Gender gap in perspectives of the impacts of degradation and restoration on ecosystem services in Ethiopia

The importance of land restoration has garnered increasing attention on the global stage through large‐scale initiatives such as the Bonn Challenge. However, policies and strategies are often gender‐blind and designed in compliance with entrenched social hierarchies, exacerbating pre‐existing social inequalities that affect restoration initiatives. The challenge of developing gender‐responsive policies and initiatives is accentuated by a persistent lack of sex‐disaggregated data concerning men's and women's differing perceptions and experiences. This study aims to help fill this gap by capturing the differences in men and women's perceptions of ecosystem services before and after restoration interventions in Ethiopia. Towards that end, in October 2021, we collected data from fifty‐nine (59) paired husband‐wife households and six gender‐segregated focus group discussions in two regions of Ethiopia: Amhara and Southern Nations, and Nationalities and Peoples (SNNP). Kendall's concordance analysis established a strong degree of disagreement between men's and women's ratings of restoration outcomes for most indicators. Men attributed degradation to landscape conditions and natural forces, while women considered the lack of appropriate restoration strategies as a precursor for accelerated degradation. The study also reveals that men tended to benefit more than women from enhanced ecosystem services post‐restoration, with increased labour and land management burdens often falling on the shoulders of women. Based on these findings, we argue that including men's and women's perspectives from the earliest planning phases of restoration initiatives is essential to ensure greater equity in benefit‐sharing, mitigate trade‐offs for women, and build more nuanced, just and successful approaches to restoration.


| INTRODUCTION
Despite considerable success in achieving restoration goals, women's stark under-representation in goal setting and benefit-sharing remains an impediment (Bayu, 2020;Kumasi & Asenso-Okyere, 2011). The links between gender inequalities, degraded ecosystems, and land and water management interventionsas well as between gender equality and enhanced ecosystem servicesare increasingly recognised (Broeckhoven & Cliquet, 2015;Collantes et al., 2018;Elias et al., 2021). A systematic review by Yang et al. (2018) established that worldwide rural women and men who are affected by and practice restoration have uneven access to resources and differing perceptions and knowledge of ecosystem services. Research gaps exist as to how women's and men's differential perceptions can be determined.
A recent study of gender dynamics in eastern Kenya found that employing an intrahousehold approach to land restoration increased both the uptake of restoration practices and the success and equity of restoration efforts (Crossland et al., 2021). Another study found that perceptions of degradation and priority areas for restoration varied with gender, but did not show how men and women value ecosystem services (Crossland et al., 2018), which underpin restoration motives (Stern et al., 1999). The question: 'how do men and women perceive benefits from their engagement?' remains unanswered.
Studies have sought to draw gendered comparisons using survey data collected from different household heads to explore differences in perceptions of decision-making power that shape land restoration in Kenya (Crossland et al., 2021), gender dynamics within agroforestry home gardens in Ethiopia (Gebrehiwot et al., 2018) and even on health issues such as family planning preferences between men and women in Ethiopia (Short & Kiros, 2002). In the absence of a reference criteria, these attempts to use norm-reference by comparing gender groups sampled from different households fell short of comparability due to differences in lived experiences among the households (Burkett, 2018).
We acknowledge that in addition to gender equality being a goal in its own right, it is essential for achieving sustainable and fair restoration outcomes. Yet, potential synergies between restoration initiatives and gender equality outcomes are rarely addressed in restoration research and implementation (Basnett et al., 2017). Moreover, interventions tend to focus on physical and biological measures while neglecting social and institutional dimensions (including gender inclusivity), which have a strong bearing on management (Crossland et al., 2018(Crossland et al., , 2021Gebreselassie et al., 2016;Stern et al., 1999). Notwithstanding, it has been found that while international and national legal documents recognise women's contribution and their human and civil rights, in practice, rural women are still disadvantaged, as customary institutions restrict women's access to land, markets and trading, and decision-making processes at the household and community levels (Gebrehiwot et al., 2018).
The paucity of systematic, consistent global and national data on women's land and resource rights and tenure security has been acknowledged (Collantes et al., 2018). In parallel, several studies have revealed that enhanced tenure security correlates positively with women's likelihood of engaging in restoration (Basnett et al., 2017).
However, such shifts in tenure could weaken men's tenure rights.
Hence, a better understanding of how gendered tenure arrangements influence motivations for restoration and the distribution of benefits from restored land would enhance dialogue and foster gender equity.
We considered the value-belief-norm (VBN) theory as a robust theory for landscape restoration projects and programs. The VBN theory builds on: (i) moral norms, (ii) personal values, and (iii) a new ecological paradigm, which are critical pillars to achieve landscape restoration outcomes (Stern et al., 1999). The theory assumes that moral norms and personal values are critical to restore degraded social-ecological systems (Canlas et al., 2022).
The study objective is to examine intrahousehold differences in men's and women's perceptions and perceived values regarding ecosystem services before and after restoration interventions in northern and southern rural regions of Ethiopia, thereby bridging the gender data gap. Understanding differences among spouses is critical to ensuring that efforts to enhance access to and use of ecosystem services can be equally beneficial to all genders. We investigated the different needs, priorities, and interests of women and men, and the challenges they face, in relation to landscape restoration. It is important to note that social, political, and cultural factors influence perceptions and perceived values of women and men towards land degradation and restoration on ecosystem services. However, empirical evidence to evaluate such factors was outside the scope of the two case studies.

| Land degradation and restoration in Ethiopia
Over the past century, Ethiopia has experienced severe land degradation coupled with booming population growth, leading to national and regional food insecurity. These significant challenges were exacerbated by a succession of droughts in the 1970s and 1980s that severely impacted social and ecological wellbeing across Ethiopia (Gashaw et al., 2014;Gebreselassie et al., 2016;Taddese, 2001 (Gebreselassie et al., 2016;Le et al., 2016). Despite the widely reported notion that population pressure was and is the main driver of land degradation in Ethiopia (Gashaw et al., 2014;Gebreselassie et al., 2016;Taddese, 2001), historical evidence from the northern region shows that in some land parcels, vegetation, soil and water conservation have improved even as the population has increased by a factor of 10 (Nyssen et al., 2009). Restoration efforts in the last 20 years have enhanced ecosystem services, including soil retention, crop yield, biomass production, and groundwater recharge (Nyssen et al., 2009).
The recovery of severely degraded lands in northern Ethiopia challenges the hypothesis of irreversible degradation in semi-arid regions (Dregne, 1991) and showcases the potential for rural communities to improve heavily degraded lands through restoration and revegetation efforts grounded in strong social cohesion and collective action (Nyssen et al., 2009). The success of this approach has spurred the global call for collective restoration action that culminated in two major global restoration initiatives: the Bonn Challenge and the UN Decade on Ecosystem Restoration (UN, 2019;UNEP and FAO, 2020).
As part of the Bonn Challenge, which aims to restore 150 million hectares of the world's degraded and deforested lands by 2020 and 350 million hectares by 2030, Ethiopia has boldly committed to restoring 15 million hectares of degraded and deforested areas to productivity by 2025, representing 10% of the total area to be restored (IUCN, 2020;Pistorius et al., 2017). In Ethiopia, the main restoration activities include collective action to carry out reforestation, constructing physical structures such as terracing and soil bunds to control soil erosion on steep slopes as well as allowing natural regeneration through enclosing areas degraded by grazing, cultivation, and indiscriminate tree and grass extraction (Descheemaeker et al., 2006;Tamene, 2005).

| Study sites
The study was conducted in two landscape restoration sites: Gudo-  (Zeleke & Vidal, 2020). In the Gudoberet site, the local Agriculture Bureaus, along with the communities, play an important role in the implementation and management of restoration practices whereas non-governmental organisation Inter Aide France and local communities lead restoration activities in the Lemisuticho sites (Tadesse et al., 2018). At both sites, the Alliance of Bioversity International and the International Center for Tropical Agriculture (CIAT) provides research support ranging from planning and implementation to evidence generation.
Notably, women's land tenure differs between the two study sites; while both Amhara and SNNP were included in Ethiopia's First Level Land Certification process (in 2003 and 2004, respectively), registration of women as individual or joint landholders is much higher in Amhara than in SNNP, where polygamy is more common and registration is for individuals rather than households (Deininger et al., 2011).
Cultural norms concerning women's access to land also differ across these regions, at times in conflict with national laws. For example, in Amhara, women are culturally prohibited from ploughing land, and in SNNP, women culturally have no right to own land and must access it through their husbands (Monterroso et al., 2021). The Gudoberet site has a cool sub-humid moist climate with an extended rainfall period receiving 1449 mm (±16% variability according to weather station data collected on site) that supports high agricultural potential with 66%, 25% and 6% of land used for cultivation, grazing and woodlot, respectively (Yaekob et al., 2020). After a longduration crop, short maturing highland crops are planted. However, low nutrient content (especially available phosphorus), gully erosion in some places, and rainfall variability greatly affect productivity. Integrated soil and water conservation measures (enclosures [i.e., areas where some or all human activity is prohibited or limited to favour their restoration], terraces, soil bunds, stone bund trenches, percolation pits, check-dams, and gully rehabilitation) have been implemented by 1164 households (580 male-headed and 584 female-headed) covering over 1700 hectares in the Gudoberet restoration area to reduce runoff and enhance infiltration. An impressive effort was also made to rehabilitate gully erosion through gabion terraces and re-shaping in the Adisghe Kebele.

| Data collection
In the context of this study, a household survey was conducted in October 2021 with paired samples of spouses to collect robust information on intrahousehold gender perceptions related to land degradation and restoration. In Ethiopia, although the law provides for women to be independent, the patriarchal customary system confers husbands' decision-making power over household matters. As such, in addition to obtaining women's free, prior and informed consent as participants, we obtained men's consent to interview their wives, who were then interviewed by a female researcher. Sex-disaggregated data were collected on drivers of land degradation and the perceived benefits and challenges from restoration faced by women and men. The survey tool was structured according to the gender in restoration design and evaluation framework by Basnett et al. (2017) to capture degradation-restoration themes.
We randomly selected 59 married households (30 from Basona-Worena and 29 from Doyogena) who participated in landscape restoration initiatives in the two sites. In each household, we interviewed both husbands and wives. Since this was an explanatory case study, an effective sample of around 30 households was targeted for the interviews, which has been found to be sufficient for the ratings to reach the saturation point (Kumar, 1989), when the additional data no longer sparks new insights (Creswell & Cresswell, 2018). Before data collection, the household survey questions were pre-tested by enumerators. We used the Open Data Kit (ODK) survey tool to collect data (Hartung et al., 2010). The survey was conducted in the Amharic, Kenbata and Oromifa languages.
In the survey, respondents were asked to identify the primary causes of (i) land degradation, (ii) its impact on 15 ecosystem services and (iii) the impact of land restoration efforts on these same 15 ecosystem services. To do this, participants rated the perceived degree to which land degradation impacted ecosystem services on a 4-point Likert scale: (1) greatly, (2) moderately, (3) least, (4) not at all, or (5) N/A. Thereafter, participants were asked to estimate the perceived impact that restoration efforts had on each ecosystem service on an inverse 5-point Likert scale as: (1) increased a lot, (2) increased a little, (3) did not increase or decrease ('no change'), (4) reduced a little, or (5) reduced a lot.
Six focus group discussions (FGDs) were conducted to complement household surveys. Open-ended questions enabled a more indepth understanding of degradation and restoration impacts on women and men. FGD participants were separated by gender to capture dynamics in perceptions of degradation, participation in community-level restoration interventions and parity in benefitsharing from ecosystem services. Within each site, FGD participants were selected from households who live at the upper, middle, and lower part of the study landscapes and at different parts of the project sites, with the aim of capturing their perceptions along the local landscape gradient.

| Framework, analysis and limitations
Perceptions are defined as the ability to sense (see, hear, feel, smell and taste), memorise, understand, consider and interpret external stimuli (Cambridge Dictionary, 2016). A meta-analysis by Yang et al. (2018) established that gender differentiated perceptions of ecosystem services could arise due to the intersection of multiple socio-economic factors that differentiate men and women, including educational backgrounds, culture, socioeconomic status, age, religious beliefs and access to information. These perceived values are embodied in the value-belief-norm theory of environmentalism by Stern et al. (1999), which postulates that environmental behaviours including awareness of consequences and ascription of responsibility follow pro-environmental norms that underpin gender differentiated restoration efforts (Bratt et al., 2015;Oreg & Katz-Gerro, 2006). In a typical family farming setting, farmers live and work on their fields. Hence, perceptions are founded on lived experiences.
A comparative study was carried out to examine women's and men's perceptions of root causes of degradation and of perceived pre-and post-restoration ecosystem services among paired husbands and wives. Since there were no prior and validated evaluations against which the respondents would base their perceived ratings on, the ratings by men and women have no reference criterion and are assessed using norm-referenced rating (Burkett, 2018). Results are presented using radar diagrams and bar graphs whose interpretations are supported by participant narratives from FDGs. Further correlational and concordance analyses were conducted. Spearman's rank correlation coefficient was used to measure the ordinal association between the pairs of sets of ratings of the impact of restoration activities. The rank correlation is used to measure the (dis)agreement between men and women living in same households.
Kendall's concordance analysis, W t , calculated using the irr package in R (Gamer et al., 2019), is used to measure the degree of agreement or disagreement among m sets of men and women for n ratings of restoration outcomes. This index measures the ratio of observed variance of the sum of the ratings to the maximum possible variance of a sum of ratings and ranges from zero to one. A lower score shows a higher degree of disagreement among men's and women's ratings, signifying that the men and/or women raters have dissimilar experiences with respect to restoration outcomes. In contrast, a higher score shows that the groups experiences are similar. Like the correlation, W t , is a measure of opinions and value judgements that sheds light on gender dynamics in restoration activities. Following Edwards (1964) and Martey et al. (2014), W t is determined as: where T = sum of ratings for each degradation cause or restoration outcome, m = number of ratings (husbands and/or wives), and n = number of restoration outcomes.
Given the data collection methods used, this study is based on perceptions of participants in the sample. Working with perceptions also presents challenges, however, as perceptions are underpinned by socio-political factors and unequal power dynamics which are often unspoken, unrecognised and even subconscious. In relation to this study, reported perceptions may be part of pre-existing debate and competition over contested gendered access rights to land and resources (Fairhead, 1992) that are embodied in, among other sociopolitical issues, inexplicable emotional and spatial connections (Mandondo, 1997). Mapping and analysing the full scope of the underlying power struggles that shape and colour the perceptions presented in this paper would require targeted methodologies outside the scope of this study.

| Perceived factors and drivers of land degradation
Perceptions of factors and drivers of land degradation varied according to landscape scale, gender, and site. At the level of a managed plot, in both SNNP and Amhara, poor soil fertility was regarded by both men and women as one of the major root causes of land degradation, whilst at the landscape level, heavy rainfall was perceived to be the root cause (Figure 2). In Amhara, men perceived the cultivation of sloping land as the main factor leading to land degradation (soil erosion) while women believed that the absence of crop rotation causes land degradation at the plot level. Both men and women also perceived weeds to be causational factors of land degradation at the plot level. In Amhara, at the landscape level, more men indicated biophysical factors (e.g., erosion on cultivated slopes) while women considered land management (e.g., lack of crop rotation/continuous cultivation) as the key drivers of degradation. In SNNP, men attributed shallow soils to being prone to land degradation at plot level. In Amhara, women indicated in focus group discussions (FGDs) that a lack of strategies to control erosion led to degradation, noting that erosion is less severe in areas with conservation measures. They expressed that some areas are prone to erosion due to overgrazing, lack of cover crops and removal of crop residues. Amhara receives heavy rainfall that washes down the fertile topsoil thereby depleting the nutrients.

| Gendered perceptions of impacts of land degradation and restoration on ecosystem services
Of the 15 ecosystem services (ES) that were perceived to be impacted by degradation and whose restoration was prioritised, the concordance analysis shows disagreements (W t < 0.5) in perceptions among the 59 men and 59 women ( Table 1). The greatest disagreement was on economic ES and tenure. Notably, women largely have convergent views (W t = 0.63) on regulatory (e.g., wildlife and pollinator habitat, reduction in weeds) and cultural (e.g., landscape beauty, religious and cultural sites) ES. Men, in contrast, have convergent views mostly on provisioning ES (e.g., crop production, irrigation water, household/ livestock water, fodder or grazing grass, fuelwood or poles or timber).
In Ethiopia's restoration areas, there was higher agreement on men's rating of provisioning services than their rating of regulatory and socio-economic ES.
Intra-gender comparisons for each ES revealed divergent perceptions between men and women on access to land and availability of water for households and livestock, indicating differences in lived experiences, aspirations and opinions of land degradation, restoration, and their perceptions of these ES (Table 2). However, there is convergence in men's and women's rating of availability of water for irrigation and fodder or grazing grass, condition of religious and cultural sites and availability of natural resource-based jobs and business opportunities, indicating shared experiences among these ES. On average, restoration activities tend to lower women's access to communal land, especially in Amhara (Table 2). Compared to their husbands, women of SNNP more strongly perceived that restoration activities increased water availability for household and livestock use and income from farming, but women gave a relatively lower rating of costs and expenditures on the farm following restoration.
T A B L E 1 Kendall's coefficient of concordance (Wt) for gendered agreements in restoration impacts. Note: NB: Provisioning (crop production, irrigation water, household/livestock water, fodder or grazing grass, fuelwood or poles or timber); regulatory (landscape beauty, wildlife and pollinator habitat, reduction in weeds); cultural (religious and cultural sites); economic (income, costs, time and energy from or on farm and availability of jobs and businesses); tenure (access to communal land).

| Provisioning ecosystem services: Land productivity
The study found that men and women have different perceptions of the contribution of restoration to land productivity ( Figure 4).
Although both groups gave positive ratings, men gave a much higher rating than women for increased production. Ten percent of women in Basona-Worena indicated that restoration reduced productivity. Men and women were also found to have different perceptions of what is driving productivity loss, a symptom of land degradation, in their areas (Figure 3). Around 50% of both women and men considered the ongoing mining and/or loss of nutrients and topsoil loss due to erosion as top-ranked drivers of low productivity (Figure 4). In

Men in Amhara indicated in
Doyogena, more women than men considered that soil fertility decline greatly contributes to a decline in productivity, whereas this was the case for more men than women in Basona-Worena. In both sites, more men than women considered that erosion greatly impacts land productivity.

| Provisioning ecosystem services: Natural resource availability
Both men and women perceived land degradation and restoration to have a profound impact on provisioning ES. In Amhara, women considered that they were greatly impacted by degradation in terms of water scarcity for household and livestock use, whereas men focused on water for irrigation. In SNNP, both men and women reported greater scarcity of water for their livestock due to degradation ( Figure 3). Likewise, both men and women reported positive effects from restoration interventions on water for livestock, which significantly increased in both sites. Concerning irrigation water, one-third of men indicated that restoration had a moderate to a significant positive impact on water for irrigation while women perceived it as insignificant.
In FGDs, men in Amhara reported a substantial shortage of firewood due to degradation. Post-restoration, both women and men perceived a significant improvement in fuelwood and pole availability.
In SNNP, most respondents indicated that the change, though positive, was minimal. In both sites, women and men indicated that restoration significantly improved the availability of fodder and grazing grass. During FGDs, men recognised that women are very worried when there is a livestock feed shortage. During the female FDG discussions in SNNP, women indicated that as men move to search for employment and income, women remain at home to take care of livestock. In SNNP, men also expressed that communities are highly satisfied with the availability of grass for the cut and carry system. They noted that the practice of planting improved grass varieties on soil bunds helps farmers to integrate crop and livestock farming.
T A B L E 2 Mean ratings and correlations of perceptions of restoration impacts among paired husband and wife ES respondents. Reversed rating. *Sig. < 0.1; mean and correlations estimated from ratings: 2 = more increased, 1 = increased but not much 0 = no change, À1 = reduced but not much, À2 = reduced a lot.
Some participants reported converting their land from food production to other uses that provide economic returns in response to degradation. In Amhara, it was noted that marginal lands are converted to tree plantations which have higher economic value than if the land is used for food crop production, given the soil's low fertility.
Likewise, in pursuit of higher income from smaller plots, farmers in the lower landscape positions tend to grow high-value plants including Khat (Catha edulis (Vahl) Forssk.) stimulant shrub. Others leave their fields fallow and use enclosures for controlled grazing for 3 to 5 years, to be recultivated once fertility is regained.
F I G U R E 3 Gendered perceptions of impacts of degradation on ecosystem services. NB: erosion, fertility loss, grazing and deforestation impact land productivity. Degradation impacts water availability for home (household and livestock) and irrigation, on landscape scenic beauty, habitat for wildlife, pollinators and beneficial plants. Degradation impacts weed incidence and pest and disease infestation. Pre-restoration costs and labour are used to source firewood, poles and timber and forage or grazing grass. [Colour figure can be viewed at wileyonlinelibrary.com] F I G U R E 4 Gendered perceptions of impacts of restoration on ecosystem services. NB: restoration impacts on land productivity; availability of water for home (household and livestock) and irrigation, of fodder or grazing grass and firewood, poles and timber; landscape scenic beauty and habit quality for wildlife and pollinators; weed reduction; job and business opportunities; land tenure security; and on farm income, input costs and labour. [Colour figure can be viewed at wileyonlinelibrary.com]

| Regulatory ecosystem services: Habitat quality
Our study reveals gender differences in perceptions of regulatory ecosystem service indicators in a landscape, that is, habitat quality for wildlife; pollinators; beneficial plants; weeds; and pests and diseases (Figures 3 and 4). Men in Amhara perceived that pre-restoration, wildlife habitat was greatly impacted by degradation, while women rated the impact as moderate. In SNNP, both men and women gave a moderate rating of degradation's impact on wildlife habitat.
In SNNP, degradation was overall rated to have a moderate impact on pollinators and beneficial plants, but a large impact on the proliferation of weeds and pests and diseases. In Amhara, these ratings were inversed, with a higher impact rated for pollinators and beneficial plants. Most women in Amhara perceived a minimal impact on beneficial plants, but men observed that increased use of herbicides harmed bees and other pollinators. According to respondents, habitat improvements leading to an increase in wildlife and pollinators have been substantial in Amhara but moderate in SNNP. In SNNP, respondents indicated that there has been a significant reduction in weeds, but in Amhara, most women and a third of men reported an increased occurrence of weeds post-restoration. Men and women of Amhara also reported divergent experiences concerning pests and diseases.
More women perceived least to moderate pest and disease occurrence in degraded landscapes, while men rated the problem as least to none (Figure 3). In SNNP, both genders reported a higher occurrence of pests and diseases post-restoration.

| Cultural and economic ecosystem services
At both sites, men and women indicated that landscape beauty was greatly impacted by degradation, but more women than men perceived that landscape restoration enhanced the scenic beauty ( Figure 4).
We observed site differences in tenure system changes across sites. Following the land ownership reforms, women (co)own and engage more in restoration and men felt that their tenure rights over restored family and communal land were reduced. One-third of men in Amhara reported that restoration activities reduced their tenure rights while women (20%) reported greater access to land. Some men (15%) in Amhara also reported increased tenure over family and communal lands. In SNNP, no change to tenure systems post-restoration was reported.
In both sites, degradation impacted both men and women in terms of labour and costs of sourcing fuelwood (Figures 3 and 4).
Without restoration, women indicated being acutely constrained by the lack of feed and forage availability and time deficits. To support livelihoods through livestock, women reported spending much of their time and income acquiring forage. The study revealed significant gender differences in time and labour demand with mixed impacts associated with restoration. In Amhara, restoration was reported to have a disproportionate impact on women as they experienced a significantly increased demand for their time and labour on the farm while most men experienced a decrease in both time and labour requirements. In SNNP, women and men indicated that restoration slightly reduced the time and labour demands on the farm.
In terms of nature-based business and job opportunities, in Amhara, more women than men indicated an increase post-restoration, while in SNNP respondents reported a slight change. With increased opportunities for women such as selling firewood to a perfume factory in Amhara, most indicated that their incomes and associated purchase and transportation costs were slightly increased, while another 20 percent reported a reduction in both income and costs. In contrast, men in Amhara reported a significant increase in income as well as in associated costs. Similarly, in SNNP men reported a significant increase in income from restored areas and a small reduction in restoration costs, while women reported a slight increase in income from restored lands. The women of SNNP reported selling fuelwood, construction poles, and forage grass for feeding animals. Men also indicated that selling forage grass has become the main source of household income but also supports livestock in the restoration areas.

| DISCUSSION
The study has revealed gendered perceptions of changes in ecosystem services that shape men's and women's restoration strategies for addressing degradation. It is not surprising that both groups considered erosion as the most pressing source of degradation. Erosion by water is the most common form of land degradation in Ethiopia, with reports indicating annual soil loss of between 42 and 300 Mg/ha/year (Gebreselassie et al., 2016). Although deforestation and overgrazing still cause long-term productivity loss in Ethiopia (Gebreselassie et al., 2016;Kumasi & Asenso-Okyere, 2011), both men and women in Amhara and SNNP rated these as moderate or insignificant causes of productivity loss. In a 2018 study with a similar focus in northern Ethiopia, Crossland et al. (2018) found that the cultivation of sloping land was considered by male farmers in Amhara to be the major cause of degradation, a finding which is in agreement with the aforementioned perceptions of men in Amhara. Men's perception of slope of land as a primary factor leading to degradation points to the greater underlying degradation drivers of population growth and a shortage of optimal agricultural land, which have pushed farmers into suboptimal sloping plots. Crossland et al. (2018) also noted gender differences in the discussion of degradation factors and found that women mapped more and larger areas as degraded and mapped larger parcels for restoration as compared to men.
Our study expands on such gender differences by exploring how women and men perceive the root causes of degradation and the value of corresponding restoration strategies. We found that men largely attribute degradation to external factors beyond farmers' control such as high rainfall and slopes as leading causes of erosion, whilst women regard anthropogenic causes of such degradation, such as a lack of measures to prevent erosion. These narratives reinforce the findings that men tend to focus on physical factors as root causes of degradation, while women focus on human-induced degradation pathways.
Our findings of higher perceived values of provisioning ES in restoration activities by women support established notions that men value extractable products, while women value intangible regulatory services (Fortnam et al., 2019;Kalaba et al., 2013). Women tend to focus on farmland and attribute a moderate impact of land degradation on habitat quality, while men attribute enhancement of biodiversity and pollinators to restoration. A biodiversity study by Terefe et al. (2020) in central Ethiopia reported that farmers perceived an increased abundance of pollinators and re-appearance of new species in restoration sites compared to neighbouring unrestored sites.
Our study echoed the aforementioned findings but to differing degrees, as we found that perception variations are mediated by gender roles. Gender differences are a result of spatial visioning as farming and landscape restoration activities are implemented at different spatial locations. With women spending more time on homesteads, they pay attention to crop production while men focus on communal grazing land and high-value irrigation parcels (Crossland et al., 2018). ship (Teklu, 2005) or separation (Holden & Tefera, 2008).
However, enhancing the value of land through restoration can affect the functionality of the (land) resource. It is not yet known whether enhancing the value of land through restoration will trigger land dispossession or translate to equity in benefit sharing (Girma & Giovarelli, 2013). The restoration initiatives would then impact livelihoods especially for the asset-poor groups, including women in patrilocal systems, who are dependent on natural resources. Moreover, as Lavers (2017)  Women's insecure land tenure continues to be one of the biggest challenges for gender equality in restoration (Girma & Giovarelli, 2013).
The uneven impacts of land tenure interventions illustrate that enabling the equal participation of women in land management is as complex as it is necessary. Addressing such deep-seated socioeconomic and ideological issues as gender roles, rights, and representation requires culturally and context-specific understandings and approaches. By adding to the repertoire of sex-disaggregated empirical data concerning gender and restoration at the household level, this study has contributed to such a nuanced understanding (Zgang et al., 2019).
In Ethiopia, fuelwood collection is primarily a woman's task, and due to the scarcity of resources near homesteads, women travel for up to 2 hours per day (12 km/day) to fetch fuelwood (CSA, 2014) with large economic consequences (Gebru & Bezu, 2014;Mosa et al., 2020). Studies in Kenya and Tanzania have found that fuelwood collection is affected by degradation, which increases women's burden to search for cooking fuel (Njenga et al., 2021) and affects their time to pursue other economic activities and childcare (Levison et al., 2018). Another recent study in Ethiopia found that although enclosures supply close to 10% of households' fuelwood demand and are proven effective at controlling grazing and restoring degraded grasslands and woodlands, women still travel 10 km/day to collect fuelwood from degraded forests (Mekuria et al., 2019). A restoration workload study in Kenya found that although structural restoration measures increased the cost and time for land preparation, the labour investment paid off through increased productivity and income from sales of vegetative products (Crossland et al., 2021). An increase in household income has been recorded for restoration programs in sloping landscapes of China (Lin & Yao, 2014). However, as indicated by our findings, the overall economic benefit at the household level can be on the backs of women, whose labour and already significant time burdens may be disproportionately increased.

| CONCLUSION
Pre-and post-restoration, women and men have differing perspectives on ES reflecting differential lived experiences and visions with significant impact on restoration. Through this study, we found that women's more management-focused perspectives on drivers of degradation could better inform actionable restoration activities than men's more fatalistic perceptions that degradation is caused by factors beyond human control. Further differential perception revealed on restoration outcomes indicates the divergent appreciation and valuation of ecosystem services. There is a substantial gendered division of labour with men engaged in and attaching value to extractable products to support immediate family needs while women value intangible regulatory services that focus on sustainability.
This research can also be used as a jumping-off point for more nuanced and intersectional research which looks at how other axes of marginalisation and identity (e.g., ethnicity, religion, age, and socioeconomic status) further impact understandings of degradation and priorities for restoration, thus furthering the triple goals of ecological and economic restoration as well as social inclusion. While we have seen the importance of women's inclusion in restoration activities at all stages, we can also see that such involvement can have considerable trade-offs, with the risk that disadvantaged groups, including women, may disproportionately carry the burdens of restoration in the form of (often unpaid) labour.
Beyond the significant progress made on entitling land to women, women's relative absence in decision making indicates lack of an enabling policy environment for successful implementation and enforcement of gender-responsive restoration strategies. Moreover, with the discriminatory cultural norms and societal roles limits women's active engagement. The results from this study can guide improvements in institutional and policy support for social inclusion, including in decision making, implementation design and benefit sharing, to arrest degradation and ensure equity in ES post-restoration.