Appendix Three

Objectives 1 and 2: to quantify the optimum spatial scale and temporal provision of winter food resources in representative agricultural landscapes; to quantify dispersal patterns for a representative range of species and link them with winter/early spring food provision.

Movements of resident farmland passerines between winter and breeding locations: East Anglian Experiment

Introduction

The scale of the dispersal between wintering and breeding locations of non-migratory farmland birds is a key unknown in the ecology of such species, if we are to link winter habitat and resource requirements to breeding populations. Scales of movement between winter and spring are also important, along with within-winter movement propensities, in determining the optimum separation for food resource patches in winter (see also Appendix 2). The key piece of information in this context is the average movement of an individual between its summer and winter locations, which, in practice, requires identifying and locating the same individuals in both seasons, i.e. catching birds in one season and recapturing or resighting in the other. The winter feeding experiment provides an unparalleled opportunity to capture large samples of birds in winter and to colour-mark them individually. This then potentially allows the birds concerned to be identified in breeding locations in summer, when large-scale recapture effort would be impractical.

Winter ringing and summer resighting provide two general approaches to investigating winter-to-breeding movements. First, estimates of total local population size in each season and the proportion of those that are ringed allows estimation of rates of dispersal out of the focal area, in the context of gross changes in abundance. Second, distances between ringing locations and breeding territories show absolute extents of movement and allow estimation of probabilities of occurrence of breeding ringed birds with respect to distance from winter locations. The latter estimates can then be transformed to produce estimates of proportions of the winter population that travel given distances and of distances at which a breeding bird has a given probability of using a winter food resource.

The results assist in identifying the scale over which winter resource provision operates in terms of benefits provided to local breeding birds.

Methods

Winter surveys

All farmland habitats in each study tetrad were surveyed in October or November each winter and all potentially seed-rich habitats (crop stubbles, game cover/bird seed crops and set-asides) and other “hotspots” where flocks of the target species were found were then subsequently re-surveyed each month from December to March (inclusive). In 2005 and 2006, additional surveys of seed-rich habitats were conducted in November in a subset of tetrads. The survey method was modelled on the BTO’s Winter Farmland Bird Survey (Gillings et al. 2005), walking the boundaries of all non-seed rich habitat fields and carrying out whole area searches in stubble, cover and set-aside fields.

Counts from January to March each winter were used as a measure of the size of the local winter populations sampled for ringing because almost all trapping activity took place during this period. Counts from earlier in the winter were likely to differ both because of mortality and because the availability of high quality habitats, such as stubble fields, changed considerably from month to month. Maxima of the three late winter monthly counts were used as the best available guide to the population sampled. This was because it was clear during ringing sessions that flocks were mobile such that “snapshots” of local abundance were unlikely to consider all birds using a feeding location and that a proportion of the local population were likely to be outside the focal tetrad at any given time. This pattern was supported by occasional catches of more birds than had been found on the most recent survey of the local area concerned.

Winter ringing

Ringing took place each winter from 2004-05 to 2006-07, often at several ringing sites within each study tetrad in each winter. Ringing activity was designed to catch a good proportion of the chaffinches and yellowhammers present in a tetrad in winter, so effort was generally roughly proportional to local abundance.

A number of different trapping methods were used, adapted according to habitat context and bird behaviour in each area, but principally, mist-nets placed parallel to or in gaps in hedges, baited walk-in traps and elastic-powered “whoosh” nets. In fed areas, most trapping activity was focused on food patches themselves, which acted as long-term bait for bird flocks and facilitated trapping. In control areas, flocks of target species were located during the monthly surveys and short-term targeted baiting (using the same seed types as used for the main feeding activity) was used to concentrate these flocks in particular locations that allowed trapping. Seed bait was only supplied for long enough for the birds to find and use it and was not replenished after trapping occurred to minimize the blurring of the fed-unfed experimental dichotomy. Similar targeted baiting and trapping was also used in some fed areas, where concentrations of birds were found away from permanent food patches.

Trapping was initiated once flocks had built up at feeding sites, mostly from January to March each winter. A few trapping days also occurred in December and the first week in April. Some species will have begun to breed towards the end of the period used for ringing, potentially biasing some analyses of movements. However, trapping was only conducted where birds were found in flocks foraging on seed, i.e. exhibiting what can be considered to be “winter behaviour”. Further, although chaffinches were regularly observed exhibiting “breeding season behaviour” in March in the study areas, this was rare for yellowhammers, which seemed to start breeding in late April or May.

All chaffinches, yellowhammers, goldfinches and reed buntings caught were fitted with individual colour-ring combinations consisting of one plastic colour-ring and a BTO metal ring on one leg and two colour-rings on the other. In almost all cases, combinations were organized such that the colours and ring positions on the “non-metal” leg were a unique identifier, alone, for the site concerned. As well as the birds ringed during the current study (2004-05 to 2006-07), additional, individually colour-ringed individuals were present in the local populations at three sites as a result of previous research there in 2002-03 and 2003-04 (BD1616).

Breeding season surveys and resighting

Transect surveys based on the BTO/RSPB/JNCC Breeding Bird Survey (BBS) were used to identify overall changes in relative abundance using a standardized technique (see Appendix 9 for details). Registrations of target species from these surveys were mapped for each tetrad and combined with data on territory locations derived from intensive searches for ringed birds. The latter consisted of survey visits to each tetrad in which observers walked along all field boundaries in each tetrad during May and June, recording the locations and behaviour of all yellowhammers, chaffinches, goldfinches and reed buntings encountered, with the aim of locating all breeding pairs present. Suitable habitats in each tetrad were covered in this way twice each year. Every effort was made to record whether each individual seen was ringed and, if so, what its individual colour-ring combination was. This process often involved focal observation of individuals for up to 30 minutes, during which time the presence of any other pairs nearby was often revealed (e.g. by other males singing). In addition, song playback was used to draw out concealed territorial birds, which also occasionally had the effect of eliciting responses from other birds nearby. Territory locations revealed by these surveys were mapped, in combination with locations found during the breeding bird survey visits to measure total, absolute, local population sizes in each study area. (The breeding bird survey bird locations were also used to inform the targeting of intensive search effort during the later visits to tetrads, indicating, for example, where birds found during transect surveys that had been seen during the initial intensive searches represented real, additional territories.) After the end of the fieldwork period each year, the composite maps of bird locations from all tetrads were analysed by a single observer, making use of clustering of bird locations from different visits, simultaneous records of birds on adjacent territories, relevant information on behaviour (singing, carrying food or nesting material, etc.) and records of individually colour-ringed birds to estimate the total number of territories present. Despite the intensive search effort employed (per visit, greater than that used in the BTO’s Common Birds Census [Marchant et al. 1990], for example), it was inevitable that uncertainties remained in total territory numbers: for example, adjacent unringed males might never have been heard singing simultaneously or a bird might have been recorded only during an early-season transect survey (perhaps subsequently moving, dying or simply not being detected). Total local territory numbers were, therefore, calculated as minima and maxima possible, together with the “best estimates”.

The intensive searches of each study area revealed the ringed/unringed status of as many as possible of the birds present and the colour-ring combinations work by the ringed birds revealed where they had been ringed. However, it was apparent in the first year of the study that, unless late winter mortality was extremely high, many ringed birds had dispersed out of each focal tetrad. Intensive searches outside the study areas were logistically impossible (search area scales as the square of the distance from origin points), so a sampling approach was undertaken in 2006 and 2007. This involved following two or more transect routes of up to 10km in length through suitable (farmland) habitat in the areas around each tetrad in June-July each year. Yellowhammers and (especially) chaffinches are likely to have been less detectable at this time of year, but this was accounted for implicitly in the methods used for analysis (see below) and was unavoidable because of logistical constraints on fieldworker time. While following each transect walk, all yellowhammers, chaffinches, goldfinches and reed buntings encountered were recorded on a map and watched until ringed status and (if applicable) colour-ring combination had been determined or the bird had been lost. Search effort was separated from time spent trying to determine status and combinations.

These out-of-tetrad transects, in combination with the within-tetrad records, provided data on ringed versus unringed status at distances up to around 8km from a focal tetrad. However, data from other nearby tetrads also provided information on potential movements from a focal tetrad because birds ringed in the latter would have been identifiable as such if they were sighted in another tetrad. This meant that sampling for ringed birds outside tetrads effectively covered distances up to around 100km from each study area (although the probability of locating ringed birds will, a priori, have fallen rapidly towards zero at the larger distances).

Analysis

i)Proportions ringed by season

The monthly surveys between October/November and March provided estimates of total winter populations of chaffinches and yellowhammers in each tetrad. Total numbers of individuals trapped or resighted provided an estimate of the proportion of each local population that was carrying rings in each winter. Total breeding population estimates derived from the annual transect surveys and intensive searches, combined with totals of ringed birds present in each tetrad also derived from the latter, then provided estimates of the proportion ringed during the breeding season. Differences between absolute winter and spring population sizes showed gross apparent emigration (or mortality) or immigration. The differences between the two proportions constituted an estimate of proportions “migrating” out of an average 2×2km tetrad between seasons in each year, as follows: the actual number of ringed individuals present in spring RactS can be estimated as

RactS = RobsS × (Ts / Nseen),

where RobsS is the number of ringed individuals observed in spring, Ts is the total number of individuals in a tetrad in spring and Nseen is the number of individuals with known ringed status in spring.

Also, assuming zero dispersal,

RactS = Pw × Ts,

where Pw is the proportion of the local population that was ringed in winter.

Combining these equations,

RobsS = Pw × Ts × (Nseen / Ts) = Pw × Nseen = RpredW,

where RpredW is therefore the number of ringed birds present, as predicted by winter ringing.

If RpredW were greater than the observed number of ringed birds present RobsS, the difference would reflect the number of birds dispersing out of a focal square between winter and spring. If the observed value were lower, it would suggest a disproportionate “emigration” of unringed birds, which, in practice, would probably reflect biased sampling by the winter trapping, for example if winter counts were bolstered by transient flocks or if two independent flocks with different dispersal behaviours were present in winter but only one of these were subject to trapping.

ii)Absolute seasonal “migration” distances

All breeding season locations of ringed, unringed and unknown status individuals and all ringing sites and locations where resighting was conducted (mostly permanent food patches) were digitized using the ArcMap GIS system (ESRI 2004). Absolute distances moved were measured by comparing locations of winter ringing or sighting with breeding territory locations. Breeding season observations were only used as indications of a movement from a winter location if associated with a previous winter location, i.e. birds located in two breeding seasons were only used as two data points if they were seen in both preceding winters. Birds ringed, for example, in winter one and not seen in breeding season one or winter two, but then found in breeding season two were considered to provide a single movement data point. Repeat observations of individuals in paired winters and breeding seasons were treated as repeated measures in analyses.

A second analysis considered variation in the probability that a bird sighted in summer would be ringed with respect to distance from a ringing site. This probability can be considered to be equivalent to that that a bird found in winter will establish a breeding territory at a given distance away. Each record of an individual in the breeding season constituted a data point for a distance from each individual ringing site, so ring statuses of summer records were re-assigned with respect to each ringing site in turn: birds not ringed at a given focal site were assigned an “unringed” status for that site and all others apart from the one at which it had been ringed. Note that only one individual was caught or sighted in two different tetrads in winter.

These data were analysed in two ways, each method being used separately for each year. First, distances were calculated from all summer locations (ringed or unringed) to all ringing sites. This meant that each individual contributed to the data many times, because there were often multiple ringing sites within single tetrads, but that all movements of ringed individuals were real, i.e. corresponding to actual ringing sites. The second treatment of the results involved calculating a single, average centroid ringing location for each tetrad as the mean of the x- and y-coordinates of the ringing sites in the tetrad, weighted by the number of birds of the focal species ringed at each site in the year in question. All distances were then calculated with respect to these centroids and separate analyses of variation with respect to distance were conducted for each tetrad.

Both forms of the status-distance data were analysed using non-linear logistic regression, assuming a binomial error distribution, using the NLMIXED procedure of SAS, with checks also being run using logit-linear models (using the GENMOD procedure: SAS Institute, Inc. 2001). The logit link function allowed the modelling of probabilities directly from binary data on ringing status. Non-linear modelling allowed a range of function forms for the variation in the probability that a bird would be ringed with respect to distance whilst still producing defined functions that would allow prediction in other contexts (unlike a non-parametric fit). The functions considered were a linear decline, an exponential decay function, a logistic curve and a sigmoidal curve. These curves were defined using the following formulae, where y denotes the logit of the probability that a bird at distance x from a ringing site had been ringed at that site and each of a, b, c and d denote parameters estimated to fit each function:

Linear:y = c + (b×x);

Exponential:y = c + (a× e(b× x));