5 Discussion
5.1 Translocation and subsequent survival and movements
Out of fifteen lofts at which attempts were made to trap Sparrowhawks, seven individuals were caught at five lofts. The total number of hours or days that traps were open was unknown, so capture rates could not be calculated.
The radiotracking demonstrated that post release, all seven translocated Sparrowhawks survived in the short term (up to 9 days). Survival of the birds is likely to have been longer though, assuming the bird (Sparrowhawk 3) that was caught and translocated twice was not exceptional. No information on the physical condition of the birds post release was available however (except for Sparrowhawk 3, which returned to the loft after 33 days with a similar weight as at first capture).
Most of the translocated Sparrowhawks did not return to the loft of origin during the periods for which battery life allowed each to be radiotracked. Only one of the translocated Sparrowhawks (an adult female) was known to return to the loft of origin, and that same bird was known to return to a park in the same town as its loft of origin when moved for a second time (after being translocated over 80km in each case).
At two of the five lofts from which Sparrowhawks were translocated, more than one Sparrowhawk was captured, and, in both instances, an adult female was caught initially, followed by an immature female. This suggested that following the removal of adult birds, sub adult Sparrowhawks could move in rapidly to utilise the apparently vacant winter home range. The sample of lofts and duration of trapping was not sufficient to test to what extent a pool of Sparrowhawks existed to replace birds translocated away from lofts but it is known that immature birds in particular could disperse over distances great enough to fill vacant home ranges and that resident Sparrowhawk numbers are likely to augmented by wintering continental birds (Newton 1986, 2002). Unfortunately loft owners did not routinely record the age of Sparrowhawks making attacks at their lofts, so data from this trial cannot be used to assess whether naïve immature birds are more or less likely to attempt attacks (and whether they differ from adults in their success rates) at pigeon lofts.
Conclusion
The trapping and translocation of Sparrowhawks, and subsequent radiotacking, showed that it was feasible to carry out this form of management and that the translocated Sparrowhawks survived, at least in the short-term. However, at least one of the seven female Sparrowhawks that was translocated showed the ability to return to the loft of initial capture or the close proximity (twice), over a distance of more than 80km. In two further cases, immature female Sparrowhawks were captured at lofts following translocation of adult females. The limited life of the transmitters (maximum 11 days) made them unsuitable for longer term surveillance to assess whether additional translocated Sparrowhawks returned to their lofts of origin subsequently, or whether their subsequent welfare ( e.g. body condition) and survival matched that of non-translocated individuals.
5.2 Rates of attacks (and outcomes)
The number of log books returned at the end of the trial was low, with 23 useable sets submitted out of the 45 lofts involved in the trial. In particular, log books were not returned from three of the five lofts from which Sparrowhawks were actually translocated. The proportion of days that lofts were opened was lower than anticipated for all treatment groups, reducing observation periods for all groups. Therefore the trial had substantially reduced samples sizes compared to the three original treatment groups. This issue was compounded further by the need to subdivide the Larsen trap group into those lofts that successfully captured Sparrowhawks and those that did not. In addition, it was necessary to consider separately the periods of pre-capture and post-capture of Sparrowhawks at the lofts where hawks were captured. Also, importantly, trapping (and thus the main trial period) did not start until the middle of January despite some loft owners making observations from 1 November. These resultant small sample sizes imposed major constraints on making valid comparisons of attack rates and attack outcomes between the different treatment groups.
The detailed behavioural observations of Sparrowhawk attacks and outcomes for pigeons at the loft were relatively difficult to make. Although basic instructions were supplied with the log books, the volunteer loft owners were not given formal training in data recording (detailed observations of this nature might be made at ecology graduate or postgraduate levels). Records of attacks were descriptive rather than quantitative or categorical, and the level of detail of recording varied substantially between loft owners. Loft owners would have been helped greatly by a more formal recording form, with observations recorded in a more systematic format and by being provided with more detailed written guidance.
Observer effort during the period in which lofts were open was not documented; it was clear, however, that loft owners were not always present (for example because there were records of "suspected attacks"). This meant that attacks were not always observed, and loft owners were only able to report dead, injured or missing pigeons in such instances. Therefore some attacks that did not result in negative outcomes for pigeons may have gone undetected at some lofts. The lack of recording of observation effort had the potential to introduce significant bias into the data collected, for example if loft owners with higher pre-trial Sparrowhawk attack rates were motivated to commit more time to observing their lofts during the trial itself. The lack of recording of observer effort during loft opening times meant that attack rates could only be reported per hour of pigeon exposure rather than more rigorously as per hour of observation. In the absence of any other information, for reporting purposes it had to be assumed that there was no difference in the proportion of time for which observations were made across the three treatment groups. However, it was probably unrealistic to expect loft owners that operate the natural system to be able to make detailed observations throughout the whole day for the duration of the entire trial period. If a formal scientific trial were to be carried out, formal guidance on making observations during sample periods and the recording of the duration of such observations would markedly improve the trial protocols. There were no systematically collected independent observations at the lofts of Sparrowhawk attack rates and their outcomes. Given that in this exploratory trial observations were made by volunteers with little formal guidance on recording of attacks or observation durations, some systematic independent collection of observations at lofts would have been useful for validation purposes.
It is possible that there were unintentional biases in the way that lofts were allocated. Attack rates at individual lofts were not quantified prior to allocating the lofts to the three treatment groups, and therefore the natural variation in rates could not be factored into the allocation of lofts to treatment groups. This may have been less of a problem if a larger random sample of lofts had submitted observational information during the trial but, given the small sample, variation between lofts in pre-trial Sparrowhawk attack rates could have significantly biased the results observed. Other factors which could have also been influential because of the small resultant sample of lofts that submitted observations but which were not formally taken into account in the trial design included: (1) exercise regimes, which are likely to have led to different levels of exposure of pigeons to attacks; (2) the landscape, which could have varied significantly in the habitat quality for Sparrowhawks and could have influenced their attack approach to the loft; (3) the presence of aviaries, which may act as attractants to Sparrowhawks; and (4) loft size (as represented by number of birds released for exercising), which tended to smaller in the control group (Table 2). There may also have been geographical bias, particularly with respect to the locations of the control lofts. For example, all six control lofts that submitted diaries were located away from the Central Belt (see Figure 1b). Given that Sparrowhawk densities vary across Scotland a non random allocation of sites, in spatial terms, could have affected the observed attack rates. Up to date however, data on potential variation in Sparrowhawk densities does not exist.
Attack rates at translocation lofts
Rates of recorded attacks for No capture Larsen, Pre-capture and Post-capture Larsen control group were similar to rates at the control lofts where no management was carried out. Although attack rates at the two lofts from which Sparrowhawks were translocated showed a drop from pre-capture to post capture, the statistical significance of this could not be tested formally because of the small sample size.
Attack rates at visual deterrent lofts
Attack rates at lofts with visual deterrents were not significantly different from those at control lofts without any form of management. Therefore this trial has not shown any evidence that the use of mylar tape reduced the frequency of attacks at lofts, although the statistical power to detect any difference was low due to the small sample sizes.