Received 19 November 2003 Accepted 17 December 2003 Published online 22 March 2004 Nestling immunocompetence and testosterone covary with brood size in a songbird Marc Naguib1*, Katharina Riebel2, Alfonso Marzal3 and Diego Gil4 1Department of Animal Behaviour, University of Bielefeld, PO Box 100131, 33501 Bielefeld, Germany 2Behavioural Biology, Institute of Biology, Leiden University, PO Box 9516, 2300 RA Leiden, The Netherlands 3Departamento de Biolog��a �� Animal, Facultad de Ciencias, Universidad de Extremadura, Avenida Elvas s/n 06071 Badajoz, Spain 4Departamento de Ecolog��a �� Evolutiva, Museo Nacional de Ciencias Naturales (CSIC), Jose �� Gutierrez �� Abascal, 2, 28006 Madrid, Spain The social and ecological conditions that individuals experience during early development have marked effects on their developmental trajectory. In songbirds, brood size is a key environmental factor affecting development, and experimental increases in brood size have been shown to have negative effects on growth, condition and fitness. Possible causes of decreased growth in chicks from enlarged broods are nutritional stress, crowding and increased social competition, i.e. environmental factors known to affect adult steroid levels (especially of testosterone and corticosteroids) in mammals and birds. Little, however, is known about environmental effects on steroid synthesis in nestlings. We addressed this question by following the development of zebra finch (Taeniopygia guttata) chicks that were cross-fostered and raised in different brood sizes. In line with previous findings, nestling growth and cell-mediated immuno- competence were negatively affected by brood size. Moreover, nestling testosterone levels covaried with treatment: plasma testosterone increased with experimental brood size. This result provides experimental evidence that levels of circulating testosterone in nestlings can be influenced by their physiological response to environmental conditions. Keywords: challenge hypothesis immunocompetence nutritional stress sibling competition testosterone zebra finch 1. INTRODUCTION The social and ecological conditions that individuals experience during early development have profound effects on their development, survival and reproduction (Roff 1992 Stearns 1992). Environmental factors affect both an organism���s development and its adult condition, often through physiological changes triggered by a stress response. Nutritional stress during early development has been shown to affect growth, physiology, social behaviour and reproduction (Metcalfe & Monaghan 2001) thereby affecting fitness via a large array of traits. Songbird chicks undergo extremely fast development during the first two weeks in the nest, and experimental enlargements of brood sizes result in reduced growth, immunocompetence and survival (Tinbergen & Boerlijst 1990 Saino et al. 1997 Brinkhof et al. 1999). Such effects could result from nutritional stress through ecological competition (fewer resources for each nestling) or from increased social com- petition, such as increased begging and aggression (Neuenschwander et al. 2003). Indeed, recent experi- ments showed an energetic cost of begging (Kilner 2001) and that increased brood size leads to increased begging and social competition (Neuenschwander et al. 2003). This combined evidence suggests that nestlings in larger broods are exposed to increased competition both indirectly and directly because of the presence of their nest mates. * Author for correspondence (marc.naguib@uni-bielefeld.de). Proc. R. Soc. Lond. B (2004) 271, 833���838 833 ��� 2004 The Royal Society DOI 10.1098/rspb.2003.2673 In adult vertebrates, competitive abilities and the regu- lation of aggression are linked to levels of circulating tes- tosterone (T), which in turn have been shown to covary with the levels of social competition experienced (Harding & Follett 1979 Wingfield et al. 1990 Oliveira et al. 1996). T is already present in nestlings in passerines (Woods et al. 1975 Adkins-Regan et al. 1990) and thus, like adult T levels, may covary with the social and ecologi- cal environment, i.e. experimental brood size. Studies so far have shown that experimental increases in T in the egg increase post-hatching growth and begging behaviour (Schwabl 1996). However, little is known about the direct and indirect causes of variation in levels of T produced by developing nestlings and to what extent T relates to brood size and associated levels of competition during nestling development. Competition between siblings for parental care is the first instance of social competition an organism encounters after birth (Mock & Parker 1997). Conse- quently, physiological adaptations that allow nestlings to compete in conditions of nutritional deprivation should be strongly selected for, as has been shown for effects of nutritional stress on secretion of corticosterone (Kitaysky et al. 1999, 2001). Here, we examine the effects of brood size on nestling biometry, immunology and circulating T levels by manip- ulating brood sizes in zebra finch nestlings. We cross-fos- tered nestlings to randomize genetic and parental effects and to create broods containing different numbers of nes- tlings within the naturally observed range for brood size in this species (Zann 1996). Based on previous findings, we predicted that chicks should show reduced growth and

834 M. Naguib and others Immunocompetence, testosterone and brood size in a songbird immunocompetence with increasing brood size (de Kogel & Prijs 1996 Saino et al. 1997). If the increased brood size results in increased social stress and social com- petition, then T levels should increase with increasing brood size. 2. METHODS (a) General We conducted the experiment at the University of Bielefeld, Germany, in 2001. From a laboratory stock of wild-caught Aus- tralian zebra finches (F4 generation) groups of four non-sib males and females were distributed across 21 indoor aviaries (0.92 m �� 1.8 m �� 1.85 m or 1 m �� 3 m �� 3.3 m). These were made from plywood with mesh wire roofs and front panels, thus allowing acoustic but not visual contact between birds in adjacent aviaries. All rooms had an air temperature of 22���25 ��C, 75���85% relative humidity and a 16 L : 8 D regime. Birds were kept under ad libitum feeding conditions throughout the experi- ment. They were supplied daily with both dried and germinated senegal, plata and red millet seeds, and fresh water (with added vitamins three times a week). Each aviary was equipped with five wooden nest-boxes (12.5 cm �� 12 cm �� 14 cm) and with coconut fibres as nesting material on the floor before the birds were introduced. We regularly observed the adult birds to deter- mine which pair bred in which nest-box. During egg laying, egg shells were supplied as an additional calcium source. After the chicks hatched, egg food mixture was supplied daily. (b) Cross-fostering and brood-size manipulations Nests were checked daily between 10.00 and 13.00, and eggs were labelled with a felt pen on the day of laying, allowing us to keep track of egg laying order. Chicks were individually marked on the day they hatched by cutting their down feathers to individual patterns. These remained recognizable until day 10 when the chicks were colour ringed. At an age of 2 �� 1 days, chicks from the same brood were distributed across one to five different broods. These experimental broods consisted of two to six cross-fostered chicks with an age difference of maximally 2 days. Chicks in an experimental brood came from one to six different original broods. Broods of two (n = 10 broods) or three (n = 6 broods) nestlings were defined as small broods, broods of four nestlings (n = 10 broods) were treated as medium broods, and broods of five (n = 2 broods) or six nestlings (n = 6 broods) as large broods. Experimental brood sizes were evenly distrib- uted among original brood sizes. Overall 34 manipulated (experimental) broods (brood size 3.6 �� 1.4 chicks, mean �� s.d.) with a total of 123 nestlings were created from the 50 original broods (brood size 3.2 �� 1.2 chicks). Large broods are more likely to contain some extremely undernourished nestlings. As these birds may contribute disproportionately to differences among brood sizes, we ran the analyses including all chicks and then repeated the analyses excluding chicks (n = 12) that died before the age of 35 days, i.e. before reaching nutritional inde- pendence. The results of both analyses showed the same signifi- cant effects. Thus, we report the outcome of only the second more conservative analysis. (c) Data collection (i) Biometry Biometric measurements were taken throughout development and from adult birds: on the day of brood-size manipulation, at 5 and 10 days post-hatching, at 35 days (independence), 90 days Proc. R. Soc. Lond. B (2004) (sexual maturation) and nine months. Nestling body mass was measured with an electronic scale (Sartorius PT 120) to the nearest 0.01 g and adult body mass with a Pesola scale to the nearest 0.5 g. Tarsus length was measured with callipers to the nearest 0.01 mm and wing length with a ruler (flattened wing) to the nearest 1 mm. At the age of 10 days individuals were also marked with unique combinations of white or orange numbered plastic rings. To keep disturbance of nests to a minimum we calculated the days on which measurements had to be taken based on the average age for all nestlings within a given nest. However, for statistical analysis, the exact age of each chick was included in the model. Body condition was defined as the effect of body mass when a measure of size (i.e. tarsus length) was considered as a covariate in the model (Garc��a-Berthou �� 2001). Thus, this measure of body condition reflects the relative weight at a given size. (ii) Immune response Cell-mediated immune response at an age of 10 days was esti- mated using a standard phytohaemagglutinin (PHA) injection. We injected 0.05 ml of a buffered solution of PHA (0.025 mg) into the left wing web, and measured its thickness using a micrometer before the injection and again after 24 h (Smits et al. 1999). The strength of the immune response is reflected by the thickness of the wing-web swelling (Lochmiller et al. 1993 Merino et al. 1999). Blood smears were used to estimate the total numbers and the proportions of different types of leucocytes under magnification, ��1000. The proportions of different types of leucocytes were assessed on the basis of an examination of a total of 50 leuco- cytes other than thrombocytes (thrombocytes present an aggre- gated distribution). The number was quantified as number of leucocytes per 50 fields examined. The heterophyle : lymphocyte (H : L) ratio was estimated from the percentages of heterophyles and lymphocytes per 100 leucocytes obtained in these counts (Merino et al. 1999). All leucocyte proportions were Cox���Box transformed to obtain normal distributions (Sokal & Rohlf 1995). (iii) Levels of testosterone We collected blood samples from the brachial vein at the age of 11 days. Chicks were taken from the nest as a group and placed in an artificial nest from which chicks could be taken one by one without disturbing the remaining chicks. The handling for blood sampling took ca. 30 s, and immediately afterwards chicks were placed back into their original nest. To minimize overall disturbance to the brood we took blood samples from nestlings only once. Blood samples were centrifuged at 3000 r.p.m. for 10 min. Plasma was stored separately from the blood cells at 18 ��C for later analysis of hormone levels. Nes- tling plasma T levels were determined using a highly sensitive ELISA kit according to the manufacturer���s instructions (Cayman Chemicals, Ann Arbor, MI, USA). We obtained plasma samples from a total of 77 nestlings across the three experimental treatments. Plasma samples of 25 ��l diluted in 25 ��l of buffer were analysed with respect to a standard curve (between 3.2 and 100 pg ml 1). The antibody used was highly specific to T with little cross-reactivity to other steroids (less than 5%), except for 5��-dihydrotestosterone (21%). Owing to the small quantities of plasma available from the nestlings, we were not able to run duplicate samples. However, intra-assay and inter-assay validation were small enough to ensure reliability without duplicates (intra-assay coefficient of variation = 9.1