New paper led by @Diversity_Turn MSc student @DomiSchwab out in @JAppliedEcology!

Predation is a key ecosystem function – but which factors influence predation rate and predator composition in #vanilla #agroforestry?

A thread about our findings 👇👇👇 1/15
In north-eastern #Madagascar 🇲🇬, vanilla agroforests, rice paddies, land under shifting cultivation, forest fragments and old-growth forests form a diverse mosaic landscape. 2/15
Importantly, #vanilla agroforests differ in land-use history: they can be established on fallow land or inside forests, with strong implications for #Biodiversity and #EcosystemFunctions

See this thread about our paper in @ConLetters: 3/15 https://twitter.com/Dominic_Mart/status/1272429539713519623
. @DomiSwb visited all our 80 @Diversity_Turn plots across seven land-use types to measure predation rate. 4/15
But how to measure predation rate? 5/15
He used fake caterpillars 🐛 – standardized plasticine dummies – to measure 1) predation rate and 2) predator composition which he assessed by bitemark identification.

36 dummies per plot were exposed for 48 hours – summing up to 2880 exposures in total. 6/15
Predation rates where highest in the unburned land-use types old-growth forest, forest fragment and forest-derived vanilla. Rice paddies showed the smallest rates, but heterogeneity within land-use was high throughout. 7/15
In-depth analysis revealed that on-plot stem- and vegetation density were both positively associated with predation rate cross land-use types. 8/15
However, landscape-scale forest cover was also positively associated with predation rate. Predation rate differed between unburned and burned plots (that were once used for shifting hill rice cultivation). 9/15
The bitemark analysis further revealed differences in predator composition: Ants 🐜 were the most important predators throughout, but predation by other groups, especially Gryllacrididae, differed between land-use types. 10/15
These differences in predator composition also became apparent in an NMDS. Particularly old-growth forest and forest fragments featured a different set of predators than all other land-use types. 11/15
Our results show that:
- Measuring predation rates is possible with dummy caterpillars
- Predation rates – a key ecosystem function - decline with tree loss on the plot- and landscape-scale.
- The plots’ land-use history determines the predator community 12/15
We also show that it is possible to reliably identify various predators from bitemarks – even in a tropical location.

This showed a shift in predator composition across the land-use gradient & showed that Malagassy Gryllacrididae are particularly vulnerable to LUC. 13/15
You can follow @Dominic_Mart.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: