An investigation into the decline of breeding kestrels in Pembrokeshire.

An investigation into the decline of breeding kestrels in Pembrokeshire. Paddy Jenks, 22 St James Street, Narberth, SA67 7BU. paddyjenks@fsmail.net Tansy Knight, The Water Margins, Lampeter Velfrey, Narberth, SA34 0RB. tansyknight@hotmail.com

Contents: Executive summary 3 Future monitoring 3 Introduction 4 Previous status of Kestrels in Pembrokeshire 4-5 Methods 6 Results 7 Nest box results 8 Discussion 9 Habitat change 9 Productivity 9 Predation 10-11 Survival between breeding seasons 11 Weather and climate 12 The use of nest boxes 13 Rodenticide Use 13 Conclusions 14 Bibliography 14

Executive Summary Status and distribution 2012 The current distribution of breeding kestrels is shown above; a total of 40 sites were used during 2008 to 2012, though the peak in any single year was 25 breeding pairs in 2011. Kestrel numbers within the county have declined and are now considered to be at their lowest level since 1894. The reasons for the decline are discussed in detail. Much of the decline is attributable to loss of suitable breeding habitat due to changes in farming practices, but there are also other contributing factors. Productivity seems sufficient to maintain a viable breeding population, but many successful breeding sites become unoccupied in subsequent years. This indicates that either adult survival between breeding seasons is too low, or insufficient recruitment is taking place due to low survival of first winter birds following independence, or perhaps both. A likely cause of the low survival of first winters is again change in farming practices reducing the foraging quality of the arable landscape but this assumes that Pembrokeshire kestrels disperse to lowland arable areas to over winter in line with the national trend (Shrubb 1993), an assumption for which there are no data to help validate. Predation cannot be ruled out as a cause of poor over winter or post fledging survival, but during the breeding season it appears to be insignificant. Competition for nest sites may occur inland and the provision of artificial sites in some areas may help, but it is concluded that nest site competition in not a significant factor that is driving the population decline, and until other factors are understood and mitigated for, then the kestrel population will not increase significantly in response to provision of artificial nest sites. Future monitoring To help assess the decline with a view to perhaps taking measures to reverse or stabilise it, then much more needs to be discovered about kestrels in winter; in particular where they disperse to and their survival. The recently started colourringing scheme which is now being expanded to cover all of Wales is now being promoted by the Welsh Kite Trust. Early results indicate that colour ringing increases the report rate by 500% and also provides a chance to accumulate multiple sightings of individuals. This will help build up an understanding of dispersal patterns, and may provide survival data. A basic winter census will also provide valuable clues as to the winter distribution of kestrels, particularly if marked individuals can be recorded.

Introduction Kestrels are a familiar sight to many as they hover over open ground hunting for small prey such as rodents, large invertebrates and ground nesting birds. Whilst they take quite a range of prey, they are specialists in hunting voles. They are one of the smallest breeding raptors and were formerly the commonest with an estimated 50,000 pairs breeding in the UK at the time of the BTO national atlas (1988 to1991). They have declined at a national level since then by 20% (BBS recent trend 1994-2008) and they appear to be declining in the Southwestern parts of their range more so than the East. There is evidence of a reduction in the breeding range of Kestrels within Pembrokeshire since 1970, but the only population estimates published for the county were of 50 pairs in 1984 to 1988 (Rees and Donovan 1994) and 31 to 35 pairs in 2003-2007 (Rees et al 2009).This is further confirmation of a continuing decline in breeding numbers. The aims of this study are to establish the reasons behind the decline of the Pembrokeshire population and use this information to inform conservation management of Kestrels. Previous status of Kestrels in Pembrokeshire The earliest reference to Pembrokeshire s breeding Kestrels is in Matthew 1894; 'The most numerous of all our Hawks, to be met with all over the county, nesting in woods, in old ruins, and in many places on the cliffs all round the coast. Whilst Matthew does not attempt to place a figure on the breeding population his description clearly indicates a more widespread breeding population than of recent years. Rees and Donovan (1994) placed a county estimate of 50 pairs based on fieldwork carried out between 1984 and 1988 and noted that Breeding Kestrels are now confined to the less intensively farmed areas, the offshore islands and coastal strip, the Preseli Mountains, industrial sites and remnant bogs and moors. Clearly there has been a decline over the hundred years between those assessments. Red = breeding confirmed Orange = breeding probable Yellow = breeding possible Figure 2 Distribution of breeding Kestrels 1984-1988 (taken from Rees & Donovan 1994)

1984-1988 2003 2007. - Breeding possible - Breeding probable - Breeding confirmed Figure 3 Comparison of breeding distribution 1984-1988 and 2003-2007 Figure 3 indicates that there has been further decline and range reduction since 1988. By 2007 an estimate of 30-35 pairs was made based on the latest atlas fieldwork. The most obvious losses are from farm land in the South and Northeast and to some extent the coastal section between Abereiddy and Fishgaurd. Over all there has been a 38% decline in distribution between the two survey periods. Since then this study has revealed an even lower figure of a fluctuating breeding population of 15 to 25 pairs. However this does not necessarily represent a decline since 2007 because different methodologies were used. This point is discussed in detail later in the report (page 10). They are now confined to the coast, Preseli and the industrial areas in the haven. They are now absent from commercial farmland.

Methods During the period 2008-2012 an attempt was made to locate Kestrel pairs and their nest sites across the whole county. To achieve this extensive survey work was carried out by the authors of all apparently suitable habitat as well as making use of information posted on the local Pembrokeshire bird blog and the results of the 2003-2007 Pembrokeshire breeding atlas. At least two visits were made to all potential habitats with the first visit during April to early May to establish the presence of Kestrels and at least one follow up visit, but usually more, to try and find the nest. Due to the inaccessibility of most of the nest sites, it was not possible to confirm incubation by inspection of the nest cavity, but the presence of an adult male delivering a food item to the female ( a change over ) was considered proof of a breeding attempt even when the nest cavity remained invisible. Nests were sometimes difficult to locate and some pairs required several visits to confirm a breeding attempt. If it was possible to access the nest the chicks were ringed and from 2011 onwards were fitted with individually numbered colour rings. Productivity figures were obtained only from nests which were directly inspected or closely studied at the time of fledging. All locations were recorded on GIS software. During 2007 and 2008 a set of 60 nest boxes were placed widely within the county with a further 20 added in 2010 and these were monitored annually. In addition extensive anecdotal notes were recorded on habitat use, interactions with predators and other Kestrels. These notes are not included in this report but were drawn on in the discussion. Adult male kestrel at Guttle Hole

Results Figure 4 The Breeding Distribution of Kestrels in Pembrokeshire 2008-2012. Total number of territories Unpaired birds holding territory Pair, outcome unknown Successful breeding at least one chick fledged Failed breeding attempt Productivity per successful pair 2008 26* 5 9 9 3 2.6 (n = 7) 2009 20* 2 7 10 1 4.1 (n = 7) 2010 21 6 1 12 2 3.5 (n = 8) 2011 26 1 5 19 1 3.6 (n=11) 2012 29 5 0 14 10 4.4 (n=10) Table 1 A summary of breeding numbers and productivity 2008 to 2012 * Partial survey for initial visits. Figure 4 shows that distribution of breeding Kestrels is mainly coastal with a few pairs in the Preseli hills and the industrial areas in Milford Haven. Excluding unpaired individuals the number of territories has fluctuated between 15 in 2010 to 25 in 2011. Some of the apparent fluctuations may be explained by variations in survey effort and experience but survey effort was consistent between 2010 and 2012 and the figures from these three years can be considered accurate. A total of 40 nest sites were used over the 5 year period with an average of 21 occupied in any one year (range15-25). Productivity ranged from 1 to six chicks in a successful nest with an overall average of 3.72 (n = 43). This compares to a national average 3.9 (BTO website).

Nest box results Year 2008 2009 2010 2011 2012 No of 1 2 2 5 6 occupied nest boxes Total no of boxes 50 50 68 65 62 Table 2 Uptake of boxes The take up of nest boxes increased from 1 in 2008 to 6 in 2012. Occupied boxes were all inland sites in ideal foraging habitat. Several boxes set in good habitat remained unoccupied throughout. All of the boxes that were initially put up in marginal habitat were never occupied. Boxes were used at eight different sites. Figure 5 Map of nest boxes 2008-2012 The pole-mounted box at Pentre Galar was residence to six chicks

Discussion The results show that the Kestrel population and breeding range is now at its lowest since the time of Matthew 1894. There are several possible contributing factors which are discussed below. Habitat change Since the time of Matthew (1894) the landscape has changed significantly. The extent of unimproved pastures is now far lower and evidence for this is given by Matthew when he describes the Snipe as being widespread and numerous as a breeding species. Mixed farming practices which leave a mosaic of arable and pasture with fallow land provide good foraging habitat for Kestrels, but specialisation of farms and the abandoning the practice of leaving fallow land and winter stubbles have left large areas of the county unsuitable to Kestrels. This trend has continued right up to present where the only suitable habitat for Kestrels is now found in conservation managed land. Further evidence of habitat loss due to changes in farming practices is revealed by comparing the recent breeding atlases. Losses have mainly come from inland areas from South of the Milford Haven and Northeast of the Preseli hills where there has been a change to specialised dairy farming. Even where arable land still persists, the practise of leaving over winter stubbles has largely ceased. However despite the significant change in habitat being an obvious cause of decline it does not explain why suitable territories are sometimes unoccupied. Figure 4 shows that the theoretical kestrel population should be at least 40 pairs if change in breeding habitat is the sole factor driving the decline. The average population of 21 pairs (2008-2012) uses just over half of the proven suitable breeding territories with in this period. This can only be explained by insufficient recruitment either due to poor productivity, predation or low survival between breeding seasons. Productivity can be measured but gathering evidence that helps assess the other two factors is more complicated and requires specialised fieldwork. Productivity The productivity of 43 successful nests (3.72) is very close to the national average of 3.9 and the percentage of successful breeding attempts lies between 69% and 80% depending on the actual outcome for those recorded as unknown. It is likely to be nearer the lower limit as it is more difficult to prove failure than success. Even if all outcome unknowns are regarded as failures then there is an overall productivity figure of 2.61 per nesting attempt. This compares favourably with the 2.39 figure from (BTO nest record cards, 1950-1987). It would therefore seem unlikely that poor productivity is a factor in driving the decline.

Predation If the adult male of a pair dies during the breeding season before the chicks are a few days old, the nesting attempt invariably fails (Shrubb 1993). This is because the male provisions the female for a six week period from egg-laying until the chicks are about a week old. After this both adults feed the young for a further five to six weeks. During the three year period 2010 to 2012, 64 breeding attempts included 13 definite failures and an additional 6 for which the outcome is unknown which are probable failures. It was not possible to determine the reason for failure for most breeding attempts, though males were apparently absent in three of these, and in 2008, an additional 3 failed nesting attempts were also associated with absence of males. It is safe to assume that a male only disappears if he dies, but it is usually impossible to eliminate the possibility that he hasn t drifted off after failing for another reason, and there is only one record of a male definitely dying, and even in this case, it is not known how he died. However, unusually high losses of males due to predation will be associated with a reduction in productivity, for which there is no evidence. Using this information it can be concluded that predation of adult males in the breeding season is not significant even if it does happen occasionally. It is also apparent that chick survival from a week old to fledging is good, implying that adult females are not routinely predated when they are most active. However we have no information to help assess predation levels of birds that are newly independent, or predation outside of the breeding season. A long-term colour-ringing study will add to our knowledge in this poorly understood area. The kestrel was ringed as a chick on 10/06/2011 at St Justinian, Pembrokeshire. (Photo by Janet Baxter, Ynyslas, Tuesday 17 th April 2012) The relationship between Kestrels and Peregrines seems complex. Several pairs of Kestrels have successfully bred near to Peregrine nests, some of which have been also successful. At Rosebush a female Kestrel laid a clutch of four eggs on an exposed quarry ledge within 30m of an incubating Peregrine. The Peregrine nest failed at the egg stage, but both adults stayed in the quarry for several weeks. The Kestrels went on to fledge all four young and at least three of these were seen in the nearby clear fell two weeks after fledging. A quick glance at the breeding distribution of Kestrels and Peregrines shows a good correlation, and this is almost certainly a result arising from abundant nest site availability and good foraging habitat for both species. However it is very evident that both young and adult Kestrels regard Peregrines as a serious threat. On many occasions during the fieldwork, evasive action was recorded where individual Kestrels dive to the ground and remain

motionless until the marauder has moved on. The usual choice of a hiding place is the top of a cliff where there is exposed soil and stones providing the perfect background for camouflage. They usually remain in full view though an adult female was seen to dive into vegetation. Adult male Kestrels were also recorded mobbing Peregrines on several occasions using their supreme agility to dance around the more powerful but less manoeuvrable Peregrine. Goshawks have been known to drive population declines in Kestrels at Kielder Forest, Northumberland (Petty et al 2004). A quick glance at the distribution of Goshawk territories in the county reveals very little overlap with the breeding Kestrel distribution making it appear as if there is a strong negative correlation. However, in Pembrokeshire their different distributions could be explained by their habitat preferences which have limited overlap. This was not the case in Kielder. Goshawks have colonised Pembrokeshire from the East between the two breeding atlas surveys and they are still very scarce in the areas where most of the Kestrel decline has taken place. This indicates that predation by Goshawks is probably not a cause of decline in the breeding population, though it is not possible to evaluate levels of predation outside of the breeding season if Kestrels were perhaps to disperse to areas with established Goshawk populations. Although a buzzard is unlikely to predate a Kestrel in the air they are well known for taking both adult birds and nestlings from nests and have been witnessed to predate Red Kites (pers obs). It is notable that during this study Kestrels were not recorded to breed in Crows nests, a habit which is very common in Southern England (Shrubb 1994) and in Pembrokeshire at the time of Matthew ( Matthew 1894). Matthew also noted that the Buzzard population was about 6 pairs compared to an estimated 650-900 pairs by 2007. The implications are that Buzzards potentially limit the number of nest sites available to breeding Kestrels. It is possible in the past that they have played a part in the Kestrel decline but currently overall nest productivity figures indicate that nest predation by Buzzards of the cavity nesting population is insignificant. Survival between breeding seasons Accurately measuring post fledging and adult survival between the breeding seasons is impossible without extensive fieldwork. To obtain enough detail it is necessary to record the life histories of many individuals over a long period of time and this is impractical without very expensive equipment. There are, however, a few observations to be made regarding survival. Firstly, assuming that adults will return to their previous territories then adult survival can be measured by colour-marking individuals and then recording their presence in subsequent years. Even if individuals cannot be recognised, then the absence of a breeding pair in a territory where successful breeding took place in the previous season indicates that either both adults have died, or at least one has died and the other has shifted territory to find another mate. Adult survival has been measured in other studies and is around 70% (village 1990), in which case both adults may be expected to die between seasons in 10% of territories and at least one in an additional 40%. Whilst there are currently insufficient data to make a realistic assessment, territory vacancies have been a regular feature during the study, implying that adult survival is potentially an issue though it could equally well be explained by insufficient recruitment due to poor survival of first winter birds during their first two years following fledging. In either case than the situation is worthy of further study. The assumption that juveniles will return to their exact natal area to breed is not backed up by evidence, and so measuring first year survival remains a problem, as emigration appears as mortality. Not only this, but 1 st winters disperse a much greater distance in the autumn than adults who have far greater tendency to remain in their breeding areas all year, especially so in adult males (Shrubb 1993, Village 1990). This leaves them potentially vulnerable to effects that well outside the study

area. The only way of assessing juvenile survival is to mark many individuals over a wide area covering a protracted period and gather a large data set of resightings. Of the fledged broods that have been observed, they appear to survive well in the first few days whilst still dependent, but after that it is impossible to know whether they have died or dispersed. Of the 28 nestlings that were colour-marked in 2011, only three were seen in 2012, one of which was 80km from its natal site, another was 15km whilst the third was probably only 1km, but the ring wasn t read fully so this cannot be confirmed. It is not known where the Pembrokeshire kestrels spend the winter, but Shrubb draws attention to the fact that most recoveries in the national database from the winter period are from lowland areas where arable farming predominates, a landscape that offers far less available foraging than it used to. In their first winter the survival rate of stable populations is measured at 35 to 40% followed by 60% during their second year, but a large sample is needed to measure this. In this respect, colour ringing becomes highly significant, as it has already been shown to increase the rate of resightings, albeit in a tiny sample. Expanding the colouring ringing to other counties will help speed up the information gathering process To summarise, poor survival is almost certainly playing a major role in supressing the population, but there are insufficient data to determine at what stage or stages or by what mechanism it is having an effect. It is potentially due to lack of food at the sites to which kestrels disperse during their first autumn, or perhaps predation during the first few months, or even wetter weather which is thought to have a negative influence on kestrel survival. Future research to address this would be well worthwhile. Weather and climate Nobody spending the 2012 summer in south west Wales can fail to mention the weather; the coldest April, the wettest June and July, not to mention one of the severest summer gales for many years on June 8 th. This latter event caused complete nest failure among the Red Kites in the county, and it is likely to have caused many kestrel nest failures. Figure 4 shows that most of the ten nest failures were in the west and on Preseli, which took the brunt of the storm. The failure rate in 2012 is much higher than in the previous two years but unexpectedly, the mean productivity of successful nests (4.4) is higher. This is perhaps due to more experienced pairs which tend to nest earlier and have larger clutches and broods had reached the stage where their nestlings were well feathered by the time of the severest storm, whilst the later, smaller broods were downy and more susceptible to chilling. Another explanation may simply relate to the availability of prey being higher but this has not been verified. A one-off poor season cannot be regarded as a factor driving long-term steady population declines, but a run of poor seasons may exacerbate a decline driven by other factors. The low population recorded in 2010 is possibly due to the harsh winter of 2009-2010 causing a lower survival rate than usual. However this is not backed up by national BBS figures which show a 36% decline 2008-2009, followed by 3 % increase between 2009 and 2010. Unfortunately no figures are available for Wales due to the sample size of BBS squares with kestrels being too small, but the SW region measured a 49% decrease between 2009 and 2010, mimicking the Pembrokeshire decline. Whatever the interpretation of this discrepancy, by 2011 the Pembrokeshire population had recovered to 25 pairs, which is the highest recorded during the 5 years of fieldwork. This is at least is evidence that the population has the capacity to recover following one off weather events.

The use of nest boxes Six nest boxes were occupied in 2012, four of which held successful broods. As in previous years several others showed evidence of regular use by kestrels, particularly over the late winter period. The result of the nest box scheme is encouraging as it has undoubtedly contributed to a range expansion inland. To increase the chances of a nest box being used, we have observed that a large area (eg > 100Ha) of prime feeding habitat must be very close to the box site. This is because the female hunts within sight of the nest when the chicks are growing. Also the male kestrel makes extensive use of additional roost sites nearby, including other nest boxes and barns, so several boxes in one area may be beneficial. For the coastal pairs there appears to be an abundance of sites for both nesting and roosting, but inland ledges and crevices are restricted to quarry sites, many of which now have breeding peregrines, buzzards and/or ravens. Other potential nest sites such as old corvid nests are abundant inland but higher exposure to the Pembrokeshire wind and rain together with increased predation risk must make these sites unattractive to kestrels, as we have yet to record such a site being used in the county. All in all, nest boxes which are suitably sited, inland and in prime habitat, are needed to sustain a kestrel population away from the coast. It is possible that Kestrels breeding inland are limited by nest site availability. Habitats can be very open with no tree cavities or cliffs, limiting their choice to crows nests which may be vulnerable to predation and exposure. It is well established that Kestrels readily take to artificial sites, so a nest box scheme may help consolidate a population in areas where natural sites are limited. The Pembrokeshire scheme established nest boxes across a wide area of inland and a few coastal habitats. Figure 5 shows that a total of 8 nest boxes were used at some point during the study. All boxes that were occupied were sited within prime habitat and no boxes in improved farm land were ever used even if there was good habitat within the 3km range that males are prepared to travel to forage (PJ &TK radio tracking study unpublished) This is because females hunt near their nest and don t like to leave the young out of sight. Hence the location of any nest box which stands a chance of being used is critical. Over all it seems that the nest box scheme has offered modest benefits to the breeding population. Although there is no evidence that the population has increased numerically due to the establishment of nest boxes there is evidence that some of their former range has been reoccupied inland. Also many boxes are used as roost sites outside the breeding season and this may well offer benefits in terms of survival. However if as suspected the population decline is being driven by poor over wintering survival rate any benefits that a box scheme may offer will have a limited overall effect. Rodenticide Use Concern has been raised over the new generation rodenticide use, which is considerably more toxic than traditional rodenticides. There is currently no information as to whether or how this affects Kestrels.

Conclusions Kestrel numbers within the county have declined and are now considered to be at their lowest level since 1894. The reasons for the decline are discussed in detail. Much of the decline is attributable to loss of suitable breeding habitat due to changes in farming practices, but there are also other contributing factors. Productivity seems sufficient to maintain a viable breeding population, but many successful breeding sites become unoccupied in subsequent years. This indicates that recruitment is being suppressed by either low adult survival between breeding seasons or low survival of first winter birds following independence (or perhaps both). A likely cause of the low survival of first winters is again change in farming practices reducing the foraging quality of the arable landscape but this assumes that Pembrokeshire Kestrels disperse to lowland arable areas to over winter in line with the national trend (Shrubb 1993), an assumption for which there are no data to help validate. Predation cannot be ruled out as a cause of poor over winter or post fledging survival, but during the breeding season it appears to be insignificant. Competition for nest sites may occur inland and the provision of artificial sites in some areas may help, but it is concluded that nest site competition in not a significant factor that is driving the population decline, and until other factors are understood and mitigated for, then the kestrel population will not increase significantly. Bibliography Mathew M.A. 1894, Birds of Pembrokeshire and it's Islands Donovan J.W. & Rees G.H (1994), Birds of Pembrokeshire Altlas Of Breeding Birds In Pembrokeshire 2003-2007 (Rees et al 2009) PETTY. S. J, ANDERSON. D. I. K, DAISON. M, LITTLE. B, SHERRATT. T. N, THOMAS. C. J, LAMBIN. X 2003,The decline of Common Kestrels falco tinnunculus in forested area of northern Engalnd : The role of predation by Northern Goshawks Accipiter gentilis, Ibis Vol 145, 472 483 WWW.BTO.ORG Village A. The Kestrel T and A D Poyser 1990