Figure 1. Map of the study area
The cool sub-humid moist climate (bimodal rainfall of 1449 mm± 16% variability) of Gudoberet supports higher agricultural potential with 66%, 25% and 6% of land used for cultivation, grazing and woodlot, respectively. Low nutrient content (especially available phosphorus), gully erosion in some places and rainfall variability greatly affect productivity. Integrated soil and water conservation measures (enclosures, terraces, soil bunds, stone bund trenches, percolation pits, check-dams, gully rehabilitation have been implemented by 1164 households (580 male and 584 female) covering over 1700 ha 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.
Receiving bimodal rainfall of 1000-1400mm/annum, the sub-humid highland (2280-2640 m.a.s.l) cool (12-20°C) climate of Doyogena supports diverse livelihood strategies including cereal-livestock integration and agroforestry. Despite the favourable climatic conditions, high population pressure (highest in rural Africa of up to 600 inhabitants/km2), soil erosion and nutrient depletion affect productivity. Restoration interventions being implemented include physical structures (soil bunds and fanya juu terraces) integrated with biological measures (desho grass). Agroforestry around homesteads is a common land use and restoration practice in the Doyogena and surrounding sites.

Data collection and analysis

The household survey was conducted with paired samples of couples. The pairing provided robust intrahousehold gender dynamics including the preferences, roles, benefits and challenges for women (wives) and men (husbands). We used stratified random sampling to pick 59 husbands and 59 wives. Before data collection, the household survey questions were pre-tested by enumerators. We used Open Data Kit (ODK) to conduct household surveys. The survey tool was framed based on gender design and evaluation framework by Basnett et al. (2017) capturing the degradation-restoration themes. We captured gender differentiated data on drivers of land degradation, and perceived benefits and challenges faced by women and men.
The focus group discussions (FGDs) were organised to complement the household interviews and the open-ended questions enabled gathering of more in-depth evaluations of the degradation and restoration impacts on women and men. The FGD members selected to represent men and women groups participating in community level restoration interventions. All the FGD members also represent the three-landscape positions (the upper, middle, and bottom), to capture households’ feedback on different restoration interventions across the landscape position.
From the household survey, a comparative analysis was done on gender-disaggregated ratings by paired husbands and wives representing men and women voices, respectively. The results are presented using radar diagrams and bar graphs and supported by participant narratives from FDG. Further correlational and concordance analyses were done. Spearmans’s rank correlation coefficient was used to measure ordinal association between the pairs of sets of ratings (see note on Table 2) of impact of restoration activities. Since there were no experimental evaluations against which the raters would rank the attributes, the ratings by men and women had no reference criterion. Hence, the perceptions were founded on their respective lived experiences and the rank correlation is the measure of agreement between men and women.
We run Kendall’s concordance, Wt , analysis using the irr package in R to measure the degree of agreement or disagreements among m sets of men and/or women for n ranks of restoration outcomes. It is an index that measures the ratio of observed variance of the sum of the ranks to the maximum possible variance of a sum of ranks and ranges from 0 to 1. The least rank score shows that men and/or women have divergent experiences of degradation impacts and restoration outcomes and hence use a different standard when rating. On the other hand, a higher score shows that men’s and/or women’s experiences are similar. Like the correlation,Wt is a measure of opinions and value judgements useful to establish the gender dynamics in restoration activities.Wt is estimated as (Edwards, 1964; Martey et al., 2014):
\begin{equation} W_{t}=\ \frac{12\left[\sum{T^{2}-\left(\sum T\right)^{2}/n}\right]}{nm^{2}(n^{2}-1)}\nonumber \\ \end{equation}
Where;
T = sum of ranks for each degradation cause or restoration outcome being ranked. m = number of rankings (husbands and/or wives) and n = number of restoration outcomes being ranked The following hypothesis was tested: H0: K= 0; there is no agreement among experiences of men and/or women regarding land restoration benefits. H1: K≠0 There is agreement among experiences of men and/or women regarding land restoration benefits.