Abstract

The ecology of Madagascar's grasslands is under-investigated and the dearth of ecological understanding of how disturbance by fire and grazing shapes these grasslands stems from a perception that disturbance shaped Malagasy grasslands only after human arrival. However, worldwide, fire and grazing shape tropical grasslands over ecological and evolutionary timescales, and it is curious Madagascar should be a global anomaly. We examined the functional and community ecology of Madagascar's grasslands across 71 communities in the Central Highlands. Combining multivariate abundance models of community composition and clustering of grass functional traits, we identified distinct grass assemblages each shaped by fire or grazing. The fire-maintained assemblage is primarily composed of tall caespitose species with narrow leaves and low bulk density. By contrast, the grazer-maintained assemblage is characterized by mat-forming, high bulk density grasses with wide leaves. Within each assemblage, levels of endemism, diversity and grass ages support these as ancient assemblages. Grazer-dependent grasses can only have co-evolved with a now-extinct megafauna. Ironically, the human introduction of cattle probably introduced a megafaunal substitute facilitating modern day persistence of a grazer-maintained grass assemblage in an otherwise defaunated landscape, where these landscapes now support the livelihoods of millions of people.

Highlights

  • The grasslands of Madagascar have long been considered degraded wastelands (e.g. [1,2,3])

  • The fire group comprises 23 species, all of which are tall caespitose grasses with thicker leaves, low bulk density and low leaf width to length ratios compared to the grazing group

  • The species richness is similar between the two assemblage groups, and phylogenetic diversity within the grazing-maintained assemblage is significantly higher than the fire-maintained assemblage

Read more

Summary

Introduction

The grasslands of Madagascar have long been considered degraded wastelands (e.g. [1,2,3]). Some main roads through the Central Highlands follow river valleys and can reflect landscape water availability and soil properties that is important to shaping potential cattle densities Values of these environmental variables across our 71 studied sites are given in the electronic supplementary material, figure S2. Generalized latent variable models were used to determine whether distinct grass assemblages could be identified across sites based on the patterns of species co-occurrences [42] across 71 communities. Each test respectively used: (i) a generalized linear model with a Poisson distribution and log link function; (ii) a two-proportions z-test; (iii) a phylogenetic ANOVA using ‘phytools’ package [53]; and (iv) an estimation of Blomberg’s K [54] with the ‘phylosig’ function using 999 numbers of tree shuffling randomization

Results
Findings
Discussion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call