Oecologia (2001) 127:435–443 DOI 10.1007/s004420000596 Ola Olsson · Ulf Wiktander · Andreas Malmqvist Sven G. Nilsson Variability of patch type preferences in relation to resource availability and breeding success in a bird Received: 22 February 2000 / Accepted: 13 November 2000 / Published online: 8 February 2001 © Springer-Verlag 2001 Abstract This paper investigates how variability in par- overall food availability. This conclusion is strengthened tial foraging preferences for patch types can be used as a by two additional facts: the preference for lime corre- behavioral indicator of the energetic value of that patch lates positively (1) with the average giving-up density of type, and of overall food availability in the territory. The food, which has previously been shown to estimate over- species studied was the lesser spotted woodpecker all food availability in the territories, and (2) with repro- (Dendrocopos minor) and the patch types it uses are four ductive success, at least during the early stages of repro- groups of tree species (oak Quercus robur, birch Betula duction. pendula, B. pubescens, alder Alnus glutinosa, and lime Tilia cordata), in which it feeds upon wood-living insect Keywords Foraging preferences · Giving-up density · larvae. We partition the variation in foraging preferences Food availability · Reproduction · Dendrocopos minor into three scales. Firstly, within territories, the foraging preference for a tree species group was positively related to the prey density in that species group. That is, the Introduction preferences measure the patch types’ energetic profit- abilities. This result should be general in cases like the It is common to find that animals do not use resources in present, where the costs of using different alternatives do relation to their availability. Some are used more than not differ substantially. It may therefore be the preferred expected by chance, i.e., preferred, whereas others are behavioral indicator in determining the relative benefits used less (e.g., McDonald et al. 1990; Podolsky and associated with different alternatives. Secondly, between Price 1990; Crist and MacMahon 1992; Whitehead et al. the seven years of study, much of the variation in tree 1995). Many of the patterns of resource use can obvious- species group preferences was attributable to measured ly be thought of as species-specific (e.g., McDonald et fluctuations in the density of one important prey species al. 1990; Moser et al. 1990), and some are considered as (Argyresthia goedarthella, Argyresthidae, Lepidoptera), individual-specific characteristics (Sherratt and Mac- which occurred in some years on birch, in others on al- Dougall 1995). However, there is ample evidence that der, and in one year was virtually absent. Thus, in con- foraging preferences are variable within both species and cordance with the previous result, the values of these individuals in both space and time (Schooley 1994; tree species groups fluctuated between years according Arthur et al. 1996). Based on predictions from foraging to prey density. Thirdly, between territories, we found theory (MacArthur and Pianka 1966; Lucas 1983; that the preference for one tree species, lime, was higher Stephens et al. 1986; Mitchell 1989) we should expect in areas where it was more abundant. We attribute this to the extent to which a resource is used to be related to the the fact that the density (per patch) of at least one impor- fitness inputs an animal is expecting from it. If so, ecolo- tant prey species (Stenostola dubia, Cerambycidae, Cole- gists should be able to use foraging preferences as a be- optera) on lime increased with the abundance of its host havioral indicator of how the individual values different tree species in the territory. That is, the overall food alternatives. If both costs and gains may differ between availability was higher in territories where lime was alternatives, foraging preferences are likely to reflect the more common. Hence, the preference for lime estimates trade-off between these (Holbrook and Schmitt 1988) and may yield a complicated picture. In such cases, these O. Olsson (✉) · U. Wiktander · A. Malmqvist · S.G. Nilsson costs may best be evaluated applying patch use theory Department of Ecology, Animal Ecology, Lund University, Ecology Building, 223-62 Lund, Sweden (Brown 1988). However, in cases where costs are instead e-mail: [email protected] likely to be similar among alternatives the preferences Tel.: +46-46-2223817, Fax: +46-46-2224716 should straightforwardly reflect the foragers’ expected 436 energetic gains from using that alternative (Mitchell Olsson 1998; Wiktander et al., in press a). This food, 1989). Thus, foraging preferences could be a powerful however, seems to be relatively plentiful in all territories, and general way to estimate the energetic rewards that and does not seem to constrain food supply or create be- foragers gain from specific foods. So far, they have rare- tween-territory differences in breeding success. ly been used to make such inferences. In this paper, we will show how the patch type (tree In this paper we study the lesser spotted woodpeck- species group) preferences vary in a population of lesser er’s (Dendrocopos minor) preferences among groups of spotted woodpeckers. We will consider how the prefer- tree species (oak Quercus robur, birch Betula pendula, ences (within territories) for four tree species groups are B. pubescens, alder Alnus glutinosa, and lime Tilia cor- related to their respective prey densities, how these pref- data), that are likely to vary in energetic reward, but not erences vary between years with different food availabil- in energetic or predation-related costs associated with ity, and between territories between which reproductive using them. success varies. We will evaluate to what extent the forag- The lesser spotted woodpecker inhabits deciduous ing preferences can be used as a behavioral indicator to and mixed deciduous/coniferous forests in the Palearctic. the energetic value a tree species has to the woodpeck- From early autumn to late spring (August to May), it ers. In addition we will evaluate to what extent they can feeds almost exclusively on wood-living insect larvae in be used to estimate overall food availability in the terri- thin dead tree branches, mostly in living trees (Cramp tories, and how well they indicate territory quality, i.e., 1985; Olsson 1998). Most of the prey are larvae of Cer- reproductive success. ambycid beetles, such as Stenostola dubia (Ceramby- cidae, Coleoptera), and these are to a great extent special- ists on their host tree species (Palm 1959). An exception Methods to this is when it feeds on larvae of the moth Argyresthia goedarthella (Aryresthidae, Lepidoptera). These occur Study area abundantly in the years when either birch or alder bloom (Agassiz 1987). A. goedarthella eggs are laid in the male The study was conducted in a 150-km2 study area in the hemibo- catkins of the trees, but in spring when the flowers de- real vegetation zone in southern Sweden (56° 40′N, 14°10′E) be- tween 1990 and 1997. The landscape is dominated by coniferous velop the larvae move down to the lower regions of the forest, mainly spruce (Picea abies), but with deciduous forests and trees and bore into the trunk bark or into dead branches. agricultural land occurring patchily, particularly on richer soils Here they are accessible to the woodpeckers. and close to lakes. The forested parts of the study area are divided The variable occurrence of A. goedarthella, and per- into homogeneous stands (average size: 3.6 ha, SD 3.3), according to dominating tree species and age of the trees. Most non-conifer- haps also population fluctuations of other prey species, ous stands contain a mix of several tree species. The forests in the inevitably lead to the value of tree species fluctuating be- study area are generally managed for timber production, although tween years. However, prey density within and between many of the deciduous forests are cautiously managed, with some tree species may also vary between territories. consideration of conservation. In a previous study (Olsson et al. 1999), we used a technique based on giving-up densities (GUDs, i.e., the Estimation of foraging preferences number of prey left in exploited patches, cf. Brown 1988) to infer that the most important component of ter- Foraging behavior and breeding success were observed in 12 non- ritory quality of this species is the availability of food overlapping territories (average size 205 ha; Wiktander et al., in press b), in which a total of 68 different color-banded lesser spot- several weeks before egg laying. In depleting patches, ted woodpeckers occurred during the 7 years (1990–1996) when GUDs measure the forager’s intake rate when it decides foraging observations were made. Data collection started each to leave the patch. Therefore, GUDs contain information year on 21 March and lasted until the day when oaks burst into about the animal’s foraging costs (e.g., Brown 1988, leaf (3–20 May; Wiktander et al., in press a) as at that time the woodpeckers started to glean surface-living insects from the fo- 1999). By combining measures of GUDs with measures liage (Török 1990; Olsson 1998). of patch residence times and fitness it is possible to de- We collected the great majority of data from birds that were lo- termine which is the most important environmental or in- calized by radio tracking. We localized birds that remained with- dividual factor determining the variation in fitness out a radio transmitter by carefully walking through all parts of the territory, and adjacent areas. There we listened for drummings and among individuals (Olsson and Holmgren 1999; Olsson vocalizations, which are performed to the same extent by both et al. 1999). The foraging variables were measured dur- sexes. Once found, the birds can usually be followed for a period ing the pre-breeding period, i.e., when the woodpeckers (several minutes to 1 h), as they are moving through the forest.