Abstract
Background: The extensive spatial configurations of breeding seabirds are increasingly understood as metapopulations, comprising connected subpopulations, each with its unique demography, population dynamics and exposure to risk. However, until now, the viability of each subpopulations has been treated in isolation from other neighbouring subpopulations, and under assumptions of density independence, made for simplicity and generally considered both parsimonious and precautionary. However, both connectivity and density dependence are widespread in nature and it is unclear whether ignoring such well-documented mechanisms leads to unbiased population assessments. This issue is of particular concern for the black-legged kittiwake (Rissa tridactyla, hereafter kittiwake) metapopulation, a red-listed species in the UK, which is assessed as being in decline. Conservation to date has focused on individual Special Protection Areas (SPAs) which enclose subpopulations of interest, but it is not clear how these are connected within the network of all kittiwake colonies, if non-SPA colonies supplement or drain SPA colonies and how the whole system of kittiwake colonies is responding to disturbance.
Objectives: Here, we tested the assumptions of parsimony and precaution by developing the first seabird metapopulation model to contain connectivity, three hypothesised forms of density dependence and the existence of floater birds that can act as buffers to population perturbations. We aimed to fit this sophisticated model to all available kittiwake data in the UK and Ireland, hence examining the role played by density dependence and connectivity. Using kittiwakes as an example, we also aimed to develop the next-generation fast and realistic metapopulation PVA for seabirds and explore the need for global and localised mitigation and compensation measures.
Data Collation: We collated all available relevant data, compiling the most complete and up-to-date data set on kittiwake demography and population sizes across the Britain and Ireland. We highlight limitations and assumptions and propose approaches to the treatment of spatial subdivisions of the kittiwake metapopulation network. The dataset encompasses both SPA and non-SPA colonies, covering population size surveys from 1985 to 2023, data surveys on breeding success, survival of adults and immatures, and observations and ringing data on dispersal.
Model development: Our modelling approach aimed to achieve 1) A biologically realistic representation of demography by incorporating time-varying demographic rates (with potential to include environmental covariates) and density dependence of three types (Allee effect, crowding at colonies and scramble competition at sea), 2) A metapopulation structure of sufficient detail, incorporating 89 colonies (a mixture of 33 SPA and 56 non-SPA locations), 3) Integrated use of multiple sources of data, including population sizes together with survival and breeding success data, along with dedicated sub-modelling used to extract the most informative priors possible. We examined the performance of three models. Model 1 (regional summaries model) allowed each colony to have independent time series of survival and breeding success. Model 2 (regional dynamics model) aggregated demographic performance (survival and breeding) into six spatial regions (NW,NE,W,E,SW,SE points of the compass). Model 3 (regional trends model) was the same as model 2, but with the addition of time trends in demographic performance. The three models were fitted to successful convergence to the complete colony network and available data. A crucial limitation in the process were the lack of any direct or indirect information on the localised carrying capacities of breeding colonies. The models produced adequate fits, but with extensive associated levels of uncertainty on the strength of density dependence. The fitted models presented similar patterns, indicating that only the middle and eastern part of the British isles is characterised by persistent populations.
Forecasts and counterfactuals: Using the most parsimonious model 2, we generated forecasts of population trends. All colonies in the metapopulation are predicted to tend towards extinction over a period of 50 years. Geographically, there are no clear patterns in the speed with which populations are declining. Running the model with connectivity switched off shows that connectivity does not make much difference to these trends since the entire metapopulation comprises sink populations. To understand possible mitigation/compensation options, we explored the level of improvements needed in demographic rates for the decline to be reversed. We considered combinations of incremental improvements and deteriorations for four demographic rates: Breeding success, Adult survival, Pre-breeder survival and Floater survival. The most important demographic rate, by far, was found to be adult survival. We also investigated possible mitigating effects of artifical nesting structures (ANS). Our approach illustrates how the positioning and benefits of ANS can be investigated. We illustrate a practical approach investigating positioning and benefits of ANS as mitigating conservation management action. For assessing the feasibility of such recovery strategies the features of connectivity and density dependence examined here are highly relevant, especially the Allee form of density dependence that refers to colonies whose size is well below their carrying capacity.
Conclusions: For a declining population in which all subpopulations are sinks, we conclude that the advanced metapopulation PVA developed here adds very little in predicting the time to extirpation of each subpopulation. This is mainly because the levels of donations between subpopulations are not greatly affecting their fate. However, because covariates are not included it is unclear how the metapopulation may redistribute in response to environmental change, and how this may change local and national persistence. We illustrate a practical approach investigating positioning and benefits of ANS as mitigating conservation management action. For assessing the feasibility of such recovery strategies the features of connectivity and density dependence examined here are highly relevant, especially the Allee form of density dependence that refers to colonies whose size is well below their carrying capacity.