sizes as auxiliary variables and form the joint posterior distribution over both the parameters and unknown population sizes. We can then use MCMC to sample from the posterior distribution, whereby we update the parameters and the auxiliary variables at each iteration of the MCMC algorithm and thus obtain marginal posterior distributions for both. It is then straightforward to obtain estimates of the true population sizes in the form of marginal posterior means or medians. It is equally simple to calculate error bands, in the form of posterior credible intervals, whereas this can be difficult or time consuming in the classical paradigm (Brooks et al. 2004). Further mathematical and implementational details of the Bayesian approach are provided for specific applications in later chapters (see, in particular, Sections 2.2.2 and 2.2.3), and see also King et al.
(2009) for some additional, more general information.
1.4
Application: common guillemots in the UK
The UK is host to internationally important numbers of breeding seabirds— including around 12% of the world common guillemot Uria aalge population
(Harris & Wanless 2004)—making them an important component of the nation’s
biodiversity. Furthermore, the position of seabirds at the top of the marine food chain makes them useful indicators of both the state of the marine environment and the effects of human activities upon it (Parsons et al. 2008). Thus, there is an incentive to collect and analyse data on seabird populations, both to assess their conservation status, and to monitor aspects of the health of the wider marine environment. In this thesis we focus on the integrated analysis of data collected on several UK populations of common guillemot. Our primary aim is the integration of multiple sources of data from a single colony, but we also consider how this may be beneficially combined with similar data from other populations.
1.4.1
Guillemot biology
The common guillemot (hereafter guillemot) is a medium-sized marine bird of the auk (Alcidae) family. The guillemot is one of the most abundant seabirds in temperate and colder parts of the northorn hemisphere, with very large popula- tions in the Atlantic, Pacific and Arctic Oceans (Harris & Wanless 2004). It is primarily a pelagic species, but during the breeding season (April–September) birds return to land, forming large, dense colonies on coastal cliffs and rocky offshore islands. At this time they are highly visible and accessible, making
monitoring and data collection relatively straightforward and inexpensive com- pared to some terrestrial mammals, for example. Adult guillemots are highly territorial and site-faithful, returning to the same small area of cliff ledge year upon year (Harris et al. 1996b)—a characteristic that makes them particularly suitable for long-term, individual-based studies such as mark-recapture. Young guillemots also have a tendency to return to their natal areas during their pre- breeding years and many subsequently recruit nearby (Harris et al. 1996a), although a proportion of birds are known to recruit to other colonies (Halley &
Harris 1993).
The guillemot is a typical ‘K-selected’ species (MacArthur & Wilson 1967), its life history being characterised by a long lifespan (expected 24 years:Robin-
son 2005), low reproductive rate (maximum clutch size 1 egg:Robinson 2005)
and delayed maturity (median age of first breeding 5–7 years:Harris et al. 1994). Consequently, guillemots have low annual recruitment rates so populations tend to change slowly over time. In common with other species at the ‘slow’ end of the life-history continuum (Sæther & Bakke 2000), adult survival has a high contribution to the population growth rate.
1.4.2
Status and trends
The guillemot is Britain and Ireland’s most abundant breeding seabird with one million pairs estimated in the Seabird 2000 census, the main concentrations of these being in the north and west (Harris & Wanless 2004). The total population increased substantially between 1969–70 and 1998–2002, although the rate of increase slowed from 4–5% per annum during the 1970s and ’80s, to 2% during the 1990s (Harris & Wanless 2004). More recently, the population has levelled off or even started to decline (2% decline 2000–2008: JNCC 2009).
A greater cause for concern has been the recent decline in breeding per- formance of many UK seabird populations—the common guillemot included— which has made national headlines and featured prominently in high-profile publications (e.g.Eaton et al. 2005,2007). Guillemot productivity in 2004 was by far the worst on record for many colonies in the North Sea and Northern Isles, with no chicks fledged at all from the large colony on Fair Isle (Mavor
et al. 2005). In 2005 there was some improvement, although productivity was
still markedly below the long-term mean and breeding failures were observed for the first time along the west coast of Scotland (Swann 2005, Mavor et al. 2006). The trend continued in 2006, with low levels of breeding success recorded throughout Britain: guillemots on Handa (northwest Scotland) experienced al-
1.4 Application: common guillemots in the UK 15
most complete breeding failure, and record lows were observed at colonies as far apart as the Isle of May (southeast Scotland) and Skomer (Wales) (Mavor et al. 2008). Further decreases in mean UK guillemot productivity were recorded in 2007 and 2008 (JNCC 2009). Overall, annual productivity in guillemots declined by nearly 50% during the period 1989–2007, with most of that fall occurring since 2002 (Eaton et al. 2009).
The main reason for the poor breeding success appears to have been low availability and poor quality of the guillemot’s main prey species, especially the lesser sandeel Ammodytes marinus (Mavor et al. 2005, Wanless et al. 2005); this, in turn, is thought to be linked through complex mechanisms to climate change (MCCIP 2009). Sea surface temperatures in UK coastal waters have been rising since the early 1980s by around 0.2–0.9◦C per decade (Holliday
et al. 2008), and warmer sea temperature has been correlated with poorer than
average sandeel recruitment (Arnott & Ruxton 2002). This is presumed to be the mechanism linking high winter sea surface temperature to poor breeding success (and survival) in another seabird species, the black-legged kittiwake
Rissa tridactyla, during the last two decades (Frederiksen et al. 2004b). Un- til recently the guillemot appears to have been largely buffered against these changes, possibly because it dives and thus may gain access to a wider variety of prey than surface feeders such as kittiwakes. The fact that guillemot breed- ing success is also now being affected thus points towards more severe food shortages in 2004 and subsequent years (Mavor et al. 2005).
Due to the low annual recruitment rate of most seabirds, even dramatic changes in productivity may take a number of years to manifest themselves as changes in population growth rates (Eaton et al. 2007). Changes in adult survival, on the other hand, have a more direct and immediate effect on breeding population size, and because population growth of long-lived species is most sensitive to variation in adult survival (Lebreton & Clobert 1991, Sæther &
Bakke 2000) even small reductions can have large effects on population trends.
A strong negative relationship has been identified between autumn sea surface temperature in the North Sea and adult survival of common guillemots from the colony of Hornøya, northern Norway, which are known to winter in the North Sea (Sandvik et al. 2005). Although no such relationship has yet been found among any UK guillemot populations, predictions from climate change scenarios of further increases in sea surface temperature (Lowe et al. 2009) must inevitably raise serious concerns about the future of common guillemots and other UK seabird populations.
1.4.3
Data collection
The Joint Nature Conservation Committee (JNCC)’s Seabird Monitoring Pro- gramme coordinates seabird monitoring on a UK-wide basis (Mavor et al. 2006). Under the Seabird Monitoring Programme, a variety of species have been rou- tinely assessed for breeding numbers and breeding success at a representative sample of UK colonies since 1986. More intensive, individual-based monitor- ing schemes, providing information on survival rates, for example, have been conducted at a few geographically dispersed ‘key sites’: Isle of May (southeast Scotland), Fair Isle (Shetland), Canna (west Scotland), Colonsay (west Scot- land), and Skomer (Wales). It is these detailed data that we are interested in here.
The Isle of May long-term study (IMLOTS; Centre for Ecology and Hy-
drology 2009) has provided a particularly rich dataset for common guillemots,
including: annual counts, productivity estimates, mark-recapture and ring- recovery time series, data on chick diet and growth rates, laying dates, colony attendance patterns, etc. As a result this colony has been extensively stud- ied (see, for example, references above and in later chapters), but most analyses have focused on a single aspect of guillemot biology and, until now, no attempts have been made to integrate the different data sources and model the complex dynamics of the population.