Study area
This study was conducted in Ethiopia (3°–15°N, 33°–48°E), a landlocked country located in the horn of Africa and encompassing an area of 1.13 million km2. With a population of about 110.14 million and annual population growth rate is about 3.09 percent, the country the 2nd most populous country in Africa after Nigeria although it is the 10th largest country in Africa (Planning and Development Commission 2021). Altitudinal of the country ranges from 116 m bsl in the Danakil Depression to the highest peak of 4,620 m asl on Mount Ras Dashen. The Great Rift Valley runs from Northeast to Southwest of the country and separates the Western and South-eastern highlands. Extensive semi-arid lowlands in the East, South and West are extensions of these highlands. Because of the combined effects of the above factors, the country contains diverse ecosystems and exceptionally high diversity and endemic species of flora and fauna. As such the country spans two of the global biodiversity hotspots – Eastern Highlands and Horn lowlands (Williams et al. 2004). Overall, the country harbours 6,500-7,500 vascular plant species (including 1,600 endemic species), 322 mammal species (63 endemic species), 881 bird species (18 endemic), 253 reptile species (26 endemic), 78 amphibian species (39 endemic species), 175 fish species (41 endemic) (Tessema et al. 2022a). Although the conservation status and population trends have not known for many of these species, available data show that there are, at least, 109 globally threatened vertebrate species in Ethiopia, including 14 critically endangered, 42 endangered and 53 vulnerable species (Tessema et al. 2022a; see also IUCN 2020).
Agriculture is the mainstay of the Ethiopian economy, employing about 83% of the total population and contributing about 45% to the GDP, but agriculture in the country is rainfed agricultural practice. Ethiopia has the largest number of livestock in Africa and the 10th largest in the world (Mamo 2019). Exports almost entirely rely on agricultural commodities, such as coffee, oil seeds, dried pulses, hide and skin as well as live animals. Rapid human population growth with increasing food demand, fragile ecosystems, increasing energy and construction demand and eminent dangers from climate change are among key problems in the biodiversity sector of the country [Ministry of Finance and Economic Development 2011; Ethiopian Biodiversity Institute (EBI) 2014a; Planning and Development Commission 2021]. As a result, the current national development plan has focused on improving agricultural production, productivity and commercialization to reduce the impacts from the sector on biodiversity, and reducing degradation of natural resources and improving its productivity through protection and restoration actions (Planning and Development Commission 2021). Consequently, the issues of identifying and reducing threats to biodiversity conservation via threat abatement and restoration have been included in many of national environmental and development plans (see Ministry of Finance and Economic Development 2011; EBI 2014a; Planning and Development Commission 2021) and global plans (e.g., BIODEV2030; Vergez 2023).
STAR data acquisition and cleaning
For STAR score calculation at any spatial scale, four species-specific datasets are required: 1) the species extinction risk category in the IUCN Red List category reflected in a weight for this species s (Near threatened [NT] = 1, Vulnerable [VU] = 2, endangered [EN] = 3 and Critically endangered [CR] = 4); 2) the share of their current Area of Habitat (AOH) of each species s within location i compared to their global current AOH; 3) the historical Area of Habitat (AOH) for species s within location i that is expressed as a percentage of current global AOH; and 4) the relative contribution of each individual threat t to the extinction risk of species s calculated as the percentage global population decline from that threat (CBD 2019; Mair et al. 2021). Currently, all these datasets and STAR scores at global level are available from the International Biodiversity Assessment Tool (IBAT 2021b; http://www.ibat-alliance.org/) for 195 countries encompassing 15,000 species for which the necessary data are available.
For Ethiopia, STAR score data are available for a total of 113 vertebrate species (12 amphibians, 51 birds and 50 mammals), including 42 endemic species, by IUCN extinction-risk categories, which were included in the global STAR analysis. We downloaded these data from IBAT under specific permission and used for national STAR calculation. Information on list of species and their corresponding IUCN extinction-risk category and STAR score are provided in Table SI1–3 of additional data are given in Online Resource 1.
Prior to using these data for calculating national STAR score for Ethiopia and conducting detailed individual threat level analysis, we undertook a preliminary data quality assurance analysis in order to ensure that the global data (STAR scores, species and threats) were relevant to Ethiopian context. First, we examined the STAR scores and found that the habitat restoration (STAR-R) component of the STAR metric represented more than 15 times the total STAR-T score for Ethiopia. However, those surprising figures, very different from other countries’s profiles, were due to six species (Arvicanthis blicki, Ptychadena nana, Crocidura Lucina, Tachyoryctes macrocephalus, Leptopelis yaldeni and Crithagra ankoberensis) with a very high STAR-R score (higher than 3000). These high restoration values seemingly show that all species had very large ranges but are now very restricted and highly threatened (Vergez 2022). However, we found that the reason for the extremely large values was an error in calculation of the original and current area of habitat of each species. For example, for Tachyoryctes macrocephalus, a species restricted to grassland and alpine moorlands of the Bale mountains, above 3000 m asl (Asefa et al. 2022), the current AOH was estimated in the original STAR calculation as 5,140 km2 and historical range was estimated as 8,738km2, with restoration area equals to 3,598km2 (70% of current AOH). However, these extents are far greater than the total area of the Afroalpine moorlands and grasslands in the Bale mountains range above 300m asl, where the species’ distribution is confined, in fact is about 1,200km2 (Asefa et al. 2022; Wraase et al. 2022). Thus, instead of omitting these species with outlier data, we extracted historical and current areas of habitat types of each species from land use/land cover studies (see Table S1) and recalculated their STAR-R component scores based on literature review, data from IUCN Red List and our experience (Table S2 and S3). To recalculate STAR-R scores for the six species with outlier score values based on the global scale analysis, we derived original and current extent of area of habitats based on information on their known distribution, suitable habitat types and land use and land cover changes in their known range. Detailed methodological discussions on how historical and current areas of habitat for each species were calculated and the calculated score values are given in Table SI 1–3 in Online Resource 1.
Secondly, we reviewed the initial list of amphibian, bird and mammal species used in the global STAR analysis. Although all the proposed species are relevant, we found that two endemic species with known Red List categories, were missing in the global STAR analysis. One of these species, the Amphibious Rat (Nilopegamys plumbeus), is currently listed as Critically Endangered, but its distribution range map is not available on the IUCN Red List (Peterhans & Lavrenchenko 2008). The other species is the Sheko Forest Brush-furred Rat (Lophuromys pseudosikapusi), a species known to occur only in Sheko forest in the south-western Ethiopia, listed as Endangered (Dano et al. 2018). Thus, based on literature review and our experiences on the suitable habitat, distribution range and threats of each species, we derived STAR scores for these two species and included them in the national STAR calculation (for detail see Data analysis). In total, we analysed the STAR metric for Ethiopia based on 115 (including two more species we added to the original proposed list) species: 12 amphibians, 51 birds and 52 mammals. Forty-three (36.5% of the total) of these species are endemic (Table 1).
Thirdly, we checked for the relevance for Ethiopia of threat types used for the global STAR analysis. We found that the threat ‘4.3 Shipping lanes’ was not relevant to Ethiopia and thus excluded it from the analysis, since Ethiopia is a landlocked country and has not shipping transportation.
Formal calibration guidelines are currently not available, but are being developed by IUCN and will be published probably in the future (IBAT 2021a). The three steps discussed above therefore reflect a calibration strategy to derive national scale of the Estimated STAR scores from IBAT for Ethiopia.
Table 1
Number of all species (i.e. non-endemic and endemic species) and endemic species of the three taxonomic groups with diffrent IUCN Red List categories (CR = Critically Endangered; EN = Endangered; VU = Vulnerable; NT = Near Threatened) included in the STAR calculation for Ethiopia
| All species | | Endemic species | |
| NT | VU | EN | CR | Total | | NT | VU | EN | CR | Total |
Amphibians | 1 | 3 | 5 | 3 | 12 | | 1 | 3 | 5 | 3 | 12 |
Birds | 21 | 14 | 9 | 7 | 51 | | 2 | 4 | 3 | 1 | 10 |
Mammals | 14 | 20 | 14 | 4 | 52 | | 4 | 8 | 7 | 2 | 21 |
Total | 36 | 37 | 28 | 14 | 115 | | 7 | 15 | 14 | 5 | 43 |
Data for verification of the STAR metric
To examine the reliability of the STAR metric for appropriately ranking threats to biodiversity in Ethiopia, we undertook a questionnaire survey with national biodiversity experts (referred to as “Expert-based Threat Assessment”) to assess the impact of direct human threats on six target biological taxonomic groups (plants, mammals, birds, reptiles, amphibians and fish), following Gudka (2020). We chose these six target biological taxonomic group to obtain a general picture of biodiversity status in the country, as there are few experts specialized in the three taxonomic groups used in the STAR analysis. National biodiversity experts considered for the interviews were those with good experiences in practical biodiversity conservation and/or research from academic institutions (e.g., Addis Ababa University and Wondo Genet College of Forestry & Natural Resources) and non-academic organizations working in the biodiversity, agriculture, investment, fisheries sectors. Regardless of their speciality in specific taxonomic group, we asked each expert to assess threat impacts on each of the target biological taxonomic groups. We sent, via email, a questionnaire containing four questions to 40 expert assessors in August 2021. For each target of six taxonomic groups (plants, mammals, birds, reptiles, amphibians and fish), assessors were asked to: 1) assess the relevance of the 25 level 2 threats from the IUCN - CMP Unified Classification of Direct Threats version 3.2 (the first version was proposed by Salafsky et al. 2008) to the Ethiopian context, 2) remove irrelevant global-level threats by assigning a ‘not applicable to Ethiopia’ label, 3) record existing local threats if missing from the IUCN global threat list, and 4) to rank each threat to each biodiversity taxonomic groups. The questionnaires were accompanied by guidance instructions. Relevant threats were ranked on a scale of Low, Medium, High, and Very High, based on ‘contribution’ and ‘irreversibility’. Here ‘contribution’ is the contribution from a particular threat to population declines and/or habitat degradation of a target taxonomic group, while ‘irreversibility’ was the difficulty of reversing those declines or degradation (Salafsky et al. 2008). We received completely filled in/answered questionnaires from 17 (43%) experts.
Given the COVID-19 outbreak that hit Ethiopia during the study, we were unable to include other stakeholders, especially local communities, during this process as these groups, who have limited computing and internet capacity and no access to online communication platforms, would have required a face-to-face approach.
Data analysis
STAR score calculation for species
To recalculate STAR-R scores for the six species with outlier score values based on the global scale analysis, we derived historical and current extent of area of habitats based on information on their known distribution, suitable habitat types and land use and land cover changes in their known range. While Tachyoryctes macrocephalus is endemic to alpine grasslands and moorlands of the Bale Mountains, Arvicanthis blicki, Ptychadena nana and Crocidura Lucina are restricted to the Bale Mountains and Arsi mountains (Tessema et al. 2022a). Earlier reports (Hillman 1986) show that 2.1. Annual & perennial non-timber crops and 1.1. Housing & urban areas were rare in forest ecosystem of the Bale Mountains prior to 1960s. We obtained land use and land cover analysis for the year 1986 and 2015 (Nigussie 2016). From these, we observed that the increase in the area of farmland has been at the expense of grasslands, wetlands and shrub lands. Thus, the summed areas of cultivation land, settlement, grassland, shrub land and wetland in the year 1986 has been used as a proxy for the historical habitat of the species, while the area of grassland, wetland and shrub land for the year 2015 is assumed to represent current area of habitat. We followed similar approach for Arsi Mountains based on land cover and land use analysis by Hailu (2014). For the two other species − Leptopelis yaldeni and Crithagra ankoberensis –, which are restricted to the northern western highlands of the country, we were unable to reliably estimate their ranges. Thus, based on several literature reviews and our experience, we assumed that about 75% of the historical habitats have been lost and then used this value to recalculate their STAR values. Finally, we also estimated, as above, historical and current area of habitat for the two additional species (N. plumbeus and L. pseudosikapusi) we included in the national analysis.
Detailed methodological discussions on how the STAR scores were calculated for each of the above eight species are given in Online Resource 1. Below, we presented simple version of the formula modified from Mair et al. (2021) for one species s, L. pseudosikapusi, as an example.
The STAR-T for species s of threat t at site i is calculated as:
where Ps,i is the current Area of Habitat (AOH) of species s within location i, expressed as a percentage of the global species’ current AOH; Ws is the IUCN Red List category weight of species s (varies from 1 for Near Threatened, 2 for Vulnerable, 3 for Endangered and 4 for Critically Endangered); Cs,t is the relative contribution of threat t to the extinction risk of species s calculated as the percentage global population decline from that threat.
The STAR restoration score (R) for the potential contribution of habitat restoration (and threat abatement therein) for species s at location i for threat t is calculated as:
where: Hs,i is the area of original habitat (AOH) of species s that has been lost and is potentially restorable at location i, expressed as a percentage of current area of habitat; Mi is a multiplier (= 0.29) to discount restoration scores.
Given the above formula, we derived the STAR scores for the species using compiled necessary information from literature review and experts consultations, as follows.
L. pseudosikapusi is an Ethiopian endemic rodent species known to occur only in Sheko forest of south-western Ethiopia. It is listed as Endangered in the IUCN Red List (Dando et al. 2018). Thus, its weighting ratio (in the STAR-T and STAR-R scores formula) is 300. The species’ historical AOH is unknown, but its estimated current area of occurrence (EOO) is 2,185 km2. Here, assuming that the area of Sheka Key Biodiersity Area (3,723.3 km2; see Dando et al., 2018; IBAT 2021b) represents the species’ historical AOH and the EOO as its current AOH, we estimated extent of lost (potentially restorable) habitat of the species 1,538.3 km2 [((3723.3–2185)/2185))*100], which accounts for about 70% of its current AOH. Based on this, the STAR-R score for the species is 61.25 (0.704*300*0.29).
For L. pseudosikapusi, two IUCN CMP level 3 threats are listed, one for 2. Agriculture & aquaculture -> 2.1. Annual & perennial non-timber crops -> 2.1.2. Small-holder farming, and one for 5. Biological resource use -> 5.3. Logging & wood harvesting -> 5.3.3. Unintentional effects: (subsistence/small scale) [harvest]. These threats are on-going, but the Scope, Severity and Impact Score of both threats are Unknown. Based on literature review (e.g., Dando et al. 2018) and experts consultations, for 2.1. Annual & perennial non-timber crops threat, we assigned Timing to Ongoing (score: 3), Scope to be Majority (threat affecting majority of the population; score: 2), Severity (Rapid; score: 2) and Impact Level (in terms of contribution to population decline) of Medium (Score = 7). Similarly, for 5.3. Logging & wood harvesting, we assigned Timing as Ongoing (score: 3), Scope (Low, score: 1), Severity (Slow, score: 1), Impact Score of 5 (Low Impact). Threat score was done following the guideline provided in “Threat Impact Scoring System (based on additive scores (Salzer 2007). Accordingly, the STAR-T score for 2. Agriculture & aquaculture was estimated 175 (7/12*300) and for 5. Biological resource use threat 125 (5/12*300).
STAR score calculation for Ethiopia
Once we obtained STAR scores for the two species added to the original list (L. pseudosikapusi and N. plumbeus) and revised scores for the other six species, we then calculated the total national STAR-T component as the sum of STAR-T values across all 115 species, which is assumed to represent the national threat abatement effort needed for all species to become Least Concern. The same procedure was used to calculate STAR-R component score. We also calculated both STAR-T and STAR-R scores by each individual threat (IUCN level 2 threat types) across the species impacted from a particular threat.
STAR metric verification: Analysis of expert-based threat assessment data
We followed a three-step procedure to combine the data and assess the severity of impact of each threat to each taxonomic group. First, we re-coded each assessor’s rank score given for each threat to each taxon by assigning numerical score values as: Low = 1, Medium = 2, High = 3, Very High = 4. Second, we calculated the weighted average impact rank score of each threat to each target taxon. Average score for each threat to each taxon was calculated using weighted severity score algorithm. For example, out of the total number of 14 assessors who perceived that Housing and Urban areas impacts amphibians, 4 assessors ranked Low, 3 Medium, 5 High and 2 Very High. The weighted average score for the impact of Housing and Urban areas on amphibians was then equal to 1.64 [(4*1 + 3*2 + 5*3 + 2*4)/14]. Assuming that average values falling between 1.5 and 2.5 to be Medium, this average rank score (= 1.64) suggests that the impact of Housing and Urban areas on amphibians to be Medium.
To assess the reliability of STAR metric for identifying and ranking top threats to threatened species in Ethiopia, we examined the relationships between threat ranks obtained based on STAR-T score values of each threat category and threat ranks obtained from expert-based assessment data as sums of rank scores of each threat type across the six taxonomic groups. Then, to examine how closer (consistent) were the STAR-T scores with those obtained via the expert-based rank scores of each threat, we conducted a rank-based correlation analysis between the STAR-T scores of each threat and the threat rank score values obtained from the expert-based data.