Plant diversity in Oceanic archipelagos: realistic patterns emulated by an agent‐based computer simulation

Abstract

Although islands as natural laboratories have held the attention of scientists for centuries, they continue to offer new study questions, especially in the context of the current biodiversity crisis. To date, habitat diversity on islands and spatial configuration of archipelagos have received less attention than classical island area and isolation. Moreover, in the field where experiments are impossible, correlative methods have dominated, despite the call for more mechanistic approaches. We developed an agentbased computer simulation to study the effect of habitat diversity and archipelago configuration on plant species richness and composition in five archipelagos worldwide (Hawaii, Galapagos, Canary Islands, Cape Verde and Azores) and compared simulated diversity patterns to the empirical data. Habitat diversity proved to be an important factor to achieve realistic simulation results in all five archipelagos, whereas spatial structure of archipelagos was important in more elongated archipelagos. In most cases, simulation results correlate stronger with spermatophyte than with pteridophyte data, which we suggest can be attributed to the different dispersal and evolution rates of the two species groups. Correlation strength between simulated and observed diversity also varied among archipelagos, suggesting that geological and biogeographic histories of archipelagos have affected the species richness and composition on the islands. Our study demonstrates that a relatively simple computer simulation involving just a few essential processes can largely emulate patterns of archipelagic species richness and composition and serve as a powerful additional method to complement empirical approaches.

Estimating probabilistic site‐specific species pools and dark diversity from co‐occurrence data

“…Here, we created a virtual landscape containing different habitats and a set of species with different suitability for these habitats and allowed communities to develop following simple rules for a period of time (see below). Simulations were based on Jõks and Pärtel (2019), with the difference that our agents represented individuals of a species rather than populations. We created a 100 × 100 grid divided into 100 sites (each encompassing 10 × 10 cells); cells could either contain an individual or be empty.…”

Legacy of archipelago history in modern island biodiversity – An agent‐based simulation model

“…Present archipelago maps were derived from elevation–bathymetry raster maps (Amante & Eakins, 2009), where one raster cell represents an area of 0.0167° × 0.0167° (2–3 km 2 , depending on the latitude of the archipelago). Advantages and possible limitations of such map resolution are discussed by Jõks and Pärtel (2019). The mean elevation in each raster cell was converted into the variable “habitat number” (see Section 2.2).…”

“…Simulations have gained popularity as a helpful tool to understand the mechanisms behind biodiversity patterns (for review, see Leidinger & Cabral, 2017), but until now they have focused mainly on the effects of present island properties, with only a few exceptions (Borregaard et al., 2016; Cabral, Whittaker, et al, 2019; Cabral, Wiegand, et al, 2019; Chalmandrier et al., 2018; Valente et al., 2014). Recently, Jõks and Pärtel (2019) presented an agent‐based island simulation for addressing the main archipelago properties that affect development of biodiversity. They found habitat diversity to be very important, whereas archipelago configuration was significant only in linearly arranged archipelagos (i.e., the Hawaiian Islands and Azores).…”

“…Here, we extend the agent‐based simulation model by Jõks and Pärtel (2019) to study the effects of geological history and Pleistocene sea‐level fluctuations on present biodiversity and apply the model to three hotspot archipelagos: the Hawaiian, Galapagos and Canary Islands. The simulation approach allows us to evaluate the effect of different types of historical information on the archipelago diversity patterns.…”

Macroecology in the age of Big Data – Where to go from here?

“…Brown, 1999), and the continuing diminution of computational limitations has made it possible to include various processes into macroecological analyses. These processes include, for example, physiology-related mechanisms (Kearney & Porter, 2004), microevolutionary dynamics of populations via explicit simulation of the genetic architecture of phenotypes (Schiffers et al, 2014), metapopulation dynamics via explicit simulation of dispersal and local demography across changing environment in distribution models (Juliano S Cabral & Schurr, 2010;Zurell et al, 2016), metacommunity dynamics via inclusion of resource competition and other biotic interactions (Juliano Sarmento Cabral & Kreft, 2012;Münkemüller et al, 2012), macroevolutionary processes (Aguilée, Gascuel, Lambert, & Ferriere, 2018;Cabral, Wiegand, & Kreft, 2019;Jõks & Pärtel, 2018;Rangel et al, 2018) and plate tectonics (Descombes et al, 2018;Leprieur et al, 2016). The current trend in mechanistic macroecology is to include the manifold processes into an integrative modelling framework (Cabral, Valente, & Hartig, 2017;Leidinger & Cabral, 2017;Methorst, Böhning-Gaese, Khaliq, & Hof, 2017;Pontarp et al, 2018;Thuiller et al, 2013;Urban et al, 2016).…”