The co-evolution of multiply-informed dispersal: information transfer across landscapes from neighbors and immigrants

Abstract

Dispersal plays a key role in natural systems by shaping spatial population and evolutionary dynamics. Dispersal has been largely treated as a population process with little attention to individual decisions and the influence of information use on the fitness benefits of dispersal despite clear empirical evidence that dispersal behavior varies among individuals. While information on local density is common, more controversial is the notion that indirect information use can easily evolve. We used an individual-based model to ask under what conditions indirect information use in dispersal will evolve. We modeled indirect information provided by immigrant arrival into a population which should be linked to overall metapopulation density. We also modeled direct information use of density which directly impacts fitness. We show that immigrant-dependent dispersal evolves and does so even when density dependent information is available. Use of two sources of information also provides benefits at the metapopulation level by reducing extinction risk and prolonging the persistence of populations. Our results suggest that use of indirect information in dispersal can evolve under conservative conditions and thus could be widespread.

Keywords: Adaptive dynamics; Density-dependent; Dispersal; Evolution; Immigrant-dependent; Meta-population; Social information.

Figure 1. Life cycle of organisms in the model.

Diagram of the basic life cycle of individuals in the model. The two age classes of reproductive individuals (subadults aged 1 year, and adults aged 2 years and more) are described by their age-specific survival (s) and fecundity (f). Individuals disperse during the juvenile stage from age 0 to 1, indicated by *.

Figure 2. Temporal dynamics of the evolution of informed dispersal.

Temporal dynamics of the evolution of information based dispersal due to local density (d_D in red) and the number of arriving immigrants (d_I in blue). Trajectories reflect average dispersal rates for 100 Monte Carlo simulations. (A) Dynamics of immigrant number information use alone (d_I). (B) Dynamics of density dependent information use alone (d_D). (C) Dynamics of both density dependent and immigrant dependent information when used simultaneously (D&I) with no cost of dispersal. Uninformed dispersal is fixed at 10% and does not evolve. The 95% confidence interval is shown for the last time step on each trajectory.

Figure 3. Evolution of information use.

Probability that each form of information use evolves. Plotted are the proportion of simulations where dispersal evolved based on density dependent information (D, red), immigrant information (I, blue), both density and immigrant information (D + I, red and blue hatch), or where dispersal did not evolve (None, white) across 100 Monte Carlo simulations. D-alone and I-alone are for models with just one source of information available (plus U fixed at 10%). D&I is a model with both density and immigrant dependent dispersal present.

Figure 4. Behavioral reaction norms of informed dispersal.

Reaction norms for informed dispersal behavior. Solid lines show the reaction norms (black) and 95% CL (grey) for each form of dispersal. Dashed lines reflect uninformed baseline dispersal. Reaction norms were created using the Informed Dispersal equations with the mean evolved coefficient after 100000 generations. Lines for the 95% CL were constructed using the variance in evolved coefficients among 100 Monte Carlo runs. Top panels are for models where only one source of information is possible and show dispersal due to (A) density dependent dispersal (D-only) and (B) immigrant dependent dispersal (I-only). Bottom panels are for models where only both sources of information are possible (D&I) and show dispersal due to (C) density dependent dispersal and (D) immigrant dependent dispersal.

Figure 5. Benefits of informed dispersal.

The relative benefit of dispersal behavior to an individual is estimated by how much better a disperser did by moving (i.e. old pop density/new pop density, both at reproduction). Shown is the dispersal benefit over the first 100000 time steps for models with low environmental stochasticity (5% of populations hit) in models (A) I-only (D-only is similar) or (B) D&I. Benefits of multiply-informed dispersal (D&I) relative to using no information or a single source of information (D or I-only) is also observed at the meta-population level by reducing global extinction risk (proportion of 100 Monte Carlo simulations where the metapopulation goes extinct) as stochasticity increases due to (C) random environmental stochasticity or (D) small population size.