Willow tit behavior
We studied free-living willow tits in Oulu, northern Finland (65°N, 25°30'E), collecting data on 79 days between August 10 2005 and May 3 2006. We collected behavioral data on average every other day, except for the period from mid-December to late February when no observations were made. The sample sizes for the different months were (in days): 14 (August), 14 (September), 14 (October), 11 (November), 2 (December), 0 (January), 0 (February), 9 (March), 13 (April) and 2 (May). The study population was color-banded, thus mostly allowing individual identification, at least for the winter period. Birds were banded under Finnish Ringing Centre License number 180. This is part of MO's long-term ongoing population study on willow tits. No birds were banded especially for our study. The population study's protocols are to put metal bands on local pulli in the nest during the breeding season. From August to November, a constant banding effort uses mist nets to color band every willow tit they can catch on the field site, thus keeping all birds individually marked. Due to the color banding effort, the number of individually identifiable birds increased as the season progressed. At banding, age was scored as “adult” or “juvenile” using the shape of the rectrices as criterion (Svensson, 1992). The behavioral data were collected by seeking out bird flocks in their natural habitats in an area of mixed woodland approximately 3 x 2 km in size, without the aid of food supplements or territorial song playback.
Observations took place between 0800 and 1500 hours with the majority before 1300 hours. Once located, a flock was followed for as long as possible, usually for about 10 to 20 minutes. Behavioral observations focused on a single individual (or occasionally two), following it in the binoculars (10x42) for as long as possible. Observations of under 15s were discarded from the analysis. The mean valid observation duration was 51s, with a range of 15 to 420s. The observations were dictated into a Sony IC recorder and later transcribed into spreadsheets. We strived to sample the behavior of all individuals in a flock (2-6) where possible; however, unmarked individuals were scored as 'unknown' and treated as one individual. During an observation we counted the number of occurrences of the different behaviors during that observation bout. An individual's behavior was scored as 'foraging' (searching), 'eating', 'hoarding' or 'other' (e.g. preening). We recorded single food handling events as either 'eating' or 'hoarding'.
We collected brains from adult willow tits and great tits between July 2006 and September 2007. Birds were captured under a license from the North Osthrobothnian Regional Environmental Centre. Willow tits were caught using mist nets, song playback and decoy birds in the woods around Oulu from adjacent populations that were not part of the long-term ringed study population. Great tits were caught in the town and suburbs of Oulu, using funnel traps baited with food. A few great tits were caught in the woods (1 in September, 2 in November and 2 in March) and a few willow tits were caught in town (2 in September and 3 in April). At capture, birds were kept in cloth bags for max 2h45min (mean: 1h09min, SD:31min) in preparation for processing. Birds were weighed and fat scored, and their wing length and tarsometatarsal length measured (Redfern & Clark, 2001). All birds were aged in the hand based on plumage. Great tits were sexed using the color and pattern of their plumage, and wing length if necessary, and willow tits by the observation of song production and wing length. Sex was confirmed after the dissection of the gonads. Birds were collected at 5 different times of the year: July 2006/2007 (WT: 3F/5M, GT: 4F/4M), August 2006/2007 (WT: 4F/3M, GT: 4F/5M), September 2006/2007 (WT: 4F/6M, GT: 4F/4M), November 2006 (WT: 3F/4M, GT: 2F/6M), and spring 2007. In the spring breeding season, great tits were collected between 24 March and 17 April 2007 (4F/4M), and willow tits were collected between 16 April and 22 April 2007 (2F/6M). The average timing of the first clutches in 2007 was 15 May for great tits, and 10 May for willow tits.
Birds were processed in the field in the back of a Land Rover®, especially equipped for this purpose. They were humanely killed by rapid decapitation. One hemisphere of the brain was immersed in 4% paraformaldehyde in PBS, while the other hemisphere was fresh frozen on dry ice. Which hemisphere was fixed and which was fresh frozen was alternated among birds, so half the birds for each time point, species and sex had the right hemisphere fixed, and half the left hemisphere. We only report on the fixed hemisphere in this study. After 48 hours of fixation, the hemispheres were cryoprotected in 30% sucrose solution, embedded in O.C.T. (Optimal Cutting Temperature compound for cryosectioning), frozen on dry ice and stored at -80°C. After all of the samples had been collected, they were shipped from Oulu to Newcastle. 70 µm coronal sections were cut on a cryostat (Microm HM560), and every other section was thaw-mounted onto gelatin-coated slides. The sections were stained with cresyl violet and coverslipped with Histomount®.
Microscopy and quantification
The person quantifying the slides (HL) was blind to the identity of the birds he was working on (although the species is easily determined from just looking at the sections). Slides were viewed using a Leica DM-LB microscope with a motorized stage, and connected to a computer running StereoInvestigator v7.5 (MBF Bioscience, Williston, VT, USA). The telencephalon was outlined at 2.5X magnification on every 8th section (560 µm apart). Nucleus rotundus was outlined every 4th section (280 µm apart). Finally, the HF was outlined every 4th section (280 µm apart), and cells counted using the Optical Fractionator in StereoInvestigator at 100X (oil immersion) magnification. The counting parameters were as follows: counting frame was 30x30 µm, sampling grid was 450 x 450 µm, and dissector height was 25 µm. To calculate volumes, measured areas were multiplied by the distance between measured sections and added up. We counted 2 cell types in the HF, identified as follows: neurons are defined as large cells with a clear (low staining) nucleus and obvious, darkly-stained nucleoli; small cells are smaller than neurons and have a darkly stained nucleus with or without obvious nucleoli. The small cells could be neurons or non-neuronal cells, like glia.
For the behavioral data, sometimes the same individual was observed for more than one observation in a given day. For the analysis, we combined all the observations of any given individual on a particular day and calculated the rate of eating and hoarding per hour of observation for that individual by dividing its total number of hoarding or eating events by its total observation time for that day. The per-hour rates for all individuals observed in a given day (on average about 9 individuals: 3 individuals in each of 3 flocks) were then averaged to give a daily hoarding and eating rate. These rates per hour were rounded to the nearest whole number to give a number of items per hour. Statistical analyses were performed with these daily rates as the unit of analysis using Generalized linear models (GzLM) in IBM SPSS 26® with a Poisson error distribution and the log link function. To analyze the neurobiological variables, we also used GzLM, using linear response variables. All anatomical variables were natural-log transformed to better reflect the log-log allometric relationships between parts of body and brain. Analysis of brain structures is always controlled for individual variation in either body size (tarsometatarsus length) or in telencephalon size in the analysis of Rt, HF volume, and cell numbers, by using that variable as a linear co-variate in the analysis. No interactions with the co-variate are included in the model. For the cell type analyses, we used Generalized Estimating Equations, with cell type as a within-subject factor. Otherwise, this was treated exactly as the other neurobiological variable analyses. The test statistic for these models is Wald’s Χ2.
BBSRC grant BB/C006186/1
- Psychology, Psychiatry and Neuroscience