The importance of realistic dispersal models in conservation planning: application of a novel modelling platform to evaluate management scenarios in an Afrotropical biodiversity hotspot

Abstract

As biodiversity hotspots are often characterized by high human population densities, implementation of conservation management practices that focus only on the protection and enlargement of pristine habitats is potentially unrealistic. An alternative approach to curb species extinction risk involves improving connectivity among existing habitat patches. However, evaluation of spatially explicit management strategies is challenging, as predictive models must account for the process of dispersal, which is difficult in terms of both empirical data collection and modelling.Here, we use a novel, individual-based modelling platform that couples demographic and mechanistic dispersal models to evaluate the effectiveness of realistic management scenarios tailored to conserve forest birds in a highly fragmented biodiversity hotspot. Scenario performance is evaluated based on the spatial population dynamics of a well-studied forest bird species.The largest population increase was predicted to occur under scenarios increasing habitat area. However, the effectiveness was sensitive to spatial planning. Compared to adding one large patch to the habitat network, adding several small patches yielded mixed benefits: although overall population sizes increased, specific newly created patches acted as dispersal sinks, which compromised population persistence in some existing patches. Increasing matrix connectivity by the creation of stepping stones is likely to result in enhanced dispersal success and occupancy of smaller patches. Synthesis and applications. We show that the effectiveness of spatial management is strongly driven by patterns of individual dispersal across landscapes. For species conservation planning, we advocate the use of models that incorporate adequate realism in demography and, particularly, in dispersal behaviours.

Keywords: Eastern Arc Mountains; Phyllastrephus cabanisi; RangeShifter; SEPM; connectivity; conservation planning; demography; dispersal; fragmentation; habitat network.

Figure 1

Location of forest fragments and exotic plantations which were converted to habitat patches under scenarios EXOTIC‐S (1 [MB], 2 [SU], 3 [WM], 4 [WR]) and EXOTIC‐L (5 [YA]). For clarity, we show relatively permeable matrix land‐cover types (indigenous forest, exotic plantation, agroforestry and bush) in grey and less permeable types in white (agricultural fields) (see Fig. S1 for a colour map). The transparent zones connecting patches indicate corridors used for the creation of stepping stones under scenario MATRIX‐C. The total area considered for the simulations was 145·6 km² (dark grey in lower inset). Map of EAM modified from Platts et al. (2011).

Figure 2

Percentage change in mean projected population size relative to the baseline under five management scenarios: habitat restoration of degraded forest patches (RESTORE), creating stepping stones in the matrix randomly (MATRIX‐R) or within corridors connecting patches (MATRIX‐C), and converting exotic plantations to forest habitat of either several small patches (EXOTIC‐S) or one large patch (EXOTIC‐L).

Figure 3

Mean dispersal success (proportion of emigrants successfully settling in a non‐natal patch) in each scenario relative to that of the baseline scenario (i.e. 0·232).

Figure 4

Mean yearly immigration rates and probabilities of patch occupancy for each patch under five management scenarios relative to the baseline scenario (see Table S1). Only patches considered in every simulation are shown.

Figure 5

Mean yearly immigration rates for each patch under six scenarios (A [baseline], B [RESTORE], C [MATRIX‐R], D [MATRIX‐C], E [EXOTIC‐S], F [EXOTIC‐L]) and for the four patches converted under the EXOTIC‐S scenario (1 [MB], 2 [SU], 3 [WM], 4 [WR]). Each bar depicts the mean number of immigrants per year originating from a large patch (CH or NG), from one of the smaller existing patches and from a converted plantation.

Figure 6

Mean yearly population sizes predicted for the newly created habitat patches under the EXOTIC‐S (a) and EXOTIC‐L (b) scenarios. The dashed line represents the population size expected for each patch assuming that converted plantations can support populations at a density of 1·09 individuals per ha (i.e. the estimated current density in low‐quality forest patches).