Diet of Black-legged Kittiwakes (Rissa tridactyla) feeding chicks at two Irish colonies highlights the importance of clupeids

Abstract

Capsule We examined regurgitates from Black-legged Kittiwakes during the early chick-rearing period over two breeding seasons at two colonies in Ireland where diet has not been studied previously. Clupeids were the dominant food source at both colonies, which contrasts with other studies throughout the British Isles that suggest Kittiwakes feed mainly on sandeels during this period. Our study is limited by sample size and restricted to the early breeding season, but suggests a link between reproductive success and dietary composition which should be investigated further.

The Black-legged Kittiwake Rissa tridactyla (hereafter Kittiwake) has had low reproductive success at many colonies throughout the British Isles in recent years (JNCC 2011a). Low productivity has been principally attributed to low food availability (Frederiksen et al. 2004, JNCC 2011a). Research into the effects of food availability on Kittiwake breeding has occurred primarily in the northwestern North Sea (e.g. Hamer et al. 1993, Frederiksen et al. 2006) where birds feed mainly on the Lesser Sandeel Ammodytes marinus (Furness & Tasker 2000, Lewis et al. 2001). Elsewhere in the British Isles there is a paucity of data on Kittiwake diets (Frederiksen et al. 2007, Swann et al. 2008) and in regions such as western Scotland or the Irish Sea, where aggregations of sandeels Ammodytes spp. are smaller and more isolated than those in the North Sea, Kittiwakes may exploit other prey types (Frederiksen et al. 2005).

We examined the diet of Kittiwakes during the early chick-rearing period at two colonies in Ireland where diet has not been studied previously. At Rathlin Island, Northern Ireland (55^o17′32′′N, 06^o11′30′′W), the colony has exhibited variable breeding success since 2005, with seasons of complete breeding failure (Chivers 2008, Chivers et al. 2012a). In contrast, at Lambay Island, Republic of Ireland (53^o29′30′′N, 06^o01′00′′W), the colony has exhibited high reproductive success that has remained stable during the same period (JNCC 2011b, Chivers et al. 2012a). It is suggested that this difference is related to variability in food availability (Chivers et al. 2012a) and thus we discuss chick diet in relation to reproductive success.

Birds were caught at the nest using a noose-pole during the early chick-rearing period when chicks were < 3 weeks old between 3 and 19 July during 2009 and 2010. Regurgitates produced in response to being caught and handled were stored in 70% ethanol. To assess frequency of occurrence, all prey remains were identified by means of their vertebrae following the methods of Watt et al. (1997). We were unable to distinguish between species using vertebrae and so categorized prey as either (i) clupeids, (ii) sandeels (Ammodytes spp.) or (iii) gadoids (cod fishes) for comparison with similar studies (Harris & Wanless 1997, Lewis et al. 2001, Bull et al. 2004, Swann et al. 2008). Prey was classed as ‘present’ when vertebrae were identified in a sample. Intact partial sections of clupeid or sandeel vertebral column were used to calculate fish length by measuring the most anterior caudal vertebrae and using the formulae: lnTL = 4.252 + 0.962*lnVL for clupeids and TL = 100(VL)/1.5 for sandeels, where TL equalled total length (mm) and VL equalled vertebrae length (mm) while ln represented a transformation by natural logarithm (Watt et al. 1997). All whole fish were identified to species level (Muus & Dahlstrom 1974).

Provided sampling biases are equal, the comparative success of obtaining a regurgitate from a bird can indicate food availability (Lewis et al. 2001). Thus we compared differences between Rathlin and Lambay during 2009 and 2010 using Chi-squared tests adjusted for Yates' correction. Dietary data were expressed as the frequency of occurrence of each fish type within regurgitates and by birds (where birds were caught on more than one occasion). Variability in the occurrence of clupeids was examined between colonies and years using a Generalized Linear Mixed Model (GLMM) assuming a binomial error structure and a logit link function. We fitted the occurrence of clupeids (a binary response of present or absent) as the dependent variable, colony, year and their interaction (colony*year) as fixed factors, the time of day at which birds were caught (minutes from midnight) as a covariate and individual bird treated as a random factor to account for multiple regurgitates per bird. Variability in clupeid size was also examined between colonies and years using a GLMM assuming a Gamma error structure and a reciprocal link function. We fitted clupeid length (mm) as the dependent variable, colony, year and their interaction (colony*year) as fixed factors and individual bird as a random factor. In all models the influence of each term was described by the F statistic taken as the Wald value divided by the degrees of freedom which corresponded to a χ² distribution (Quinn & Keough 2002). All statistical analyses were performed using Genstat 6.0^© (VSN International Ltd, Hemel Hempsted, UK). Statistical analyses were not performed on sandeels or gadoids because sample sizes were small.

In total, 48 regurgitate samples were produced by 95 Kittiwakes caught. At Rathlin Island, a total of 21 regurgitates were obtained from 34 catches during 2009. Significantly fewer birds produced regurgitates during 2010, when only 4 regurgitates were obtained from 17 catches (χ² ₁ = 5.2, P = 0.023). The number of regurgitates produced by birds at Lambay did not differ between years with a total of 8 regurgitates obtained from 17 catches during 2009 and 15 regurgitates obtained from 27 catches during 2010 (χ² ₁ = 0.06, P = 0.81). Overall there was no difference in the number of regurgitates produced by birds at Rathlin or Lambay (χ² ₁ = 0.01, P = 0.91).

Clupeids were the most frequently recorded prey item at the two colonies (Table 1). Clupeids occurred at a higher frequency in regurgitates recovered from birds at Rathlin compared to Lambay (F _1,43 = 9.38, P = 0.002, n = 48). There was no effect of Year (F _1,43 = 0.7, P = 0.41), time (F _1,43 = 0.02, P = 0.90) or interaction of Colony*Year (F _1,43 = 1.04, P = 0.31). Sandeels were recorded in samples from just two birds at Rathlin during 2009 and did not occur at all during 2010. Conversely, the frequency of sandeels in the diet of birds from Lambay increased substantially between 2009 and 2010. Gadoids were absent from the diet samples of birds at Rathlin in both years but were recovered from approximately 30% of birds (and regurgitates) at Lambay.

Table 1. Frequency of occurrence of prey items in regurgitates of Kittiwakes during the early chick-rearing period at (a) Rathlin Island, Northern Ireland and (b) Lambay Island, Republic of Ireland in (i) 2009 and (ii) 2010. Sample sizes are given as numbers of birds (b) and numbers of regurgitates (N). Values are given in each column for numbers and percentages are given in parentheses. Data on reproductive success have been extracted from Chivers et al. (2012a).

				Frequency of occurrence (%)
				Clupeids		Sandeels		Gadoids
Colony	Year	(b)	(N)	b (%)	N (%)	b (%)	N(%)	b (%)	N (%)	Reproductive success (chicks fledged nest^−1)
(a) Rathlin	(i) 2009	20	21	20 (100.0)	21 (100.0)	2 (10.0)	2 (9.5)	0 (0.0)	0 (0.0)	1.11 ± 0.11 (N = 37)
	(ii) 2010	3	4	3 (100.0)	4 (100.0)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0.45 ± 0.08 (N = 38)
	Subtotal	23	25	23 (100.0)	25 (100.0)	2 (8.7)	2 (8.0)	0 (0.0)	0 (0.0)	0.78 ± 0.10 ( N = 75)
(b) Lambay	(i) 2009	8	8	8 (100.0)	8 (100.0)	1 (12.5)	1 (12.5)	3 (37.5)	3 (37.5)	1.62 ± 0.08 (N = 50)
	(ii) 2010	12	15	8 (66.7)	11 (73.3)	9 (75.0)	10 (67.7)	3 (25.0)	4 (26.7)	0.93 ± 0.05 (N = 390)
	Subtotal	20	23	16 (80.0)	19 (82.6)	10 (50.0)	11 (47.8)	6 (30.0)	7 (30.4)	1.28 ± 0.07 ( N = 440)
	Grand total	43	48	39 (90.7)	44 (91.7)	12 (27.9)	13 (27.1)	6 (14.0)	7 (14.6)	1.03 ± 0.08 ( N = 515)

All 19 whole fish recovered were European Sprat Sprattus sprattus. The calculated length of clupeids ranged from 10 to 100 mm with the majority being 40–50 mm. However, there was a significant interaction of Colony*Year (F _2,347 = 3.98, P = 0.046). Specifically, the length of clupeids taken by birds at Rathlin during 2009 was shorter than those taken by birds at Lambay during the same year, but the converse was true during 2010. The mean length of clupeids at Rathlin during 2009 was 49.6 ± 0.1 mm (n = 214) and during 2010 it was 47.9 ± 1.3 mm (n = 50) compared to Lambay during 2009 where it was 51.9 ± 2.0 mm (n = 33) and during 2010 it was 44.0 ± 1.7 mm (n = 54). Sandeels varied from 20 to 60 mm with the majority being 30–55 mm. The mean length of sandeels at Rathlin during 2009 was 40.0 ± 2.1 mm (n = 5) and at Lambay during 2009 it was 20.0 mm (n =1) and during 2010 it was 41.3 ± 1.7 mm (n = 39).

Black-legged Kittiwakes fed on a range of prey including clupeids, sandeels and gadoids during the early chick-rearing period, but their frequency in the diet varied between colonies and between years within colonies. Clupeids, particularly European Sprat, dominated the diet of Kittiwakes breeding at Rathlin and Lambay Island during 2009 and 2010. Clupeids can form the main constituent of Kittiwake chick diets (during mid-June to early July) at estuarine colonies in the British Isles (80.2 % clupeids, N = 85 regurgitates, from two sites in the inner Firth of Forth, data from Bull et al. 2004), whereas at marine colonies, sandeels appear to be the predominant dietary item during the early chick-rearing period with a minor component consisting of clupeids in the majority of years (14.9 % clupeids, N = 470 regurgitates, data from the Isle of May: Harris & Wanless 1997, Lewis et al. 2001; the Farne Islands: Bull et al. 2004; and Canna: Swann et al. 2008). Thus, we suggest that Kittiwakes breeding at Rathlin and Lambay Islands are more reliant on clupeids during the early chick-rearing period than at marine colonies elsewhere in the British Isles.

Fewer regurgitates were recovered from birds caught at Rathlin during 2010 than 2009. This may be a result of factors such as the time that birds were caught after the last chick feed, but could also suggest that prey densities may have been low in 2010 (Lewis et al. 2001), which would support the supposition that poor breeding success was caused by resource availability (Chivers et al. 2012a). The diet of birds at Lambay was more varied than that of birds at Rathlin and the availability of prey sources other than clupeids during 2010 (sandeels and gadoids) may account for the higher reproductive success recorded at Lambay compared to Rathlin in this year.

Clupeids taken by Kittiwakes at Lambay during 2010 were smaller than those taken during 2009 and those taken at Rathlin in either year. It has been postulated that Kittiwakes select fish within a preferred size range (Coulson 2011) and thus if clupeids available during 2010 were not of the preferred size, it may have prompted the shift in diet. Sandeels taken by birds at both colonies were generally < 60 mm in length, indicating that they were probably 0 group (less than 1-year-old) sandeels which would be expected during the early chick-rearing period (Lewis et al. 2001, Wanless et al. 2004). Sandeels made up a large proportion of prey delivered to chicks of Razorbill Alca torda (Linneaus, 1758) and Common Guillemot Uria aalge (Pontoppidan, 1763) breeding at Rathlin during 2009 and 2010 (Chivers et al. 2012b). Sandeels were therefore clearly present but unavailable to Kittiwakes. This may be because sandeels were deeper in the water column and therefore inaccessible to Kittiwakes that feed in the top metre (Coulson 2011), or because Kittiwakes were foraging in different locations to auks.

Our results indicate that there may be a relationship between reproductive success and dietary composition. However, our study was based on data from only two colonies for a limited part of the breeding season and thus general conclusions are precluded. To thoroughly assess the relationship between reproductive success and dietary composition, similar studies need to be replicated over the entire range of the species and throughout the breeding season.

ACKNOWLEDGEMENTS

This project was funded by the Natural Heritage Research Partnership (NHRP) between the Northern Ireland Environment Agency (NIEA) and Quercus, Queen's University Belfast and supported through collaboration with the Royal Society for the Protection of Birds, Northern Ireland (RSPB NI) and BirdWatch Ireland (BWI). All fieldwork was conducted under Government licence and complied with the Queen's University Belfast Ethical Code of Conduct. We are particularly grateful to Maggie Hall (BWI) for assistance in the field and the Lambay Estate Trustees for permission to work on the island. Thanks also to Neil McCulloch who acted as NIEA Client Officer.