Natural and anthropogenic pressures shape spatial distribution and niche partitioning in fruit-feeding butterflies across Madagascar's fragmented landscapes

Natural and anthropogenic pressures shape spatial distribution and habitat preference partitioning of fruit-feeding butterflies across a forest-cover gradient in Madagascar's central highlands These data support all results reported in the associated manuscript. Fruit-feedingbutterflies were sampled with Van Someren-Rydon bait traps across 45 sampling units (SUs),nested within 9 landscapes (6x6 km), nested within 3 municipalities (Antsirabe,Tsiroanomandidy, Manjakandriana) in the Antananarivo region, Madagascar, spanning a20-60% forest-cover gradient. Sampling was conducted over three periods: June 2023,April 2024 and June 2024 (an intended April 2023 round was lost to flooding). Five files are provided, at four levels of aggregation. All files share the `Landscape`and, where applicable, `SamplingUnit` identifiers, so records can be joined across files. | File | Level | Rows | Description ||---|---|---|---|| `butterfly_sampling_units.csv` | Sampling unit (pooled across periods) | 45 | Habitat/anthropogenic variables and species abundances per SU, summed across all 3 sampling periods || `butterfly_sampling_units_by_period.csv` | Sampling unit x period | 135 | As above, disaggregated by sampling period || `butterfly_community_matrix_by_landscape.csv` | Landscape | 9 | Species abundance matrix aggregated to landscape level, with diversity metrics || `butterfly_data_long.csv` | Landscape x species (long format) | 144 | One row per landscape-species combination, with habitat-preference guild, rarity class, and frequency of occurrence || `butterfly_data_long_by_period.csv` | Sampling unit x period x species (long format) | 2,160 | Full-resolution long-format record of every count, for reproducing GLMM/db-RDA analyses | Site codes Nine landscapes, coded by forest-cover percentage and municipality identifier, as usedthroughout the manuscript: | Landscape | Municipality | Forest cover (%) | Cover category ||---|---|---|---|| 20A | Manjakandriana | 20 | Low (C20) || 26D | Tsiroanomandidy | 26 | Low (C20) || 29G | Antsirabe | 29 | Low (C20) || 34E | Tsiroanomandidy | 34 | Middle (C40) || 44B | Manjakandriana | 44 | Middle (C40) || 50H | Antsirabe | 50 | Middle (C40) || 53C | Manjakandriana | 53 | High (C60) || 56I | Antsirabe | 56 | High (C60) || 60F | Tsiroanomandidy | 60 | High (C60) | Cover categories follow Methods 2.1: C20 = 20-30%, C40 = 30-50%, C60 = >50%. Each landscape contains 5 sampling units, coded `SamplingUnit` (e.g., N6-N10 or equivalentper-landscape codes), spaced >=2 km apart within the landscape. Species list and habitat-preference guild Sixteen species were recorded. Guild assignment (`HabitatPreference` in the long-formatfiles) follows the classification described in Methods, based on documentedhost-plant and adult-habitat association records (Uehara-Prado et al., 2007; Wurz et al.,2022), independent of each species' observed distribution in this dataset. | Species | Guild ||---|---|| Hypolimnas misippus | Generalist || Danaus chrysippus | Generalist || Papilio demodocus | Generalist || Vanessa cardui | Generalist || Spoladea recurvalis | Generalist || Phalanta phalantha | Generalist || Bocchoris inspersalis | Generalist || Acraea ranavalona | Forest tree (canopy specialist) || Precis andremiaja | Forest tree (canopy specialist) || Cyligramma disturbans | Forest tree (canopy specialist) || Chrysiridia rhipheus | Forest shrub (understory specialist) || Leptomyrina phidias | Forest shrub (understory specialist) || Charaxes antamboulou | Forest shrub (understory specialist) || Charaxes andara | Grassland specialist || Zizula hylax | Grassland specialist || Junonia rhadama | Grassland specialist | File 1: `butterfly_sampling_units.csv` One row per sampling unit (n = 45), variables pooled/summed across all 3 sampling periods. | Column | Description | Units / values ||---|---|---|| Landscape | Landscape code | see Site codes table || SamplingUnit | Sampling-unit code within landscape | text || Municipality | Municipality | Antsirabe / Tsiroanomandidy / Manjakandriana || ForestCoverPct | Landscape-level forest cover | percent || CoverCategory | Forest-cover class | Low (C20) / Middle (C40) / High (C60) || Relief | Relief complexity (SD of altitude, 10 points within 0.5 km radius, SRTM) | dimensionless (m SD) || TreeShrubDensity | Mean tree + shrub density within 5 m radius of traps | individuals / plot || TreeShrubSD | SD of tree + shrub density across the SU's traps | individuals / plot || HerbCover | Herbaceous cover (1x1 m quadrat) | proportion (0-1) || LitterDepth_cm | Leaf litter depth | cm || EVI | Enhanced Vegetation Index (MODIS, 250 m) | dimensionless (0-1) || EVI_SD | SD of EVI across the SU's traps | dimensionless || Physiognomy | Local site physiognomy classification | see note below || DistWeir_m | Euclidean distance to nearest weir | m || DistResidence_m | Euclidean distance to nearest residence | m || DistRuralRoad_m | Euclidean distance to nearest rural road | m || DistRoad_m | Euclidean distance to nearest main road | m || BovineIndex | Bovine presence (frequency of feces per 15 m trail interval) | proportion (0-1) || PSava | Proportion of savanna in landscape buffer | proportion (0-1) || PGrass | Proportion of wooded grassland in landscape buffer | proportion (0-1) || PGrazi | Proportion of grazing/pasture land in landscape buffer | proportion (0-1) || PWater | Proportion of water bodies in landscape buffer | proportion (0-1) || [16 species columns] | Individual counts per species at this SU (summed across periods) | count || Abundance | Total individuals (sum of all species columns) | count || Richness | Number of species with count > 0 | count | File 2: `butterfly_sampling_units_by_period.csv` Same structure as File 1, disaggregated by sampling period (n = 135 = 45 SUs x 3 periods).Adds: | Column | Description | Units / values ||---|---|---|| Period | Sampling round | June 2023 / April 2024 / June 2024 || Year | Sampling year | 2023 / 2024 || Month | Sampling month | June / April || Season | Season classification | Rainy (April) / Post-rainy (June) | Habitat and landscape variables (Relief, TreeShrubDensity, EVI, etc.) are repeated perperiod since they reflect relatively stable site conditions measured once; only speciescounts, Abundance, and Richness vary by period. File 3: `butterfly_community_matrix_by_landscape.csv` One row per landscape (n = 9); species counts and diversity metrics aggregated acrossall 5 SUs and all 3 periods within each landscape. | Column | Description | Units / values ||---|---|---|| Landscape, Municipality, ForestCoverPct, CoverCategory | as above | || [16 species columns] | Total individuals per species in this landscape | count || Abundance | Total individuals, all species | count || Richness | Total species recorded in this landscape | count || Shannon_1D_reported | True (Hill-number) alpha diversity, Shannon (q=1), from rarefaction (iNEXT) | equivalent number of species || Pielou_J_reported | Pielou's evenness (J') | dimensionless (0-1) | File 4: `butterfly_data_long.csv` Long-format table, one row per landscape x species combination (9 landscapes x 16species = 144 rows), used for the habitat-preference partitioning and indicator-speciesanalyses (Methods). | Column | Description | Units / values ||---|---|---|| Landscape, Municipality, ForestCoverPct, CoverCategory | as above | || Species | Species name | binomial || HabitatPreference | Habitat-preference guild | Generalist / Forest tree / Forest shrub / Grassland || RarityClass | Frequency-of-occurrence class (Methods 3.3: Fo >= 0.5 Constant; 0.1 < Fo < 0.5 Common; Fo <= 0.1 Rare) | Constant / Common / Rare || Fo | Frequency of occurrence (Fo = Pa / P, landscape-level) | proportion (0-1) || Count | Total individuals of this species in this landscape | count | File 5: `butterfly_data_long_by_period.csv` Full-resolution long-format table, one row per sampling-unit x period x speciescombination (45 SUs x 3 periods x 16 species = 2,160 rows). This is the base table fromwhich all other files are aggregated, and the appropriate table for reproducing theGLMM and db-RDA/PERMANOVA analyses. | Column | Description | Units / values ||---|---|---|| Landscape, SamplingUnit, Municipality, ForestCoverPct, CoverCategory | as above | || Period, Year, Month, Season | as in File 2 | || Species, HabitatPreference, RarityClass | as in File 4 (RarityClass computed at landscape level, repeated here for convenience) | || Count | Individuals of this species recorded at this SU in this period | count |