %
Bio18 Precipitation of Warmest Quarter mm
Bio19 Precipitation of Coldest Quarter mm
Aspect Direction of slope Degree
Nitrogen_sd1 Total nitrogen content at a depth of 0–20 cm g/kg
Slope Steepens of landscapes Degree
Open in a new tab

Species distribution modeling

As the ability of different algorithms to identify and predict suitable habitats varies, a single model might not give the maximum prediction accuracy38. Thus, to obtain robust results and to minimize uncertainty when using single algorithms, an ensemble approach is recommended39. Accordingly, in this study, seven more common algorithms, namely, boosted regression trees (BRT), random forest (RF), generalized linear model (GLM), generalized additive model (GAM), maximum entropy (MaxEnt), multivariate adaptive regression splines (MARS), and support vector machines (SVM), were used. In contrast, we used six algorithms for Sorghum bicolor since the generalized linear model shows the least performance in model validation during modeling. To evaluate the model performance robustly, K-fold cross-validation (k = 5) was employed during the modeling process using the sdm package in the R programming environment version 4.4.1, and the collected occurrence data was divided into five equal subset. In each iteration 80% of the occurrence points were allocated for training the model, while the remaining 20% was used to assess the model’s performance. To ensure that the modeling conditions are environmentally realistic and to minimize background sampling bias, we randomly generated 1,000 background points within a 10 km radius of the recorded presence points40. This approach assumes that habitat near known occurrence locations is more likely to reflect environmentally suitable conditions for Striga hermonthica and Sorghum bicolor. This is especially true in agricultural landscapes where the weed is often linked to the cultivation of host crops, which helps reduce the inclusion of ecologically unrealistic background environments. This approach also helps reduce potential sampling bias associated with uneven survey effort, particularly the tendency for records of Striga hermonthica to be concentrated near roads, agricultural hubs, and grazing areas.

Model validation

In model performance evaluation we used both threshold-dependent and threshold-independent metrics. While the area under the curve (AUC) and the receiver operating curve (ROC) were used for the overall model discrimination, the threshold of the true skill statistic (TSS) and the sum of sensitivity and specificity provided threshold observation41. AUC and TSS are an effective way to measure model performance42. AUC values range from 0 to 1 and are categorized as follows: 0.5–0.7 implies poor performance, 0.7–0.9 shows good performance, and values greater than 0.9 indicate excellent performance43. On the other hand, the TSS value ranges from − 1 to 1, with values below zero indicating performance no better than random prediction, 0 to 0.3 indicating poor performance, 0.3 to 0.5 showing moderate performance, 0.5 to 0.7 indicating good performance, 0.7 to 0.9 referring to very good performance, and values greater than 0.9 suggesting excellent model performance44. TSS values obtained from the cross-validation testing data were utilized to weight individual models performance in ensemble modeling.

Spatiotemporal change assessment

To assess changes in ecologically suitable habitats over time due to climate change and human influences, four evaluation criteria were applied based on the approach of Yan, Zhang45. These criteria include: (1) Suitable areas – habitats that remain suitable under both current and future conditions; (2) Unsuitable areas – habitats that remain unsuitable across time periods; (3) Vulnerable areas – currently suitable habitats projected to become unsuitable in the future; and (4) Newly suitable areas – habitats that are currently unsuitable but are expected to become suitable under future climate scenarios. The calculations for these categories were conducted using formulas adapted from Thuiller, Georges46.

graphic file with name d33e2173.gif
graphic file with name d33e2176.gif
graphic file with name d33e2179.gif
graphic file with name d33e2182.gif
graphic file with name d33e2185.gif

The suitable classes of host and weed plants were categorized into less suitable, moderately suitable, and highly suitable habitats following the method used by47. To achieve this, we used the threshold used for binary classification, i.e., suitable and unsuitable, and then divided the suitable area into three equal parts. We used the threshold value of 0.06567 for Striga hermontica and 0.07839 for Sorghum bicolor.

Additionally, we identify the overlapping suitable habitat for both the host plant and the weed using the intersect tools in ArcGIS (version 10.8). We then calculate the areas that are shared and those that are not shared, utilizing the formula adapted by48.

graphic file with name d33e2200.gif

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (875.5KB, docx)

Acknowledgements

We would like to thank our field assistants, including Seid Adem, Seid Mujeh, Melake Eshete, Abdu Seid, Seid Ali, Keder Mohamed, Jemal Ahemd, and Bekalu Assefa. We gratefully acknowledge the Department of Plant Biology and Biodiversity Management of Addis Ababa University and the National Herbarium (ETH) for letting us use the available facilities.

Abbreviations

AUC

Area under curve

VIF

Variance Inflation Factor

SDM

Species distribution model

ROC

Receiver operating characteristics

TSS

True skill statistic

Author contributions

Mikiyas Abebe: Conceptualization, data collection, curation, and analysis, R-software, writing the draft and the final version. Zerihun Woldu: Conceptualization, Methodology, Supervision, Validation, Review and Editing, analysis with R-software. Zemede Asfaw: Conceptualization, Methodology, Supervision, Validation, Review and Editing.  Bikila Warkineh: Conceptualization, Methodology, Supervision, Validation, Review, and Editing.

Data availability

All data generate during this study are included in this article and supplementary information.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.IPCC, b. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the IPCC (Cambridge University Press, 2021).
  • 2.IPCC Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri & L.A. Meyer (eds.)].. Geneva, Switzerland, 151 pp.: IPCC. (2014).
  • 3.Muluneh, M. G. Impact of climate change on biodiversity and food security: A global perspective—a review article. Agriculture & Food Security10(1), 1–25 (2021). [Google Scholar]
  • 4.Dafaallah, A. B. Biology and physiology of witchweed (Striga spp.): A review. Int. J. Acad. Multidiscip.3(10), 42–51 (2019). [Google Scholar]
  • 5.Teressa, T. Review on Striga distribution, infestation and genetic potential in Ethiopian sorghum (Sorghum bicolor (L.) Moench). Int. J. Res.5(2), 23–31 (2019). [Google Scholar]
  • 6.FischerInga, E. Flora ofEthiopia and Eritrea Vol. 5 (Uppsala University, 2006). [Google Scholar]
  • 7.Degebasa, L. et al. Prevalence and socioeconomic impact of Striga (Striga hermonthica) in sorghum producing areas of east and west Hararghe zones, Oromia, Ethiopia. Int. J. Agron.2022 (1), 8402280 (2022). [Google Scholar]
  • 8.Tadesse, F. et al. Effect of Striga trap crops and nitrogen fertilizer application on yield and yield related traits of Sorghum [Sorghum bicolor (L.) Moench] at Fedis District, Eastern Ethiopia. Open. Access. Libr. J.5 (4), 1–17 (2018). [Google Scholar]
  • 9.Teferi, M. G. Ethiopian Sorghum (Sorghum bicolor (L.) Moench) genotpyes: Striga (Striga hermonthica (Del.) Benth.) resistance. (2019).
  • 10.David, O. A. et al. Habitat suitability and dispersal of invasive Striga species under climate change in Africa. Afr. J. Ecol.60(4), 1143–1154 (2022). [Google Scholar]
  • 11.Gebreslasie, A. et al. Abundance and distribution of Striga (Striga hermonthica (Del.) Benth.) infestation in selected sorghum (Sorghum bicolour L. Moench) growing areas of Tigray Region, Ethiopia. Afr. J. Agric. Res.11 (45), 4674–4682 (2016). [Google Scholar]
  • 12.Gomes, V. H. et al. Species Distribution Modelling: Contrasting presence-only models with plot abundance data. Sci. Rep.8 (1), 1003 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Williams, J. J. & Newbold, T. Local climatic changes affect biodiversity responses to land use: A review. Divers. Distrib.26 (1), 76–92 (2020). [Google Scholar]
  • 14.Zewudie, D. et al. Spatiotemporal dynamics of habitat suitability for the Ethiopian staple crop, Eragrostis tef (teff), under changing climate. PeerJ9, e10965 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Wagesho, N., Goel, N. K. & Jain, M. K. Temporal and spatial variability of annual and seasonal rainfall over Ethiopia. Hydrol. Sci. J.58 (2), 354–373 (2013). [Google Scholar]
  • 16.Shigute, M. et al. Analysis of rainfall and temperature variability for agricultural water management in the upper Genale river basin, Ethiopia. Sci. Afr.20, e01635 (2023). [Google Scholar]
  • 17.Derese, S. A. et al. The impact of drought on sorghum production, and farmer’s varietal and trait preferences, in the north eastern Ethiopia: implications for breeding Vol. 68, p. 424–436 (Acta Agriculturae Scandinavica, Section B—Soil & Plant Science, 2018). 5.
  • 18.Kimathi, E. et al. Ecological determinants and risk areas of Striga hermonthica infestation in western Kenya under changing climate. Weed Res.63 (1), 45–56 (2023). [Google Scholar]
  • 19.Sharma, A. K. et al. Assessment of land use change and climate change impact on biodiversity and environment, in Environmental pollution and natural resource management p. 73–89 (Springer, 2022).
  • 20.Beyene, S. et al. The soils of Ethiopia (Springer, 2023).
  • 21.Fazzini, M., Bisci, C. & Billi, P. The climate of Ethiopia. Landscapes and landforms of Ethiopia, : pp. 65–87. (2015).
  • 22.Doggett, H. Striga hermonthica on sorghum in East Africa. J. Agricultural Sci.65 (2), 183–194 (1965). [Google Scholar]
  • 23.Ebbisa, A. F. et al. Effect of vermicompost, intercropping, and seedbed types on Striga hermonthica infestation and sorghum production in East Hararghe, Ethiopia. Weed Biology Manage.24 (4), 167–181 (2024). [Google Scholar]
  • 24.Gufi, Y. et al. Modeling impacts of climate change on the geographic distribution and abundances of Tamarindus indica in Tigray region, Ethiopia. Heliyon, 9(7). (2023). [DOI] [PMC free article] [PubMed]
  • 25.Zhao, L. et al. Effect of rainfall intensity and duration on soil erosion on slopes with different microrelief patterns. Geoderma396, 115085 (2021). [Google Scholar]
  • 26.Umkulzhum, F. et al. Weeds and their response to changing climate: A review. Int. J. Environ. Clim. change. 14 (4), 768–779 (2024). [Google Scholar]
  • 27.Woldemariam, Z. G., Damot, G. A. & Zewdie, D. A. Nitrogen fertilizer and cattle manure for Striga (Striga hermonthica) management and enhancement of sorghum productivity in northwest Ethiopia. Nutrition45 (2), 232–245 (2021). [Google Scholar]
  • 28.Fashing, P. J. et al. Ecology, evolution, and conservation of Ethiopia’s biodiversity. Proc. Natl. Acad. Sci.119 (50), e2206635119 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Hernández-Lambraño, R. E., Carbonell, R. & Sánchez-Agudo, J. Á. Making the most of scarce data: Mapping distribution range and variation in population abundance of a threatened narrow-range endemic plant. J. Nat. Conserv.57, 125889 (2020). [Google Scholar]
  • 30.Fick, S. E. & Hijmans, R. J. WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int. J. Climatol.37 (12), 4302–4315 (2017). [Google Scholar]
  • 31.Farr, T. G. et al. The shuttle radar topography mission. Rev. Geophys., 45(2). (2007).
  • 32.Poggio, L. et al. SoilGrids 2.0: producing soil information for the globe with quantified spatial uncertainty. Soil7 (1), 217–240 (2021). [Google Scholar]
  • 33.Yebeyen, D. et al. Modeling and mapping habitat suitability of highland bamboo under climate change in Ethiopia. Forests13 (6), 859 (2022). [Google Scholar]
  • 34.De Marco, P. & Nóbrega, C. C. Evaluating collinearity effects on species distribution models: An approach based on virtual species simulation. PloS one. 13 (9), e0202403 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Irving, K., Jähnig, S. C. & Kuemmerlen, M. Identifying and applying an optimum set of environmental variables in species distribution models. Inland. Waters. 10 (1), 11–28 (2020). [Google Scholar]
  • 36.Brun, P. et al. Model complexity affects species distribution projections under climate change. J. Biogeogr.47 (1), 130–142 (2020). [Google Scholar]
  • 37.Olivoto, T. et al. Multicollinearity in path analysis: a simple method to reduce its effects. Agron. J.109 (1), 131–142 (2017). [Google Scholar]
  • 38.Norberg, A. et al. A comprehensive evaluation of predictive performance of 33 species distribution models at species and community levels. Ecol. Monogr.89 (3), e01370 (2019). [Google Scholar]
  • 39.Iturbide, M., Bedia, J. & Gutiérrez, J. M. Background sampling and transferability of species distribution model ensembles under climate change. Glob. Planet Change. 166, 19–29 (2018). [Google Scholar]
  • 40.Barbet-Massin, M. et al. Selecting pseudo‐absences for species distribution models: How, where and how many? Methods Ecol. Evol.3 (2), 327–338 (2012). [Google Scholar]
  • 41.Shabani, F., Kumar, L. & Ahmadi, M. Assessing accuracy methods of species distribution models: AUC, specificity, sensitivity and the true skill statistic. Global Journal of Human-Social Science: B Geography, Geo-Sciences, Environmental Science & Disaster Management, 18(1). (2018).
  • 42.Abrego, N. & Ovaskainen, O. Evaluating the predictive performance of presence–absence models: Why can the same model appear excellent or poor? Ecol. Evol.13 (12), e10784 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Salako, V. K. et al. Predicting the potential impact of climate change on the declining agroforestry species Borassus aethiopum Mart. in Benin: a mixture of geostatistical and SDM approach. Agroforest Syst.93, 1513–1530 (2019). [Google Scholar]
  • 44.Hanssen, A. & Kuipers, W. On the relationship between the frequncy of rain and various mateorological parameters: with reference to the problem of objective forecasting (na, 1965).
  • 45.Yan, Q. et al. Global climate change promotes aboveground growth and high-latitude species distribution in medicinal plants: Insights from the genus Panax. J. Clean. Prod.478, 143863 (2024). [Google Scholar]
  • 46.Thuiller, W. et al. Package ‘biomod2’. Species distribution modeling within an ensemble forecasting framework, 10: pp. 1600-0587.2008. (2016).
  • 47.Abebe, M. et al. Habitat suitability dynamics of the Ethiopian endemic multipurpose Millettia ferruginea subspecies under current and future climate change scenarios. Ecol. Processes. 15 (1), 32 (2026). [Google Scholar]
  • 48.Sahin, E. K. Assessing the predictive capability of ensemble tree methods for landslide susceptibility mapping using XGBoost, gradient boosting machine, and random forest. SN Appl. Sci.2 (7), 1308 (2020). [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (875.5KB, docx)

Data Availability Statement

All data generate during this study are included in this article and supplementary information.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES