WCP and BOMP Network sites

In the first year of BOMP, WCP selected 125 nest sites that they undertook to monitor annually. The WCP area of operations, most of southern and central England, was divided into five regions on the basis of administrative boundaries and 25 sites were selected from each region. Each site consists of a single nest box or pair of adjacent boxes, the latter designed to allow females to lay a repeat brood before the young from the first brood have fledged. Three nest box designs were employed: ‘Pole’ boxes mounted on stand-alone poles and ‘A-Frame’ boxes mounted on trees (Dewar & Shawyer 1996), plus a hybrid design used at sites in the south-west of England that was similar in terms of both box volume and siting location (on poles or trees). With the exception of the south-west, the relative proportion of box types across regions was held constant. Starting in 2002, an additional set of 75 sites were also selected, using similar criteria, and have been monitored by the WCP for BOMP each year.

Since 2002, a further 320–380 BOMP ‘Network’ sites have been monitored each year by volunteer recorders, who guarantee to participate in the project for a minimum of three years. Again, sites consist of either single or paired nest boxes, although the designs and dimensions are more variable than at WCP sites. Sites are selected by recorders and therefore reflect the distribution of the human population, but despite some resulting aggregation, most regions and habitats included in the species' UK range as indicated in the last Breeding Atlas (Gibbons et al. 1993) are reasonably well represented, with the possible exception of Wales and parts of northern Scotland (Fig. 1). While there are potential biases associated with monitoring schemes based on non-random site selection (Freeman et al. 2007), the geographical extent of coverage and the high proportion of the UK population included in the BOMP sample (c. 10%, assuming current population estimates are accurate) suggest that the results are likely to be representative of the population as a whole.

Another potential criticism of BOMP methodology is the focus on artificial nest sites, which, while they permit collation of accurate breeding data and facilitate the capture of both juveniles and adults, may not be representative of natural sites in terms of occupancy probability or productivity. However, many ‘natural’ nest sites utilised by Barn Owls in the recent past have been provided by human activity. In addition, the increasing trend for barn conversions and the loss of mature trees from the British countryside is potentially reducing the availability of ‘natural’ sites and increasing the species' dependency on boxes. Project Barn Owl estimated that there were some 25 000 boxes in the UK (Toms et al. 2000) and noted that 40% of Barn Owl breeding attempts located by fieldworkers involved pairs using nest boxes, compared with 30% in buildings and 30% in trees, although the sample size was small (n = 97). Again, trends identified by BOMP are therefore likely to be broadly representative of the UK population.

Fieldwork

A minimum of two visits to each site are made per year, the first between April and mid-May, when initial breeding attempts are likely to be underway, and the second between mid-July and early August, when first broods are at the point of fledging and some second broods may be initiated. Observers are encouraged to visit sites more frequently where possible, in order to increase the accuracy of any productivity data collected. Barn Owls tend not to be easily disturbed by careful fieldwork (Percival 1990, Taylor 1991). Several long-term studies of the breeding biology of Barn Owls indicate that monitoring active nest sites is unlikely to bring about desertion (de Bruijn 1994, Taylor 1994). Taylor (1991) found no significant differences in productivity between nests visited only at the late chick stage and those that received multiple visits, and Percival (1990) also failed to find any effect of visit rate on the number of fledged chicks when analysing BTO Nest Record Scheme data. Taylor (1994) also noted that site fidelity was high, with only 0.9% of males and 5.6% of females changing nest sites between consecutive breeding seasons. Nest site inspections are therefore unlikely to compromise the welfare of Barn Owls, or the integrity of the data gathered, provided that they are carried out carefully following the protocol outlined in the BOMP Barn Owl Fieldwork guidance notes.

Figure 1.

Distribution of Wildlife Conservation Partnership (black triangle) and Bam Owl Monitoring Program Network (grey circle) sites monitored in 2005 in the United Kingdom.

At each site the presence/absence of Barn Owls and their breeding status are recorded. Nest site occupancy provides a minimum estimate of Barn Owl abundance in a specified area, as non-breeding individuals or pairs and those breeding in unmonitored sites may be overlooked. However, given their high degree of nest site fidelity (Taylor 1994), it is reasonable to assume that changes in observed occupancy rates can provide useful information about the species' status and population trends. In years of particularly abundant prey or favourable weather conditions, birds may additionally breed in areas of poor quality habitat where nest box provision, and monitoring, is lower, resulting in a nonlinear relationship between the size of the ‘true’ population size and that of the monitored population. There was relatively little evidence of this happening in the area intensively surveyed by Shawyer, however, and variation in nest site quality within the UK is generally less marked than in some continental populations. This said, determining the extent to which inter-annual variation in occupancy is related to site quality would be a useful further validation of this method

Once an active nest has been located, the number of live and dead eggs and chicks present at each visit are recorded following BTO Nest Record Scheme (NRS) protocol (Crick et al. 2003), and the stage of nestling development is noted using standardised NRS status codes. At WCP sites, additional measurements of nestling primaries are taken to help establish accurate laying dates. The majority of BOMP Network participants are trained ringers, and are therefore able to ring young and adults encountered.

At WCP sites, the dominant habitat in the area surrounding the nest site is recorded using the BTO habitat coding system developed by Crick (1992). At BOMP Network sites, participants are asked to collect data in a more quantified manner, recording the proportion of the area in 0.5 km radius around the nest box that can be attributed to each of 16 major habitat types. To allow direct comparison of habitat trends identified using the two datasets, each BOMP Network site was assigned a dominant habitat. WCP and BOMP Network sites were then grouped into three principal habitat types; arable, pastoral and natural/semi-natural grassland.

Calculation of occupancy rates

A site was classed as ‘used for nesting’ if a breeding attempt had been made, defined as the presence of one or more eggs or chicks on at least one visit made during the season. If a Barn Owl(s) was encountered or if fresh pellets were present, but no eggs or chicks were recorded during the season, the site was classed as ‘used for roosting’. Sites that were not visited and those at which Barn Owls were prevented from nesting, e.g. by the presence of other species, were excluded from all analyses.

For analysis, boxes within 1 km of each other were treated as a single site and if a breeding attempt was initiated in either box then the site was classed as ‘Used for nesting’. However, in a few cases overlapping breeding attempts did occur in these ‘paired boxes’. If this occurred during any season, the paired boxes were treated as two separate sites in all years as there was the potential for simultaneous breeding. As males were seldom caught at the nest site, it is impossible to exclude the fact that such situations may have been due to a polygamous individual fathering multiple broods. However, the incidence of paired boxes being occupied simultaneously was very low (<0.5% of breeding attempts per annum) and therefore this is unlikely to significantly affect any results presented.

Calculation of breeding parameters

Few nests were found sufficiently early for the laying date of the first egg (FED) to be recorded directly. Laying dates were therefore estimated by back-calculation using information on clutch size, stage of nest contents and primary feather growth was used to estimate laying dates (Shawyer 1998, Crick et al. 2003). The key factor to ascertain in determining clutch size is whether laying had been completed. Thus, records were omitted from these analyses if nests were only visited once, if they were only visited when the eggs were cold (suggesting the nest had failed before the first visit), if laying may still have been in progress on the last visit or if the maximum recorded brood size exceeded the maximum number of recorded young (Crick et al. 2003).

Clutch and brood sizes reported in these analyses refer to the maximum number of eggs and chicks respectively recorded at any visit to the nest (Crick et al. 2003). Clutch sizes of a single egg were excluded from the analysis as this sample is likely to include clutch sizes estimated at ‘1+’ where eggs were present but no count was made. Records were excluded from the analysis of brood size if no visit was made while any of the young were alive.

Analytical methods

Factors influencing the proportion of sites at which Barn Owls were present, whether breeding or nonbreeding, and the proportion of occupied sites at which Barn Owls bred were investigated in the current analysis. As the dataset included information from the same nest sites in several different years, a repeated measures GENMOD procedure was used in SAS v9.1 (SAS Institute 2001), with a site identifier as the repeated variable and specifying an autoregressive correlation function. As the mean life-expectancy of an individual is 3 years (maximum 13 years) (Robinson 2005), the use of a repeated-measures approach allows us to control for the fact that the same pair might be breeding at a specific site in successive years. In all models of occupancy rates, a binomial error distribution was assumed and a logit link function was specified. Terms for northing, easting, year of data collection and primary habitat type were included as independent variables in all models of occupancy rate.

As productivity may vary between first and second broods, any breeding attempts identified as repeats by observers were removed from the dataset prior to analysis of laying date, clutch size and brood size. Analyses of laying date were performed using a repeated measures GENMOD procedure in SAS, assuming a normal error distribution and specifying an identity link function, while analyses of clutch and brood sizes were performed using a repeated measures GENMOD procedure, assuming a Poisson error distribution and specifying a log link. Terms for northing, easting, year of data collection and primary habitat type were included as independent variables in all productivity models.

BOMP coverage

Data from the first six years of BOMP, 2000–05, were used in all of the following analyses. Since 2002, 200 WCP sites have been monitored in each year, but the totals were slightly lower for the initial two field seasons (159 in 2000, 170 in 2001). Data were collected at BOMP Network sites from 2002 onwards. While the importance of consistent coverage is emphasised to participants, all efforts are voluntary and the annual totals of monitored sites are therefore more variable, ranging from a low of 327 sites in 2004 to a high of 386 in 2003. Sites were distributed across the whole of the UK (Fig. 1), but again, due to the voluntary nature of the survey, coverage did to some extent reflect human population density. The number of monitored sites was greatest in the south-east England, and lowest in south-west England, Wales and northern Scotland.

Occupancy rates

The proportion of WCP sites at which Barn Owls were recorded as present (Fig. 2), whether breeding or merely roosting, declined significantly over the period 2000–05 (Table 1). However, no significant temporal trend was detected in the BOMP Network data over the period 2002–05 (Table 2), and when the analysis of the WCP data was limited to the same time span, the effect of year was no longer significant (χ²₁ = 2.30, P = 0.129). The proportion of observed pairs that attempted to breed decreased over time at BOMP Network sites (Table 2), and while a similar trend was reported from WCP sites, it was not statistically significant.

Figure 2.

Proportion of Wildlife Conservation Partnership (solid line) and Barn Owl Monitoring Program Network (dashed line) sites at which Barn Owls were recorded as present, irrespective of breeding status, in each survey year.

Occupancy rates at WCP sites varied significantly between habitats (Fig. 3), with Barn Owls present at a greater proportion of sites in areas of natural or semi-natural grassland (Table 1). However, habitat did not appear to influence occupancy rates at BOMP Network sites (Fig. 3A), and no relationship between habitat type and the proportion of occupied sites at which birds attempted to breed was identified for either dataset.

Table 1.

Factors influencing A) the proportion of Wildlife Conservation Partnership sites at which Barn Owls were observed to be present, and B) the proportion of these sites from which evidence of breeding was reported for the years 2000–05. Statistically significant results in bold. Sample sizes given are number of site × year combinations.

Figure 3.

Predicted occupancy rates (A) and brood sizes (B) at Wildlife Conservation Partnership (dark grey) and Barn Owl Monitoring Program Network (light grey) sites located in different habitats. Values presented were calculated using the parameter estimates generated by the repeated measures GENMOD models, using the mean values for northings and eastings calculated from the WCP dataset (specifying an area in Cambridgeshire, OS Grid Ref TL1487) with the year set to 2003, the mid-point of the data run.

Table 2.

Factors influencing A) the proportion of Barn Owl Monitoring Program Network sites at which Barn Owls were observed to be present and B) the proportion of these sites from which evidence of breeding was reported for the years 2002–05. Statistically significant results in bold. Sample sizes given are number of site × year combinations.

Barn Owl occupancy rates at WCP sites increased significantly with latitude (Table 1). At BOMP Network sites (Table 2), occupancy rates were significantly higher towards the west of the country, as was the proportion of occupied WCP sites at which birds attempted to breed. However, both relationships are potentially confounded by a significant negative correlation between longitude and the prevalence of rough grassland habitats (logistic regression, n = 685, parameter estimate = -0.366, χ² = 13.16, P < 0.001).

Barn Owl productivity

Analyses of laying dates could only be performed for WCP sites, where chick age could be calculated by measuring the developing flight feathers. While mean laying dates became progressively earlier, no significant effect of latitude, longitude or habitat type was identified (Table 3). Analysis of clutch size data was only possible for BOMP Network sites and no significant patterns of temporal or spatial variation in terms of the number of eggs produced per attempt were identified (Table 4).

Sufficient data were collected to allow brood sizes to be analysed for both WCP and BOMP Network sites. The mean number of chicks per brood increased significantly over time at the WCP sites (Table 3). However, this was solely due to a very productive season in 2005, during which the mean brood size was 0.5 chicks greater than the average over the preceding five years. If the analysis is restricted to the period 2000–04, the temporal trend is no longer significant (χ²₁ = 0.13, P = 0.715). At BOMP Network sites, brood sizes produced by pairs nesting in areas of natural or semi-natural grassland were significantly larger than those produced by pairs breeding in arable or pastoral areas (Table 4), on average by approximately 0.5 chicks (Fig. 3B).

Table 3.

Factors influencing Barn Owl laying dates and brood sizes at Wildlife Conservation Partnership sites 2000–05. Statistically significant results in bold. Sample sizes given are number of site × year combinations.