The extinction of all Madagascar’s megafrugivores ca. 1000 years ago, may have left its signature on the current distribution of vertebrate-dispersed plants across the island, due to the loss of effective seed dispersal. In this study, we dissect the roles of extinct and extant frugivore distributions, abiotic variables, human impact and spatial predictors on the compositional turnover, or beta-diversity, of palm (Arecaceae) species and their fruit sizes across 40 assemblages on Madagascar. Variation partitioning showed that palm beta-diversity is mostly shaped by the distribution of extant frugivores (8 lemur, 3 bird, 2 rodent and 1 bat species) and the abiotic environment (e.g., forest cover, slope and temperature), and to a lesser extent by the distribution of extinct frugivores (5 giant lemur and 3 elephant bird species). However, the contribution of these variables differed between dry western assemblages and wet eastern assemblages, with a more prominent role, albeit still small, of extinct megafrugivores in the west. These results suggest that palm distributions in the dry west of Madagascar, where megafrugivores were probably most abundant in the past, still show signatures of past interactions. With a fourth-corner analysis we observed that the distribution of palm species with relatively large fruits and seeds was negatively associated with home range of extant mammalian frugivores and frugivore richness of both past and extant communities, and positively associated with the hand-wing index (HWI) – a proxy for dispersal ability - of extant bird communities. This suggests that palm species with relatively large fruits tend to occur in places with fewer, small-ranged mammalian frugivores, which may indicate dysfunctional seed dispersal, although a few wide-ranging bird species may compensate this loss by dispersing the seeds of small-to medium-sized palm fruits. Our results shed new light on anachronisms in Madagascar, and how defaunation and past interactions may underlie current plant distributions.

Please refer to the Methods section and Supplementary Information of the published article.

This data repository consists of the following folders and files:

• Main folder
  • Word document containing information on each file and instructions for use (ReadMe.docx)
  • Excel document with 4 different sheets, containing supplementary tables S1, S2 and S3 and their references (TableS1_S2_S3.xlsx)
• • Supporting information for the manuscript (SupportingInformation.pdf)
• “data” folder
  • Palm presence-absence data for each of the forty 0.3x0.3 grid cells (Palms_in_grid03.csv)
  • Palm dispersal-related trait data (maximum stem height in meters, average fruit length in millimeters, average fruit width in millimeters, average seed length in millimeters and average seed width in millimeters) for the 165 palm species appearing in the 40 grid cells (PalmTraits.csv)
  • Average climatic, soil and human variables per grid cell (Environment_grid03.csv):
  • Extant and extinct frugivore presence-absence data for each of the forty 0.3x0.3 grid cell (frugivore_extant_extinct_in_grid03.csv)
  • Extant and extinct frugivore dispersal-related trait data (frugivore_traits.csv): Body mass in kilograms, hand-wing index (HWI) and home range in hectares.
  • Geographic coordinates in Latitude and Longitude decimal degrees for the centroids of each of the forty 0.3x0.3 grid cell (centroids_grid03.csv)
  • Shapefile of the forty 0.3x0.3 grid cell created in QGIS 3.10.7 used to extract all the other values (0.3grid_20records)

• Metadata

  Average climatic, soil and human variables per grid cell (Environment_grid03.csv)

• BIO1: Annual mean temperature (°C x 10)
• BIO4: Temperature seasonality (standard deviation ×100)
• BIO6: Minimum temperature of the coldest month (°C x 10)
• BIO12: Annual mean precipitation (mm x month-1)
• pet: Annual potential evapotranspiration from the Thornthwaite equation (mm)
• cwd: Annual climatic water deficit (mm)
• alt: Altitude (in m)
• slop: Slope in degrees
• solar: Solar radiation (in Wh.m-2.day-1) 
• percfor2010: Madagascar’s forest in 2010 was derived from the 30 m resolution 2000 forest map by Harper et al. (2007).
• Clay: Soil clay content (0-2 micrometre) in g/100g (w%)
• CationEC: Cation exchange capacity (CEC measured in 1 M NH4OAc buffered at pH 7) in cmolc/kg (fine earth) 
• Extractable_Aluminum: Extractable aluminium content (Al measured by Mehlich 3) in mg/kg (fine earth)
• Total_Nitrogen: Total nitrogen (N) content in g/kg of the fine earth fraction
• Total_Phosphorus: Total Phosphorus (P) content of the soil fine earth fraction in mg/kg (ppm) 
• Popdensity2010: “For all locations with more than 1000 people per km−2, we assigned a pressure score of 10. For more sparsely populated areas with densities lower than 1000 people·km−2, we logarithmically scaled the pressure score using: Pressure score = 3.333 × log(populationdensity+1)” (Venter et al. 2016).
• HFP2009: Human footprint for 2009, calculated as a summary value from other several human-related variables such as built environments, population density, night-time lights, croplands, pasture, roads, railways, navigable waterways.
• Roads: “We mapped the direct and indirect influence of roads by assigning a pressure score of 8 for 0.5 km out for either side of roads, and access pressures were awarded a score of 4 at 0.5 km and decaying exponentially out to 15 km either side of the road” (Venter et al. 2016)