Stakeholder perspectives on nature, people and sustainability at Mount Kilimanjaro

1. Effective approaches towards sustainability need to be informed by a diverse array of stakeholder perspectives. However, capturing these perspectives in a way that can be integrated with other forms of knowledge can represent a challenge. 2. Here we present the first application of the conceptual framework of the Intergovernmental Science- Policy Platform on Biodiversity and Ecosystem Services (IPBES) to a participatory assessment of local perspectives on nature, people and sustainability on Mount Kilimanjaro, Tanzania. This assessment was organized in the form of a participatory workshop with five different groups of stakeholders. 3. Following this framework, we assembled information on the state of and trends in species diversity, Nature's Contributions to People (NCP), and on the main drivers of changes in species and habitats. Additionally, we gathered perspectives on the needs and opportunities for the sustainable management and conservation of natural resources from the individual to the international level. The habitats climate human


| INTRODUC TI ON
Effective approaches to sustainability challenges require both a global perspective and an in-depth understanding of local socialecological systems (Martín-López et al., 2020), which are traditionally accessed through literature and field measurements. Both global and local understandings and perspectives also reside with local stakeholders and can be elicited through dialogues and social assessments using different forms of knowledge mobilization and sharing as well as various methods from the social sciences and humanities (Lakerveld et al., 2015;Orenstein & Groner, 2014;Tengö et al., 2017). By placing stakeholders at the centre of the research, social assessments complement the predominantly ecological and economic approaches typically adopted in ecosystem-related assessments and that might overlook relevant social aspects (Orenstein & Groner, 2014;Raymond et al., 2013;Scholte et al., 2015). They further enable to embrace the diversity of perspectives on nature and its management residing among different social groups and stakeholder segments (Brondízio et al., 2021), whose views are affected by different social, economic and cultural factors, including prior knowledge and involvement in local management and sustainability initiatives (e.g. Caballero-Serrano et al., 2017;Cuni-Sanchez et al., 2019;Lamarque et al., 2011;Lewan & Söderqvist, 2002).
Local perceptions are particularly relevant as they may capture knowledge that is not yet recorded and because they are often influential in guiding decision-making, justifying actions (Orenstein & Groner, 2014) and resolving conservation trade-offs (e.g. Cuni-Sanchez et al., 2019). Accordingly, the integration and weaving of knowledge from academic and non-academic origins through a process of co-production is key to addressing complex sustainability questions Tengö et al., 2017). The collection of contextualized information that accurately captures a diverse array of local stakeholder perspectives on social-ecological systems and their conservation and that can be weaved with quantitative desktop findings is methodologically challenging. Yet it offers powerful perspectives for comparisons across scales, geographies and contexts, and for the just accounting of local grassroot knowledge and needs. In this context, the conceptual framework of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES, Díaz et al., 2015) represents a unique opportunity to take up the challenge.
Here we report on a first attempt to apply the IPBES framework in a participatory assessment of nature, people and sustainability on Mount Kilimanjaro as a structured tool for capturing stakeholder perspectives. We further highlight how these perspectives reveal salient differences between stakeholder groups and enrich current narratives on the relationship between nature and people around Mount Kilimanjaro as well as on options for decision-making and action towards a more sustainable future.

| Mount Kilimanjaro social-ecological system
Mount Kilimanjaro is the highest free-standing mountain in the world (Newark, 1991) and hosts approximately 2,500 plant and a few hundreds bird species. The classification of its elevational zones is a matter of perspective (Hemp, 2006a;Hemp & Hemp, 2008;Liseki, 2015;Misana, 2012;Soini, 2005a).
Habitat classification based on agro-ecological systems and local perceptions delineates a highland zone including the coffee-banana belt and (Chagga) home garden area (1,200-1,800 m a.s.l.), a midland zone characterized by the maize-bean belt (900-1,200 m a.s.l.) and 6. The most frequently suggested measures to address the observed decline in species diversity and its drivers were related to land and water management and to education and awareness raising. Yet, the stakeholder groups differed in the measures they suggested. 7. The willingness of a diversity of knowledge holders to systematically engage in a structured discussion around all the elements of the IPBES framework provides support for its applicability in participatory workshops aimed at capturing nuanced and context-based perspectives on social-ecological systems from informed stakeholders. 8. The application of the IPBES framework enabled the comparability needed for developing narratives of stakeholder visions that can help identify new pathways towards sustainability and guide planning while retaining the context-based nuances that remain unresolved with non-participatory methods.
Read the free Plain Language Summary for this article on the Journal blog.

K E Y W O R D S
participatory workshops, IPBES, Tanzania, East African mountains, mountain biodiversity, Nature's Contributions to People, ecosystem services, social-ecological system a lowland zone (<900 m a.s.l.) characterized by extensive livestock grazing. Scientists in turn distinguish up to 12 major habitat types along the elevational gradient (colline savanna, submontane-lower montane rainforest, Ocotea forest, Podocarpus forest, Erica forest, alpine Helichrysum vegetation, Erica forest shifted to lower elevations due to disturbance, logged Ocotea forest, meadow, Chagga home gardens, coffee plantations and maize fields, Hemp, 2006b).
Mount Kilimanjaro is home to approximately 1.3 million people (United Republic of Tanzania, 2013), the majority of which lives in the Chagga home gardens (Fernandes et al., 1985). These home gardens are unique, sustainable and well-developed agroforestry systems that have been in use for centuries (Fernandes et al., 1985;Hemp, 2006a;Hemp, 2006b;Misana, 2012;Newark, 1991;Sébastien, 2010;Soini, 2005a). However, over the last decades, these systems have been faced with numerous challenges associated with human population growth and changes in land use (Hemp & Hemp, 2008;Misana, 2012;Sébastien, 2010;Soini, 2005a (Hemp, 2005;Sébastien, 2010) and an important reservoir of useful plants ( Molina-Venegas, Fischer, Mollel, et al., 2020;Mollel et al., 2017)) and its importance as a popular destination for economically important international tourism (Adili & Robert, 2016) calls for efforts to ensure its conservation and sustainable management.
The recent conservation history of Kilimanjaro started in 1904 with the protection of the 'Mount Kilimanjaro Forest' (Newark, 1991) and continued with its classification as a natural reserve in 1921 (Sébastien, 2010) and the establishment of a half-mile forest strip below the reserve in 1941. This forest strip was established as a social buffer to provide local people with wood and other forest products (Newark, 1991). Major milestones in recent history include the reclassification of the area above the upper forest border (i.e. > ~2,700 m a.s.l.) as a national park in 1973, the listing of the National Park as a UNESCO World Heritage Site in 1987 and the establishment of community-based forest management practices for the halfmile forest strip in 1998. Under its current status, Mount Kilimanjaro National Park belongs to the 40% of Tanzanian land that is already under protection, and which the international community is committed to further increase (Keane et al., 2020). Despite numerous conservation measures, the Mount Kilimanjaro National Park and its adjacent forest belt and habitats face many threats, including logging, quarries, livestock grazing, cultivation within the forests, poaching, charcoal production and outbreak of fires (Hemp, 2006c;Lambrechts et al., 2002;Liseki, 2015;Misana, 2012;Noe, 2014;Soini, 2005a).

| The IPBES conceptual framework
The IPBES conceptual framework is a simplified model of the complex interactions between the natural world and human societies.
It identifies six components representing the natural and social systems and specific links between them: (a) nature (biodiversity and ecosystems, herein species diversity and habitats), (b) Nature's Contributions to People (NCP, the contributions that people derive from nature (Díaz et al., 2018)), (c) human well-being (good quality of life), (d) direct and (e) indirect drivers of change and (f) anthropogenic assets and their interrelations. Direct drivers include climate-and land-use change, over-exploitation, invasive species and pollution, while indirect drivers include demographic, economic, technological, policy and institutional as well as cultural factors. IPBES explicitly calls for the inclusion of multiple knowledge systems, such as Indigenous and Local Knowledge (ILK), in analysing and assessing social-ecological systems (Díaz et al., 2015). As such, it addresses the fact that representations of human-nature relationships may vary across cultures and knowledge systems according to specific worldviews and cosmologies, including between scientific and ILK systems, as well as among indigenous cultures.

| Stakeholder workshop
We conducted a 3-day stakeholder workshop in Moshi, Tanzania, in September 2018. The workshop was attended by 73 participants (16 women and 57 men), whom we invited to represent various sectors and local communities. We established the list of invitees through an extensive online search validated and complemented by key local informants. We divided registered participants into five groups based on their sectoral affiliation: 16 residents of local communities, including farmers (herein 'Community'), 14 researchers and scientists ('Research'), 16 professionals in conservation and management ('Conservation'), 17 professionals in forestry, agriculture and water management and governance ('Resources') and 10 other professionals mainly drawn from the tourism sector ('Other'). Each of these groups was moderated by one facilitator and a chair elected among the members of the respective groups. The primary language was English, but workshop material was also available in Swahili. Swahili was also used for discussions, particularly in the 'Community' group.

| Data collection
We used three data collection methods: (a) moderated group discussions, (b) individual questionnaires and (c) a 'carousel-like' session.
We conducted regular plenaries to set, explain and recall the workshop context as well as the elements of the IPBES framework, to report back from group discussions and to gather feedback.

| Moderated group discussions
The group discussions aimed to establish a common understanding within the five stakeholder groups of the various sections and elements covered by the questionnaires (see Section 2.2.2). In the first discussion, each group distinguished different types of habitats and agreed on a typology that was subsequently used in the questionnaires. This discussion was based on a preliminary mapping exercise during which each participant was asked to identify her/ his geographic area of expertise on a printed map of the region depicting major habitat types. While we recorded the different areas of Mount Kilimanjaro for which participants conducted the assessments, we did not specifically interrogate their definitions of the concept of 'nature', which can vary between cultures and languages (Coscieme et al., 2020). During three subsequent discussions, the groups listed and valued important species (IPBES component 1), NCP (IPBES component 2) and drivers of change (IPBES components 4 and 5). During the discussion on species, participants were asked to indicate the use of individual species, whereas during the discussion on NCP, they were also asked to discuss the habitats providing them as well as their status, trends and future provision. During a last discussion before the carousel session, each group formulated a shared vision for a sustainable future for Mount Kilimanjaro where people live in harmony with nature.

| Questionnaires
We used two questionnaires-herein 'habitat' and 'ecosystem Participants had between 45 and 60 min for each of the two questionnaires and were encouraged to fill in as many forms as possible.
Although we used the more familiar concept of ecosystem services during the workshop, we classified these into NCP for our analyses using the classification included in IPBES/5/INF/24 (IPBES, 2017).

| Carousel session
The aim of this session was to offer a space for participants to suggest actions and actors that could contribute towards achieving their vision for a sustainable future for Mount Kilimanjaro. During the session, the groups moved along the five stops of the carousel (each identified by a flipchart and a facilitator) that represented five levels of governance and action: (a) individual/household, (b) community, (c) regional, (d) national and (e) international. At the start, each group was assigned to one of the five stops of the carousel and invited to propose actions at the corresponding level and identify actors who could be entrusted with these actions. Answers were recorded on the flipchart. After 10 min, the groups moved to the next stop, discussed the results of their predecessors and suggested additional actions. After four changes, all groups had had the opportunity to suggest actions at each of the five levels of governance.

| Research ethics
There was no local research ethics committee that could have approved the proposed protocols. Nevertheless, we took very good care that the rights and well-being of all participants were respected during the whole process and that there was no coercion whatsoever. To ensure that there was no sense of injustice or of preferential treatment and that participation was fully vol- terests in the discussed questions as well as a very positive and cordial atmosphere, it did not appear necessary to seek consent in writing. All participants were very pleased by the prospect of contributing to this research and several of them expressed their appreciation for the broad scope of the workshop, which they found especially useful compared with earlier experiences they had with thematically more limited and more sectoral meetings run in other contexts. All participants were also very keen to see the results of this workshop synthesized in one or more scientific publications. They all consented orally with the use of their anonymized responses in one or more scientific publications, again at workshop inception. Anonymization of data from oral discussions was achieved by refraining from recording the identity of the contributors. Moreover, a numeric ID rather than any personal identifying information was included when completing the questionnaires, leaving the data completely anonymous.
To further ensure that the rights and well-being of participants were respected, each discussion group at the workshop designated one local participant as a group leader responsible to facilitate the group discussion and one local note-taker. The designation of both was a participatory process within each group and both the leader and the note-taker were given the opportunity to turn down their responsibilities. By doing so, the workshop organizers ensured that they merely played the role of moderators and that discussions were facilitated on the basis of shared local understandings and cultural values.
Group leaders were available to collect and relay any concerns and discomfort of the participants to the organizers, but no such feedback

| Coding habitat responses
Despite a preliminary group discussion about the habitats of Mount Kilimanjaro (see Section 2.2.1), there was a substantial variation among respondents to the 'habitat' questionnaire in the naming of these habitats. Some organized habitats into highlands, midlands and lowlands, whereas others named habitats with a higher level of detail and made distinctions between land use and land cover (e.g. Chagga home gardens, natural springs and coffee plantations; Table   S1 in Supplementary Material S2). To enable comparison among responses, we standardized the habitat classification to seven categories-alpine, forest (including both lowland and montane forest), agroforestry, cropland, urban, freshwater and grassland-and assigned each of the reported habitats to one of these categories using the digitized maps to check for consistency (see Table S1 and Figure S1 in Supplementary Material S2 for correspondence between reported and standardized habitats). Digitized maps ( Figure S1 in Supplementary Material S2) were generated using a geographical information system (QGIS Development Team, 2019) based on the printed maps that participants used to delineate their geographic area of expertise (see above). For the figures and tables, we excluded responses for the urban habitat because of the low sample size (n = 2).

| Coding ecosystem services into NCP categories
We coded open questions relating to ecosystem services from the 'ecosystem services' questionnaire using IPBES standard terminologies for NCP (Table S2 in Table   S3 in Supplementary Material S2). We introduced an additional 'Livelihoods' category for reported benefits that did not point to a specific NCP, but rather named health or economic benefits in general, including earnings from selling crops and charcoal, and tourism.

| Coding direct and indirect drivers, and recommended actions
We classified responses on direct and indirect drivers of change using the categories adopted by the IPBES Regional Assessment for  Table S4 in Supplementary Material S2). We followed a similar protocol for indirect drivers, for example population growth was categorized as 'Demographic'; corruption, poor land management and enforcement of conservation laws were categorized as 'Institutional' (Table S4 in Supplementary Material S2). We coded recommended actions using the same standard. For example, 'Land/Water management' included suggestions such as 'afforestation', as well as 'integrated water resources management' and 'institutional development' included suggestions such as 'establish environmental groups/committees'.

| Comparison between the groups
We processed and visualized data using the tidyverse package (Wickham et al., 2019) in the R version 3.6.2 (R Core Team, 2019).
We used the networkd3 r package (Allaire et al., 2017) to visualize flows between different components of the IPBES framework.
We tested whether the responses given for each of the components of the IPBES framework (species diversity and habitats, NCP, direct drivers, indirect drivers and actions) differed between the five stakeholder groups using a permutational multivariate analysis of variance (PERMANOVA; Anderson, 2017). Specifically, we tested whether there were significant differences between (i) the groups and habitats in reported changes in habitat extent over the last 10 years; (ii) the groups in listed NCP; (iii) the groups and NCP in reported changes in access and provision over the last 10 years; (iv) the groups and NCP in responses on how NCP access and provision will change we also tested for interactions between the predictors.
We tested questions (i) and (v-vii) using data from the 'habitat' questionnaire, and questions (ii-iv) on the 'ecosystem services' questionnaire. We implemented PERMANOVA using the adonis function with 1,000 runs in the vegan r package (Oksanen et al., 2019).
To account for multiple comparisons and control for false discovery rate, we adjusted p-values using the Benjamini-Hochberg (BH) procedure (Benjamini & Hochberg, 1995). We applied a conservative 5% significance threshold to the BH-adjusted p-values, but also report BH-adjusted p-values (p-value adjusted in Table 1) >0.05 and <0.1 as 'marginally significant' at the 10 % significance threshold.

| Individual contributions and habitat mapping
The 73 participants filled between one and four forms each (except for one participant who filled in 11 habitat forms), resulting in a total of 143 and 144 responses to the 'habitat' and 'ecosystem services' questionnaires, respectively. Of all the participants, 68 (16 women and 52 men) submitted responses to the 'habitat' questionnaire. This corresponds to 15 participants per group except for the 'Research' and 'Other' groups, where only 14 and 9 participants, respectively, contributed. The same number of participants submitted responses to the 'ecosystem services' questionnaire, which in this case corresponded to 16 participants from the 'Community' and 'Resource' groups, 13 and 14 from the 'Research' and 'Conservation' groups, respectively, and nine from the 'Other' group.
The mapping exercise, during which the participants indicated on a map of Kilimanjaro which areas they were conducting the assessment for (see 'Moderated group discussion' above and Figure   S1 in Supplementary Material S2), highlighted key differences.   (Benjamini & Hochberg, 1995) 3.2 | Species diversity and habitats (IPBES component 1) During the group discussion on species, participants identified individual types of organisms at different taxonomic resolution.
Altogether participants identified 45 different types of plants, 29 mammal species or groups, nine reptile and amphibian species or groups, eight bird species, three categories of fish, five arthropod taxa, plankton, algae and worms (see Table S5 in Supplementary Material S2). The taxonomic resolution of the listed species varied by taxa, with mammals and birds typically listed at species level, plants to genus and invertebrates at coarser resolution. The 'Community' and 'Conservation' groups listed the highest number of species (37 each), with the former group listing the most plants (23 types) and the latter the most mammals (15 types). As these species lists resulted from group discussions, there is only one data point per group, and we could not test for statistical differences between the groups.
Based on the 'habitat' questionnaire, stakeholder groups did not differ significantly in their assessment of changes in habitat area (Table 1) and the interaction between stakeholder group and habitat was not significant either. However, the reported trends in area differed significantly across habitats. That is, there was a broad consensus among participants that the habitat area was decreasing, but that this decline differed across habitat types (Figure 1 left, Table S6 in Supplementary Material S2). In particular, grassland and freshwater had almost unanimously negative assessments, with over 90% of responses indicating declines in extent and condition (Table   S6 in Supplementary Material S2). While a higher proportion of responses indicated positive trends for forest condition than for grassland and freshwater (27% of 37 responses), the predominant view was also of declining forest extent (76%) and condition (68%; Figure   S2 in Supplementary Material S2). In the questionnaire, species diversity was only assessed at very coarse taxonomic level (including birds, mammals and trees) and was largely reported to be declining ( Figure S3 in Supplementary Material S2).

| Nature's Contributions to People (IPBES component 2)
Based on the 'ecosystem services' questionnaire and after clas-  Figure 2).
Access to these NCP was reported to have declined over the last 10 years by most participants in all categories except Livelihoods (where more participants reported increased than decreased access), reflecting a general trend for material NCP increasing at the expense of non-material NCP (Figure 3). In the 'ecosystem services' questionnaire, participants cited a number of direct and indirect drivers to explain the trends in NCP access over the last 10 years (Figure 4). According to their answers, increased population and competition for resources, combined with climate change, deforestation and forest degradation, has reduced access to Water, Energy and Materials, Food and Medicine, and Air and Climate NCP. However, education and awareness programs, irrigation and infrastructure improvements and tourism have contributed to improving access to Water and Livelihood NCP, countering the general declines ( Figure 4). When asked to forecast how access to these NCP might change over the next 10 years, the respondents almost unanimously predicted declines in all NCP categories ( Figure 3). The assessment of past trends in NCP provision and access differed significantly across groups (Table 1) but not across NCP categories. The interaction between stakeholder group and NCP category was marginally significant. The assessment of future trends in NCP provision and access did not differ across stakeholder groups and NCP category, and the interaction between them was not significant either.
The 'Community' group listed plant species/genera with value for medicine, food and firewood, and primarily listed domestic and edible animals, while members of the 'Conservation' group listed charismatic megafauna and birds that attract tourists. Participants also discussed the positive and negative benefits provided by some species, such as elephants enhancing tourism livelihoods but raiding crops and threatening humans.

F I G U R E 2
The NCP mentioned by different stakeholders in the 'ecosystem services' questionnaire, and the habitats with which the NCP are associated. Links show co-associations between variables in responses (e.g. which NCP was mentioned by which group member, and the habitats with which that NCP was associated), and the width of each link is proportional to the number of responses. The flows are not totally balanced because some respondents associated a particular NCP with multiple habitats (one NCP could be associated with multiple habitats) and because some respondents did not associate a habitat with an NCP. Each source node in the diagram has a different colour to make the links easier to discriminate

| Direct and Indirect Drivers of Change (IPBES components 4 and 5)
Based on the 'habitat' questionnaire, participants identified a wide range of direct and indirect drivers for the observed changes in the state of Kilimanjaro's habitats and species diversity (Figure 1). Neither the attribution of direct drivers nor that of indirect drivers differed significantly across stakeholder groups nor habitats and no interaction effect was detected between them (Table 1). The key direct drivers were land-use change (58%, number of questionnaires n = 143), climate change (31%), over-exploitation (15%) and pollution (15%), with only one respondent mentioning invasive species. Demographic change was the key indirect driver (29%), followed by cultural changes (10%) and governance (8%). These patterns held across habitats: land-use change was perceived as the main direct driver of change in non-urban habitats, except for the alpine belt, where it was tied with climate change (Figure 1), and demographic change was perceived as the main indirect driver in all non-urban habitats.

| Actions for the sustainable management of species and habitats on Mount Kilimanjaro and for a sustainable future
Based on the group discussions ( Figure 5), there was a general consensus that a happy future for Kilimanjaro depends on water security, a thriving economy and employment, good governance, education, social justice, high environmental quality, food and energy security, sustainable land use and health. A rich diversity of species, sustainable infrastructure and culture were also reported as important elements to a sustainable future, but by fewer stakeholder groups. Further differences between stakeholder groups consisted in the mere formulation of a desirable future and its elements. The 'Community' and 'Conservation' groups, respectively, described it in the following sentences: 'By 2030, Kilimanjaro people by adopting sustainable production and consumption practices, under good governance, enjoy the multiple benefits from a thriving biodiversity and healthy ecosystem which contributes to their social, economic and cultural well-being' and 'By 2030, Kilimanjaro people by adopting sustainable land-use practices under good governance enjoy the multiple benefits from a well-conserved environment and healthy ecosystems, which contribute to their culture, health and well-being; food, water and energy security; employment; education; and wealth creation'. In the 'Resources' group, no overall vision was formulated but a number of quantitative objectives were listed, such as 'an increase in hydropower production by 80%', 'climatesmart agriculture adopted by 50%' or '40% improvement in by-law enforcement'. In the remaining two groups ('Others' and 'Research'), elements were merely listed with more or less detail (e.g. 'health' in 'Research' and 'reduced child mortality' in 'Others').
The options for actions towards a sustainable future identified during the group discussions ( Figure S4 in Supplementary Material S2) and the carousel session ( Figure 6,

F I G U R E 7
The key indirect drivers, direct drivers, changes in habitat area within habitats and suggested actions mentioned by different stakeholders in the 'habitat' questionnaire. Links show co-associations between pairs of variables in responses (which indirect driver was associated with which direct driver, which direct driver was associated with which habitat area trend and which habitat trend was associated with which suggested action), and the width of each link is proportional to the number of responses. The flows are not totally balanced because one trend in habitat area could be associated with multiple (or no) drivers and recommended actions. Only links with more than five responses are displayed. Each source node in the diagram has a different colour to make the links easier to discriminate Newark, 1991;Peters et al., 2019;Soini, 2006;Stephen, 2015) and across East African mountains (IPBES, 2018b;Payne et al., 2020).
The negative trends in all non-urban habitats of Mount Kilimanjaro  et al., 2016) and to the long-persisting challenges associated with its management (e.g. Lein, 2004). It further contextualizes available evidence for increasing water resource conflicts due to growing demand as well as decreasing access and availability (Said et al., 2019). The importance given to food and energy (Figure 2) also speaks to existing evidence for the role of Mount Kilimanjaro as a primary source of food, fuel and building materials for people of north-central Tanzania (Bär et al., 2017;Sébastien, 2010). The general consensus among workshop participants that access to NCP had declined over the last decade and that these trends are likely to worsen in the future ( Based on our analysis, land-use change, and in particular the conversion of forests and grasslands into cropland in response to the shortage of arable land, is largely a result of growing demographic pressure and an increasing demand for agricultural products. Economic factors represent another important driver frequently mentioned in the literature. These factors include the timber and charcoal market responsible for much of the ongoing legal and illegal logging activities as well as the large inter-annual variation in coffee prices, which results in a transition towards the cultivation of alternative crops (e.g. maize, beans and vegetables) and horticulture (Misana et al., 2003;Soini, 2005b). Societal factors in turn, including the overall decrease in fertility rates, increased empowerment of women (Larsen & Hollos, 2003) and an increase in the marketable knowledge and skills of local populations (Soini, 2005a;Soini, 2005b), may contribute to counteracting ongoing trends. Yet, population growth in the Kilimanjaro region remains a major driver towards land-use change and causes subdivision of land into fragments that are both too small to support a family (Soini, 2002) and too numerous to allow any further expansion (Misana et al., 2003).
The conversion of natural habitats into the croplands needed to feed a growing population (Tilman et al., 2017) is a global problem and one of two broad responses to demographic and economic pressures, the second being an increase in the intensity of production per unit area (Phalan, 2018). Pursuing the former has led to alarming declines in biodiversity world-wide (Chaplin-Kramer et al., 2015;Dudley & Alexander, 2017), including in the Kilimanjaro region (Hemp & Hemp, 2018). Unless food production is outsourced to areas outside Kilimanjaro, which in turn outsources environmental problems, the sparing of Mount Kilimanjaro's biodiverse forests and grasslands will require the sustainable intensification of cropland agriculture and agroforestry systems, combined with socially acceptable land management measures (Phalan, 2018;Tilman et al., 2017).

| Options for decision-making and action towards a more sustainable future for the region
In line with recent results for all of Northern Tanzania (Kariuki et al., 2021), desirable futures for the participants of our workshop were those with high environmental integrity and stable livelihoods (thriving economy, high employment rates, water, energy and food security). Among the most frequently reported actions, both afforestation and integrated water resource management echo the importance that most participants attributed to water provision and their focus on forest and freshwater habitats. The importance of reforestation is in line with priorities set by other stakeholders in northern Tanzania (Kariuki et al., 2021). The insistence on water resource management, in turn, suggests that past water management efforts in the Pangani River Basin (IUCN, 2003) and recent recommendations to account for the welfare of local communities in watershed conservation and water governance along the Basin (Kabogo et al., 2017;Lalika et al., 2015) have not yet been fully successful. Stakeholder perceptions as collected here represent valuable information to support the complex and iterative process of forming coalitions and including participation of all stakeholders in enacting Tanzania's legal and institutional frameworks related to freshwater resources management. In the face of the growing hydrological risks, to which Mount Kilimanjaro and its populations are exposed, water and forest management are of utmost importance (Sébastien, 2010). However, the importance given to water across stakeholder groups, and in particular among the community representatives, supports recommendations to develop solutions to the water problems that are not mainly technological but mostly of social nature and based on agreed social rules and organization (e.g. social organization of access to irrigation canals; Sébastien, 2010). It further supports recommendations to encourage development projects that focus not exclusively on water as a natural capital but also as a social and cultural capital (Sébastien, 2010). Increased societal participation in the formulation of management schemes also appears important for addressing long-term forest conservation and use (Kijazi & Kant, 2011a).  Table   S7 in Supplementary Material S2) and a number of different actors.

| Insights from novel data
In most cases, the stakeholders in our study confirmed existing literature and quantitative evidence on the state of and trends in species and habitats around Mount Kilimanjaro. They also confirmed perceptions gathered during group discussions with other stakeholder groups in the region (e.g. Kariuki et al., 2021). However, in line with differences among stakeholders and ethnic groups in other Central and East African contexts such as Burundi and the Democratic Republic of Congo (Cuni-Sanchez et al., 2019;Ndayizeye et al., 2020), we detected differences between stakeholder groups.
Specifically, we found statistical differences at the 5% significance level across the questionnaire responses in the NCP listed, in NCP access and provision trends over the past 10 years, and among habitats in changes in habitat area over the last 10 years. At the 10% significance level, we also detected significant differences between the groups in recommended actions to conserve and restore species diversity and NCP. The choice of significance threshold is an arbitrary one, as there is always a trade-off between detecting false positive effects and failing to detect true positive effects. However, despite the small sample sizes and low power of the analysis, seven of 19 tests were significant at the 5% threshold for unadjusted p-values, and three of 19 were significant at the 5% threshold for BH-adjusted p-values, which is more than the ~one in 19 that would be expected by chance. Through group discussions and the carousel session, we also found qualitative differences in the actors responsible for implementing individual measures, as well as in the elements and the formulation of desirable futures.
These differences point to the necessity of adopting different discourses, entry points and levers for different stakeholder communities in order to collectively identify and ultimately achieve common sustainability objectives that align with national (e.g. the Tanzanian national development blueprint 'Vision 2025') and global agendas. Yet, developing narratives of stakeholder visions that could guide the short-and long-term planning and prioritization of management actions would require the collection of additional data and the adoption of additional participatory approaches for the exploration of alternative futures (Capitani et al., 2016;Thorn et al., 2020;Thorn et al., 2021). Whereas the challenges of species and habitat conservation are typically illustrated with the example of protected areas (Caro et al., 2009), which points to tensions between global conservation objectives and local societal and economic needs, our results highlight the importance of resolving discrepancies and aligning visions among local stakeholders at subnational scale.

| The IPBES framework in participatory workshops
Participatory workshops and focused group discussions effectively serve to make stakeholders the focal point of sustainability research and action. This is essential in reconnecting top-down management and policy mechanisms with grassroot knowledge and needs and in identifying socially acceptable pathways towards sustainability that are owned and endorsed locally across stakeholder groups (Capitani et al., 2016;Orenstein & Groner, 2014). Accordingly, landuse and land-cover change scenarios for northern Tanzania based on stakeholder discussions revealed conflicting objectives of wildlife conservation and agricultural expansion and highlighted the need for an integration of local governance in sustainable landscape management (Kariuki et al., 2021). Putting stakeholders at the centre of research is further essential in validating the results of strictly ecological approaches to biodiversity and ecosystem service assessments and overcoming their limitations (Orenstein & Groner, 2014).
The IPBES framework served as a useful tool to guide the recording of perceptions on the relationship between nature and people. As such, and although it has not necessarily served to unveil fundamentally novel narratives and presents limitations for dealing with contrasting knowledge forms (Löfmarck & Lidskog, 2017), it has been useful in achieving a level of standardization and comparability between answers and stakeholder groups that is essential for a balanced understanding of individual visions and the detection of differences between the groups. Regular group discussions, in turn, were helpful in establishing a common understanding of the framework's elements. Yet, the use of specific terminologies and concepts, such as that of ecosystem services, represents a challenge that we could not fully address, that undermines the achievement of shared understandings between stakeholders with different perceptions of and relationships to nature (e.g. Lamarque et al., 2011) and that the language barrier amplified. For example, depending on the language, the word 'nature' can implicitly include humans, exclude humans or have a spiritual component, as is the case in the Swahili term 'Asili' (Coscieme et al., 2020). Targeted efforts to establish a mutual understanding of individual concepts can help improve the results and ultimately save time spent interpreting ambiguous answers after the workshop. However, the cultural and emotional experiences behind 'experiential knowledge' and the social dimensions associated with 'value-based knowledge' (Glicken, 2000) remain inherent to people's vocabulary and perceptions and are important for interpretation.
Moreover, the language of science can generate a feeling of exclusion among non-scientific participants (Glicken, 2000). These challenges highlight the importance of workshop facilitation (Reed, 2008 lating qualitative storylines into quantitative information (Walz et al., 2007) and of codifying stakeholder knowledge that is often varied and sometimes conflicting (Reed et al., 2013). It confirms that standardized categories do not necessarily correspond to individual perceptions or worldviews and are mostly conceptually useful.
Other participatory approaches, ranging from participatory drawing (e.g. O'Donovan et al., 2020) and GIS (Brown & Fagerholm, 2015;McCall, 2003) to participatory scenario development (e.g. Capitani et al., 2016) and planning (Thorn et al., 2020;Thorn et al., 2021), could have elicited different understandings and visions, in particular in terms of possible futures (Thorn et al., 2020). Moreover, other conceptualizations of mountain social-ecological systems (e.g. A challenge in applying participatory approaches to the generation of experiential or value-based knowledge resides in stakeholder representation and the identification of social groups likely to differ in their understandings, perceptions and representations (Caballero-Serrano et al., 2017). For this workshop, we specifically selected participants covering a broad range of sectors and local communities and subsequently grouped them into stakeholder groups but did not seek to constitute representative samples by gender and age group.
Thus, some caution is warranted when interpreting both the absence of differences between stakeholders in their assessments of species diversity, NCP and drivers and the detected differences in the suggested actions towards long-term sustainability. Future analyses seeking to more explicitly explore the perceptions of stakeholder groups, as opposed to individual perceptions, will benefit from a more systematic stakeholder mapping (Glicken, 2000). Moreover, a number of prerequisites need to be met to ensure engagement and to manage communication, including an understanding of the potential social dynamics at play and a clear agreement among stakeholders on the objectives for the participatory process at the onset (Reed, 2008).

| Implications
The message from the stakeholders was clear. Land-use and climate changes threaten Mount Kilimanjaro's unique species diversity, and its contributions to people (NCP). Participants broadly identified improved natural resource management as important to achieve a sustainable future for Mount Kilimanjaro's people and its nature. The overall consensus across stakeholder groups of the problems facing Mount Kilimanjaro's species diversity and habitats and of possible solutions is encouraging as it suggests broad support for the conservation actions that need to be taken. Looking ahead, we recognize the need to further our understanding of the nuanced ways in which people value nature and its contributions.
We further recognize the need to better understand the implica-

| CON CLUS ION
Despite the challenges associated with concepts and terminologies and with achieving a shared understanding across various stakeholder groups, we successfully tackled the many dimensions of the IPBES framework within only 3 days, and participants were favourably disposed and eager to engage in a discussion addressing different aspects of the social-ecological system of Mount Kilimanjaro.
Based on this experience we propose that the IPBES framework can be effectively adopted for the mobilization of non-academic knowledge on the relationship between nature and people and that it represents a useful methodological tool to scale up the participatory assessments of local perspectives on social-ecological systems.
Future applications of this framework in participatory assessments of social-ecological systems will further show its potential in eliciting stakeholder perceptions and mobilizing the non-academic knowledge needed for the co-design of pathways towards a sustainable future.

ACK N OWLED G EM ENTS
The research was supported by the German Science Foundation (DFG) through the research unit KiLi (FOR1246). G.W.P. was supported by SNF Grant IZ08Z0_177386. The authors thank C. Penone, H. Saiz and N. Schenk for advice on the analysis.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.