Complementary ecosystem services from multiple land uses highlight the importance of tropical mosaic landscapes

Tropical agricultural landscapes often consist of a mosaic of different land uses, yet little is known about the spectrum of ecosystem service bundles and materials they provide to rural households. We interviewed 320 households on the different benefits received from prevalent land-use types in north-eastern Madagascar (old-growth forests, forest fragments, vanilla agroforests, woody fallows, herbaceous fallows, and rice paddies) in terms of ecosystem services and plant uses. Old-growth forests and forest fragments were reported as important for regulating services (e.g. water regulation), whilst fallow lands and vanilla agroforests as important for provisioning services (food, medicine, fodder). Households reported the usage of 285 plant species (56% non-endemics) and collected plants from woody fallows for varying purposes, whilst plants from forest fragments, predominantly endemics, were used for construction and weaving. Multiple land-use types are thus complementary for providing ecosystem services, with fallow lands being particularly important. Hence, balancing societal needs and conservation goals should be based on diversified and comprehensive land management. Supplementary Information The online version contains supplementary material available at 10.1007/s13280-023-01888-3.


Details about village selection
To select these 10 out of the 60 villages from Hänke et al., (2018), we first did a cartographic stratification by eliminating villages where the landscapes within a 2 km radius were dominated by wetlands or lakes, rice fields, and coconut plantations, which were too close to the ocean or harboured no forest fragments.From the resulting 17 villages, we considered the distance between them and randomly removed one of the villages for village pairs with <4 km distance.After this second step, we had 14 villages that we visited; one of them did not have all the prevalent land-use types and the community of another village did not want to participate in our study.With the 12 remaining villages, we then randomly selected the 10 villages as a compromise between the number of villages and logistical constraints.

Details about data analysis:
To determine the value of land-use types for ecosystem services, we calculated the mean of household percentages that reported each ecosystem service type within each landuse type with standard errors.Then, we visualised the mean percentage of the households benefiting from each type of ecosystem services per land-use type, using polar area charts with R-package ggplot2 (Wickham 2016).Additionally, we also characterised the households relying on different land uses for each type of ecosystem services by categorising households' heads by gender (male and female), age group (young adults <36 years old; middle-aged adults: 36-55 years old and older adults > 55 years old) and highest attained educational level (no school education, primary school, secondary school, and high school or university).For this, we analysed the variation of the groups for each type of ecosystem service within each land-use type, using generalised linear models with a binomial distribution.
To depict the differences between land-use types for each ecosystem service type, we analysed the variation of the percentages of households reporting enjoying each type of ecosystem service across land-use types using generalised linear models with binomial distribution followed by post-hoc tests using the R-package multcomp (Hothorn et al. 2008).
To evaluate the importance of each land-use type for each use category, we calculated the percentage of households that reported collecting plants from the respective land-use type along with their frequency of use as well as their purposes (use, selling, or both).
Subsequently, we used the Sankey diagram from the ggforce R-package for visualisation.We also analysed the variation in the percentage of households that collected plants across landuse types by applying generalised linear models with binomial distribution followed by post-hoc tests.
To assess the number of species used, we first calculated the total number of species reported by all interviewed households, and the total number of species from each land-use type.To detect the importance of each land-use type in terms of the number of species used at the village level, we calculated: (1) the number of species used; (2) the percentage of endemic, native, or exotic species used; and (3) the percentage of growth forms (tree, shrub, liana and herb) used.To evaluate the variation in the species used across land-use types, we also applied a generalised linear model with generalised Poisson distribution for the number of species used and binomial distribution for species origin and growth form groups.We also did post-hoc tests to determine the differences between land-use types and between species origin and growth form groups within each land-use type.
To determine where most cited species were collected across the land-use types, we calculated the number of households naming each species (hereafter called citation number) per land-use type.For this, we used the fractional citation number, i.e. if a species was cited once for one use category by one household but collected from three different land-use types, the citation for each land-use type is one-third.However, when the species is mentioned once and collected from a single land-use type then the citation is one.Then, we visualised the species by summing up their fractional citation number per land-use type for all households using ranked bar plots separated by the growth forms (tree, shrub, liana and herb).
References: Table S1.Overview of the questionnaire used to evaluate the importance of land-use types in terms of ecosystem services types (I) and the collection of plants for the 7 categories of use (II).Full questionnaire of the survey is available at Open Science Framework data repository (https://osf.io/7ca5g/).Prior to each interview, ClB started with a self-introduction, along with the objective and structure of the survey, the approximate length of the interview, and the use of the data (academic and research purposes).The interviewer also ensured the respondent the anonymity of their identity in any use of the data and then asked verbally for their consent to participate in the survey.-Animals' habitat: shelter for any type of animal from an ecological point of view -Unmanaged cultivated food: plants cultivated but do not require any management as they grow with wild plants -Wild food: wild and edible plants -Plants for construction: plants used for construction via their woods or leaves, bark etc… -Land for descendants: land that is reserved to be used by their children or grandchildren in the future

Figure S4 .Figure S5 .
Figure S4.Total number of species used across land-use types in north-eastern Madagascar

. Reference for icons in Figure 2
Amrita Mayuri (2020) Weaving icon retrieved from https://thenounproject.com/search/?q=Weaving&i=858481 What do you use for what or what benefits do you get from the land-use type (other than the main crop for rice paddies and vanilla agroforests)?1. Please name plant species you use for each of the following categories: medicine, food, construction, firewood, charcoal, fodder and weaving For each named species: 2. Is it for use only, selling only or for both use and selling? 3.Where do you collect this plant and how often?
I) Importance of land-use types for farmers (for each land-use type): old-growth forests, forest fragments, vanilla agroforests, woody fallow, herbaceous fallows, and rice paddies.II) Importance of land-use type to collect plants for medicine/ food/ construction/ firewood/ charcoal/ fodder/ weaving 1. Do you own or have access to old-growth forests, forest fragments, vanilla agroforests, woody fallows, herbaceous fallows and rice paddies?• No access • I have access but not on my own (it can belong to their relatives or neighbour who granted access) • I possess this land-use type 2. • Old-growth forests • Forest fragments • Vanilla agroforests • Woody fallows • Herbaceous fallows • Paddy rice • Other (e.g.border of the river, lake, around the village, market)

Table S6 .
Mean percentage with standard error of households that are collecting plants for different categories of use with their frequency of use for households across land use types in north-eastern Madagascar

Table S7 .
Mean percentage with standard error households that are collecting plants for different categories of use with their purposes for households across land-use types in north-eastern Madagascar