4 Results
4.1 Translocation and subsequent radiotracking
4.1.1 Capture
The use of pigeons previously killed by Sparrowhawks was the only successful attractant in this trial. Neither the dead feral pigeons nor the remote control decoys were successful in attracting Sparrowhawks for capture. Loft owners reported that although the decoys appeared to be initially attractive to Sparrowhawks, the hawks appeared unwilling to approach the traps closely.
Seven female Sparrowhawks were caught and subsequently translocated on dates between 16 January and 24 March 2009. All birds were removed within the specified time limit of 2 hours from capture to removal (See Table 1) and none had suffered from feather damage or cuts/abrasions. The range of weights of birds (Table 1) was higher than the known mean weight of Sparrowhawks for that time of year (Fig. 6, p.89 in Newton 1986, I Newton pers. comm.), indicating that birds were likely to be in good condition. At three lofts, only single birds were translocated during the trial period. At another two lofts, two birds were translocated from each. One of these Sparrowhawks returned to the loft, as identified by the presence of the radiotag and matching up of band markings on the tail from photographs, where it was subsequently recaught and translocated for a second time. No birds were captured from the remaining ten lofts in the translocation treatment group. Loft owners did not consistently record the days that traps were opened. Out of a possible 497 days during which the traps could be operated legally (71 days at each of 7 lofts), information was provided only for 90 days. The opening and closing times of the traps were also not reported, rendering it impossible to calculate capture rates of Sparrowhawks per day or hour of trap opening.
4.1.2 Post release survival and movements
Survival
Translocated Sparrowhawks were tracked for a mean of 7 days (range 3-9 days, see Table 1). All birds survived in the short term but information was of limited duration due to the short life of the radiotag batteries. As one Sparrowhawk was caught at the same loft and translocated twice (Sparrowhawk 3 at loft number L06; see Tables 1 and 2), it must be possible for birds to survive for longer periods (minimum of 42 days based on dates of first capture at the loft and last date that the Sparrowhawk was successfully tracked following the second translocation; see Table 1).
Distances travelled by Sparrowhawks
Locations of the lofts at which Sparrowhawks were captured, sites of release and any subsequent fixes of translocated Sparrowhawks are shown in Figures 2-9. Excluding the Sparrowhawk that was known to return to the loft or close to the loft (twice), the maximum recorded distance that an individual Sparrowhawk travelled from the release site during radiotracking was 42 km (Table 1). Within the short life span of the radiotag batteries, there was no obvious tendency of the other Sparrowhawks to return back to the loft, based on locations by date (see Figures 2-9).
For the single Sparrowhawk that was caught at the same loft twice, the distances between the loft and the release sites were 86 and 87 km for the first and second releases, respectively. Following the first translocation, the bird was tracked successfully for 9 days reaching a maximum distance of 32 km in a north to north-west direction from the release site. It was last located on 11 February before being recaptured on 6 March at the loft at which it had previously been caught. Following the second translocation, the bird was tracked for 8 days and had returned to the town in which the loft was located by the fourth day. Although it was not sighted at the loft again, it was located at a neighbouring park for a further four days until the battery eventually lost power. Hence at least one Sparrowhawk showed the ability to return back to the loft site of capture over a distance of 87 km. It is also conceivable that other translocated Sparrowhawks could have returned but were undetected as the radiotag batteries had failed by that time, and the Sparrowhawks were not subsequently recaught at the Larsen traps.
Occupancy of winter home ranges
At two of the five lofts from which Sparrowhawks were translocated, more than one Sparrowhawk was captured (Tables 1 and 2). In both instances, an adult female was caught initially, followed by the capture of an immature female (Tables 1 and 2). This could indicate that the translocation of a single Sparrowhawk away from a pigeon loft may only reduce the potential for attacks very temporarily because other Sparrowhawks can quickly move in to exploit a feeding opportunity. Immature females were caught initially at three further lofts.
4.2 Observations at the loft
In total, useable log books were returned from 23 of 45 lofts in the trial comprising of 6 from control lofts, 7 from lofts using Larsen traps, and 10 from lofts using visual deterrents. Supporting information was also provided for another five lofts but this was not sufficiently comprehensive to be included in the data summaries: one set of diaries had been completed but subsequently had been lost and the loft owner was able to provide only brief summary information; one loft owner submitted only a short account of the season instead of diaries; one loft owner (due to illness) was unable to participate but suggested information was available from the contract falconer (which it was not); another loft owner had moved house during the trial so provided anecdotal information on attack rates at their new and former property; and one loft owner, who after reporting diaries had been lost in the post, stated that three kills had been made but was unable to provide other supporting information. Unfortunately, three of these lofts with no log books were from the group from which Sparrowhawks were actually caught and translocated (See Table 2). This resulted in there being only two out of the total of five lofts from which Sparrowhawks were translocated, for which observations made by loft owners were also available.
Summaries of observer effort and attacks along with their outcomes from the lofts were calculated by treatment. Lofts which operated traps but did not successfully catch a Sparrowhawk might be viewed as another form of controls. As we cannot discount the possibility that the presence of baited traps may have had an effect on the behaviour of Sparrowhawks, it was not appropriate to combine them with the true control group and subsequently they are referred to as the 'No capture Larsen group' (see Tables 3 to 4). Moreover, data from the two lofts at which Sparrowhawks were caught, were separated according to pre and post capture of Sparrowhawks.
4.2.1 Observer effort
Variability between lofts
The total number of days that individual lofts were opened during the trial period ranged widely from 1.4% of trial days (2 days) to 75% (84 days). There was also a large degree of variability in the total number of hours in the trial that individual lofts were opened, ranging from 14 to 506 hours (see Table 3). It is important to note that although the vast majority of observers recorded opening and closing times of lofts, this was not always done consistently. Therefore for 4 lofts, the calculated total amount of time was a minimum estimate only (see Table 3), and for 1 loft the times of opening and closing were never recorded throughout the trial (L05).
Variability between treatments
Daily opening rates during the trial period were 41, 37 and 48% for the Control, Larsen and Visual Deterrent groups respectively (Table 4). Subdividing the Larson group, indicated a higher daily opening rate for pre capture at 54% compared to post capture 22%. Reasons for not opening lofts for exercise included illness, harsh weather conditions, and family commitments (but reasons were not always consistently reported). It was also apparent that following successful Sparrowhawk attacks that resulted in the death of a pigeon, loft owners sometimes would not exercise their pigeons during the following day. Moreover, when there were a number of consecutive successful Sparrowhawk attacks at a loft within a relatively short time, some lofts were closed for multiple days ( e.g. one loft owner closed the loft for most of March following a succession of successful attacks). In addition, it was also noted within the log books that the weather during the winter of 2008/2009 was poor and therefore loft owners were reluctant to release birds for exercise in very cold periods.
The total numbers of hours that lofts were opened were 1052, 1301 and 1991 for for the Control, Larsen and Visual Deterrent groups respectively (Table 4), indicating that the number of hours that pigeons were exposed to Sparrowhawk predation were very variable not only between individual lofts but also between different treatments. It was therefore important to express attacks and their outcomes as an hourly rate as well as absolute numbers of events.
4.2.2 Attacks and outcomes
Variability between lofts
Comparison of observations between lofts over the trial period revealed a high variability in Sparrowhawk attacks in terms of absolute numbers, rates and their outcomes (see Table 3). Attack rates per hour were also highly variable between lofts, with individual lofts ranging from 0 to 0.3 attacks per hour. The total number of pigeons killed at individual lofts ranged from 0 to 18 but high numbers (>5) were only found at one loft (L03). The total number of pigeons injured ranged from 0 to 5 birds at lofts. Death and injury rates per hour were also variable between lofts ranging from 0 to 0.1 and 0 to 0.1 respectively. The total numbers of birds that went missing permanently (due to unknown causes) ranged from 0 to 16 but only one loft lost more than 5 birds (D06).
Variability between treatments
The total number of Sparrowhawk attacks that were recorded by the loft owners throughout the trial for the Control, Larsen trap and Visual deterrent treatments were 9, 74 and 113 respectively. Looking at the total number of attacks for the lofts with traps, this subdivided into 66, 4 and 4 for the No capture Larsen, Pre-capture Larsen and Post-capture Larsen groups respectively (see Table 4). After controlling for the differing amounts of time during which the lofts were open, by expressing attacks as a mean hourly rate, the control lofts (mean attack rate per h = 0.008; lcl-ucl = 0.004-0.056) were not significantly different to the values found for the Larsen trap (mean attack rate per h = 0.050; lcl-ucl = 0.012-0.111) and the visual deterrent (mean attack rate per h = 0.056 ; lcl-ucl = 0.018-0.112) groups. The attack rates fell for both lofts from pre-capture (0.120 and 0.019 for lofts L04 and L06 respectively) to post-capture of Sparrowhawks within the Larsen trap treatment (0.042 and 0.012 for lofts L04 and L06 respectively) but all of these values are within the range of values shown by the other treatments. Moreover this apparent fall in rates cannot be tested statistically because only two lofts were involved.
Total numbers of pigeons recorded as killed during the trial period in the control, Larsen trap and visual deterrent treatments were 2, 31 and 13 respectively. As a total of the loft population (the sum of the maximum of number of birds released from individual lofts) this represented 4% of pigeons being killed during the trial period. Closer inspection of the lofts with traps showed that the numbers of pigeons killed subdivided into 22, 4 and 5 for the No capture Larsen, Pre-capture Larsen and Post-capture Larsen groups respectively (see Table 4). Even when expressing mortality as an hourly rate, the control lofts (mean death rate per h = 0.000; lcl-ucl = 0.000-0.003) were not significantly different to the Larson (mean death rate per h = 0.023; lcl-ucl = 0.000-0.071) and visual deterrent treatment group (mean death rate per h = 0.000; lcl-ucl = 0.000-0.002).
The total numbers of pigeons recorded as injured during the whole trial period for the control, Larsen trap and visual deterrent treatments were 6, 13 and 13 respectively (see Table 4). Breaking down the lofts with traps further revealed that the total number of injured pigeons split into 9, 3 and 1 for the No capture Larsen, Pre-capture Larsen and Post-capture Larsen groups respectively. Mean hourly rates for control lofts (mean injury rate per h = 0.008; lcl-ucl = 0.000-0.040) were not significantly different to those for the Larson (mean injury rate per h = 0.012; lcl-ucl = 0.000-0.053) and Visual deterrent groups (mean injury rate per h = 0.003: 0.000-0.009).
Totals numbers of pigeons reported as missing permanently during the whole trial period were 1, 2 and 33 for the control, Larsen trap and visual deterrent treatments respectively (see Table 4). Those lofts with traps operating subdivided into 1, 0 and 1 pigeons missing for the No capture Larsen, pre-capture Larsen and post-capture Larsen groups respectively. The total numbers of pigeons that went missing but returned later were 12, 11 and 199 for the control, Larsen trap and visual deterrent treatments. The high total of birds that went temporarily missing in the visual deterrent group (see Table 4) was accounted for by three individual lofts (D01, D03 and D06: see Table 4). Dividing the Larsen lofts as before into No capture Larsen, Pre-capture Larsen and Post-capture Larsen groups resulted in respective values of 11, 0 and 0 pigeons missing on a temporary basis.